The Archaeology of Geological Catastrophes
Geological Society Special Publications
Series Editors A. J. HARTLEY R. E. HOLDSWORTH
A. C. MORTON M. S. STOKER
Special Publication reviewing procedures The Society makes every effort to ensure that the scientific and production quality of its books matches that of its journals. Since 1997, all book proposals have been refereed by specialist reviewers as well as by the Society's Publications Committee. If the referees identify weaknesses in the proposal, these must be addressed before the proposal is accepted. Once the book is accepted, the Society has a team of series editors (listed above) who ensure that the volume editors follow strict guidelines on refereeing and quality control. We insist that individual papers can only be accepted after satisfactory review by two independent referees. The questions on the review forms are similar to those for the Journal of the Geological Society. The referees' forms and comments must be available to the Society's series editors on request. Although many of the books result from meetings, the editors are expected to commission papers that were not presented at the meeting to ensure that the book provides a balanced coverage of the subject. Being accepted for presentation at the meeting does not guarantee inclusion in the book. Geological Society Special Publications are included in the ISI Science Citation Index, but they do not have an impact factor, the latter being applicable only to journals. More information about submitting a proposal and producing a Special Publication can be found on the Society's web site: www.geolsoc.org.uk
It is recommended that reference to all or part of this book should be made in one of the following ways: McGuiRE, W. J., GRIFFITHS, D. R., HANCOCK, P. L. & STEWART, I. S. (eds) 2000. The Archaeology of Geological Catastrophes. Geological Society, London, Special Publications, 171. GUIDOBONI, E., MUGGIA, A. & VALENSISE, G. 2000. Aims and methods in Territorial Archaeology: possible clues to a strong IV century AD earthquake in the Straits of Messina (Southern Italy) In: McGuiRE, W. J., GRIFFITHS, D. R., HANCOCK, P. L. & STEWART, I. S. (eds) The Archaeology of Geological Catastrophes. Geological Society, London, Special Publications, 171, 45-70.
GEOLOGICAL SOCIETY SPECIAL PUBLICATION NO. 171
The Archaeology of Geological Catastrophes
EDITED BY
W. J. MCGUIRE
University College London, UK
D. R. GRIFFITHS
University College London/Institute of Archaeology, UK
P. L. HANCOCK University of Bristol, UK
I. S. STEWART Brunei University, UK
2000
Published by The Geological Society London
THE GEOLOGICAL SOCIETY The Geological Society of London was founded in 1807 and is the oldest geological society in the world. It received its Royal Charter in 1825 for the purpose of 'investigating the mineral structure of the Earth' and is now Britain's national society for geology. Both a learned society and a professional body, the Geological Society is recognized by the Department of Trade and Industry (DTI) as the chartering authority for geoscience, able to award Chartered Geologist status upon appropriately qualified Fellows. The Society has a membership of 8600, of whom about 1500 live outside the UK. Fellowship of the Society is open to those holding a recognized honours degree in geology or cognate subject and who have at least two years' relevant postgraduate experience, or who have not less than six years' relevant experience in geology or a cognate subject. A Fellow with a minimum of five years' relevant postgraduate experience in the practice of geology may apply for chartered status. Successful applicants are entitled to use the designatory postnominal CGeol (Chartered Geologist). Fellows of the Society may use the letters FGS. Other grades of membership are available to members not yet qualifying for Fellowship. The Society has its own publishing house based in Bath, UK. It produces the Society's international journals, books and maps, and is the European distributor for publications of the American Association of Petroleum Geologists, (AAPG), the Society for Sedimentary Geology (SEPM) and the Geological Society of America (GSA). Members of the Society can buy books at considerable discounts. The publishing House has an online bookshop (http:Jlbookshop.geolsoc.org.uk). Further information on Society membership may be obtained from the Membership Services Manager, The Geological Society, Burlington House, Piccadilly, London W1V OJU, UK. (Email:
[email protected]: tel: +44 (0)171 434 9944). The Society's Web Site can be found at http://www.geolsoc.org.uk/. The Society is a Registered Charity, number 210161. Published by The Geological Society from: The Geological Society Publishing House Unit 7, Brassmill Enterprise Centre Brassmill Lane Bath BA1 3JN, UK (Orders: Tel. +44 (0)1225 445046 Fax +44 (0)1225 442836) Online bookshop: http://bookshop.geolsoc.org.uk First published 2000 The publishers make no representation, express or implied, with regard to the accuracy of the information contained in this book and cannot accept any legal responsibility for any errors or omissions that may be made. © The Geological Society of London 2000. All rights reserved. No reproduction, copy or transmission of this publication may be made without written permission. No paragraph of this publication may be reproduced, copied or transmitted save with the provisions of the Copyright Licensing Agency, 90 Tottenham Court Road, London W1P 9HE. Users registered with the Copyright Clearance Center, 27 Congress Street, Salem, MA 01970, USA: the item-fee code for this publication is 0305-8719/99/S 15.00. British Library Cataloguing in Publication Data A catalogue record for this book is available from the British Library. ISBN 1-86239-062-2
Typeset by Aarontype Ltd, Bristol, UK Printed by The Alden Press, Osney Mead, Oxford, UK
Distributors
USA
AAPG Bookstore PO Box 979 Tulsa OK 74101-0979 USA Orders: Tel. +1 918 584-2555 Fax +1 918 560-2652 e-mail:
[email protected] Australia Australian Mineral Foundation Bookshop 63 Conyngham Street Glenside South Australia 5065 Australia Orders: Tel. +61 88 379-0444 Fax +61 88 379-4634 e-mail:
[email protected] India Affiliated East-West Press PVT Ltd G-l/16 Ansari Road, Daryaganj, New Delhi 110002 India Orders: Tel. +91 11 327-9113 Fax+91 11 326-0538 e-mail:
[email protected] Japan Kanda Book Trading Co. Cityhouse Tama 204 Tsurumaki 1-3-10 Tama-shi Tokyo 206-0034 Japan Orders: Tel. +81 (0)423 57-7650 Fax +81 (0)423 57-7651
TContents
Preface HANCOCK, P. L., CHALMERS, R. M. L., ALTUNEL, E., £AKIR, Z. & BECKER-HANCOCK, A. Creation and destruction of travertine monumental stone by earthquake faulting at Hierapolis, Turkey GRIFFITHS, D. R. Uses of volcanic products in antiquity JONES, R. E & STIROS, S. C. The advent of archaeoseismology in the Mediterranean BUCK, V. & STEWART, I. A critical reappraisal of the classical texts and archaeological evidence for earthquakes in the Atalanti region, central mainland Greece GUIDOBONI, E., MUGGIA, A. & VALENSISE, G. Aims and methods in territorial archaeology: possible clues to a strong fourth-century AD earthquake in the Straits of Messina (southern Italy) FRIEDRICH, W. L., SEIDENKRANTZ, M.-S. & NIELSEN, O. B. Santorini (Greece) before the Minoan eruption: a reconstruction of the ring-island, natural resources and clay deposits from the Akrotiri excavation DRIESSEN, J. & MACDONALD, C. F. The eruption of the Santorini volcano and its effect on Minoan Crete BICKNELL, P. Late Minoan IB marine ware, the marine environment of the Aegean, and the Bronze Age eruption of the Thera volcano RUSSELL, J. K. & STASIUK, M. V. Ground-penetrating radar mapping of Minoan volcanic deposits and the Late Bronze Age palaeotopography, Thera, Greece CIONI, R., GURIOLI, L., SBRANA, A. & VOUGIOUKALAKIS, G. Precursory phenomena and destructive events related to the Late Bronze Age Minoan (Thera, Greece) and AD 79 (Vesuvius, Italy) Plinian eruptions; inferences from the stratigraphy in the archaeological areas PARESCHI, M. T., STEFANI, G., VARONE, A., CAVARRA, L., GIANNINI, F. & MERIGGI, A. A geographical information system for the archaeological area of Pompeii CIONI, R., LEVI, S. & SULPIZIO, R. Apulian Bronze Age pottery as a long distance indicator of the Avellino Pumice eruption (Vesuvius, Italy) CHESTER, D. K., DUNCAN, A. M., GUEST, J. E., JOHNSTON, P. A. & SMOLENAARS, J. J. L. Human response to Etna volcano during the classical period KIRK, W. L., SIDDALL, R. & STEAD, S. The Johnston-Lavis collection: a unique record of Italian volcanism PLUNKET, P. & URUNUELA, G. The archaeology of a Plinian eruption of the Popocatepetl volcano GONZALEZ, S., PASTRANA, A., SIEBE, C. & DULLER, G. Timing of the prehistoric eruption of Xitle Volcano and the abandonment of Cuicuilco Pyramid, Southern Basin of Mexico TORRENCE, R., PAVLIDES, C., JACKSON, P. & WEBB, J. Volcanic disasters and cultural discontinuities in Holocene time, in West New Britain, Papua New Guinea RIEHLE, J. R., DUMOND, D. E., MEYER, C. E. & SCHAAF, J. M. Tephrochronology of the Brooks River Archaeological District, Katmai National Park and Preserve, Alaska: what can and cannot be done with tephra deposits DODGSHON, R. A., GILBERTSON, D. D. & GRATTAN, J. P. Endemic stress, farming communities and the influence of Icelandic volcanic eruptions in the Scottish Highlands
vii 1 15 25 33 45 71 81 95 105 123
143 159 179 189 195 205 225 245 267
vi
CONTENTS
DAY, S. J., CARRACEDO, J. C, GUILLOU, H., PAIS PAIS, F. J., BADIOLA, E. R., FONSECA, J. F. B. D. & HELENO, S. I. N. Comparison and cross-checking of historical, archaeological and geological evidence for the location and type of historical and sub-historical eruptions of multiple-vent oceanic island volcanoes GRATTAN, J. P., GILBERTSON, D. D. & DILL, A. 'A fire spitting volcano in our dear Germany': documentary evidence for a low-intensity volcanic eruption of the Gleichberg in 1783? JAMES, P., CHESTER, D. & DUNCAN, A. Volcanic soils: their nature and significance for archaeology SIDDALL, R. The use of volcaniclastic material in Roman hydraulic concretes: a brief review HUNT, P. Olmec stone sculpture: selection criteria for basalt HUGHES, R. & COLLINGS, A. Seismic and volcano hazards affecting the vulnerability of the Sana'a area of Yemen WAELKENS, M, SINTUBIN, M., MUCHEZ, P. & PAULISSEN, E. Archaeological, geomorphological and geological evidence for a major earthquake at Sagalassos (SW Turkey) around the middle of the seventh century AD STIROS, S. C. Fault pattern of Nisyros Island volcano (Aegean Sea, Greece): structural, coastal and archaeological evidence DE BOER, J. Z. & HALE, J. R. The geological origins of the oracle at Delphi, Greece Index
281
307 317 339 345 355 373 385 399 413
Preface The Archaeology of Geological Catastrophes brings together a diverse collection of papers that address the archaeological identification and cultural significance of large-scale geological events, mainly earthquakes and volcanic eruptions. Major earthquakes and volcanic eruptions typically recur at intervals of anything between a few decades to many tens to hundreds of thousands of years. Yet the instrumentation by which we record and monitor them has only been around for little over a century. To reduce the hazard posed by earthquakes and volcanism, we require a longer record of them than can be provided from modern instrumental snapshots. On the assumption that future earthquake and volcanic activPaul L. Hancock who sadly died during the ity will be like that of the recent past, completion of this volume. we need to understand the history of earthquakes and volcanism over millennial timescales. For the geologist, therefore, archaeology presents a potential tool to illuminate this time window, lying astride the documentary archives of historians and the geological archives of the surficial rock record. For the archaeologist, recognizing the impact of earthquakes or volcanic activity on a site or region more often provides a missing piece of the human history of that area, often explaining the conditions for cultural development or demise. In this context, an individual earthquake or volcanic eruption may often be the solution to an archaeologist's interpretation of inferred local or regional upheavals. By contrast, for the volcanologist or earthquake geologist, the identification of a major prehistorical seismic or volcanic event is generally the starting point from which to go on to derive other parameters (e.g. event magnitude, source etc.), or fit into regional models or datasets. For example, for archaeologists, the 464 BC earthquake at Sparta, Greece, was the trigger for a major change in political conditions in the Peloponnese region; for earthquake geologists it provides key evidence to estimate the earthquake energy released on a major fault that has since been seismically quiescent (Armijo et al. 1991). Furthermore, this difference is more than simply one of perspective, since it is generally underpinned by contrasting philosophical and theoretical frameworks, by varying methodological approaches, and by practitioners using distinct terminologies and presenting results in widely differing forums. In short, in many ways, archaeology and geology are fundamentally distinct disciplines. The distinction leads to uncertainty, or even suspicion, about studies that seek to integrate the two. Many seismologists, for example, will no doubt still
viii
PREFACE
empathize with comments made by Charles Richter 40 years ago, complaining that 'Ancient accounts of earthquakes do not help us much; they are incomplete, and accuracy is usually sacrificed to make the most of a good story' (Richter 1958, cited by Vita-Finzi 1986, p. 8). Guidoboni (1996) is more generous though equally cautious, noting that 'Earthquakes are not kind, and they do not care for researchers. Their traces can travel through strata and upset methods for dating in unexpected ways. This is one reason why so many important pieces of archaeological evidence are lost for seismology'. To some extent, these criticisms may be less resonant for volcanic archaeology investigations, since eruptive activity frequently leaves geochemically or petrographically distinctive 'event horizons' (e.g. Riehle et al. this volume), or even volcanic deposits that preserve the archaeological record more or less intact (e.g. Gonzalez et al. this volume). By comparison, destruction horizons produced by seismic shaking must compete with the often comparable debris traces of warfare and natural collapse of poor constructions. In this regard, the focus in this volume on 'catastrophes' is less to do with the assumption that these events are inherently more important (or interesting) for our understanding of recent geological history or of our cultural heritage. Instead, it reflects the recognition that it is the large-magnitude geophysical events that are most likely to leave the clearest signals in the archaeological record. The diverse ways in which investigators may interpret those signals is arguably the main theme of this volume. The Archaeology of Geological Catastrophes presents a broad spectrum of papers on the geoarchaeology of earthquakes and volcanoes, and here we draw attention only to a few general themes. Although earthquakes and volcanic eruptions are generally viewed as agents of destruction, numerous papers discuss their potential benefits to past cultures - providing materials for tools, building and sculpture, and even the fertile environmental conditions on which societies depended. Perhaps the most intriguing proposal is the suggestion that the power of the Delphic oracle to the ancient Greeks derived from the geological setting of the site, specifically from gaseous emissions from an underlying active fault. The bulk of contributions, however, focus on the destructive power of earthquakes and volcanoes. Several papers deal specifically with 'archaeoseismology' - the study of pre-instrumental earthquakes that, by affecting locations and their environments, have left their mark in the archaeological record. An important debate to emerge from these papers is whether major past earthquakes are more effectively recognized through regional disturbances in occupation or settlement patterns (territorial archaeology) or through the identification of 'diagnostic' structural indicators at individual sites. A suite of papers tackle different facets of arguably the most prominent geological catastrophe in the archaeological record - the Bronze Age eruption of Thera (Santorini, Greece) and its consequent regional impacts on Minoan culture. Human responses to major volcanic eruptions are also discussed, both in terms of local reactions to volcanism in Sicily and Mexico, and far-field effects, such as the impacts of Icelandic eruptive activity on agricultural demise in the Scottish Highlands. In turn, the value (and potential pitfalls) of historical records of past eruptive activity in documenting the capricious character of volcanism in an area are assessed in case studies from Italy, Germany, the Canary Islands and the Cape Verde islands. Other themes covered within the volume include the application of tephrachronology in volcanic archaeology, the value of volcanic soils in archaeological research, the use of geographic information systems in preserving vulnerable archaeological information
PREFACE
ix
at key cultural sites and the assessment of the vulnerability of important cultural centres to seismic and volcanic threats. To those that may dislike the eclectic character of this volume, the editors would argue that this only serves to reflect the rather disparate state-of-play within the burgeoning fields of earthquake and volcanic archaeology. Furthermore, the papers presented here show varying degrees of cross-disciplinary co-operation, but the bulk of the research is still largely being undertaken by archaeologists or by geologists working in relative isolation. It is hoped that by raising some important research questions, volumes like The Archaeology of Geological Catastrophes, will accelerate the move towards the type of interdisciplinary research advocated by Van Andel (1991, p. 324), in which historians, archaeologists and geologists (among others!) participate in a ' ... collaboration which assumes intensive exchange of information, ideas and procedures from the planning stage through to final publication'. Such collaborations are likely to be essential if the past societal impacts of earthquake and volcanic activity are to be effectively unravelled. References ARMIJO, R., LYON-CAEN, H. & PAPANASTASSIOU, D. 1991. A possible fault rupture for the 464 BC Sparta earthquake, Nature, v, 351. GUIDOBONI, E. 1996. Archaeology and historical seismology: the need for collaboration in the Mediterranean Area. In: STIROS, S. & JONES, R. E. (eds.) Archaeoseismology, Fitch Laboratory Occasional Paper 7 British School at Athens, Athens, Greece, 7-13.
VAN ANDEL
> T- H- 1991.Geo-archaeology and archaeological science. In: NICK-KARDULIAS, P. (ed.) Beyond the Site: regional Studies in the Aegean Area, University Press of America Inc. Maryland, 25-44. ViTA-FiNZi, C. 1986. Recent Earth Movements: an introduction to Neotectonics, Academic Press, London. j
in Stewart
London, December 1999
This page intentionally left blank
Creation and destruction of travertine monumental stone by earthquake faulting at Hierapolis, Turkey P. L. HANCOCK*'1, R. M. L. CHALMERS1, E. ALTUNEL2, Z. £AKIR3 & A. BECKER-HANCOCK4 1
Department of Geology, University of Bristol, Wills Memorial Building, Queen's Road, Bristol BS8 1RJ, UK Jeoloji Muhendisligi Bolumu, Muhendislik Mimarlik Fakiiltesi, Osmangazi University, Eskisehir, Turkey 3 Engineering Faculty, Mersin University, Ciftlik Koyu, Mersin, Turkey 4 Department of English, University of Bristol, 3-5 Woodland Road, Bristol BS8 1TB, UK Abstract: The presence of travertines adjacent to the city and their value for construction was well known to the Greek, Roman and Byzantine residents of Hierapolis (modern Pamukkale). The travertines were mainly extracted from quarries on the outer slopes of a low plateau below the city. The distinctive attribute of most of the quarries is that they are narrow but deep vertical-sided trenches. Each trench is the site of a nearly vertical fissure that was filled by banded fissure travertine, one type of so-called Phrygian marble. Trench walls, formerly the contacts between vertical banded travertines and outward dipping bedded travertines, bear a well-defined herringbone pattern of tool marks identical to those on many of the stone blocks that were used for building Hierapolis. Deposition of the travertines in 21 major fissure-ridges was a consequence of precipitation following ' degassing of carbonate-rich hot waters emerging from springs aligned along active faults and associated fissures. Whereas the dense and attractively banded travertine in fissures was principally used as an ornamental stone, the bedded travertines of ridge sides were mainly employed as a dimension stone and for making columns. After many of the monuments at Hierapolis had been constructed from travertine, itself a faulting-related material, some of them were subsequently destroyed or damaged by earthquake fault reactivation, which caused them to be either shaken or displaced. The zone of greatest seismic damage coincides with the trace of the Hierapolis fault zone, whose location was detected from an alignment of offsets of walls and petrified irrigation channels. The kinematic class of this fault zone could be deduced because offsets of the linear archaeological features permitted opening directions to be determined, thus allowing the fault zone to be reinterpreted as a normal fault zone achieving a small downthrow to the southwest. The knowledge that the Hierapolis fault zone is a structure across which there is active stretching and increased hydrothermal flow helps to explain why the present-day area of hot pools and travertine deposition is situated immediately downslope of the fault trace. If this relationship between displaced features and recent travertine deposits occurs elsewhere it might be employed for finding the locations of earthquake faults.
The purpose of this paper is to explain how earthquake faulting in the city of Hierapolis is associated with the deposition of a large body of travertine that was quarried for stone in Greek and Roman times, and how subsequent reactivation of faults in the same area during these periods and later was responsible for damaging the city. The site of Hierapolis, one of several * Deceased, reprint requests should be addressed to A. Becher-Hancock.
Greek and Roman cities in the Maeander River valley (the present Menderes River) (Fig. 1), is roughly coincident with the present tourist village of Pamukkale, a settlement that in the last 50 years has served the needs of visitors not only to Hierapolis but also to the famous white travertine deposits that give Pamukkale its name, 'cotton castle'. Excluding the extensive northern necropolis and smaller southern necropolis, much of Hierapolis lies within its late Roman city wall (Peres 1987) (Fig. 2).
From: McGuiRE, W. G., GRIFFITHS, D. R., HANCOCK, P. L. & STEWART, I. S. (eds) The Archaeology of Geological Catastrophes. Geological Society, London, Special Publications, 171, 1-14. 1-86239-062-2/OO/S 15.00 © The Geological Society of London 2000.
Fig. 1. (a) Turkish sector of the Aegean extensional province, (b) Geological, topographic and historical setting of Hierapolis within the Denizli basin (modified after Altunel & Hancock 19930).
CREATION AND DESTRUCTION OF TRAVERTINE MONUMENTAL STONE
3
Fig. 2. Sketch map of selected principal monuments within Hierapolis.
Neotectonic, topographic and historical setting Many landforms in regions of active extensional tectonics directly reflect recent earth movements and hence they can have a great influence on routeways and settlements. This is especially true of western Anatolia, which is situated in the east of the Aegean extensional province (Fig. la), a region currently experiencing normal faulting and the formation of rift valleys (grabens) and intervening horst-block mountains as a consequence of roughly NNE-SSW stretching (Jackson 1994). Hierapolis is sited on the northeastern edge of the Denizli basin, a structure within the Menderes graben but close to its confluence with the Gediz graben. The Denizli basin has been subsiding since Miocene time (Westaway 1993). It is framed to the south by a major E-W trending normal fault, which is part of the Menderes system, but to the northeast the principal faults
trend NW-SE, that is, they follow the Gediz graben trend. The Denizli basin and the Gediz graben are separated by a zone, to the northwest of Buldan (Fig. Ib), that does not contain large normal faults and hence is not expressed by a graben or basin. The floor of the Denizli basin is mainly underlain by Neogene and Quaternary clastic sediments. The Quaternary travertine masses of the basin, of which the Pamukkale mass is only one, rest on these clastic sediments. External to the Denizli basin there are outcrops of metamorphic and igneous basement rocks unconformably overlain by Neogene clastic deposits. The Pamukkale range-front fault separates the sediments of the Denizli basin from the basement rocks to the northeast, with a downthrow southwest of at least 450m (Altunel & Hancock 1993^). The large-scale topography of the region is a direct expression of Neogene and Quaternary tectonic activity. For example, the Denizli basin
4
P. L. HANCOCK ET AL.
is an area of low relief, the nearly level Quaternary flood plains of the Maeander and Lykus (modern £uruksu) rivers being at about 280m above sea level and about 60m lower than the surrounding incised plateau underlain by Neogene sediments. The mountain massif of Kuciik £6kelezdag to the northeast of the basin rises to a maximum of 1739m, whereas to the south of Denizli the higher mountain of Babadag reaches
2300 m and the even higher peak of Honazdag, south of Colossae, achieves 2571 m. Between the Denizli basin and the Alasehir valley of the Gediz graben is an area of high but not rugged ground, rising to about 1324m. The Pamukkale travertines, which were used for building Hierapolis, are situated within what we call the Pamukkale plateau but some parts of northeastern Hierapolis are sited on the
Fig. 3. Map of the distribution of morphological varieties of travertine and active faults in the Hierapolis area (based on Altunel & Hancock 19930). It should be noted that the large outcrop of actively depositing terraced-mound travertine near Pamukkale is sited in the hanging wall of the Hierapolis fault zone.
CREATION AND DESTRUCTION OF TRAVERTINE MONUMENTAL STONE
Fig. 4. Vertically banded fissure travertine cutting horizontal bedded travertine in the Yarikkaya fissure-ridge, about 1400m north of Develi.
lower slopes of the Pamukkale range front. The plateau is bounded to the northeast by the 300m-high Pamukkale range front that defines the southwest edge of the Kiiciik £6kelezdag massif. There are, except in the south, two levels within the Pamukkale plateau. Hierapolis is situated on the upper level, adjacent to the
5
Pamukkale range front, and about 30m above the lower level, which contains most of the travertine deposits (Fig. 3). The southern segment of the slope between the two terraces of this divided plateau is the site of the most spectacularly white of the actively depositing travertines. The major valleys, which coincide with the Menderes and Gediz grabens, were important routeways for peoples and armies travelling either west or east between the Aegean coastlands and the central Anatolian plateau, a region that gave access to the upper Euphrates valley and from there to Persia and further east (see Ramsay's (1890) map of routes in ancient Asia Minor). Although the Menderes graben provided the easiest pathway from the Aegean coast via Tralles (modern Aydin) and Nyssa to the interior, the Gediz graben was also a vital route connecting Magnesia (modern Manisa), King Croesus' city of Sardis and Philadelphia (modern Alasehir) in the Gediz graben to the cities of Tripolis, Hierapolis, Laodicea and Colossae in the Denizli basin. Because the Denizli basin lies at the confluence of two major routes from the coast to the interior and because it is a large area of relatively level and well-watered ground close to the Anatolian plateau it is no surprise that cities developed within it in ancient times. In addition to Hierapolis (of Greek foundation, although mainly Roman and Byzantine monuments remain), there were Tripolis, Laodicea and Colossae (Fig. Ib). Hierapolis was probably occupied before Classical times, its hot springs, both then and later, being a great attraction. Furthermore, the presence of a holy spring (the Plutonium next to the Temple of Apollo; see later discussion) in Hierapolis was critical to
Fig. 5. A vertical crestal fissure of approximately 5 m width within an inactive NW-trending fissure-ridge about 1500m north of Develi. (Note the gentle dip of the bedded ridge travertines away from the fissure, which is the site of a Roman quarry from which Phrygian marble was obtained.)
6
P. L. HANCOCK ET AL.
Fig. 6. Terraced-mound travertine with metre-scale pools, 800 m NE of Pamukkale village. Water supplying these pools issues from springs sited on the Hierapolis fault zone. (Note the palisades of stalactites fringing each pool.)
Fig. 7. A perched self-built and petrified water channel of approximately 10m height that is now ruptured, possibly as a result of earthquake ground shaking; about 1600m east of Develi.
maintaining the city's continuing importance throughout Classical times. It is also noteworthy that Herodotus (484-420 BC), who in Book 7 writes about this area, refers to a city in the present neighbourhood of Hierapolis as Cydrara, and later mention is also made by him to Hydrela, both places possibly being the city that we now think of as Hierapolis. Laodicea was a Greek city, mainly built of travertine, but with important Roman modifications in the form of aqueduct pipes bringing water from a spring in Denizli, 8km away, across a valley and into a water tower within the city centre. The pipes of this damaged water tower are now furred-up with calcareous deposits, testifying to the widespread presence of dissolved calcium carbonate in the ground waters of the entire Denizli basin. Colossae, which is much less well preserved than Hierapolis or Laodicea, was, again, both a Greek
and Roman city. Tripolis, a mainly Greek city in the extreme northwest of the Denizli basin, is also built of travertine quarried from a small mass within the footwall block of the rangefront fault. Until excavated, the city was largely covered by slope deposits derived from the range front. Herodotus (1954), in Book 7 states that this is the route taken by Xerxes, who is reputed to have discovered, near what was probably Tripolis, a plane tree so beautiful that he decorated it with golden ornaments. Travertine deposits In this paper we use the term 'travertine' to embrace all 'freshwater' limestone products of deposition from hot carbonate-rich spring waters irrespective of whether they are compact (i.e. travertine as often defined) or whether they are
CREATION AND DESTRUCTION OF TRAVERTINE MONUMENTAL STONE
7
Fig. 8. Plan and profiles of the £ukurbag ridge (900 m east of Develi), the profiles emphasizing that Roman (and possibly later) quarrying has given rise to trench-like excavations where vertically banded fissure-ridge travertine has been selectively extracted.
porous, and might be called 'tufa' (Ford & Pedley 1996). Neogene clastic sediments are the most abundant materials underlying the Pamukkale plateau but the most distinctive rocks are travertines of Quaternary age, mainly less than 400 ka (Altunel & Hancock 19930). Although the older travertines are of middle Pleistocene age (>400ka), travertines are still being deposited, testifying to the continued action of hydrothermal flow in this area of active faulting. All the travertines are products of the degassing and
consequential precipitation from hot carbonaterich waters that emerge from springs aligned along fissures and faults that opened during the late Quaternary stretching of the Denizli basin. Stretching was also responsible for the increments of slip on the normal faults that frame the basin. The Pamukkale range-front fault is the closest of these faults to the travertines of the Pamukkale plateau. Of the five types of landform constructed of travertine (Altunel & Hancock I993a,b, 1996),
Fig. 9. A well-defined trench-like quarry corresponding to the central fissure of a ridge 900m NNW of Pamukkale (visible in the background). The bedded travertines of the ridge are higher on the east side of the fissure because it expresses the location of an underlying normal fault downthrowing west.
8
P. L. HANCOCK ET AL.
Fig. 10. Part of an abandoned Roman column cut from bedded travertine in the £ukurbag ridge, about 900 m east of Develi.
Fig. 11. A toppled wall constructed of travertine blocks quarried from bedded travertine. Toppling of this wall to the northeastern side of the Colonnaded Street probably occurred during an earthquake. (Note that a small petrified water channel (arrows) formed on the upstanding edges of one row of fallen blocks.) three are especially important from the perspective of the construction and destruction of Hierapolis (Fig. 3): (1) The 21 fissure-ridges, mainly younger than 80 ka, each comprise a crestal fissure filled by vertical colour-banded travertine cutting white- to yellow-bedded travertine dipping away from ridge crests (Figs 4 and 5). Ridges range in length from about 100 to 1500m in width from about 5 to 500m, and in height they rise up to 25m above the surrounding nearly flat land of the lower level of the Pamukkale plateau. (2) Terraced-mound travertines, many of which are still accumulating, mantle the hillslopes between the upper and lower levels of the Pamukkale plateau down which the carbonaterich waters have flowed and cascaded. Where they are being actively deposited these travertines are snow-white but where they are dry they
have weathered to an unattractive dark brown to black colour. Numerous hot pools encircled by palisades of travertine stalactites characterize the southern slopes of the main mass of terraced-mound deposits near Pamukkale village (Figs 3 and 6). (3) Self-built channel travertines, the sites of most of which were artificially determined, are wall-like features of l-2m width developed where carbonate-rich, hot spring waters flow in a confined channel used for irrigation or other purposes. Degassing during turbulent flow leads to the precipitation of travertine on the floors and walls of these channels, which grow in height until the channel becomes perched far above its original level. Some of these petrified water channels date from Roman times because: (a) Vitruvius described them at the time of Augustus (27BC-AD14) (D'Adria, in Peres
CREATION AND DESTRUCTION OF TRAVERTINE MONUMENTAL STONE
9
subsurface water flow, as Muir Wood (1993) has suggested happens elsewhere in the world, is not known. Thus whether travertine deposition at Hierapolis was greater or less at such times is a major question requiring an answer. Travertine as an ancient building material
Fig. 12. Fissures cutting the walls of part of the Northern Baths, a monument built of bedded travertine blocks. Fissuring is a characteristic form of damage in dry masonry walls shaken during an earthquake. 1987), and (b) Hierapolis is the centre of the network they define. Some channels, such as those passing through the Northern City Gate or the Byzantine Basilica, are clearly younger; indeed, some of them are still in use. On the outer and steeper slopes of the Pamukkale plateau some petrified irrigation channels have grown as high as 10m (Fig. 7). The genetic connection between travertine deposition and earthquake faulting is that the stretching responsible for normal faulting also opened associated vertical fissures striking parallel to the faults (Altunel & Hancock 19930). It is mainly these fissures that have allowed the carbonate-rich hydrothermal waters to rise to the surface. In addition, some waters ascend via the steeply inclined fault zones, some of which curve to become nearly vertical fissures close to the ground surface. The Pamukkale area and surrounding region are characterized by abnormally high heat-flow values. This is vividly reflected in the geothermal field near Cubukdagi, about 15 km to the west of Pamukkale (Simsek & Okandan 1990). Whether individual slip increments on faults are accompanied by changes in
Much of Roman and Byzantine Hierapolis is built of travertine extracted from quarries in the fissure-ridge deposits situated below the city. The distinctive attribute of these quarries is that many of them are narrow (2-10m) but deep (5-20 m) vertical-sided trenches. Each trench is the site of a nearly vertical fissure that was filled by banded fissure travertine, one type of the socalled Phrygian marble (Figs 5, 8 and 9). Trench walls, formerly the contacts between vertical banded travertines and outward dipping bedded travertines, display scaffolding holes and many are decorated by a herringbone pattern of tool marks identical to those on sarcophagi and many of the blocks that were used for constructing buildings at Hierapolis. The dense and attractively banded travertine from fissures was principally used as an ornamental stone, whereas the bedded travertine from ridge sides was mainly employed as a dimension stone and for making columns (Fig. 10). Bean's (1971) remark that local marble was not widely used in Hierapolis does not accord with our experience, unless he was referring only to the banded travertine from fissures. Notable buildings constructed of bedded travertine blocks include, from north to south (Fig. 2): (1) the Northern Roman Baths dating from the second-third centuries and including a Byzantine (fifth-century) Basilica built within it; (2) the Hellenistic Theatre, which was largely demolished in the first century AD when the Roman Theatre was built; (3) the Monumental (Frontinus) Gateway dedicated to Domitian, dating from the end of the first century AD; a structure that was rebuilt after a devastating earthquake; (4) the Nymphaeum adjacent to the Monumental Gateway; (5) the early Christian Martyrion in honour of St Philip from the end of the fourth or beginning of the fifth century; (6) the Northern Gate through the city wall; (7) the Temple of Apollo; (8) the Roman Theatre; (9) the Southern Baths (now a museum); (10) a Byzantine Basilica of the sixth century; (10) a 12th-13th century Byzantine fort; (11) the sixth-century Southern City Gate. In addition, most tombs in both the Northern and Southern Necropolis, beyond the city walls, are built of travertine blocks, as is the city wall itself. According to D'Adria (in Peres 1987)
Fig. 13. Plan of the Hierapolis fault zone. The shapes of buildings are schematic and not all modern buildings are shown (after Hancock & Altunel 1997). It should be noted that in section A-B the petrified water channel is downfaulted in a mini-graben where it has been stretched over two normal faults that are oriented roughly at right angles to the line of the channel; the appearance of this mini-graben is illustrated in Fig. 15.
CREATION AND DESTRUCTION OF TRAVERTINE MONUMENTAL STONE
11
Earthquake history of Hierapolis
0
50
Fig. 14. Sketch plan of a petrified irrigation channel that has been offset at two places by the Hierapolis fault about 90 m SSE of the northern city wall. It should be noted that where the channel locally trends E-W the offset as determined from piercing points combines horizontal opening with sinistral motion, and where it locally trends N-S opening is combined with a dextral sense of motion. These apparently contradictory senses of horizontal offset have arisen because the observed 'strike-slip' sense is controlled by the angle between a channel long axis and the opening direction across the trace of the fault. There is also an overall downthrow of about 7 cm to the west across the fault (after Hancock & Altunel 1997).
blocks from many buildings were cannibalized to construct the city wall, which was built as a result of a law introduced in AD 396, that is, at a time late in the Roman Empire. Furthermore, many of the Doric columns lining the sides of the Colonnaded Street (Via Domizianea) are of travertine and directly comparable with the stone of the abandoned column at the £ukurbag quarry (Fig. 10).
Although faulted archaeological features are readily analysed indicators of deformation patterns associated with earthquakes they may be more difficult to employ as guides to their timing. Historical earthquake catalogues are needed for this aspect of their analysis. The scholarly catalogue of pre-tenth century AD earthquakes by Guidoboni et al. (1994) indicates that the following earthquakes after the birth of Christ but before the 10th century were destructive at Hierapolis: 47, 60, an unknown date in the third century, an unknown date in the fourth century, 494, and early in the seventh century. It is also possible that an earthquake recorded by Guidoboni et al. (1994) as having occurred in about 27 BC, which was responsible for rebuilding work at Laodicea, might have also damaged Hierapolis, less than 10km from Laodicea. Soysal et al. (1981) also reported the AD 60 earthquake in their catalogue, and added to it events of MSK intensities VIII and VII in 65 and 20 BC, respectively. Most archaeological and historical writers focus on the AD 60 earthquake; for example, Peres (1987) recorded that Hierapolis was rebuilt after it, as did Bean (1971), and McDonagh (1989) also reported that Laodicea was rebuilt after the event. Earthquakes of MSK intensity VII or more that affected the Hierapolis area between the 10th and 20th centuries include those of 1354, 1651, 1703, 1887 and 1899, according to Soysal et al. (1981), Ates & Bayiilke (1982) and Ambraseys (1988). In the 20th century, the only event of Ms greater than 6.0 to have affected the Hierapolis area is that of 1900 reported by Ergin et al. (1967) and Gencoglu et al. (1990).
Earthquake damage at Hierapolis Many of the monuments that had been constructed of travertine related to older Quaternary episodes of faulting were then destroyed or damaged by reactivation of the faults, which caused continued earthquake shaking or ground displacement. For example, the toppled wall (Fig. 11) part of the Colonnaded Street, collapsed without losing its essential form and then became the foundation of a later petrified water channel, which follows part of its length. Vertical fissures, which elsewhere are regarded as a characteristic form of earthquake shaking damage (Stiros 1996), rupture the walls of the thirdcentury Northern Baths (Fig. 12). Examples of buildings reported by D'Adria (in Peres 1987) as having been reconstructed after the
12
P. L. HANCOCK ET AL.
Fig. 15. A slightly raised petrified irrigation channel cut by two fissure-faults (arrows) reflecting the locations of underlying normal faults that dip towards each other. The nearer of the faults, which achieves the greater displacement, downthrows west; the further but subordinate fault downthrows east. A section along the channel close to the gendarme post is shown in section A-B in Fig. 13.
Fig. 16. The so-called 'sacred pool' containing a submerged section of the Roman Colonnaded Street and fallen columns. The pool is sited on the trace of the Hierapolis fault zone; the flooding of the hollow possibly is a result of the subsidence of a small graben within the fault zone. A thin veneer of travertine is being unconformably deposited on the columns, which, being aligned, might have toppled during an earthquake.
AD 60 earthquake in Nero's reign include the Temple of Apollo and the Roman Theatre. In the southeastern Necropolis a sarcophagus made of travertine was overturned by earthquake shaking (presumed here to be the AD 60 event because Ronchetta (in Peres 1987) described it as 'the' earthquake). The zone of greatest earthquake damage coincides with the trace of the Hierapolis fault zone, whose location was detected from an alignment of offset walls and petrified irrigation channels (Fig. 13) (Hancock & Altunel 1997). The kinematic class of this fault zone could be determined precisely from the many offsets of the linear archaeological features, which are cut by vertical fissures expressing faults that are steeply
inclined a few metres below the surface (Hancock & Altunel 1997) (Fig. 13). The offset archaeological features permitted piercing points, and hence opening directions, to be determined, thus allowing the fault zone, previously thought by Altunel & Hancock (19930) to be a sinistral strike-slip, to be reinterpreted as a normal fault zone achieving a small downthrow to the southwest. In plan, the sense of the angle between a channel and the trace of a fault determines whether a channel is offset horizontally in a dextral or sinistral sense (Fig. 14). Where the line of a channel is subparallel to the opening direction there will be horizontal opening and a normal component of displacement of the channel. This gives rise to a mini-graben, where a
CREATION AND DESTRUCTION OF TRAVERTINE MONUMENTAL STONE subsidiary normal fault that is antithetic to the main one also cuts the channel (Figs 13 (section line A-B) and 15). The knowledge that the Hierapolis fault zone is a structure across which there is active stretching and increased hydrothermal flow, as reflected by the concentration of hot springs in the zone, helps to explain why the area of greatest present-day travertine deposition is situated in its immediate hanging wall and just downslope of the fault trace (Fig. 3). In the once-beautiful 'sacred pool' (now within the Pamukkale Motel and not be confused with the holy cavern of the Plutomium) a thin veneer of travertine is being unconformably deposited on columns that have fallen alongside a submerged paved area, possibly a continuation of the Colonnaded Street. The pool, possibly sited on a small graben within the Hierapolis fault zone, is being fed from a hot spring within the fault zone. The nearly uniform direction of toppling of the fallen columns might be a reflection of earthquake shaking (Fig. 16) (Nur & Ron 1996), and, if this is so, the deposition of travertine on the columns testifies to the intimate relationship between faulting, ground rupture, earthquake shaking and travertine deposition. Hierapolis, like Delphi (Greece), possesses a holy cavern about which legends have grown up. At Delphi the Oracle's mantic sessions might have been a result of the Oracle inhaling light hydrocarbon gases arising from buried limestones rich in bitumens (De Boer 1999). At Hierapolis, the legend that animals and men, other than certain priests, who entered the holy cavern, known as the Plutomium (adjacent to the Temple of Apollo) died in the so-called strong-smelling 'steams' we think is likely to be a consequence of the concentration of carbon dioxide in addition to other gases in such a small subsurface chamber into which hot waters flowed. The Plutomium's proximity to the eastern branch of the Hierapolis fault zone means that beneath it there is likely to be a higher than normal concentration of vertical fissures that are the conduits for such waters (Fig. 13). Summary (1) The travertine deposits at Hierapolis are secondary products of the area having been stretched during earthquake normal faulting for at least 400 000 years. (2) The bedded travertines of fissure-ridges have been quarried since before Roman times for dimension stone and were the principal building materials used by Greeks, Romans and Byzantines in the construction of monu-
13
ments. Finely banded travertines quarried from the vertical fissures cutting ridges were mainly used for ornamental purposes. (3) Many of the monuments and other features built of travertine have been damaged by earthquake shaking or faulting during the period since at least AD 60. Damage is concentrated along a narrow corridor coincident with the Hierapolis fault zone. (4) Opening directions determined from piercing points defined by displaced features allow the Hierapolis fault zone to be reinterpreted as a normal fault zone. The area of contemporary greatest travertine deposition is just downslope and in the immediate hanging wall of the fault zone. If this relationship occurs elsewhere it might be employed for finding the locations of earthquake faults. On site, P. Arthur of the University of Lecce explained to us the significance of numerous archaeological monuments. We thank the Universities of Bristol, Osmangazi and Mersin, and the Natural Environment Research Council of the UK for grants supporting our research.
References ALTUNEL, E. & HANCOCK, P. L. 19930. Active fissuring and faulting in Quaternary travertines at Pamukkale, western Turkey. Zeitschrift fur Geomorphologie Supplementary Volume, 94, 285-302. 19936. Morphological features and tectonic setting of Quaternary travertines at Pamukkale, western Turkey. Geological Journal, 28, 335-346. 1996. Structural attributes of travertine-filled extensional fissures in the Pamukkale plateau, western Turkey. International Geology Review, 38, 768-777. AMBRASEYS, N. N. 1988. Engineering seismology. Earthquake Engineering and Structural Dynamics, 17, 1-105. ATES, R. C. & BAYULKE, N. 1982. The 19 August 1976 Denizli, Turkey, earthquake: evaluation of the strong accelograph record. Bulletin of the Seismological Society of America, 72, 1635-1649. BEAN, G. E. 1971. Turkey Beyond the Maeander. Ernest Benn, London. DE BOER, J. Z. 1999. Could emission of light hydrocarbon gases have played a role in the mantic sessions at Delphi (Greece)? This volume. ERGIN, K., GUCLU, U. & Uz, Z. 1967. A Catalog of Earthquakes for Turkey and Surrounding Area (HAD to 1964AD). ITU Faculty of Mining Engineering, Istanbul. FORD, T. D. & PEDLEY, H. M. 1996. A review of tufa and travertine deposits of the world. EarthScience Reviews, 41, 117-175. GEMCOGLU, S., INAN, E. & GULER, H. 1990. Tiirkiye'nin Deprem Tehlikesi (Earthquake Hazard of Turkey). Chamber of Geophysical Engineers of Turkey, Ankara.
14
P. L. HANCOCK ET AL.
GUIDOBONI, E., COMASTRI, A. & TRAINA, G. 1994. Catalogue of Ancient Earthquakes in the Mediterranean Area up to the 10th Century. Institute Nazionale di Geofisica, Rome. HANCOCK, P. L. & ALTUNEL, E. 1997. Faulted archaeological relics at Hierapolis (Pamukkale), Turkey. Journal of Geodynamics, 24, 21-38. HERODOTUS 1954. The Histories, Book 7 trans Selvincourt, Pengiun Books. JACKSON, J. 1994. Active tectonics of the Aegean region. Annual Review of Earth and Planetary Sciences, 22,239-271. McDoNAGH, B. 1989. Blue Guide: Turkey: the Aegean and Mediterranean Coasts. A. & C. Black, London. MUIR WOOD, R. 1993. Neohydrotectonics. Zeitschrift fur Geomorphologie, Supplementary Volume, 94, 275-284. NUR, A. & RON, H. 1996. And the walls came tumbling down: earthquake history in the Holyland. In: STIROS, S. & JONES, R. E. (eds) Archaeoseismology. Fitch Laboratory Occasional Paper, 7, 75-85.
PERES, A. (ed.) 1987. Hierapolis di Frigia 1957-1987. Settore Cataloghi D'Arte del Gruppo Editoriale Fabbri, Torino. RAMSAY, W. M. 1890. The Historical Geography of Asia Minor. John Murray, London. SIMSEK, S. & OKANDAN, E. 1990. Geothermal energy development in Turkey. Transactions of the Geothermal Research Council, 14, 257-266. SOYSAL, H., SlPAHIOGLU, S., KOLCAK, D. & ALTI-
NOK, Y. 1981. Turkiye ve £evresinin Tarihsel Deprem Katalogu (M.6.2100-M.S. 1900). (Historical Earthquake Catalog of Turkey and its Environment, 2100BC to 1900AD). TUBITAK Publications, Ankara. STIROS, S. C. 1996. Identification of earthquakes from archaeological data: methodology, criteria and limitations. In: STIROS, S. & JONES, R. E. (eds) Archaeoseismology. Fitch Laboratory Occasional Paper, 7, 129-152. WESTAWAY, R. 1993. Neogene evolution of the Denizli region of western Turkey. Journal of Structural Geology, 15, 37-53.
Uses of volcanic products in antiquity D. R. GRIFFITHS Institute of Archaeology, University College London, 31-34 Gordon Square, London WC1H OPY, UK Abstract: Since the advent of mankind many human societies have lived in volcanically active zones. The geological, archaeological and historical records provide a rich and diverse source of evidence for both archaeology and volcanology concerning the nature of volcanic processes and the effects of volcanism on the environment and on human society. To achieve a balanced understanding of the effects of volcanism on past cultures, it is important to consider the attractions as well as the hazards of life in an actively volcanic zone. This paper gives an overview of some of the ways in which a wide range of volcanic products were used by mankind in antiquity. These include the use of volcanic rocks as stone tools, as substrates for rock carvings, as materials for building and sculpture, as millstones, as additives to make cements that set under water, and, more indirectly, as precursors of fertile earths for agriculture and as sources of metals and semi-precious stones. The paper also considers some of the properties of volcanic products that may have made them attractive. Assessment of the interaction of past cultures with volcanism is highly relevant to the present: it can provide the temporal perspective needed to deal appropriately with the human aspects of contemporary and future volcanic hazards.
Much consideration has been given to the effects of volcanic events on past societies. This ranges from investigating legends of Atlantis and lost civilizations to the detailed debates in recent decades on the effects and dates of eruptions of the volcano Thera (or Santorini). In these and other considerations of the interplay between mankind and volcanoes, the dramatic aspects of volcanic activity understandably receive the most attention and volcanoes are seen primarily as agents of destruction. Dramatic and destructive volcanic events tend to leave their mark clearly in the geological, archaeological and, on occasion, the historical record, whereas the effects of day-to-day normal levels of volcanic activity are more subtle and harder to discern. Destruction is, however, far from being the only aspect of volcanic activity worthy of attention in studying the past relationships of mankind with volcanoes. Dramatic and destructive volcanic events may leave their mark but they are comparatively rare in day-to-day human experience, even in highly volcanically active areas. Generations might live their entire lives reaping the benefits of their volcanic location while being aware of the potential for destruction only through history or folklore. This paper attempts to provide a brief overview of the ways in which mankind has in the past used the natural resources made available by volcanic activity. It will also consider some of
the properties of these volcanic products that have made them seem attractive and useful. Awareness of the benefits mankind has derived from the use of volcanic products may encourage a more positive awareness of the beneficial aspects of living in a volcanic environment when the complex interactions of past societies with volcanic activity are being considered. Achieving a proper balance between our awareness of the beneficial aspects of volcanism and of its dangers is obviously important in interpreting the interaction of past societies and volcanism. This balance is important not only for our understanding of the past: it is also highly relevant to the present. A detailed and balanced understanding of past events can provide the temporal perspective needed to deal appropriately with the human aspects of contemporary and future volcanic hazards. Although dramatic volcanic events will cer-, tainly disrupt and sometimes end the lives of those who live within reach of their effects, it is worth considering that in the greater scheme of things these disasters may be seen primarily as triggers of change. The destructive effects of a volcanic event may partly or wholly destroy a particular culture but at the same time it may allow the remnants of that culture to re-evolve in a new direction. The more profound the destruction, the less certain the outcome, at least in terms of the survival or re-evolution of the
From: McGuiRE, W. G., GRIFFITHS, D. R., HANCOCK, P. L. & STEWART, I. S. (eds) The Archaeology of Geological Catastrophes. Geological Society, London, Special Publications, 171, 15-23. l-86239-062-2/00/$ 15.00 © The Geological Society of London 2000.
16
D. GRIFFITHS
culture immediately affected by the volcanic event. It is also the case that the destruction or dispersal of one human group by a volcanic event may make way for another to rise to new heights in the new circumstances. A loss for an individual person, a family, a community or a culture may present an opportunity for gain to another such entity. Volcanic events may thus be seen as triggers of change, natural events (natural disasters to those who suffer) that cause a sudden discontinuity in the rise and fall of communities or civilizations. In the, longer term, such disasters and discontinuities may have provided the possibility of more rapid advancement of human civilization than might have been possible under the more restrictive circumstances of an established order. It may not be absurd to ask whether, in the development of mankind as a whole, volcanic events might not on occasion have accelerated the process. Taken together with some of the observations presented here on the usefulness of volcanic products, it is not inconceivable that living in a volcanic zone might be conducive to the development of human civilization. Although it is worth while to indicate the greater questions that might be in part elucidated by the subject of the paper, they are not the central concern here. Where volcanic and human activity are interrelated, the volcanic materials and events can provide a potentially rich source of archaeological information on the human procurement of raw materials (quarrying practice, routes and means of transport, etc.) and the dates of archaeological events. This in turn may serve to assist our understanding of past human behaviour and society (trade and exchange, organization of society, rates and geographical direction of change, etc.). Similarly, there may be occasions when archaeologically datable assemblages of artefacts in association with volcanic deposits can assist in the dating and improved understanding of volcanic processes, such as the effects of ashfalls on the development of soils and agriculture (Olson 1983). Although the immediate aim is to draw attention to the wide variety of use that mankind has made of natural resources that derive from volcanic activity, one of the greater goals of this paper (and many others in this volume) is to improve our understanding of the complex interactions of past societies with the effects and products of volcanism. In this area of endeavour at least, understanding of the past should provide far more than purely academic rewards: understanding of the effects of volcanic events on past societies should also improve enormously our ability to plan for and deal with the
human aspects of volcanic hazards now and in the future. The way of life and the material possessions of some modern people may seem very different from those of the past but, in the event of a major volcanic disaster, the basic human needs may seem remarkably unchanged. Knowledge of the past may well provide the vital information to enable us to make an informed decision on how best to adapt to volcanically driven societal and environmental change.
Criteria for the selection of volcanic materials in the past In general, the criteria used for selecting a particular raw material will depend on the role the material it is intended to fulfil in a particular time and place. These criteria may include the accessibility or availability of the material, its workability, durability, aesthetic qualities and the like. Physical properties and aesthetic qualities are often intricately intertwined in the perception of the artist-craftsman. Before considering individually the volcanic resources that have been used and the more obvious aspects of their properties that may have made them attractive, it is worth noting the possibility that in the past the criteria directing the selection of materials for a particular purpose may not have been the same as they might be now. Methods available for extracting, working or transporting a material may have strongly influenced the material chosen. It should also be noted that there might in the past have been constraints on access to given geographical regions imposed by the extents of political, economic or military control. It is also true that particular sources or uses of materials might not yet have been discovered. Slightly less obviously, social or religious taboo might effectively restrict accessibility. Furthermore, it is possible that the desirability of given materials might be influenced by fashion or by symbolic significance, be that symbolism political, cultural, mystical, magical, religious or some combination of these. Archaeological evidence for the influence of such criteria in determining the selection of materials is usually circumstantial. It may thus be appropriate to consider at the outset the possible symbolic or metaphysical significance of volcanic resources so that this possibility may be borne in mind when considering more physically based criteria for selecting volcanic materials for particular functions. Volcanic activity has probably always been an object of awe in the minds of all those
USES OF VOLCANIC PRODUCTS IN ANTIQUITY who behold it, awe which may have manifested itself as fear and reverence. It would not seem unreasonable that past cultures might have perceived volcanoes as all-providing, all-engulfing instruments of mighty power and, to some, vengeance. The act of witnessing that some materials were the products of volcanic activity may have caused those materials to assume mystical or symbolic significance in the perceptions of eyewitnesses and their society and descendants. Even if volcanic events are rare on the scale of a human lifetime, they are often dramatic enough to ensure the preservation of their memory in history, be that in a formal written sense, as oral tradition or as what might be termed folklore. In some cases visual depictions may also survive. The choice of volcanic materials for use in a given context may thus have been influenced by the very fact that they were known to be of volcanic origin. It may be worth distinguishing between symbolic significance that is based on and associated with the knowledge that a given material is of volcanic origin, and symbolic significance that is based on some other aspect such as the appearance of the material, the circumstances of its discovery or its geographical origin. A material that is difficult to acquire might be intended to symbolize wealth or power rather than anything religious or mystical. In these cases, the user may be wholly ignorant of the fact that the material is more or less directly the product of volcanic activity. Despite our attempts at rationalizing motives for choosing particular volcanic materials, the reality may often be far more complex than our interpretation, and combinations of practical and symbolic criteria may influence selection of a particular material. It is often difficult to determine whether or not symbolic or metaphysical factors may have had an influence in the selection of raw materials. Part of the difficulty in inferring what criteria may have influenced the selection of particular materials arises from the nature of archaeological evidence. Archaeology gathers its information from the material remains of past human activity. This evidence relates most directly to past material culture, and it is very difficult to proceed from this with any degree of certainty to draw conclusions about the thoughts associated with past actions or the motivation driving those actions. Archaeology may (generally with some difficulty) be able to discover some aspects of what people did, where they did it, when they did it and how they did it, but to discover why they did it is the hardest question of all, a question that cannot generally be answered with any certainty.
17
Although the pitfalls of too ready a recourse to ritual explanation are fairly obvious, there are a number of examples of possible ritual influence in the choice of raw material that may be worthy of note. One of the most remarkable may be the vast heads carved out of basalt boulders by the Olmecs. It is possible that the carvers or their predecessors had seen such boulders flung from erupting volcanoes and made their carvings mindful of the volcanic origin of the boulders. Another example of the choice of a volcanic material possibly being influenced by ritual or symbolic considerations may be the choice of andesite, a dark grey volcanic rock whose phenocrysts glisten in the sun, for the Inca Koricancha or Temple of the Sun in Cuzco, Peru. Field work undertaken with Hunt and Protzen in 1988 and subsequent petrographic analysis strongly suggests that this material was transported from a quarry at Rumiqolqa 35km away when ample alternative sources of building stone were near at hand and nearer sources of andesite existed (Hunt 1990). In this case, even though there might be a symbolic element in the selection of the material, it is possible that the builders were unaware of the volcanic origin of the material. Numerous other examples of a possible symbolic or ritual element in the choice of volcanic stone might be cited but the great majority would suffer from the common problem that although archaeology may be able to establish many facts about life in past times, one can never be certain of people's motivations in acting as they did.
Stone tools made of volcanic rock Many of the earliest known stone tools, believed to be about 2.3 Ma old, were made in east Africa by flaking volcanic lava (Musty 1999). The fine-grained, isotropic texture of many lava samples means that the distribution of stress within the material upon the application of a force is fairly predictable and the mechanical strength is similar at all points in all directions. The way in which the material will flake when struck is therefore fairly predictable and a good degree of control over the flaking process can be achieved with practice. The more homogeneous the mechanical properties of the material, the sharper the edges of the flakes are likely to be, as the fracture surfaces will tend towards perfect conchoidal fracture and not be deflected by intergranular weaknesses or differences in tensile
18
D. GRIFFITHS
strength (Griffiths et al 1987; Cotterell & Kamminga 1990). Perfect conchoidal fracture can be obtained in samples of volcanic glass that have solidified as a single phase. Many samples of volcanic glass do approach this ideal although some contain crystalline phases. Wholly or almost wholly vitreous acidic rock is termed obsidian. This has been a highly sought-after volcanic product for millennia, being transported over considerable distances from the sources. The excellent flaking qualities of obsidian mean that it can be used to produce extremely sharp cutting edges, although these are rather brittle. The predictability of fracture means that long blades and very fine, precisely shaped flaked stone objects can be produced. Given the high degree of skill necessary to produce some of the finest objects and their fragility, some such objects may have had more symbolic than practical use, but this possibility should not overshadow consideration of the practical aspects of the appeal of obsidian. The main use of obsidian has been for tools and weapons, but it was also used for carving figurines and for mirrors in Mexico (Kunz 1971, pp. 204-205) and for Neolithic period mirrors at £atal Huyiik in Anatolia (Shackley 1977, p. 54).
One example where knowledge of volcanic origin, mechanical properties and good visual contrast might each have had an influence may be found in the rock carvings on pahoehoe lava at Puuloa in the district of Puua on Hawaii. Topographic location may also have had an influence (Eleftheriou 1990). These carvings may show up particularly well because of textural contrast between the glassy rapidly cooled surface and the more granular more slowly cooled interior exposed by the carving. Similar contrast is seen with carvings made on the walls of lava tubes. Other rock carvings on Hawaii and neighbouring islands are made on large boulders, and here their initial visibility is attributable to the colour contrast between the weathered surface and the unweathered interior of the rock. Other rock carvings on the Hawaiian islands are, however, made on sedimentary rock. This highlights the difficulty of determining the motives behind the selection of a volcanic material for a function: was the selection fortuitous, was there an element related to the fact that the material was known to be of volcanic origin, was the selection based on the physical properties of the material or were other factors dominant in the minds of the people who chose to carve where they did?
Rock carvings on volcanic rock substrates
The use of volcanic rock as building stone
Thousands of rock carvings on more or less flat rock surfaces (as opposed to three-dimensional sculptures) survive from antiquity in many parts of the world. It is sometimes interesting to question whether particular preference may have been shown for using volcanic as opposed to other rock as a substrate for the depiction of images. Certainly, there are many examples from around the world of carvings on volcanic substrates. In some cases, the choice of substrate may have been influenced by knowledge of the volcanic origin of the rock. There are, however, many other possible criteria that might have influenced the selection of particular substrates including geographical location or orientation. Two further factors that might cause a particular rock type to be favoured are the mechanical properties of the rock and the visual contrast between the worked line or area and the pre-existing uncarved rock surface. A number of examples of the occurrence of rock carvings on volcanic substrates might be cited where one or both of these factors may have caused the rock to be favoured.
The use of andesite to build the Koricancha in Cuzco was mentioned above but there are many other andesite buildings in Cuzco and elsewhere where the possibility of a symbolic aspect in the choice of raw material may be less likely. Studies of the provenance of the rock using thin-section petrology to compare fragments from buildings with samples from known ancient quarry sources and geological exposures, showed convincingly that the stone used by the Incas to build the Koricancha and various other buildings in Cuzco was being imported from some distance over very difficult terrain (Hunt 1990). It is clear that the Incas were not using the ample stone material that was available near at hand and that had seemed acceptable to their predecessors as a material for construction of Sacsaywayman which overlooks Cuzco. What is less clear is the balance of criteria that drove them to select andesite from quarries 35km away from the town. Volcanic origin might have played a role but other qualities include the aesthetic appeal of phenocrysts glittering in the sunlight against the matt blackness of the fine-grained groundmass of the rock.
USES OF VOLCANIC PRODUCTS IN ANTIQUITY More pragmatic considerations of the workability of the rock may also have influenced the selection of andesite as the stone of choice. Initial extraction of andesite from at least some of the quarries is facilitated by the existence of large-scale fractures in the rock, perhaps arising from thermal stresses as the extruded rock originally cooled. Although andesite is hard it is considerably easier to work than one might at first expect. It is brittle and being largely fine grained exhibits fairly poor but still useful conchoidal fracture. Protzen has demonstrated that an approximately rectangular block, of andesite can be fashioned fairly quickly and without great labour from an irregular chunk extracted from a quarry. Protzen showed the author how a large hammerstone can be bounced on the surface of the andesite block removing some material by crushing on impact. If the hammerstone is given a slight flick towards one as it is about to hit the block, the transverse motion often removes flakes from the surface of the relatively brittle fine-grained andesite. This greatly enhances the rate of removal of material and hence the speed of shaping the blocks (Protzen 1986). Extraction and rough forming are thus fairly easy. Many hammer stones and partially finished blocks of andesite were found at quarry sites. The final and most impressive stage was the fashioning of the blocks so that large structures could be built without any mortar, each block sitting in an irregular but smoothly contoured depression on top of the course below, perfectly formed to receive the shape of the next block without leaving room for even a razor blade to be inserted. Thousands of other examples of ancient building and carving in andesite or basalt may be found elsewhere in the world, the vast Buddhist monument of Borobodur in Java being but one. Kempe (19836, p. 96) cited many instances of ancient buildings constructed from andesite or basalt but said of the latter: 'its use, however, has been restricted to areas where other more suitable stones are not available'. This contrasts with Hunt's suggestion (1991, and a specific instance of his wider arguments developed in a paper in this volume) that in many parts of the world, perhaps partly for metaphysical reasons, basalt and andesite may have been rocks of choice for various building and sculptural purposes. This illustrates how widely people may differ when it comes to inferring the motivation of past cultures, although Kempe himself elsewhere admitted that: 'Rocks and minerals have long appealed to man's mythical or superstitious sense, as Pliny and Agricola amply testify' (Kempe 1983a, p. 78.)
19
The use of pyroclastic deposits for making hydraulic cements Pyroclastic deposits of various forms (ash, tuff, pumice, etc.) or other crushed volcanic materials were used in antiquity for making cement that would set under water by reaction of lime with the water and with silicates and aluminates in the volcanic material (Torraca 1988). Pure lime cement hardens by drying of the slaked lime paste followed by reaction with atmospheric carbon dioxide to form a carbonate cement: drying is required for hardening. Hydraulic cements, on the other hand, set by reaction with water and no carbon dioxide is needed. This means that large blocks of hydraulic cement or concrete may be preformed or cast in situ without the concern that air needs to reach the centre of the block to permit hardening. Hydraulic cement is water resistant once set. The incorporation of volcanic material in a cement makes it hydraulic because the volcanic material provides a source of silicates and aluminates in a reactive form. Their reactivity may derive from a combination of the presence of relatively unstable and internally stressed glass phases resulting from rapid cooling and a high specific surface area resulting from the fine particulate state of some components together with the highly vesicular nature of others such as pumice and frothy lava. (The vesicles result from the exsolution of gases dissolved in the liquid rock.) Hydraulic cements incorporating volcanic earths in a slaked lime mortar were used in the Hellenistic period around the fourth century BC. The Romans used hydraulic cements widely in marine and other architecture, incorporating pumice in the hydraulic concrete used to build the dome of the Pantheon in Rome and thereby reducing its bulk density (Torraca 1988). Materials (natural and artificial) that can produce a hydraulic reaction with slaked lime are called pozzolanas after the town of Pozzuoli, near Naples, that was a famous source of pozzolanic earths used by the Romans to make hydraulic cements (Torraca 1988). Further discussion of the use of volcanic materials in Roman hydraulic cements is provided by Siddall (this volume). Other uses of pumice In addition to its use as a weight-reducing component of the Pantheon dome, pumice was also used in Roman times to remove bodily hair, as an abrasive and by scribes for smoothing the ends of rolls of parchment (Shackley 1977, pp. 30-31).
20
D. GRIFFITHS
The use of volcanic rocks as millstones and quernstones Criteria influencing the choice of material for making millstones can be seen to be practically based. The use of vesicular volcanic rocks as quernstones and millstones is widespread in many regions throughout the world. This choice of vesicular volcanic material, although far from universal, seems far too widespread in time and place to be a reflection of anything other than the selection of a particular type of raw material for a particular task. In addition to the advantages of vesicularity discussed below, factors such as ease of extraction and forming (alluded to above in describing the advantages of andesite and basalt as building stones) may have contributed to the popularity of these rocks as quernstones and millstones. Fine-grained volcanic rock may also be resistant to the plucking of mineral grains that might contaminate flour from, for example, a sandstone millstone (Hunt & Griffiths 1992). The author hypothesizes that as a vesicle (which arises from the exsolution of gases from a cooling but still liquid volcanic rock) is first worn into by one stone rubbing against the other, the vesicle will at first act like the teeth of a cheese grater, cutting slivers from the passing cereal grains. As a given vesicle becomes progressively more worn down it tends towards becoming an approximately hemispherical depression in the millstone surface, and it ceases to present an acute-angled peripheral cutting edge to passing cereal grains. The hollow may act as a holder that holds a cereal grain static with respect to one millstone while dragging it across the edges of the open vesicles on the opposing stone. As a vesicle wears down further still, it becomes a shallow depression with an oblique-angled periphery. Too shallow to hold a grain in place, the worn down vesicle may still act as a hollow into which a passing grain may partially sink to be bruised or broken as it collides with the blunt periphery of the vesicle. At any one time, each face of a pair of millstones will present many vesicles in each of these stages of wear. It is emphasized that the proposed mode of action is at present purely speculative. Irrespective of the mechanisms of milling, however, the transport of vesicular millstones far from their sources is a testament to their effectiveness (see, e.g. Williams-Thorpe & Thorpe 1988). Fertile agricultural soils derived from volcanic deposits Weathering of volcanic deposits is widely recognized as having the potential to result in highly
fertile agricultural soils. Reasons behind this may include the fact that some of the constituents of the deposits are particularly susceptible to leaching of mineral ions into solution in the soil water. Provided these dissolved ions are retained within the soil water rather than being washed away, they provide a ready source of plant mineral nutrients. The presence of relatively unstable and perhaps internally stressed glass phases together with a high ratio of solid surface area to soil water solution volume facilitates relatively rapid rates of movement of ions into solution. Many other factors are, however, important in the formation of fertile soils (Nahon 1991). It also needs to be recognized that, even where conditions are favourable, the formation of rich soils will generally take many hundreds of years of weathering and soil development. In the shorter term of a few hundreds of years, lava flows and heavy ash falls generally yield relatively sterile ground (Olson 1983). Hydrothermal deposits associated with volcanism The process of subduction of oceanic crust generates volcanoes in the overlying crust, earthquakes from below the subduction zone and the concentration of great mineral wealth through deposition from hydrothermal waters rising from the subducted crust. This mineral wealth is thus associated with the same processes that create volcanoes and earthquakes along subduction zones. Whether or not people in antiquity associated the volcanoes, earthquakes and the mineral deposits is uncertain and perhaps unlikely. Volcanoes can be long extinct whereas the mineral deposits remain, and other processes than subduction can give rise to precious metal concentration. The vast amounts of silver extracted in the Potosi region of Bolivia during and before the Spanish conquest are, however, the result of hydrothermal processes in a subduction zone, and similar rich metal deposits exist in many regions around the Pacific Ocean. Where hydrothermal waters are driven to near the surface by subduction zone processes or by other processes, extensive deposits of minerals such as noble metals and metal sulphides leached from the rocks below may form as the water cools and/or boils because of the pressure drop near the surface. Escape of hydrogen sulphide can result in precipitation of gold, cinnabar (mercury sulphide) and realgar (arsenic sulphide). Other processes can result in precipitation of gold or copper (Hibbard 1995, pp. 395-423).
USES OF VOLCANIC PRODUCTS IN ANTIQUITY Where hydrothermal waters reach the surface precipitates may form and the water may have been used for health spas or baths. Where gases and vapours reach the surface, sulphur and other minerals may be deposited and reactions may form haematite and other minerals (Hibbard 1995, pp. 424-427). Materials such as bright red cinnabar might be used medicinally, for personal adornment or as a pigment. Similar possibilities exist for a number of substances that may result from hydrothermal activity, and such activity is often associated with volcanic activity. Much use was made in antiquity of the red chalcedony semi-precious stone cornelian or carnelian. It is formed by precipitation in amygdaloidal (literally almond-shaped) cavities in vesicular lavas from silica-rich hydrothermal or meteoric waters percolating through the rock (Hibbard 1995, p. 413). Bloodstone and agates were also prized and were similarly formed. Much secondary material eroded out of the host lava may be found in rivers and along beaches, however, so users were not necessarily aware that these materials were derived from volcanic lavas.
Fire derived from volcanoes We have dealt hitherto in this paper primarily with the solid raw material resources made available to people in antiquity by the agency of volcanic eruption. In some instances, however, the gift of fire may have been one of the more apparent resources received by mankind from volcanoes in antiquity. The high temperature of volcanic emanations often gives rise to fire in trees that stand in the path of lava. Lava may have been an important early source of fire in some parts of the world, the fire once being fed with fuel to keep it burning. Fire itself came to have manifold uses for humanity, from providing warmth and light, providing power in many forms (first over animals and enemies, later to drive machines), to permitting the modification of materials (e.g. changing the colour of pigments and jewels) and the manufacture of entirely new materials with entirely new properties such as ceramics and metals. This process of development continues to this very day. The gift of fire as such may not have been the only contribution of volcanic activity to the creation and growth of pyrotechnology. Volcanoes may also have played a role as mother of invention, a source of inspiration by example. It is not at all unlikely that the effects of heat in creating ceramics may have been noted where
21
lava flows baked adjacent earth. Nor is it impossible that the creation of metals by natural smelting of ores by volcanic activity may on occasion have been observed by people in antiquity. The archaeological importance of volcanic materials The brief overview presented in this paper has sought to show that for a variety of reasons volcanic products have been very widely used by mankind throughout antiquity. It follows from this that the study of artefacts made from volcanic material provides a potentially very valuable source of archaeological information. These studies can answer basic questions concerning provenance and technology of manufacture, and, in some cases, provide dates for archaeological events. This basic information provides clues for the drawing of many higher-level inferences about such topics as cultural interaction, trade systems and criteria influencing the choice of raw materials for particular purposes, and rates of change and development. Identification of the geographical source of raw materials gives information about the movement of materials in antiquity, which in turn may indicate cultural contacts. More detailed studies of distribution may give insight into transport mechanisms (e.g. marine, riverine or by land) and mechanisms of trade (e.g. single-step procurement or multiple stages of exchange). For example, obsidian analysis undertaken with the aim of determining provenance has been the source of much archaeological attention in recent decades. Perhaps predictably, some of the earlier clear-cut attributions of provenance have been clouded by increasing awareness of variation of composition within a source and by increasing awareness of the occurrence of obsidian in secondary contexts. Improved sampling and analytical strategies are being developed to address these problems. Volcanic materials also provide many archaeologically important opportunities for dating. Some of these approaches provide a date through the stratigraphic association of archaeological artefacts or human remains with datable volcanic deposits. Volcanic activity results in the formation of new materials as molten material crystallizes or vitrifies. This new material is often free of radiogenic gases and free of radiation damage but from the moment it solidifies radiogenic argon and radiation damage start to accumulate. In such circumstances, potassium-argon dating can be used to date the solidification by
22
D. GRIFFITHS
radiogenic gas accumulation, and thermoluminescence, electron spin resonance or fission-track dating can date the solidification by accumulation of radiation damage. Radiation damage dating can also be applied to material heated by volcanic emanations sufficiently to anneal away previous radiation damage. Living organisms killed by volcanic events during the last 40 ka years or so may provide material suitable for radiocarbon dating. All of these approaches date a volcanic event and may by association provide archaeologically useful dates. Although archaeology has much to gain from studying aspects of volcanic materials and events, it is not a one-way process. Archaeological and historical evidence may provide dates that may be useful in dating associated volcanic events or material. Thermoluminescence, electron spin resonance or radiocarbon dating may again be used to date archaeological samples if circumstances permit. Although less accurate because of the influence of environmental factors such as temperature, other dating techniques involve measuring the extents of chemical changes that occur after a volcanic product such as an andesite block or an obsidian flake has a fresh surface exposed to the environment by human shaping and it starts to weather or hydrate (Hunt 1991; Shackley 1998). Further work is needed but progress is being made. Archaeological or historical information may also serve as a valuable source for improving the understanding of the effects on habitation, society, environment and agriculture of geologically recent volcanic events of various types. Understanding the past effects of volcanic events may help us to mitigate the ill effects of comparable events in the future. Conclusions It is clear that in antiquity mankind made considerable use of a wide range of volcanically derived resources. Physical, practical, economic, political, symbolic and metaphysical factors may all have played a role in influencing the selection of particular materials under particular circumstances. The extent of use of volcanic materials in the past makes them potentially a very rich area for future archaeological research. It is hoped that this brief attempt to increase our appreciation of the breadth of mankind's use of volcanic resources will stimulate further interdisciplinary studies in the future. It is clearly important that the positive and attractive aspects of volcanic activity, some of which have been noted in this paper, should be
considered alongside the destructive aspects in developing a fuller understanding of the complex influence of volcanic phenomena on human societies in antiquity. A fuller understanding of the detailed nature and effects of a variety of past volcanic events should significantly improve our ability to deal effectively with present and future volcanic disasters.
References COTTERELL, B. & KAMMINGA, J. 1990. Mechanics of Pre-industrial Technology. Cambridge University Press, Cambridge. ELEFTHERIOU, T. 1990. An endangered legacy: ancient rock art of Hawaii. BA dissertation, Institute of Archaeology, University College London. GRIFFITHS, D. R., BERGMAN, C. J., CLAYTON, C. J., OHNUMA, K., ROBINS, G. V. & SEELEY, N. J. 1987. Experimental investigation of the heat treatment of flint. In: SIEVEKING, G. DE G. & NEWCOMER, M. H. (eds) The Human Uses of Flint and Chert. Cambridge University Press, Cambridge, 43-52. HIBBARD, M. J. 1995. Petrography to Petrogenesis. Prentice-Hall, Eaglewood Cliffs, NJ. HUNT, P. N. 1990. Inca volcanic stone provenance in the Cuzco province, Peru. Papers from the Institute of Archaeology, 1, 24-36. 1991. Provenance, weathering and technology of archaeological basalt and andesite. PhD dissertation, Institute of Archaeology, University College London. 1999. Olmec stone sculpture: selection criteria for basalt. This volume. & GRIFFITHS, D. R. 1992. The suitability of basalt, andesite and other volcanic stone for querns, millstones and grinding purposes. Quern Study Group Newsletter, 2, 4-6. KEMPE, D. R. C. 19830. Raw materials and miscellaneous uses of stone. In: KEMPE, D. R. C. & HARVEY, A. P. (eds) The Petrology of Archaeological Artefacts. Clarendon Press, Oxford, 53-79. 19836. The petrology of building and sculptural stones. In: KEMPE, D. R. C. & HARVEY, A. P. (eds) The Petrology of Archaeological Artefacts. Clarendon Press, Oxford, 80-153. KUNZ, G. F. 1971. The Curious Lore of Precious Stones. Dover, New York. MUSTY, J. 1999. The earliest tool makers. Current Archaeology, 164, 312. NAHON, D. B. 1991. Introduction to the Petrology of Soils and Chemical Weathering. John Wiley, New York. OLSON, G. W. 1983. An evaluation of soil properties and potentials in different volcanic deposits. In: Sheets P. D. (ed.) Archaeology and Volcanism in Central America. University of Texas Press, Austin, 52-56.
USES OF VOLCANIC PRODUCTS IN ANTIQUITY PROTZEN, J.-P. 1986. Inca stonemasonry. Scientific American, 254, 80-88. SHACKLEY, M. 1977. Rocks and Man. George, Allen & Unwin, London. SHACKLEY, M. S. 1998. Current issues and future directions in archaeological volcanic glass studies. In: SHACKLEY, M. S. (ed.) Archaeological Obsidian Studies: Method and Theory. Advances in Archaeological and Museum Science, 3, 1-14.
23
SIDDALL, R. 1999. The use of volcaniclastic material in Roman hydraulic concretes: a brief review. This volume. TORRACA, G. 1988. Porous Building Materials, 3rd edn. ICCROM (International Centre for the study of the Preservation and Restoration of Cultural Property), Rome. WILLIAMS-THORPE, O. & THORPE, R. S. 1988. The provenance of donkey mills from Roman Britain. Archaeometry, 30, 275-290.
This page intentionally left blank
The advent of archaeoseismology in the Mediterranean R. E. JONES1 & S. C. STIROS2 1
Department of Archaeology, Gregory Building, University of Glasgow, Lilly bank Gardens, Glasgow G12 8QQ, UK 2 Department of Civil Engineering, University of Pair as, Pair as, Greece Abstract: This paper presents a brief historical overview of the development of archaeoseismology from the observations of Lanciani at ancient sites in Rome, of Kritikos in Athens, Evans at Knossos and Blegen at Troy, to the emergence in the last years of the twentieth century of archaeoseismology as a distinct sub-discipline of palaeoseismology. Some current issues are explored, beginning with major seismic events such as that in AD 365 in the Eastern Mediterranean whose effects were geographically widespread but uneven in their destructive severity; generalizations are hard to come by, and each case has to be examined on its own merits. The need to examine the suitability of building methods and materials in areas of seismic risk is emphasized. Finally, the contribution of seismic events to destruction horizons in two contrasting cases in the prehistoric Aegean is considered: at Mycenaean centres in the Argolid in the 13th-12th centuries BC, and in the Peloponnese at the end of Early Bronze II.
The seismicity of the Mediterranean needs little introduction. It is a phenomenon transcending time and affecting many parts of this region, a fact of life even, awesome and terrifying in its potential destructive power. Despite the notable progress that has been made in recent decades in calculating seismic risk, this sense of the unknown when a tremor is about to strike applies perhaps almost as much to the region today as it surely did in antiquity. In this brief non-technical paper, the impact of seismic events on the archaeological record in parts of the Mediterranean, mainly the Aegean, is examined, together with how archaeologists and seismologists have interpreted the archaeological record there for evidence of seismic damage or destruction. This well-trodden path can, we believe, bear further exploration because the implications that may follow upon an interpretation of ancient destruction by seismic or other natural, rather than anthropogenic, causes can be considerable. From sometimes uncertain but by no means always naive beginnings earlier this century has emerged the sub-discipline of palaeoseismology, archaeoseismology. This has sharpened and refined the whole approach of identifying past earthquakes during the last few millennia, and of placing tectonic activity into a broader context. For seismologists, the new subdiscipline has provided the means for expanding the historical and instrumental seismological record. As discussed below, what surely hastened
its emergence in Mediterranean archaeology was the controversy generated by two major events (amongst many others): the collapse of the Mycenaean world, and the destruction in AD 365. Historical overview Our recent research suggests that one of the earliest instances in which seismic damage was explicitly invoked to explain the archaeological record at an excavation was in central Greece in a notoriously seismically active area near Atalanti (Fig. 1), scene of the great series of shocks in April 1894, the first measuring 7.0 in magnitude in Athens. Blegen (not dated, 1926), the US excavator renowned for his work at Troy in the 1930s and later at the Mycenaean palace at Pylos in SW Greece, explored in 1911 the area around Atalanti for indications of small Greek states known from classical texts, in particular Opous and Ion. Below the hill of Oion near Kyparissi he excavated city walls that were, unbeknown to him, remarkably close to the 1894 seismic fault. What he observed were some undulations and tilting in the foundations in the walls, which he suggested, were due to earthquakes, and there was also evidence of repairs (Fig. 2). This early work was to become just a foretaste of what has emerged in the last 15 years of field work: the fifth century BC stoa at Kyparissi with its foundations deformed by the activity of the
From: McGuiRE, W. G., GRIFFITHS, D. R., HANCOCK, P. L. & STEWART, I. S. (eds) The Archaeology of Geological Catastrophes. Geological Society, London, Special Publications, 171, 25-32. l-86239-062-2/00/$15.00 © The Geological Society of London 2000.
26
R. E. JONES & S. C. STIROS
Fig. 1. Map showing locations of the sites mentioned in Greece.
Locris fault, deformed tombs at a Hellenistic cemetery at Lamia, and at Kynos a late prehistoric site (Stiros & Dakoronia 1989; Dakoronia 1996; Buck & Stewart this volume). The first person, however, to make a systematic study of seismic effects 'fossilized' in archaeological remains and, furthermore, to make a typology of the indications of earthquake activity was Lanciani. As one of the foremost Italian scholars concerned with meticulous study of ancient Rome, he wrote about earthquake activity in that city, such as the parallel arrangement of the collapsed columns and their capitals from the main forum of the port of Rome at Ostia (Fig. 3) (Lanciani 1918). This observation, taken together with his recording in an earlier publication of the inundations of the River Tiber in recent historical times and the consequent
damage caused, point to the significance he attached to natural destructive phenomena in antiquity. In Athens, meanwhile, the earliest identification of archaeoseismic events is probably that by N. Kritikos (Galanopoulos 1956), who explained the sinusoidal offset of columns in the fifth century BC temple of Hephaistion (the Thesion) in terms of seismic deformation. Decades later, this type of seismic deformation was modelled through analytical calculations (e.g. Sinopoli 1991; Augusti & Sinopoli 1992). Earthquake damage at the Minoan Palace at Knossos on Crete is well known. In the course of the excavations at the Palace from the turn of the century, Evans was struck by the notable feature of a sequence of destruction levels associated structurally with deformed or more usually collapsed walls. But it was over two decades
ARCHAEOSEISMOLOGY IN THE MEDITERRANEAN
27
Fig. 2. Blegen's excavations at Kyparissi in 1911. Left, wall on the west side of trench IV (taken from the south). Deformation of the wall is clearly visible along the course of the wall, 5-10 m from the corner. Right, trench I, SE corner from the south. Atalanti islet in the background. Both illustrations are courtesy of the American School of Classical Studies, Athens (photographic archive: N67, 5 June 1911, and N76, 9 June 1911, respectively; Kyparissi, C. Blegen).
later that he cautiously associated them with earthquakes. While working one evening on his Palace of Minos in June 1926 at his home, the Villa Ariadne close to the Palace, north-central Crete was hit by a major earthquake, causing much structural and some fire damage to houses throughout the region. Yet the Villa Ariadne, built with the best seismic protection technology
Fig. 3. The Forum at Ostia (from Lanciani 1918).
of the day, survived unscathed. The breaks in continuity that Evans had observed at the Palace now crystallized in his mind; they were equated with seismic shocks of differing intensity, and, moreover they were to form one of the foundations of the chronology of Minoan Crete during the second millennium BC to 1400 BC, and indeed the rest of the Aegean (2190BC, Middle Minoan (MM) I A, to 1500BC, Late Minoan (LM) I A, and to 1400BC, LM II end of Palace) (Evans 1928). Some 50 years later it was to be the severe earthquake of c. 1700BC signalling the end of the First Palace Period in Crete that was invoked in more dramatic manner by Sakellarakis & Sapouna-Sakellarakis (1979) at the Minoan shrine or small temple of Anemo Spilia just to the south of Knossos. In their imaginative reconstruction based on the discovery of four human skeletons, vessels and other artefacts, in the west chamber, a priest, attended by a woman, apparently cut the carotid artery of a young man and collected blood. As an earthquake struck, an attendant dropped the libation jar before reaching the deity whose altar included the rock of the sacred hillside. The falling roof crushed and entombed the four people, and the place was set on fire. At Troy, Blegen confidently ascribed the end of the sixth city in the 13th century BC to seismic destruction (Blegen 1963), a claim since reexamined by Rapp (1982). The same event was to be linked by Schaeffer (1948) to a wave
28
R. E. JONES & S. C. STIROS
of destruction (dated by him to 1365BC) that enveloped much of the Levant, extending to the site that he was excavating, Ras Shamra (Ugarit). Schaeffer's publication was to be a turning point, proposing as it did that the IndoEuropean migrations of the third and second millennia BC in the East Mediterranean and western Asia were triggered in part by the destructive effects of successive earthquakes. Destruction horizons in well-separated areas were to be associated with single seismic events. These are just some of the landmarks of the first half of this century. But despite the lack of enthusiasm with which Schaeffer's theory was held in the 1950s, as the pace of excavation throughout the Mediterranean quickened after World War II, so the seismic explanation of damaged buildings and structures (especially when the site in question and its stratigraphy were complex) became a convenience, to be used sometimes uncritically; judgements were subjective at best. It was not until the 1970s that signs of change were afoot. There was, first of all, a need for modelling, from the seismological and construction engineering viewpoints, of earthquake damage of known seismic intensity to the full range of ancient structures from temples to mud brick houses. But, equally urgently, a lead had to come from seismologists to define diagnostic criteria that could be used objectively and rigorously to identify earthquakes in the archaeological record. One of the first of these was published by Karcz & Kafri (1978), who dealt with the rich data set from Israel, particularly around Jericho, work that has since been extended throughout Israel by Nur & Ron (1996). To document an ancient seismic destruction, according to Karcz & Kafri, it is necessary to show that the observed effects can unhesitatingly be assigned to an earthquake, and that any other cause, in particular anthropogenic causes, can be excluded. Key issues were now claming more attention, such as the nature of the structure and its constructive materials, the type and extent of damage, the existence of similar damage at neighbouring sites, and the location of the site. Other researchers, such as Stiros (1996, p. 152), have since put forward their own criteria. This writer emphasized structural considerations in elucidating the response of buildings to earthquakes (main shocks and after-shocks from a single episode, as well as the cumulative effect of several earthquakes), separating the effects on buildings' foundations and superstructure, and he identified some of the natural phenomena, such as storms and rock falls, that may induce dynamic failures in a building similar to those resulting from an earthquake.
Archaeoseismology In principle, then, no longer was the identification of earthquake damage entirely in the hands of the excavator or those concerned with the conservation and restoration of ancient buildings; there had to be some dialogue with seismologists and other earth scientists, drawing where possible on comparable work carried out elsewhere in the world, notably in California and Japan. International meetings during the 1980s and especially the 1990s, and their subsequent publication, have aided the process of dialogue between seismologists, earth scientists, engineers, historians and field archaeologists, recognizing the congruence of interests in archaeology and seismology. In the Mediterranean these events have ranged from a treatment of how writers in the Greek and Roman worlds depicted seismic events and their social and economic consequences on society (what, for example, were the social consequences of the event that destroyed Sparta in 464 BC? (Ducat 1984)); to broad coverages of methods and case studies (e.g. Guidoboni 1989; Frohlich & Jacobelli 1995 (see section on Pompeii, below); Stiros & Jones (1996), and catalogues of seismic events (e.g. Guidoboni et al. 1994). In turn, this has led naturally and legitimately to the creation of a recognizable sub-discipline, archaeoseismology, by a process that has occurred often enough in other branches of archaeology, whereby the parent science has adapted to meet the needs of, and the questions arising from, archaeology. We can now briefly review a few issues that have emerged from research in this field during the last decade. Many of the case studies illustrating these issues are drawn from Greece but they are likely to have wider application to the Mediterranean as a whole.
Major seismic events That there have been major earthquakes that caused widespread damage and loss of life over a large area, and, furthermore, that they had in some cases marked socio-political consequences, whereas others decidedly did not, is undoubtedly acknowledged. The point is that each case has to be examined on its own merits and, crucially, within its own historical framework. That in AD 365 is one good but complex example that seems to have had a tectonic expression, with an epicentre to the south of Cyprus, preceded by tremors up to 5 years previously and with subsequent tsunamis that hit Egypt, Turkey and Greece: this event, which was very well
ARCHAEOSEISMOLOGY IN THE MEDITERRANEAN documented by historians of the time and has been dated accurately numismatically, can be placed within the tectonic history of the region. Some of the relevant sites that have been explored, such as Kourion and Corinth, have left signs that meet the criteria very well, whereas this may not consistently be the case at other sites, for example, of the Roman period in North Africa, for which Di Vita (1990) has described the archaeological evidence for their destruction. At Kourion in southern Cyprus, which was completely devastated and abandoned after AD 365, much attention has been paid to the destroyed mid-first century BC temple of Apollo (Soren 1981; Soren & Davis 1985). In their remarkable work in reconstructing the events at Kourion, Soren & Lane (1981) argued for a seismic shock wave from the SW, which threw down blocks north and east. However, Rapp (1987, p. 375) has argued that the shock wave may have been from the opposite direction, as it cannot be assumed that motion causes the destruction (duration of vibration is nearly always important) nor that wall collapse always results predominantly from horizontal vibration. Secondary shear waves (i.e. vertical motion) are known to cause much damage, in part by effects on soil and of the other unconsolidated material underlying walls. At Corinth, the rich commercial Late Roman and post-Roman city was hit not only by the major event of AD 365 but also by another in AD 400, destroying the large temples and other formal buildings. However, the interest here rests not with the mechanism of destruction but with the political and social implications of the calamity, for there was no attempt to rebuild: these buildings were stripped of their stone and reused (Rothaus 1996). Corinth retained its wealth but the wealthy class evidently decided not to fund the rebuilding of major civic buildings. With the consequent demise of the temples, the pagan cult centres of the city hastened the end of the late antique city, this being one factor that led to its transformation into a Byzantine city. Some other, contrasting cases are worth mentioning. The AD 1296 earthquake at Pergamon had considerable impact on the city, but, according to Rheidt (1996), was no more than a transient episode with little influence upon the overall development of the city. In comparison, the event of 464 BC that destroyed Sparta triggered a revolt of slaves and a war that led to the end of the city (Ducat 1984). And Gortyn in southern Crete, which was destroyed by the AD 620 earthquake, was fully rebuilt and flourished, yet a later earthquake, in AD 670, ruined the city for ever (Di Vita 1996).
29
Other seismic events Much more common in the East Mediterranean is the apparently less serious earthquake. Here the historical record of seismicity in a given region may tell a cautionary tale, as it does in the case of the Middle East, as reported by Ambraseys et al. (1994). Notwithstanding the incomplete historical record, the picture emerges that earthquakes in parts of this region are indeed numerous, of medium size, preceded and followed by damaging shocks causing localized destruction and relatively small loss of life. The same also applies to central Greece, about which Ambraseys (1996, p. 32) said, 'Occasionally earthquakes precipitated more than that but their lasting effects would not seem to be very significant'. The historical dimension to archaeoseismology in the work of Ambraseys and others in the Eastern Mediterranean and the Middle East has had at least two beneficial effects: first, of emphasizing the need to look more carefully at building materials and methods in the archaeological record and their suitability in areas of seismic risk (although anti-seismic building methods in Minoan Crete and Hittite Anatolia have long been recognized, if imperfectly); and, second, of resisting the temptation to claim contemporaneity of earthquakes at well-separated locations. A reminder that the absence of historical accounts of an earthquake does not weaken or preclude the case that this was the mechanism of destruction, especially if the site in question lies in a supposedly aseismic zone, is well demonstrated at the site of Pella, a capital of Macedonia. Its destruction dated to c. 90 BC, followed by wide-scale abandonment, is not reported in ancient texts, nor could it be attributed by the excavators to anthropogenic factors or to natural causes, such as foundation instability. Instead, the nature of the destruction fulfilled the criteria to be classified as seismically based; the site lies within 100 km of the epicentre of the earthquake that occurred at Grevena in 1995, this area having been regarded by seismologists as essentially aseismic (Stiros 1995, 1996, p. 143).
Seismic destructions in Late Bronze HI and Early Bronze II Greece The case in support of earthquake destruction increases proportionately as the evidence at one site is mirrored independently at neighbouring sites. The collective picture should be the goal, because of the way it strengthens the validity of identifying the traditional criteria of, typically,
30
R. E. JONES & S. C. STIROS
deformed walls and skeletons close to collapsed doorways. An excellent case in point is the Argolid in the Peloponnese at the end of the Late Bronze Age. Did a series of earthquakes catalyse the demise of the Mycenaean palace culture? The answer must be no, and yet the scepticism on the part of some Aegean prehistorians that seismic effects had any place in the late Mycenaean world must surely now be regarded as misplaced. There is an earthquake horizon at Mycenae itself: damage in the cult centre on the Citadel, and skeletons in doorways covered with fallen stones from houses outside the Citadel, all dated to 1250BC (mid Late Helladic (LH) IIIB), and yet no destruction to large, contemporary buildings such as the Lion Gate and the Treasury of Atreus (French 1996; Nur 1998). On the Unterburg at Tiryns, Kilian found evidence of probably several earthquakes, somewhat earlier as well as later than at Mycenae; curving, deformed walls of early LH IIIC (1190-1150 BC) were notable (Kilian 1996). At Midea a skeleton was found recently in a room by the East Gate dating to the end of LH IIIB2, together with collapsed, distorted curved and tilting walls (Astrom & Demakopoulou 1996). Individually, the evidence at each site is far from watertight, and it is only when the three sites are treated collectively that the picture becomes more convincing. Midea and Tiryns suffered a probably severe earthquake around 1200BC, and Mycenae a less severe one around 1250BC; at Tiryns earthquakes seem to have been frequent, and all the indications are that they were followed by rapid and repeated repair work. Limited corroboration of the multiple seismic events comes from Papadopoulos' (1996) calculations of the probability of such events at the Mycenaean centres in the Argolid (and Corinthia) based on the seismic intensity v. frequency distributions from Greek data of this century; there was a high probability of destructive shock at roughly 30 year intervals. This case study greatly helps put earthquake damage into context: there were indeed earthquakes but they were neither devastating nor 'fatal'. An interesting, contrasting situation is emerging regarding the marked destruction horizon at the end of the Early Bronze (EB) II period in southern Greece around 2500-2400 BC, which neatly illustrates the dangers of invoking single events to explain destruction horizons in a given region. The traditional explanation of this phenomenon in EB II, which has been under attack for some time, is attributed to invasion by Indo-Europeans. But it is now becoming clear that events at that crucial period of time, the end of Early Helladic (EH) II, were not uniform throughout southern and central Greece.
We may contrast the picture at the type site of the Early Helladic period, Lerna, where there was major change (the House of the Tiles burnt down) with, on the one hand, the lack of any destruction in the comparable horizon at Kolonna on Aegina, and, on the other hand, the strong evidence for earthquake destruction in the EH II levels not only at Ayios Dimitrios and Voidokilia situated on the seismically active western flank of the Peloponnese (Zachos 1996) but also probably at Thebes (Sampson 1996, p. 115).
Pompeii A feature of recent work has been the emergence of highly detailed site-specific work. At Pompeii, for example, Frohlich & Jacobelli (1995) have documented the large amount of evidence that exists for the structural damage, repairs and changes of use that occurred during the lifetime of this town. Central to their enquiry work was to ascertain whether a second earthquake occurred after that of AD 62 and before the volcanic eruption of AD 79; sadly, this was not fully resolved, Conclusions These issues and case studies have illustrated some of the progress that seismology has made in its interaction with Mediterranean archaeology concerning the last 4000-5000 years. However, it is gratifying to note that the more distant past has not been neglected: broadening the scope of archaeoseismology to consider the role of tectonic processes on a wide time scale in relation to human adaptation to the environment, work stimulated by the British excavations of the Palaeolithic rock shelter at Klithi in Epirus is surely relevant and a likely corrective to most preconceptions. In brief, Bailey et al. (1993) have drawn attention to the active tectonics of northwest Greece giving rise to landscapes that could be beneficial to human survival, and the history of these processes shows that 'Palaeolithic sites were located to take advantage of tectonically created features at both local and regional scales'. Much remains to be done, but in looking to the future, we identify archaeoseismology's need to take a more pro-active role in formally assessing the seismic risk to ancient monuments, a process already under way in some Mediterranean countries, as the subjects' greatest challenge in the coming years.
ARCHAEOSEISMOLOGY IN THE MEDITERRANEAN
References AMBRASEYS, N. N. 1996. Material for the investigation of the seismicity of Central Greece. In: STIROS, S. & JONES, R. E. (eds) Archaeoseismology. Fitch Laboratory Occasional Paper, 7, 23-36. , MELVILLE, C. & ADAMS, R. 1994. The Seismicity of Egypt, Arabia and the Red Sea. CUP, Cambridge. ASTROM, P. & DEMAKOPOULOU, K. 1996. Signs of an earthquake at Midea? In: STIROS, S. & JONES, R. E. (eds) Archaeoseismology. Fitch Laboratory Occasional Paper, 7, 37-40. AUGUSTI, G. & SINOPOLI, A. 1992. Modelling the dynamics of large block structures. Meccanica, 27, 195-211. BAILEY, G., KING, G. & STURDY, D. 1993. Active tectonics and land use strategies: a Palaeolithic example from Northwest Greece. Antiquity, 67, 292-313. BLEGEN, C. not dated. Excavations to the S.W. of Kyparissi, 1911. American School of Classical Studies (Athens), School Papers 1899-1913 (unpublished). 1926. American Journal of Archaeology, 30, 403. 1963. Troy and the Trojans. Thame & Hudson, London. DAKORONIA, PH. 1996. Earthquakes of the Late Helladic III period (12th century BC) at Kynos (Livanates, Central Greece). In: STIROS, S. & JONES, R. E. (eds) Archaeoseismology. Fitch Laboratory Occasional Paper, 7, 41-44. Di VITA, A. 1990. Sismi, urbanistica e cronologia assolute - terremoti e urbanistica nelle citta di Tripolitania fra il 1 secolo, A.C. ed il IV D.C. In: L'Afrique dans I 'Occident Romain. Collection de 1'Ecole Francaise de Rome, 134, 425-494. 1996. Earthquakes and civil life at Gortyn (Crete) in the period between Justinian and Constant II (6th-7th century AD). In: STIROS, S. & JONES, R. E. (eds) Archaeoseismology. Fitch Laboratory Occasional Paper, 7, 45-50. DUCAT, J. 1984. Le treblement de terre de 1864 et 1'histoire de Sparte. In: HELLY, B. & POLLING, A. (eds) Tremblements de terre: histoire et archeologie. Proceedings of International Meeting on Archaeology and History, Antibes. APDCA, Valbonne, 73-86. EVANS, A. 1928. The Palace of Minos II. Macmillan, London. FRENCH, A. B. 1996. Evidence for an earthquake at Mycenae. In: STIROS, S. & JONES, R. E. (eds) Archaeoseismology. Fitch Laboratory Occasional Paper, 7, 51-54. FROLICH, T. & JACOBELLI, L. (eds) 1995. Archaologische und seismologie: la regione Vesuviana dal 62 al 79 n.C. Problemi archeologici e sismologici. Biering & Brinkmann, Munich. GALANOPOULOS, A. 1956. The seismic risk at Athens. Praktika Akadimias Athinon, 31, 461-472 [in Greek]. GUIDOBONI, E. (ed.) 1989. / Terremoti Prima del Mille in Italia e nell'Area Mediterranea: Storia, Arche-
31
ologia, Sismologia. Storia Geofisica Ambiente, Bologna. , COMASTRI, A. & TRAINA, G. 1994. Catalogue of Ancient Earthquakes in the Mediterranean Area up to the 10th Century. Institute Nazionale di Geofisica, Rome KARCZ, I. & KAFRI, U. 1978. Evaluation of supposed archaeoseismic damage in Israel. Journal of Archaeological Science, 5, 237—253. KILIAN, K. 1996. Earthquakes and archaeological context at 13th century BC Tiryns. In: STIROS, S. & JONES, R. E. (eds) Archaeoseismology. Fitch Laboratory Occasional Paper, 7, 63-68. LANCIANI, R. 1918. Segni di terremoti negli edifizi di Roma antica. Bulletino della Archeologica Communale Roma, 1-30. NUR, A. & RON, H. 1996. And the walls came tumbling down: earthquake history in the Holyland. In: STIROS, S. & JONES, R. E. (eds) Archaeoseismology. Fitch Laboratory Occasional Paper, 7, 75-85. PAPADOPOULOS, G. A. 1996. An earthquake engineering approach to the collapse of the Mycenaean Palace civilisation of the Greek mainland. In: STIROS, S. & JONES, R. E. (eds) Archaeoseismology. Fitch Laboratory Occasional Paper, 7, 205-209. RAPP, G. 1982. Earthquakes in the Troad. In: RAPP, G. & GIFFORD, J. A. (eds) Troy: the Archaeological Geology. Princeton, NJ, 43-58. 1987. Assessing archaeological evidence for seismic catastrophes. Geoarchaeology, 1, 365-379. RHEIDT, K. 1996. The 1296 earthquake and its consequences for Pergamon and Chliara. In: STIROS, S. & JONES, R. E. (eds) Archaeoseismology. Fitch Laboratory Occasional Paper, 7, 93-103. ROTHAUS, R. M. 1996. Earthquakes and temples in Late Antique Corinth. In: STIROS, S. & JONES, R. E. (eds) Archaeoseismology. Fitch Laboratory Occasional Paper, 7, 104-112. SAKELLARAKIS, Y. & SAPOUNA-SAKELLARAKIS, E. 1981. Drama of death in a Minoan temple. National Geographic, 159, 204-222. SAMPSON, A. 1996. Earthquakes at Mycenaean and pre-Mycenaean Thebes. In: STIROS, S. & JONES, R. E. (eds) Archaeoseismology. Fitch Laboratory Occasional Paper, 7, 113-117. SCHAEFFER, C. F. A. 1948. Stratigraphie comparee et chronologic de VAsie Occidentale. OUP, Oxford. SINOPOLI, A. 1991. Dynamic analysis of a stone column excited by a sine wave motion. Applied Mechanics Review, 44, S246-S255. SOREN, D. 1981. Earthquake: the last days of Kourion. In: BIERS, J. C. & Soren, D. (eds) Studies in Cypriot Archaeology. Institute of Archaeology UCLA Monograph, XVIII, 117-123. & DAVIS, T. 1985. Seismic archaeology at Kourion: the 1984 campaign. Report of the Department of Antiquities of Cyprus, 193-301. & LANE, E. 1981. New Ideas About the Destruction of Paphos. Report of the Department of Antiquities of Cyprus, 178-183. STIROS, S. C. 1995. Unexpected shock rocks an 'aseismic' area. Eos Transactions, American Geophysical Union, 76, 513-514.
32
R. E. JONES & S. C. STIROS -1996. Identification of earthquakes from archaeological data: methodology, criteria and limitations. In: STIROS, S. & JONES, R. E. (eds) Archaeoseismology. Fitch Laboratory Occasional Paper, 7, 129-152. - & DAKORONIA, PH. 1989. Ruolo storico e identificazione di antichi terremoti nei siti della Grecia. In: GUIDOBONI, E. (ed.) / Terremoti Prima del Mille in Italia e neil"Area Mediterra-
nea: Storia, Archeologia, Sismologia. Bologna, 422-439. & JONES, R. E. (eds) 1996. Archaeoseismology. Fitch Laboratory Occasional Paper, 7. ZACHOS, K. 1996. Tracing a destructive earthquake in the south-western Peloponnese (Greece) during the Early Bronze Age. In: STIROS, S. & JONES, R. E. (eds) Archaeoseismology. Fitch Laboratory Occasional Paper, 7, 169-185.
A critical reappraisal of the classical texts and archaeological evidence for earthquakes in the Atalanti region, central mainland Greece VICTORIA BUCK1'2 & IAIN STEWART2 1
British School at Athens, Odos Souedias 52, GR106 76, Athens, Greece (e-mail:
[email protected]). Neotectonics Research Centre, Department of Geography and Earth Sciences, Brunei University Uxbridge, UBS 3PH, UK Abstract: Despite numerous damaging earthquakes in central Greece only the Atalanti Fault is considered to have ruptured successively, most recently in 1894 and in a historical event in 426 BC. Although the pre-Christian earthquake is now firmly entrenched in the tectonic literature, classical literary accounts are inconsistent and do not unequivocally tie the event to a particular time and place. Archaeological evidence from sites close to the Atalanti Fault similarly remains ambiguous, and fails to convincingly corroborate the rupture of the Atalanti Fault in 426 BC. In this paper, the main thrust of the argument is not to try to define the seismogenic source of the 426 BC event, but to illustrate the level of uncertainty that accompanies literary and archaeological information, and to highlight the need for caution when using interdisciplinary methods or datasets in earthquake seismology.
Archaeological records of seismic disturbance allow historical earthquake catalogues to be extended into pre-historical times or serve as an independent dataset against which to assess documentary reports of past earthquakes. Where archaeological records are long and well constrained, they are potentially a valuable tool in reconstructing the recent earthquake history of faults, particularly in terms of shedding light on the likely 'recurrence intervals' of faults, that is, the time elapsed between successive earthquakes on an individual fault. Away from plate boundaries such intervals range from many centuries to a few millennia, and therefore fall into the realm of archaeological study. Despite numerous damaging earthquakes in Greece, both in modern times and those documented from antiquity, no individual fault is unequivocally known to have ruptured twice. However, perhaps the best candidate for a fault widely considered to have ruptured successively is the Atalanti Fault, known to have ruptured in AD 1894, and also thought to have hosted a widely reported damaging event in 426 BC. This paper will examine the ancient literary accounts of the 426 BC event and the archaeological records of two excavated sites in the vicinity of the Atalanti Fault to assess independently contemporary evidence for this histori-
cal event. The paper will show that although accounts of the 426 BC event locate the earthquake to the region of the Gulf of Evia, they fail to tie the event convincingly to the same fault segment that ruptured in AD 1894, rather than a number of other potential seismic sources in this region. In addition, the paper will illustrate the problems involved in correlating seismic events with specific fault segments in this region. Background The Bay of Atalanti lies within the extensional rift system of the Gulf of Evia, central mainland Greece. The rift system is dominated by a segmented belt of large, 35-40 km long, NW-SE striking normal faults (Roberts & Jackson 1991; Ganas et al. 1996) predominantly dipping north (Fig. 1). The rift is considered active with numerous small (M < 5.5) earthquakes instrumentally recorded since the 1960s, and at least two large events (M > 6) accounted for in the ancient and recent historical literature: the AD 1894 and 426 BC earthquakes. The 1894 Atalanti earthquake, which consisted of two main shocks of M6.4 and M6.9 (Ambraseys & Jackson 1990) and hundreds of aftershocks, caused extensive damage from Agios
From: McGuiRE, W. G., GRIFFITHS, D. R., HANCOCK, P. L. & STEWART, I. S. (eds) The Archaeology of Geological Catastrophes. Geological Society, London, Special Publications, 171, 33-44. 1-86239-062-2/00/S15.00 © The Geological Society of London 2000.
Fig. 1. Map of Lokris region in relation to Greece (inset) showing the location of modern towns and villages (indicated in bold type), archaeological sites mentioned in the text (indicated by stars and italic type), and the neotectonic faults (after Ganas et al. 1996) with downthrow indicated by barbs.
Fig. 2. Panoramic view of the Atalanti plain looking north from the footwall above Kyparissi. Donkey Island, which was formed as a result of tectonic subsidence accompanying the 1894 earthquake on the Atalanti Fault, is in the middle of the picture and indicated by an arrow. Atalanti Island, on the left, is widely considered to have formed as a result of the 426 BC earthquake.
36
V. BUCK & I. STEWART
Konstantinos to Larymna with surface rupture along the Atalanti Fault (Anon. 1894a, b; Davidson 1894#, b). This was accompanied by numerous secondary surficial effects such as ground liquefaction, fissuring, marine inundation (between Arkitsa and Almyra; see Fig. 1); rockfalls (in the Kyparissi area; see Fig. 1), landslides, and a seismic sea wave (Davidson 1894&; Skouphos 1894; Mitsopoulos 1895; Ambraseys & Jackson 1990). The general tectonic subsidence of the Atalanti plain also resulted in the formation of Gaidaronisi (Donkey Island) through the immersion of the low-lying marsh area (Fig. 2). The Atalanti/Lokris Fault segment is considered to be the seismogenic source of the 1894 event; however, numerous other potential seismogenic faults lie in the vicinity (Fig. 1). In addition to the 1894 event, the Atalanti region is noted in a number of publications as having been affected by several historical earthquakes, including those of 426 BC, AD 105, AD 551 and AD 1544 (Stiros & Rondogianni 1985; Guidoboni et al 1989; Papazachos & Papazachou 1989). This paper, however, focuses only on the preChristian event of 426 BC. The contemporary account by Thucydides and the complementary descriptions by Diodorus Siculus and Strabo (both of whom wrote several centuries after the event) of the effects of the earthquake on named towns, forts and harbours suggest that the seismogenic source lay within the Gulf of Evia rift system. Although the size and location of the 426 BC event are imprecisely known, the apparent similarity of the observed surficial effects (e.g. the formation of an island and sea inundations) with those of the 1894 earthquake has led to the suggestion that both earthquakes, probably of comparable magnitude, occurred along the Atalanti Fault (Stiros & Pirazzoli 1995; Ganas et al. 1998). However, the fragmentary nature of the information, and the inconsistencies between the information provided by the three classical authors (Appendix), suggest that an independent data source is needed before the proposed 426 BC event can be unequivocally assigned to the Atalanti Fault. Classical accounts of the 426 BC event The 426 BC earthquake is dated from its inclusion in the contemporary account of the Peloponnesian War by Thucydides (Thucyd., iii. 89.1-4). It is unfortunate that Thucydides, who wrote the only contemporary account of the earthquake, provides the least geographical detail regarding the extent of destruction, mentioning damage in three separate geographical areas, the Isthmus near Corinth leading into Attica, Orobiae on the
island of Euboea, and Atalante in Lokris (see Appendix) Although Thucydides gives few details of the overall destruction, his account focuses on the effects of marine inundations on coastal sites, which reflects the subject matter of his writing: a chronological account of a historical event, i.e. the Peloponnesian War. A significant difference between this and the later derived sources of Diodorus Siculus (Diod. Sic., xii. lix) and Strabo (Strabo, I. iii. 20) is his specific reference to the island of Atalante as a pre-existing landform. Similarly, the historian Diodorus Siculus, writing in the first century, provides little geographical detail, perhaps reflecting the fact that his writing is derived from other earlier sources. The earthquake effects he briefly outlines are assumed to be those caused by the 426 BC event on the basis of similarities between this and the text of Thucydides, i.e. the invasion of Attica by the Peloponnesians and the location of their camp at the Isthmus leading to Attica. However, it is notable that in this account the formation of Atalante Island has now become the focus of the account of this earthquake, in direct contrast to Thucydides, who states that the island was already in existence before the 426 BC event. In comparison with the first two chronologically based texts, the Geographies, written by Strabo in the first century BC, is a geographically based compilation of knowledge. Quoting at least two sources, Strabo gives an undated account of a large destructive earthquake that affected the Gulf of Evia region. The first section of the relevant passage (Strabo, I. iii. 20) (Appendix) details precisely the localities and the extent of damage, including death tolls, citing the origin of the information as a written account (now lost) by Demetrius of Callatis. However, the final section (Strabo, I. iii. 20) (Appendix), which focuses on the genesis of Atalante Island, switches from a precise to a vague, almost anecdotal, account of the effects farther east in the Bay of Atalanti, and should therefore be viewed with caution as a description of the effects of the same earthquake discussed earlier in the chapter. Again, it is the association with Atalanti Island that leads subsequent workers to assume that this is an account of the 426 BC event. Geological interpretations of the literary sources Although the texts are fragmentary, some important inconsistencies emerge from a comparison of the previous accounts. First, the dating of the
CLASSICAL AND ARCHAEOLOGICAL EVIDENCE FOR EARTHQUAKES event is derived from the chronological accounts of the Peloponnesian Wars by Thucydides and Diodorus Siculus. Strabo's account is assumed to be the same event because of the inclusion of Atalanti in the description of areas experiencing damage. Second, whereas Thucydides and Diodorus Siculus discuss the 426 BC earthquake in relation to the Atalanti area, Strabo focuses most attention on the Malakios Gulf (see Fig. 1), with only minor reference to Atalanti. The question arises, therefore, whether these accounts are selective reports of a regionally extensive earthquake or testify to separate events that have in some texts been combined. Third, whereas Atalanti Island is already in existence before the 426 BC earthquake in the contemporary account of Thucydides, the two later reports of Diodorus Siculus and Strabo ascribe the formation of the island to the earthquake itself. This is of importance because the formation of the island in the earthquake strongly suggests subsidence of the Atalanti plain, probably as a result of rupture of the Atalanti Fault, as was the case with the detachment of Donkey Island in the 1894 event (Fig. 3). By contrast, Thucydides' reports of widespread marine inundation around the Atalanti coast may equally be attributed to tsunami attack by rupture on one of a number of seismogenic faults in the area (Fig. 1). Most geologists employing these ancient sources have assumed that all three texts refer to a single major earthquake in 426 BC, but their subsequent seismotectonic interpretations vary markedly. Some, using the combined destruction and geographical information, have constructed isoseismal maps with the long axis of the derived damage ellipse trending east-west, with the maximum intensity located in the Malakios Gulf (Sieberg 1932; Bousquet & Pechoux 1977) (Fig. 4). Other workers (Stiros 19880, c\ Stiros & Dakoronia 1989; Dakoronia 1993), on the basis of the occurrence of possible archaeoseismic damage in the Atalanti area, and the apparent similarity of geomorphological effects observed here during the 1894 event to those described in the ancient texts (particularly the coastal inundation and formation of islands), ascribe the event to a 70km long NW-trending fault linking Atalanti with Lamia. It should be noted that this latter view implies that the Atalanti Fault is up to 70km long (Fig. 5), something which is disputed by the recent work on fault segmentation in the Atalanti region (Roberts & Jackson 1991; Ganas et aL 1996), which suggests a maximum segment length of 3 5-40 km (in the Atalanti area) as shown in Fig. 1. In a detailed review of the ancient texts, Mouyiaris (1988) proposed that the narrative of
37
Strabo is in fact an agglomeration of several events that cannot be securely dated, and he accepted only the contemporary account by Thucydides as that of the 426 BC event. However, on the basis of the texts alone, he was unable to define securely the seismogenic origin of the earthquake, preferring to conclude that the Atalanti plain area had clearly been the subject of previous earthquake events. Clearly, the seismogenic source of the 426 EC event cannot be unequivocally identified from the ancient literature alone, because of inconsistencies between the texts, which may simply be the result of differing lexical or writing styles. However, validation of literary evidence from an independent data source, such as archaeoseismological evidence, may be able to define which scenario is the more probable.
Archaeological data A number of coastal and inland archaeological sites are located around Atalanti Bay (Fig. 1). Their proximity to the active Atalanti Fault makes it reasonable to expect that if an earthquake had occurred then diagnostic indicators (sensu Nikonov 1988; Stiros 1996) would be present in the archaeological stratigraphies of the sites. In the following section, the archaeological record of two sites (Alai and Kyparissi) is considered in relation to the validation of the 426 BC event.
Alai The acropolis site of Alai is located on the southeastern shore of Atalanti Bay in the vicinity of the modern village of Theologos, c. 6 km NE of the Atalanti Fault (Fig. 1). The site contains evidence of occupation from the Neolithic to the Byzantine period (Coleman 19926), with an occupational hiatus of around 4000 years between the Neolithic period and the foundation of Alai in the Archaic period (Wren 1996). The site was excavated annually during 1911-1914 by the American School of Classical Studies at Athens under the directorship of Hetty Goldman and Alice Walker, with four additional campaigns during 1921-1935 (Coleman 19926; Goldman 1940). 'Seismic disturbances' in the area are mentioned in a footnote to the initial excavation report (Walker & Goldman 1915, p. 419) particularly with reference to an earthquake in 1893 (possibly a misquote for 1894) producing submergence in the Almyra area of
38
V. BUCK & I. STEWART
Fig. 3. Comparative maps showing the formation of Donkey Island, (a) Pre-1894 coastline, showing Donkey Island (indicated by a bold arrow) attached to the relatively smooth shoreline of Atalanti Bay (British Admiralty 1980). (b) Post-1894 coastline, showing Donkey Island as a post-1894 earthquake landform (British Admiralty 1984).
the Atalanti shoreline. However, there is no description of any earthquake-induced damage at the site, probably because of the general nature of the report. Goldman (1940, p. 454) discussed in slightly more detail the evidence for destruction in 426 BC: The earliest building, which never had a stone entablature or cornice, was razed and buried some time after 510BC, together with its altar, under a thick covering of earth.' A tenuous link is made between the observed architectural damage and earthquakes known from the classical literature: 'In the great earthquakes of 426 and 425 BC, the latter accompanied by cata-
strophic inundations, which were recorded by ancient authors as of such exceptional violence that they tore the island of Atalante across the bay from Halae in two, the superstructure of the building collapsed, and the temple area was strewn with blocks' (Goldman 1940, p. 454). The link, which appears to provide an explanation of the architectural remains, is further supported by dating the second building through correlation of architectural styles: The capital [head of pillar or column] No. 3 and geison [cornice or ornamental moulding just below the ceiling] are very close in style to those of the Argive Heraion, and if we accept for the second temple of Hera
Fig. 4. An isoseismal map derived from the combined use of the classical texts by Thucydides, Strabo and Diodorus Siculus. It should be noted that the maximum damage isoseismal (11) is located in the Maliakos Gulf, particularly around the site of Phalara (redrawn from Sieberg (1932)).
CLASSICAL AND ARCHAEOLOGICAL EVIDENCE FOR EARTHQUAKES
39
Fig. 5. Map showing a proposed 70 km fault system linking the Atalanti Fault with Lamia (redrawn from Stiros & Dakoronia (1989)).
the date proposed by Professor Dinsmoor 423 BC - the Halae building, which is slightly later in style, was rebuilt not long after the disastrous earthquakes' (Goldman 1940, p. 454). More recently, the remains at Alai have been the focus of the Cornell Halae and East Lokris Project (CHELP), a US interdisciplinary archaeological and environmental project under the directorship of John Coleman. The ArchaicClassical remains and stratigraphy have been carefully reconsidered and a date of 480 BC has been assigned for the collapse of the first temple (Wren 1996), thus throwing into question the correlation between the earthquake described in the ancient literature and the observed archaeological field evidence. On the basis of this new date, Wren puts forward the hypothesis that the destruction level identified in Alai could equally be the result of the Persian advance towards Athens under the command of Xerxes, which is known to have taken place at this time. Further excavation in 1996 of the Archaic-Classical levels at the site, however, has revealed a toppled wall, indicative of a natural rather than a military destruction, and cultural finds from deposits associated with it tend to confirm a 480 date, although to date there is no unambiguous evi-
dence for the cause of the destruction (Coleman, pers. comm.). Shoreline constructions at Alai, identified as harbour installations such as 'ship sheds' (Fossey 1990), also have been reinterpreted as fortification walls in the light of an extensive site survey. This, coupled with the apparent lack of hinterland associated with the remains, suggests a relative rise in sea level (Coleman I992a). Although, the mechanism (e.g. eustatic rise or tectonic land subsidence) and the magnitude of change cannot be precisely determined from the data currently available, the site is known to have subsided in 1894 and so some tectonic contribution is likely (Ganas & Buck 1998).
Kypans si (Opus) At the base of the Kokkinovrachos hill in the SE end of the Atalanti plain, 2 km southeast of the modern village of Kyparissi, lies an archaeological settlement site that is now considered as the 'most likely candidate for the city of Opus' (Dakoronia 1993, p. 117). This site consists of two localities, both lying 200m from the trace of the Atalanti Fault. A fortified acropolis is well
40
V. BUCK & I. STEWART
preserved and visible at the top of Kokkinovrachos hill, and an associated settlement site is located somewhere on the north and northwestern parts of the same hill (although there are now no remains visible above ground level, there is a high density of ceramic material visible on the cultivated surface of the olive groves). The acropolis site has been surveyed and described several times since the 1880s (see Fossey (1990) for references), with no specific mention of earthquakes or any resulting seismic damage. The original excavations of the settlement site were carried out in 1911 by the American School under the directorship of Carl Blegen. Their report, too, gives no account of earthquakes or earthquake damage, particularly noteworthy given that the site was excavated less than 20 years after the 1894 event (Blegen 1926). More recently, however, two seasons of excavation, in 1978 and 1979, were carried out by the Greek Archaeological Service under the directorship of Dr Ph. Dakoronia, which exposed an Archaic (sixth-century BC) stoa [a porch or portico not attached to a larger building] considered to have been 'destroyed by an earthquake; the site is frequently afflicted by earthquakes, lying as it does near the well known Atalante seismic fault' (Dakoronia 1993, p. 117). The configuration of the damage, i.e. 'the conspicuous upward bending' (Stiros 1988c, p. 1634) of the stoa wall, has led to the site being considered as direct evidence of vertical offset by movement of the Atalanti Fault (Stiros 1988a-c, 1996; Stiros & Dakoronia 1989; Stiros & Pirazzoli 1995). 'This seems to be a plausible explanation, since the stoa lies very close to, or just on the surface trace of the Locris 1894 earthquake' (Stiros 1988c, p. 1634). A succession of studies has dated the destruction of the stoa to between 540 and 425 BC (Dakoronia 1988; Stiros 1988a-c) on the basis of typological comparison of terracottas, specifically acroteria. It would appear that the stoa was in use until c. 480 BC, when it sustained a roof repair. However, the lack of excavation material dating to the beginning of the fifth century BC, and the redating of the Alai destruction, throws into question the later possible dates for the destruction of the stoa. Reassessment of the Kyparissi sherds in the light of new excavation data from Alai may result in a more concisely constrained destruction date in the future. It should be noted that the dates of destruction quoted by researchers relate to the date of abandonment, i.e. the latest possible occupation based upon the dating of cultural material such as pottery including terracottas, or absolute dating. This may not necessarily equate to the
date of deformation of the stoa, however, because, lying close to the rupture trace of the 1894 break as it does, it is conceivable that the stoa has been affected by multiple events. Other local sites There are a number of less published sites of archaeological remains, including the Islet of Mitrou, and the sites of Palaeomagaza near Skala, and Kynos near Livanates, all located along the shore of Atalanti Bay (Fig. 1). However, these sites will not be considered further in this paper, because of the paucity of published data on either the sites as a whole or the Archaic-Classical period within the site. Discussion To summarize the archaeological data: the initial date of 426 BC assigned to the destruction of the first temple area in Alai has now been revised to 480 BC based on a reassessment of the destruction within the Archaic-Classical levels. This was originally thought to be associated with the Persian invasion by Xerxes in 480 BC, but the discovery of new evidence in the form of a toppled wall has led to the suggestion that the destruction is probably the result of an earthquake occurring around the same time. The re-dating of the Archaic-Classical finds and of the destruction at Alai throws into question the dating of the proposed seismic destruction at Kyparissi, which is based upon typological comparison of terracottas with those from Alai. Therefore, although the remains at Kyparissi have been identified by the excavators as having been seismically damaged, the dating of the destructive event is not clear, primarily because of the redating of the comparative material at Alai and the possibility of deformation during the 1894 event. Clearly, destruction event horizons that can be correlated between neighbouring sites lend confidence to the interpretation of regionally important earthquake (or other) events rather than local, possibly site-specific and not necessarily seismic, events. Conversely, a lack of correlation between positively identified inter-site destructions should not exclude the possibility of an earthquake, because there are a number of factors that influence the extent to which an earthquake is likely to be preserved in the archaeological record. The archaeological reasons for non-correlation are discussed below in the context of the Atalanti sites.
CLASSICAL AND ARCHAEOLOGICAL EVIDENCE FOR EARTHQUAKES
41
Site occupation
Dating
Where sites are established, occupied and abandoned at different chronological periods their capacity to retain evidence of damaging seismic events will vary. Alai shows occupation levels dating from the Neolithic to the Byzantine period, with an occupation hiatus of 4000 years at the end of the Neolithic period (c. 54005300BC). Archaeological finds from the Kyparissi area begin in the Late Helladic IIIB-C period (1300-1100 BC), but the actual architectural relics date only from the Archaic and Classical periods. Both sites show evidence of occupation at the time of the earthquake mentioned by Thucydides, which means that a lack of evidence of occupation should not affect potential correlation.
Deriving a date for destruction horizons is extremely difficult because they often contain only items of material culture, i.e. building materials, ceramics including tiles, etc. and not artefacts with actual dates (e.g. inscriptions). The majority of these items are not suitable for absolute dating; however, they can be used for relative dating; however, relative dates often have broad ranges, which can result in apparent correlation. Relative dating of occupation levels in the sites studies in the Atalanti region is based on typological comparisons of terracotta and other ceramics and building styles (Dakoronia 1996). Although the Alai site now has several good 14C dates, these date from the earliest Neolithic levels and pre-date the possible earthquake by nearly 5000 years (Coleman, pers. comm.). Clearly, the dating issue affects the correlation in this case study, as, although the dating of the Archaic-Classical levels in Alai has recently been revised, there are no 14C dates for these periods, and the dates for the damage of the Kyparissi stoa still rely on comparative evidence.
Identification Seismic damage may be overlooked by excavators unfamiliar with archaeoseismic criteria, and consequently not recorded in the site reports. Kyparissi was excavated by the regional Ephoreia (regional archaeologist), who is familiar with previous research involving identified earthquake-induced damage at archaeological sites. Similarly, the director at Alai, John Coleman, aware that the area is seismically active, has liaised with the regional Ephoreia on this subject throughout the excavation. Therefore, the lack of identification of seismic damage to archaeological remains is considered unlikely to affect recent studies. However, earlier excavations carried out at these sites may not record possible earthquake damage because of the relatively recent emergence of archaeoseismology. Occurrence of damage The type and quantity of seismic damage are largely a function of the construction methods employed, the quality of materials used, and their response to any physical tilting and shaking of the ground that accompany sizeable earthquakes (Stiros 1996). To date there are no published accounts of the construction methods and materials used in the varying phases of occupation at either of the sites. However, preliminary field observations indicate that there are similarities in the construction methods and materials used between the sites at corresponding time periods, and therefore the architectural remains are likely to have had a similar receptivity to seismic events hosted by adjacent faults.
Preservation The presence of seismic damage will largely depend on the vulnerability of a site and its remains to seismic damage. However, it may be possible to suggest whether the site is good or bad for preservation, on the basis of the environmental context of the remains. Conclusion In summary, neither the classical literary accounts nor the archaeological data convincingly demonstrate that the Atalanti Fault was responsible for the 426 BC earthquake. In addition to being fragmentary, classical accounts have important inconsistencies, and in particular, the well-cited account by Strabo seems to be highly misleading in terms of the regional extent of the event and its exact date. Although Thucydides and (less convincingly) Diodorus Siculus suggest that a damaging earthquake did strike the Atalanti area in 426 BC, archaeological evidence from occupation sites in the vicinity fails to corroborate this event. In particular, a revision of the chronology at Alai has led to the redating of destruction horizons, originally attributed to 426 BC, to C.480BC. In addition, the timing of the abandonment and subsequent deformation of a stoa at nearby
42
V. BUCK & I. STEWART
Kyparissi (Opus), widely cited as evidence of rupture along the Atalanti Fault, is poorly constrained and, in the light of the revised chronology at Alai, needs reappraisal. In a broader context, it is important to appreciate that the identification of an earthquake in an archaeological, historical or classical context is used primarily within a narrative as cultural information and rarely extends into understanding the earthquake as a geological phenomenon. For example, Sir Arthur Evans (1928) linked the decline of Minoan civilization to repeated earthquakes on the island of Crete, and Thucydides tells us that 'a great many earthquakes' prevented the invasion of Attica by the Peloponnesians in 426 BC. In effect, for archaeologists, identification is generally part of the conclusion of the investigation. In contrast, earthquake geologists start with the identification made by the archaeologists and historians and seek to define earthquake characteristics, such as seismogenic source and magnitude, with the ultimate goal of including them in a regional seismic catalogue and in turn, in seismic-hazard assessment scenarios. In this paper, the main thrust of the argument is not to seek the seismogenic source of the 426 BC earthquake, whether it be the Atalanti Fault or a neighbouring structure. Instead, this examination of the classical and archaeological datasets has sought to illustrate how much uncertainty accompanies both lines of evidence. The message from this study is that earthquake geologists should in general be circumspect when integrating historical and archaeological evidence. In particular, it is crucial that data reliability and coherence are critically appraised when making entries into and using seismic catalogues. In summary, if archaeoseismic evidence is to contribute significantly to our understanding of earthquake activity, practitioners need to confront the challenge of Charles Richter that 'Ancient accounts of earthquakes do not help us much; they are incomplete, and accuracy is usually sacrificed to make the most of a good story' (Richter 1958; cited by VitaFinzi 1986, p. 8).
Ph. Dakoronia. Geological permits were granted by the Institute of Geology and Mineral Exploration Athens. R. Reinders and an anonymous reviewer are thanked for constructive comments on an earlier version of the paper.
Appendix: The classical texts from the Loeb translations in English
Thucydides, Hi. 89.1-5 Tn the following summer the Peloponnesians and their allies, led by Agis son of Archidamus, king of the Lacedaemonians, advanced as far as the Isthmus with the intention of invading Attica; but a great many earthquakes occurred, causing them to turn back again, and no invasion took place. At about the same time, while the earthquakes prevailed, the sea at Orobiae in Euboea receded from what was then the shoreline, and then coming on in a great wave overran a portion of the city. One part of the flood subsided, but another engulfed the shore, so that what was land before is now sea; and it destroyed of the people as many as could not run up to the high ground in time. In the neighbourhood also of the island of Atalante, which lies off the coast of Opuntian Locris, there was a similar inundation, which carried away a part of the Athenian fort there, and wrecked one of two ships that had been drawn up on the shore. At Peparethos [Skopelos Island] likewise there was a recession of the waters, but no inundation; and there was an earthquake, which threw down a part of the wall as well as the prytaneum and a few other houses. And the cause of such a phenomenon, in my own opinion, was this: at that point where the shock of the earthquake was greatest the sea was driven back, then suddenly returning with increased violence, made the inundation; but without an earthquake is seems to me such a thing would not have happened.'
Diodorus Siculus, xii. 59.1-2
'While the Athenians were busied with these matters, the Lacedaemonians, taking with them This research was supported by a Brunei University the Peloponnesians, pitched camp at the Isthmus postgraduate scholarship. The authors would like to with the intention of invading Attica again; but acknowledge J. Coleman, director of CHELP, for 1997 when great earthquakes took place, they were field work support with unlimited access to project filled with superstitious fear and returned to records, A. Ganas for valuable field assistance and their native lands. So severe in fact were the many thought-provoking discussions, and G. Shipley for advice and expertise on the translations of the shocks in many parts of Greece that the sea ancient texts and the original draft manuscript. actually swept away and destroyed some cities Archaeological permits were obtained from the lying on the coast, while in Locris the strip of Greek Ministry of Culture, with invaluable assistance land forming a peninsula was torn through and from the British School at Athens and support from the island known as Atalante was formed.'
CLASSICAL AND ARCHAEOLOGICAL EVIDENCE FOR EARTHQUAKES
43
gie et premiers resultats. Bulletin de la Societe Geoliguqe de France, XIX(3), 679-684. BRITISH ADMIRALTY (1890) Bathymetric Chart of the 'Demetrius of Callatis, in his account of all the Evvikos Gulf (chart 1556), London. earthquakes that have ever occurred throughout COLEMAN, J. E. 1992a. Excavations at Alai in 1991. all of Greece, says that the greater part of the American Journal of Archaeology 96, 346. Lichades Islands and Cenaeum was engulfed; the 19926. Excavations at Halai 1990-1991. Hesperia, hot springs at Aedepsus and Thermopylae, after 61, 265-277. having ceased to flow for three days, began to DAKORONIA, P. 1988. Archaic ceramics of East Lokris. flow afresh, and those at Aedepsus broke forth Hesperia, 59(1), 175-180. also at another source; at Oreus the wall next to 1993. Homeric towns in East Lokris: problems the sea and about seven hundred of the houses with identification. Hesperia, 62(1), 115-127. 1996. Earthquakes of the Late Helladic III Period collapsed; and as for Echinus and Phalara and (12th century BC) at Kynos (Livanates, central Heracleia in Trachis, not only was a considerable Greece). In: STIROS, S. & JONES, R. (eds) Archaeoportion of them thrown down, but the settleseismology. Fitch Laboratory Occasional Paper, ment of Phalara was overturned, ground and all. 7, 41-44. And, says he, something quite similar happened DAVIDSON, C. 18940. Earthquakes in Greece. Nature, to the people of Lamia and of Larissa; and 50(1279), 7. Scarphia, also was flung up foundations and all, 1894&. M. Papavasilore on the Greek earthquakes and no fewer than seventeen hundred human of April 1894. Nature, 5(133), 67. beings were engulfed, and over half as many EVANS, A. 1928. The Palace of Minos at Knossos. Macmillan & Co. Ltd, London. Thronians; again, a triple-headed wave rose up, FOSSEY, J. M. 1990. The Ancient Topography of one part of which was carried in the direction of Opuntian Lokris. Gieben, Amsterdam. Tarphe and Thronium, another part to Thermopylae, and the rest into the plain as far as GANAS, A. & BUCK, V. 1998. A model for tectonic subsidence of the Allai archaeological site, Lokris, Daphnus in Phocis; fountains of rivers were dried central Greece. Bulletin of the Geological Society up for a number of days, and the Sphercheius of Greece, XXXll(l), 181-187. changed its course and made the roadways navig, ROBERTS, G. P. & MEMOU, T. 1998. Segment able, and the Boagrius was carried down a differboundaries, the 1894 ruptures and strain patterns ent ravine, and also many sections of Alope, along the Atalanti Fault, central Greece. Journal of Geodynamics, 26, 461-486. Cynus and Opus were seriously damaged, and , WADGE, G. & WHITE, K. 1996. Fault segmentaOeum, the castle above Opus, was laid in utter tion and tectonic geomorphology in eastern centruin, and a part of the wall of Elaetia was broken ral Greece from satellite data. In: llth Thematic down, and at Alponus, during the celebration of Conference and Workshop on Applied Geologic the Thesmophoria, twenty-five girls ran up into Remote Sensing, Las Vegas, NV (unpublished). one of the towers at the harbour to get a better GOLDMAN, H. 1940. The Acropolis at Halae. Hesperia, view, the tower fell, and they themselves fell with IX, 381-514. it into the sea. And they say, also, of the Atalanta GUIDOBONI, E., COMASTRI, A. & TRAINA, G. 1989. near Euboea that its middle portions, because Catalogue of Ancient Earthquakes in the Mediterranean Area up to the 10th Century. Compositori, they had been rent asunder, got a ship-canal Rome. through the rent, and that some of the plains MITSOPOULOS, K. 1895. The Lokris Mega-earthquake. were overflowed even as far as twenty stadia, and Greek Gov. Publication, Athens. that a trireme was lifted out of the docks and cast MOUYIARIS, N. K. 1988. Destructive historical earthover the wall.' quakes in n.Euboicos and Maliacos Gulfs - their significance to the evolution of the area. In: MARINOS, P. G. & KOUKIS, G. C. (eds) Proceedings of an International Symposium Organised by References the Greek National Group of IAEG. Balkema, Rotterdam, 1249-1256. ANON. 18940. Earthquakes in Greece. Levant Herald NIKONOV, A. A. 1988. On the methodology of archaeoand Eastern Express. seismic research into historical monuments. In: 18946. Severe earthquakes in Greece. The Times. MARINOS, P. G. & KOUKIS, G. C. (eds) ProceedAMBRASEYS, N. & JACKSON, J. A. 1990. Seismicity and ings of an International Symposium Organised by associated strain of central Greece between 1890 the Greek National Group of IAEG. Balkema, and 1988. Geophysical Journal International, 11, Rotterdam, 1315-132. 663-703. BLEGEN, C. W. 1926. The site of Opus. American PAPAZACHOS, B. & PAPAZACHOU, K. 1989. The Earthquakes of Greece. Ziti, Thessaloniki. Journal of Archaeology (Second Series), XXX, ROBERTS, S. & JACKSON, J. 1991. Active normal 401-404. faulting in central Greece: an overview. In: BOUSQUET, B. & PECHOUX, P. Y. 1977. La seismicite ROBERTS, A. M., YIELDING, G. & FREEMAN, B. du Basin Egeen pendant 1'antiquite. MethodoloStrabo, I. Hi. 20
44
V. BUCK & I. STEWART
The Geometry of Normal Faults. Geological Society, London, Special Publication, 56, 125-142. SIEBERG, A. 1932. Untersuchungen iiber Erdbeben und Bruchschollenbau im ostlichen Mittelmeergebiet. Denkschriften der medizinsch-naturwissenschaftlichen Gesellschaft zu Jena, 18, 161-273. SKOUPHOS, T. 1894. Die zwei grossen Erdbeben in Lokris am 8/20 und 15/27 April 1894. Zeitschrift Gesellschaft Erdkunde zu Berlin, 24, 409-474. STIROS, S. 19880. Deformations of ancient constructions: implications for the history of sites and seismotectonic research. In: MARINOS, P. G. & KOUKIS, G. C. (eds) Engineering Geology of Ancient Works, Monuments and Historical Sites. Balkema, Rotterdam, 1591-1596. 1988&. Earthquake effects on ancient constructions. In: JONES, R. E. & CATLING, H. W. New Aspects of Archaeological Science in Greece. Fitch Laboratory Occasional Paper, 3, 1-6. 1988c. Archaeology - a tool to study active tectonics. EOS Transactions, American Geophysical Union, 69(50). 1996. Identification of earthquakes from archaeological data: methodology, criteria and limitations. In: STIROS, S. & JONES, R. E. Archaeoseismology. Fitch Laboratory Occasional Paper, 7, 129-152. & DAKORONIA, P. 1989. Rulo storico e identificazione di antichi terremoti nei siti della Grecia. In: GUIDOBONI, E. (ed.) / Terremoti Prima del Mille in Italia e nell' area Mediterra-
nea. Storia Geofisica Ambiente (SGA), Bologna, 422-438. & PIRAZZOLI, P. A. 1995. Palaeoseismic studies in Greece: a review. Quaternary International, 25, 57-63. & RONDOGIANNI, T. 1985. Recent vertical movements across the Atalandi fault zone (Central Greece). Pageoph, 123, 832-848. THE HYDROGRAPHIC OFFICE (1948) Greece-East Coast Vorios Evvoi'kos Kolpos and Approaches to Volos (chart 1556), London ViTA-FiNZi, C. 1986. Recent Earth Movements: an Introduction to Neotectonics. Academic Press, London. WALKER, A. L. & GOLDMAN, H. 1915. Report on excavations at Alai of Locris. American Journal of Archaeology, XIX(4), 418-437. WREN, P. 1996. Archaic Halai. MA thesis, Cornell University, Ithaca, NY.
Ancient works cited The Geography of Strabo, vol i, trans. H. L. Jones (1932) (Loeb Classical Library). Cambridge, Mass.: Harvard UP; London: Heinemann. Diodorus of Sicily, vol i, trans. C. H. Oldfather (1933) (Loeb Classical Library). Cambridge, Mass.: Harvard UP; London: Heinemann. Thucydides, vol iv, trans. C. F. Smith (1938) (Loeb Classical Library). Cambridge, Mass.: Harvard UP; London: Heinemann.
Aims and methods in territorial archaeology: possible clues to a strong fourth-century AD earthquake in the Straits of Messina (southern Italy) EMANUELA GUIDOBONI1, ANNA MUGGIA1 & GIANLUCA VALENSISE2 1
SGA (Storia Geofisica Ambiente), Via Bellombra 24J2, 40136 Bologna, Italy ING (Istituto Nazionale di Geofisica), Via di Vigna Murata 605, 00143 Rome, Italy
2
Abstract: This research was stimulated by the need to extend in time the record of Italy's largest earthquakes, which commonly have repeat times of the same order as the length of the available historical record. As a test case we used the 1908 Straits of Messina earthquake, a large event that geologists assume to recur at intervals of roughly a millennium but whose predecessors are as yet unknown. The 1908 earthquake caused enormous territorial upheaval and left signs in the settlements that are still largely recognizable today. We hypothesized that the Straits of Messina, which were densely populated even in ancient times, may similarly retain evidence of one or more much older 'upheavals' of the settlement network, and that this evidence may be recognized through a careful analysis of archaeological observations. We found evidence that the settled area around the Straits of Messina contracted substantially around the middle of the fourth century AD, when many sites were abandoned or relocated. This contraction can hardly be justified by the then current economic and military setting. Specific archaeological findings within the cities of Messina and Reggio Calabria also suggest a serious decline of the region during the same period. The archaeological hypothesis is in good agreement with the available historical and palaeoseismological evidence and suggests that a large earthquake, perhaps similar to the 1908 event, took place in the area surrounding the Straits of Messina around the middle of the fourth century AD.
One of the most important findings of earthquake studies in Italy over the past decade is that the repeat time of individual large earthquakes is of the order of one or more millennia. Similar results had previously been obtained for many seismic areas in the world, but the frequency of large historical earthquakes in Italy had brought seismologists to the conclusion that Italian earthquakes repeated themselves every few centuries or so. Some Italian cities indeed appear to have been struck several times in their history, but results supplied by palaeoseismology (a recently developed discipline aimed at uncovering and dating any direct surface evidence of large earthquakes of the past) suggest that this is due to multiplicity of destructive seismogenic sources rather than to the frequency with which individual sources generate large earthquakes (see, e.g. Boschi et al. 1994). Individual repeat times of the order of 2000 years, such as in the case of the fault that generated the disastrous 1980 Irpinia earthquake (Pantosti et al 1993), have raised significant concerns in the seismological community as to what exactly we can learn from traditional historical catalogues of seismicity and what is bound to be missed, no matter how good the histori-
cal record is (or appears to be; see Valensise & Guidoboni (1995)). Could seismic archaeology become the answer to the need for a substantially longer record of earthquake effects on the human environment than the few centuries usually granted by conventional historical seismology? The Straits of Messina, an area struck by a catastrophic earthquake on 28 December 1908, serves as a good test site to examine the potential of this new discipline. The Straits of Messina region is known to have been occupied for over two millennia but no clear evidence exists of a predecessor of the 1908 earthquake. Current fault-segmentation models (e.g. Valensise & Pantosti 2000) suggest that the causative fault of the 1908 earthquake is the only seismogenic source capable of inducing the collapse of buildings along the shores of the Straits of Messina, and no large historical earthquake from adjacent sources has ever exceeded intensity VIII-IX in both Messina and Reggio Calabria (Boschi et al. 1995, 1997). Estimating the actual repeat time of a 1908type earthquake, and therefore the true seismic behaviour of the region, has been of critical importance in the assessment of the design for the planned permanent crossing of the Straits, to
From: McGuiRE, W. G., GRIFFITHS, D. R., HANCOCK, P. L. & STEWART, I. S. (eds) The Archaeology of Geological Catastrophes. Geological Society, London, Special Publications, 171, 45-70. 1-86239-062-2/00/S15.00 © The Geological Society of London 2000.
46
E. GUIDOBONI, A. MUGGIA & G. VALENSISE
be accomplished in the next decade through the construction of a single-span bridge. Bridge planners have assumed that earthquakes such as that of 1908 may recur roughly every 2000 years, well beyond the expected lifetime of the bridge. Our aim is to explore the possibility that historical and archaeological evidence can be brought forward through which this assumption may be substantiated on much firmer grounds. Seismic archaeology: problems of methodology Seismic archaeology (or archaeoseismology) is a young discipline at the boundary between human sciences and geophysics that has emerged in the past 20 years or so. As with many young disciplines, however, seismic archaeology still lacks an established and satisfactory methodology. On the contrary, the diversity of aims, cultures and approaches involved has led to a situation in which observations often 'involve' archaeology, but in different ways and for different purposes. Archaeology is sometimes used as an independent source, in which case it has its own interpretational framework; sometimes as confirmatory evidence of earthquake effects reported by written sources; and sometimes even as a 'springboard for speculation' about catastrophes, which are preferably of extraordinary dimensions and unknown to written history (a noble sentiment perhaps, but one that is extremely difficult to reconcile with the scientific method). In spite of the diversity of the approaches, there is a growing consensus in the seismological community that archaeological sources can indeed contribute to a better understanding of the earthquake history of a given area. Seismic archaeology is often seen as the only tool that can constrain the repeat time of major earthquake sources in areas where the written record is limited and the geological expression of large active faults is unknown or difficult to decipher; but how this is to be achieved is neither widely known nor generally agreed upon. A close scrutiny of the relevant literature shows two basic approaches to the problem of extracting information about earthquakes from archaeological sources, each with advantages and disadvantages to different degrees. Abandonment-collapse sequence approach In this approach, conclusions are drawn from architectural collapses or damage, usually occur-
ring in an abandonment-collapse sequence. This approach includes any attempt to associate earthquake effects with individual archaeological structures, in most cases with serious dating problems and often with the result of simply confirming knowledge already acquired by other means. Hence the 'discovery' and study of a collapse may in itself be of little significance for seismology although of great importance for the history of the site. Only by situating a collapse within the seismic history of the area can its scientific value be assessed. If evidence of seismic collapses is found in an area about which historical seismology already has available a great deal of information, the finding may not justify devoting human and financial resources to it. Nevertheless, careful analyses of collapses are crucial as they supply useful lists of seismic indicators and may steer future work in an appropriate direction. Combinatory method approach The second approach includes all research that adopts a 'combinatory method', making joint use of written sources and archaeological data. From an epistemological point of view, this approach is dangerous, as it may create a kind of circular argument where written sources are used to support and explain archaeological sources and the latter are used to supply what the written sources do not say. Extreme caution is always necessary in using written sources to date in situ earthquake effects, and general conclusions should never be drawn. Only in the exceptional case of Pompeii can one make direct joint use of archaeological and written sources, because of the fact that the city was buried in ash from Mt Vesuvius 17 years after the earthquake (AD 62), while rebuilding work was still going on. The 'combinatory method' is mostly used by the older generation of Mediterranean-based archaeologists. The basic interpretational factor that it seems to have neglected or underestimated is the dating system, as written sources and archaeological sources are based on two conceptually different time scales resulting from the different nature of their data. A written source may indicate not only the year, month and day on which a historical event took place, but even smaller units of time such as the hour. Conversely, archaeological evidence can only be dated by reference to the stratigraphic sequence within which it was found. Archaeological dating, in other words, is indeed based on an ad annum time scale, but in practice it is a span of years that is bound to be involved. When
AIMS AND METHODS IN TERRITORIAL ARCHAEOLOGY archaeological and historical sources are used in combination, however, the former are dated on the time scale of the latter, which may indicate a day or smaller unit of time. Hence the use of written sources to date archaeologically attested earthquake effects may lead to an interpretational error with serious potential consequences. Case study: the 21 July AD 365 earthquake. The date of an earthquake in written sources has occasionally been imposed on a number of different archaeological collapses actually caused by different earthquakes over a period of years or decades and possibly occurring thousands of miles apart. In this way, groups of such events have been interpreted as a single huge earthquake. For example, a reconstruction of the earthquake of 21 July AD 365 using the 'combinatory method' suggests that the area of destructive effects stretched from Cyprus to western Sicily, and even involved North Africa, making it appear an 'archaeological catastrophe'. This may not be of much interest to archaeologists, but it is to historical seismologists, as recent studies have shown that the main and largest earthquake occurred off the southwestern coast of Crete, and even serious damage reported for Egypt, Syria and Palestine in close temporal association with the 21 July event may have been due to smaller earthquakes triggered by the main shock (for a general review of the written sources as well as of the historical and archaeological debate, see Guidoboni et al (1994, pp. 267-274)). It has also often happened that instead of reassessing scientifically untenable positions, users of the 'combinatory method' have concealed their failure to make correct use of the archaeological dating system behind personal polemics, thereby demonstrating a neglect of basic interpretational notions (for the latest case, see Bernabo Brea (1997)). Case study: the 3 January 1117 earthquake. Another example concerns the earthquake of 3 January 1117 in northern Italy (Guidoboni 1984). From numerous studies of 12th-century Romanesque churches of the 'Lombardo' style (there were more than 360 such churches over the whole of northern Italy), art historians have attributed many changes of style, alterations, extensions and so on to destruction caused by this earthquake. The earthquake was mentioned by many written sources, although with very general descriptions of its effects. In relying solely on the written sources, however, the art historians failed to take into account that, in the first half of the 12th century, churches were often built or extended for reasons of population growth, economic change or prestige. This
47
misuse of the 'combinatory method' has resulted in the supposed area of earthquake-induced destruction being enormously enlarged. The suggestion that the event was much more severe that in reality has serious consequences for the earthquake-hazard estimates of what is normally considered a low-seismicity area. It has required long and patient historical research to reassess the true earthquake effects exclusively based on reliable data (Guidoboni 1984; Guidoboni & Boschi 1989).
Territorial evidence of major earthquakes Both the 'abandonment-collapse sequence' and the 'combinatory method' approaches place serious limitations on the use of archaeological sources for seismological purposes. For this reason, a different approach is explored here, although one that also has its limitations. The aim of this approach is to devise a congruent hypothesis, with the realization that, in our present state of knowledge, faith is essential if substantial certainties are to be achieved in seismic archaeology. The proposed approach is based on the following consideration: historical seismology tells us that an earthquake of magnitude c. 1 (substantially larger earthquakes are unlikely at least in the Mediterranean region) striking an area subject to human occupation causes considerable territorial perturbation, in the form of destruction, abandonment, urban area contraction, population shrinkage and emigration, reconstruction, and general disruption of the natural environment, often stretching over a fairly long period. For example the earthquake that took place in the Straits of Messina on 28 December 1908 (Ms 7.1), the only large event known to have occurred in the area, left a very long-lasting territorial imprint (Boschi et al. 1995, 1997; see also the map of effects, Fig. 1). It should be recalled that until this earthquake occurred, the Straits area had been considered of low seismicity, although Messina and Reggio Calabria had already been seriously damaged by earthquakes from adjacent seismogenic sources, for instance in 1783 (Baratta 1901). The question is posed: would it be possible to draw from ancient and late antique archaeological sources a territorial picture that fits the likely 'territorial perturbation' resulting from a disaster similar to that caused by the 1908 earthquake? Historical research specifically aimed at uncovering evidence for large earthquakes in the Straits area had already been carried out for the period from the seventh to the 13th centuries (Guidoboni & Traina 1996). This research allowed the
Fig. 1. Intensity distribution for the 28 December 1908, Straits of Messina earthquake (data from Boschi et al. (1995, 1997)). The dashed line encircles the area of intensity X and over.
AIMS AND METHODS IN TERRITORIAL ARCHAEOLOGY identification of six previously unknown earthquakes, but none of them had a magnitude comparable with that of the 1908 earthquake. This was not totally unexpected in view of the 1000 years average repeat time estimated by seismologists for 1908-type events (Valensise & Pantosti 1992). A systematic search for territorial traces of a great seismic disaster in the preceding millennium, that is, between the end of the sixth century BC and the fifth century AD, commenced in 1996, and lasted for a year. The findings discussed here provide an initial contribution to the seismological application of already available archaeological data. This new perspective incorporates some relatively novel concepts, such as 'archaeological landscape' and 'context' analysis, which we regard as significant in throwing light on territorial dynamics. The data came from a systematic retrieval of results published in archaeological literature on the Straits of Messina. We first established a complete census and record of published studies, which were then subjected to a careful critical scrutiny. The data from the various archaeological finds were then subdivided into a historically significant sixphase chronology (see the section 'A chronology of territorial dynamics' and the Appendix), which yielded a complete representation of our present state of knowledge at territorial level. This, together with the critical reappraisal of seismic indicators reported in the literature, made it possible to re-examine certain earthquake cases within a more precise conceptual framework. Population estimates based on the compiled data allowed a diachronic analysis of the area to be carried out, with the aim of identifying any evidence for significant perturbations of the habitational dynamics between classical and Hellenistic times and the late antique and Byzantine periods from which many historical records commence. Seismic indicators in the archaeological record: some cautionary remarks In post-processual archaeological theory, the concept of landscape as context or as a palimpsestic container of information about the past has become firmly established. In this usage, 'landscape' may be considered as the compound effect of a sequence of activities and/or events acting on different scenarios (geomorphological, demographic, cultural, political, economic) that need to be separated through space and time. From the archaeological point of view, reconstruction of events is possible only to the extent that 'actions' have left identifiable traces in the
49
ground. As a whole, these traces have in the course of time been affected by nature or humans in such a way as to transform or even remove them completely. The analysis and interpretation of an archaeological site, therefore, involves decoding these palimpsests, and the decoding must itself involve a reconstruction of the events that occurred during transformation of the ancient buried record in question (Schiffer 1987; Leonardi 19926, pp. 13-22). As items of evidence, earthquake indicators typically have the characteristics and ambiguities of all other kinds of archaeological evidence: the connection between the depositional record revealed by excavations and the processes that brought it into being can be hypothesized or inferred, but it cannot be proved in a deterministic way. As a result, there are only a few cases where a seismic hypothesis has been accepted to account for a specific situation in the archaeological record. A common characteristic of these cases is the presence of macro-indicators, such as collapses, in structures that can be described, in a functional and constructional perspective, as 'prestige buildings', such as villas, baths and sacred places of the late Imperial age. In most of these cases, moreover, the earthquake is taken to have occurred after abandonment of the site, for it is easier to find indicators of seismic activity where human reaction to seismic damage has not involved restoring the topography and structure of the site. In this way, however, many pieces of seismic evidence may have been bypassed. For example, data concerning rural settlements and modest buildings regularly escape notice, even though they were characteristic of extra-urban territorial organization in classical Greek and Roman times. The same is true of the various ways in which society reacted to seismic disasters from an economic and recovery point of view. In most of the cases we analysed, the earthquake is taken to be a macroscopic agent affecting individual inhabited buildings, the event responsible for collapses and abandonment. However, earthquake effects are never considered as potentially responsible for disturbing already existing archaeological deposits. This contrasts with Anglo-Saxon archaeological literature, in which the primary effect of an earthquake on the buried archaeological record is recognized in the formation of cracks which may be locally filled with artefacts redeposited by natural processes, usually flowing water (Wood & Johnson 1978; Butzer 1982; Schiffer 1987, pp. 231-233). In stratigraphic terms, this results in (1) the formation of deposits containing artefacts in a secondary location, (2) the horizontal and vertical movement of artefacts, and
50
E. GUIDOBONI, A. MUGGIA & G. VALENSISE
(3) the creation of possible false functional and chronological associations of deposit finds. These are important matters for archaeology, and only careful stratigraphic analysis will reveal them. But if not identified and appropriately situated in the stratigraphic sequence, they may upset the interpretation of the sequence itself. In the archaeological literature concerning the Straits of Messina, cracks and related archaeological evidence are never mentioned. In terms of the evidence that is reported, we also found a certain failure to classify seismic indicators adequately, even though at first sight they seemed to cover a broad range of stratification phenomena. The situation is in fact a complex one, for however analytical the archaeologist's approach may be, the subdivision of the stratigraphic continuum into taxonomic indicator categories will be too selective. This complexity warrants a few final words of caution. The existence of elevated sites introduces the problem of the downslope displacement of archaeological material as a result of erosion and deposition. Secondary deposits of this kind show the need for a careful analysis of possible earthquake effects on slopes. The same care ought to be applied to interpreting deformation and movement in architectural structures, such as leaning walls and sloping floors, situated on steep slopes or in landslideprone areas. A building's resistance to ground movements is obviously conditioned not only by its own structure but also by the nature of the ground itself: the frequency of structural deformation in monumental buildings (theatres, for example) highlights the problem of lack of solidity in basal strata and unstable nature of the backfill on which they were built. Seismic activity tends to exacerbate any problems of statics that may have already existed. Furthermore, abandonment, collapses and destruction may be related not only to damage caused by war, invasion, or political and social upheavals, but also to ordinary decay as a result of lack of maintenance. In turn, ancient restoration or rebuilding works may in fact be related not only to site reoccupation after military destruction, but also to population increases, changes in the organization of production, and functional modifications. Territorial archaeology in the Straits of Messina area Physiographic and archaeological setting The Straits of Messina area is situated between latitudes 37°40'N and 38°50'N, and its coastal
environs cover about 6600km2 of land. The landscape of the Sicilian side is dominated by the Monti Peloritani, which trend parallel to the coast and consist of Palaeozoic granites and crystalline schists. Their slopes are furrowed at nearly regular intervals by short, deep and steep watercourses, locally termed fiumare. The Messina area has very little agricultural hinterland, yet through the centuries the city developed as a successful maritime and commercial centre as a result of its location and its natural harbour. On the Calabrian side, the territory of Reggio Calabria consists of a series of very rough-edged plateaux corresponding to the top of technically uplifted Mid-Late Pleistocene fan deltas. The embayed form of the Straits' coastline makes it suitable for the establishment of small landing places, of which Bova Marina and Pellaro are examples. The principal lines of communication were, and still are, strongly influenced by this physiography; in antiquity there were coast roads from Messana to Panormo (Palermo) and from Messana to Catania and Syracuse. In spite of its great importance in the ancient (classical) and early Byzantine (late antique) world due to its strategic position in the Mediterranean, the area has not been the object of systematic research in territorial archaeology. Its physical characteristics do not make it particularly suitable for settlement, yet there are sound historical reasons for supposing that in ancient times it was covered by a dense settlement network. Archaeological research is hindered by the fact that the geomorphological setting of the Straits area, especially on the Sicilian side, does not favour the preservation of the buried archaeological record. The landscape of the Monti Peloritani is very active, being characterized by sudden erosion and flooding along the steep watercourses. This has frequently resulted in the removal of upland archaeological deposits and the burial of low-lying deposits, with the city of Messina itself being a case in point. In the first stage of our research we systematically scrutinized all the archaeological literature published since 1876, especially as regards those national and local journals in which accounts of excavations carried out by guardianship bodies appeared, and bibliographies. The search involved 11 titles, each spanning a period ranging from at least 30 to more than 100 years (see Appendix 2). A total of 178 contributions were analysed, and 184 finds recorded and precisely located. We also examined a variety of unpublished material made available to us by the Archaeological Superintendency of Reggio Calabria (for a complete bibliography of studies, see Storia Geofisica Ambiente (1996)).
AIMS AND METHODS IN TERRITORIAL ARCHAEOLOGY Unfortunately, these documents are by their nature discontinuous, in part because they reflect the fact that the archaeological research of the time was concerned with monuments and objects of historical and artistic value, and in part because they reveal the rather 'primitive' nature of fieldwork techniques before present-day stratigraphic excavation came into use. A new aspect of archaeological research that emerges from the most recent publications is worth noting. Since the 1980s, Superintendency reports have recorded a substantial reduction in state subsidies. This has resulted in planned research giving way to 'rescue archaeology', an increase in non-systematic field surveys designed simply to identify the buried record, and a decrease in full-scale excavations. Such a change of strategy has had a positive effect on the establishment of a better overall representation of the territory and of its demographic trends, but it also had negative effects because recent advancements in stratigraphic excavation methods have not been applied. The body of information potentially available over the last decade has therefore been insufficiently analysed. The fact that archaeological investigations have concentrated, now as in the past, on ancient towns and their immediate surroundings, has meant that there are numerous visible 'gaps' in the archaeological maps of any given area alongside dense concentrations of finds, which often simply reflect the intensity of research devoted to that area. In spite of these limitations, we consider that the settlement pattern that emerged from our archaeological 'census' over a long time span was indeed representative of real population trends and may reflect events that had a powerful impact on the area.
A chronology of territorial dynamics: fifth century EC to sixth century AD To emphasize changes in territorial dynamics we adopted a conventional six-phase system for the division of periods (Table 1), each period bounded by a major political and historical transition likely to have a lasting effect on territorial organization. This classification was made necessary by the particular nature of the available documentary material and by the aims of our research, which required a flexible system to be applied both to a regional context and to a long chronological span. Given the poverty of some finds, the frequency of clandestine excavations and the intrinsic nature of archaeological research, a large proportion of the record can be dated only to the Greek or Roman periods generically, or cannot be dated at all.
51
Table 1. Six-phase division of chronology adopted in the paper
Phase
Chronological span Historical reference
Phase 1
510/509-406 BC
Phase 2
406-273/212 BC
Phase3
273/212-27 BC
Phase 4
27BC-AD313
Phase 5
AD 313-535
Phase 6
AD 535 to the end of the sixth century
From the fall of Sybaris to the ascent of Dionysius I of Syracuse From Dionysius I to the Roman conquest The age of the Roman republic The Imperial age, from Augustus to Constantine (Edict of Milan) From Constantine to Justinian The Graeco-Gothic War and the Byzantine rule
By adopting conventional and broad periods, however, we could place all datable finds within appropriate spatial and temporal coordinates. Thus every archaeological record, however imprecise or incomplete, preserved its own historical and contextual value, and in several cases the crude find regained significance after insertion into the regional chronological picture. The final product of these broad interpretations is a series of maps, one for each phase (Fig. 2). These 'phase maps' may prove to be underestimations or overestimations. Underestimation will be due to the fact that, although a site may have survived from one phase into another, it has been 'compressed' into a single phase. Overestimation occurs when an imprecise archaeological find is duplicated and placed in more than one phase. In fact, one may reasonably guess that certain types of find assigned to a generic chronological or cultural horizon (e.g. a small 'Roman' farm or cemetery) are unlikely to persist throughout a chronological span of five or six centuries. By using all available information over the long term we inevitably lose some short- and medium-term chronological, functional and geographical variability for the area in question; however, the only alternative would have been the total elimination of general data.
Functional types and demographic factors On the basis of current excavation literature, for each site the analysed finds were classified
52
E. GUIDOBONI, A. MUGGIA & G. VALENSISE
Fig. 2. Settlement distribution during Phases 1-6, (a)-(f), respectively. Sites in boldface are mentioned in the text. Dashed line as in Fig. 1.
AIMS AND METHODS IN TERRITORIAL ARCHAEOLOGY
Fig. 2. (continued)
53
54
Fig. 2. (continued]
E. GUIDOBONI, A. MUGGIA & G. VALENSISE
AIMS AND METHODS IN TERRITORIAL ARCHAEOLOGY according to the following functional types: inscriptions; votive deposits or sacred places; settlements: remains belonging to a villa, farm, village or town; funerary contexts: cemeteries or isolated burials; infrastructures: aqueducts, wells, cisterns, bridges or roads; coins and/or jewellery hoards; sporadic finds or uninterpretable contexts; fortifications. All the finds were located through their geographical coordinates and used to create the maps described in the previous section, with the aim of establishing models and finding the settlement density for the area in question (Leonardi, 19920). This made it immediately possible to examine the general population trends during each phase. When sites are of the same size in different phases, an increase in the overall number of sites may be related to a population increase, whereas a decrease may signify a population decline (Cambi & Terrenato 1994). This condition applies at least for Phases 2, 3 and 4 and partly for Phase 5 during which the population appears to be scattered through the territories of the large towns and gathered around farms and country houses, or in occasional villages. In our relative population estimates we first have taken account of all the classified sites, then considered only those that can be considered loci of demographic attraction: settlements, funerary contexts and fortifications. The other find types provide a reliable indicator both of the extent to which a territory has acquired an infrastructure, and of the degree to which the archaeological record has been broken up in certain morphological or inhabited basins because of natural or anthropic transformations (destructive events). How to progress from a relative to an absolute estimate of demographic dynamics is the subject of lively debate in archaeological circles. Solving this problem requires a series of methodological decisions, each involving a substantial degree of arbitrariness and standardization (Cambi & Terrenato 1994). For example, the application of standard population indexes requires a careful assessment of the characteristics of the area under analysis. In our specific case, it seemed counterproductive to attempt an absolute quantitative approach to the problem of population in the period under consideration. Towns required a different procedure (Muggia 1997). Although macroscopic expansion and contraction phenomena in towns can often be identified, it is very difficult to assess small changes in the use of space. For example, how can we assess changes in house size in Roman times within the insulae, which always keep to Greek dimensions? And how can the number
55
of persons per household be established? Here too our understanding of diachronic variations depends on macroscopic and/or dimensional changes in towns. Quantification of the urban population of Messina and Reggio Calabria also required a series of preliminary calculations. First, we calculated the resident population of towns in Sicily and Magna Graecia outside our sample area and whose urban layout is already known or can be easily worked out. From estimates made for nine towns (Lipari, Megara, Hyblaea, Tindari, Eloro, Casmene, Locri, Camarina and Naxos; see also Muggia (1997)), we deduced certain demographic indices in relation to the various phases, summarized as follows: (a) 300 persons ha"1 for the Greek period, equivalent to about 30m 2 per person (Phases 1 and 2); (b) 150 persons ha~ ! , equivalent to about 70 m2 per person, for the period of the Roman republic, generally considered to be a critical stage in urban and territorial organization (Phase 3); (c) 200 persons ha" l , equivalent to about 50 m2 per person, for Imperial and late antique times (Phases 4 and 5); (d) 100 persons ha"1, equivalent to about 100m2 per person, for the Byzantine period (Phase 6). A reasonable estimate for the population of Messina and Reggio Calabria was obtained using a projection of these standard indices, expressed as the product of population density multiplied by the occupied urban space (Hassan 1981).
Archaeological landscapes and their modifications A systematic survey of archaeological literature for the Straits of Messina area allowed the identification of chronological landscape modifications resulting from human and natural activity. Landscape archaeology is a well-known aspect of the study of territorial and population dynamics, and it is here that the idea of landscape as a palimpsest has become established (Cherry 1983; Fowler 1990; Leonardi 19920). Diachronic landscape surveys are a very important instrument for decoding these palimpsests, especially in crucial areas such as the Straits of Messina, where the interaction between humans and events includes destruction brought about by nature or humans themselves. This interaction may account for population movements,
56
E. GUIDOBONI, A. MUGGIA & G. VALENSISE
changes in settlement structures, and economic and productive employment. From a strictly archaeological point of view, landscape changes have a direct effect on the survival and disappearance of buried sites. In addition, there are recent landscape changes brought about by humans, such as the introduction of new farming techniques, intensive farming and land reclamation and, especially acute in the Straits area, modern construction (buildings, railways, motorways, etc.). along the coast.
Seismic indicators from excavation reports On the basis of excavation reports, a formal classification of seismic indicators that were recognized by archaeologists was established (Table 2). This classification was then applied to the systematic reappraisal of the archaeological literature. Where excavation reports could be
appropriately reinterpreted, they led to a broadening of the database in terms of area, chronology and function. The reinterpretation of the data gathered on the basis of our preferred characteristics made it possible to bring certain typologically and functionally different sites (i.e. modest settlements, as well as farms and cemeteries) into the overall context of an area in diachronic evolution. This fresh scrutiny of the archaeological record made it possible to increase the number of earthquake indicators by adding type 6 (anomalous upward or downward displacement of habitation levels) and type 7 (absence of any archaeological record for a particular chronological phase) (Table 3). In summary, the research was based on the following criteria: (1) We drew from the available archaeological literature items suggesting possible ancient earthquakes, without recourse to the
Table 2. Seismic indicators from the literature Reference
Indicator
Depositional outcome
1 2
Surface faulting
SchifTer (1987)
Structural failure; fissuring of floors and walls
Lattanzi (1982); Cavalier & Bernabo Brea (1993-1994); Spigo (1993-1994)
3
Collapses, ground movement
Voza (1980-1981, 1984-1985); Cavalier & Bernabo Brea (1993-1994); Spigo (1993-1994)
4
Deformation, disintegration; displacement of the archaeological material
Cavalier & Bernabo Brea (1993-1994); Spigo (1993-1994)
5
Ancient restoration work
Voza (1980-1981, 1984-1985)
Table 3. Seismic indicators from reinterpreted records Indicator
Depositional outcome
Reference
1
Surface faulting
Bonanno (1993-1994)
2
Structure failure; fissuring of floors and walls
Orsi (1899); Ferri (1926)
3
Collapse, ground movements
Orsi (1899); Bernabo Brea (1964-1965); Bonanno (1993-1994)
4
Deformation; disintegration; displacement of the archaeological material
Orsi (19166); Bonanno (1993-1994)
5
Ancient restoration work
Orsi (19166); Lattanzi (1986)
6
Anomalous upward or downward displacement of habitation levels
Bacci & Rizzo (1993-1994); Bernabo Brea (1997)
7
Absence of any archaeological record for a particular chronological phase
Cavalier (1994)
AIMS AND METHODS IN TERRITORIAL ARCHAEOLOGY common game of combining deposit characteristics with the chronological process leading to the formation of the archaeological record. (2) Where possible, we sought out unpublished material, to better reconstruct the exact sequence of archaeological deposits through precise stratigraphic reanalyses. The purpose of this was to distinguish between natural and human actions in deposit formation, as well as to date the deposits (as suggested by Leonardi & Balista (1992)). Unfortunately, this approach increased our knowledge only for Reggio Calabria. (3) We assessed the reliability of archaeological indicators of earthquake effects within the historical, demographic and socioeconomic context of the area over a given period. We believe it is possible to identify traces of earthquakes in the form of 'gaps' within the general picture of a period, such as evident changes in habitational patterns that cannot be explained in terms of local historical dynamics.
Archaeological evidence for past earthquakes Evidence of seismic activity from the archaeological literature We now examine published evidence for large earthquakes that may have involved the towns of Messina and Reggio Calabria. Messina. There is no specific archaeological evidence for earthquake effects at Messina, but a general reference to these was made by Orsi (19160), who attributed the lowering of the land level at the Prefecture excavation to the high seismicity of the area. There is possible evidence for seismic deformation in the possible postdepositional disturbance of the isolato 373 cemetery (Scibona 1984-1985), where skeletons seem to be slightly broken up and shifted from their original position. Evidence for significant lowering of the harbour by settling of the underlying loose deposits is found in the accounts of the 5 February 1783 earthquake in the Gioia Tauro Plain, located 30km northeast of Messina (Boschi et al 1995, 1997). This evidence is indicative of a phenomenon that may have characterized the city during previous but unknown large events, either in the Straits or in neighbouring areas. Reggio Calabria. Unlike Messina, Reggio Calabria does provide archaeological indicators of
57
seismicity at four locations in the city, together with an inscription dated to AD 374 that mentions the rebuilding of the baths owing partly to their age and partly to earthquake damage (Guidoboni et al. 1994). The earliest, but very unconvincing, suggestion of earthquake deformation regards the via Vollaro site (06). Fiorelli (1886a) thought an earthquake could account for two sections of a column being in a collapsed position, but it is also possible that these sections were never actually used. Perhaps more convincing, Fiorelli (18866) interpreted disconnections and collapses in the Belvedere aqueduct (51) as a post-depositional effect resulting from the combined action of earthquakes, damp and agricultural work. A significant advance occurred with the Stazione Lido excavations (site 49), where two emergency campaigns were completed in 19761977 and 1979-1980 (Fig. 3). Unfortunately, when the first campaign was carried out, the stratigraphic sequence had already been altered by the works for the construction of the state railway, and no earthquake was identified with certainty (Ardovino 1978). However, on the basis of the usual indicators, we have inferred a seismic origin for a possible mid-first-century BC collapse, as well as for damage to an apsidal structure and to the buttresses of a wall. The archaeologists who carried out the subsequent 1979-1980 excavation (directed by A. Racheli and summarized by Spadea (1993)) showed greater sensitivity to the problem and identified evidence for damage, collapses and restoration works occurring at different times. The published excavation report dates the earliest evidence of a collapse to the end of the fourth century BC, but earthquake effects are offered as an explanation only as one of the many possible causes of collapse in a containment structure. Some fluvial deposits are also identified as belonging to the same period. The most significant evidence, however, belongs to the late antique period: adjacent to the second-century AD nymphaeum are some mid-third-century AD rustic buildings that are reported to have suffered 'traumatic abandonment' in the early or mid-fourth century AD. The banks of the Torrente S. Lucia are found to have been damaged in this phase, and were subsequently flattened. After the midfourth century, the complex was renovated, with added steps and a colonnade, but shortly afterwards there was a new collapse, which was subsequently buried under sandy deposits. Revised evidence of a tsunami. To account for the presence of a thick sandy deposit within the stratification of the Stazione Lido excavation in
Fig. 3. Map of excavations at Reggio Lido.
AIMS AND METHODS IN TERRITORIAL ARCHAEOLOGY Reggio Calabria, Spadea (1993), perhaps influenced by other archaeological work based on the 'combinatory method', invoked the action of a seismic wave (tsunami) caused by the earthquake of 21 July AD 365, located off the southwestern coast of Crete (see earlier discussion of this event). This hypothesis, however, runs completely counter to the chronological sequence of the site, which can be reconstructed with consistency and precision thanks to the presence of dateable material. Unpublished material made available by the Superintendency (produced by the archaeologist A. Racheli), and a detailed examination of the stratigraphic matrix of the site, shows evidence for various destructive events, which seem to have been amalgamated and somewhat neglected in Spadea's published summary (1993). As we have already indicated, the occupation phase testified by the rustic and/ or artisan buildings can be dated to the first half of the fourth century AD. The structures concerned show poor building techniques and sandy floors. The habitational area in question was abandoned as a result of a sudden collapse, as can be judged from the presence of vases in situ. The collapse is sealed by thin layers of fine, golden, micaceous sand containing floated ceramic material (see Fig. 4). This collapse is possibly contemporary with evident damage found in a large first-century AD masonry embankment, but the damage here has been identified as relating to a complex of earlier structures still in use.
Fig. 4. Byzantine wall from excavations at Reggio Lido.
59
A secondary episode of artificial levelling containing little debris and dating to the second half of the fourth and the early fifth century AD immediately precedes a monumental phase characterized by the construction of a staircase, a colonnade and a drain channel, and associated with mortared floors. Above deposits linked to the destruction of these structures there is a second, more substantial sandy layer coincident with a deposit that Spadea (1993) suggested was brought in by the tsunami of AD 365. This layer of 1.5m thickness of lenticular sand and gravel bodies displaying a random particle-size arrangement, has been found throughout the seaward side of the excavation, but the surviving masonry seems to have acted as a barrier to its landward extension. Spadea's hypothesis is based on the draping of this deposit over one of the many collapses at Stazione Lido, but the very large extent of the area under investigation is itself evidence that the sandy deposits were of artificial origin used as infill for terracing (Ardovino 1978, p. 83). Unfortunately, we lack analyses of these crucial sediments as a result of two drawbacks to the Reggio Lido excavation: (1) the urgency and pressure brought to bear on the excavation because public building site operations were not halted, which resulted in bulldozers and other machinery moving about during the excavation works; (2) the failure to realize the value of this large archaeological area overlooking the
60
E. GUIDOBONI, A. MUGGIA & G. VALENSISE
Straits, which was never considered for preservation as an archaeological park. These two factors had a decidedly unfortunate effect: not only was there a failure to secure appropriate sampling for sedimentological analyses, but the excavation was followed by the total destruction of the site, such that further investigation is now impossible. Photographic and stratigraphic documentation, of real value in view of its good quality, allows us to make certain new observations regarding, in particular, the lower thin sandy layers overlying the collapsed artisan buildings of about the mid-fourth century, which were missed in previous summaries. Their significance arises from the fact that the anomalous sandy matrix is not found in sediments identified at the site. This thin 'exotic' layer could be evidence of a strong tsunami, which violently flung marine sand onto the inhabited coast and left it there.
A large earthquake in the fourth century AD? So far, we have put together a series of 'convergent' observations concerning anomalies in the urban and territorial setting of the Straits that cannot be explained in the usual terms, along with some specific and significant finds. These together suggest that a strong earthquake was felt in the region around the middle of the fourth century (see also Appendix 1). The key observations and finds are as follows: (1) the Straits settlement network contracted substantially in Phase 5; (2) the inhabited area of Messina also contracted substantially and continued to do so until the beginning of Phase 6 to the extent that the phenomenon has been interpreted as abandonment; (3) third- and fourth-century inscriptions, found within the urban area of Reggio Calabria, were reused at the beginning of the fifth century; (4) tombs are found within the city of Reggio Calabria in Phase 5; (5) an inscription of AD 373 records the collapse of the baths at Reggio Calabria; (6) there exists evidence of collapses in the Reggio Lido locality towards the middle of the fourth century; (7) there exists a possible tsunami deposit along the shoreline of Reggio Lido. These 'indicators' are here examined in more detail. Factor (1) is the substantial contraction of the settlement network. Although the diachronic reconstruction of settlements in the Straits area presented here does not contradict current literature (Coarelli 1981; Guzzo 1981; von Falkenhausen 1982; Bejor 1986; Wilson 1990; Greco 1992), it departs from it as regards the striking decrease in available data from Phase 5 onwards. It is known that Reggio Calabria was
sacked by Alaric in AD 410, and that this military campaign created a certain amount of alarm, the manifestation of which was the mass abandonment of the central and northern coast of the Straits, as well as of inland areas. The case of Sicily, however, is different. The balance of probabilities would suggest an increase in the number of rural sites, but in fact exactly the opposite occurs, with the abandonment of the northern coast and its hinterland, and with surviving sites acquiring particular functions. We know that the Vandals raided Sicily between AD 440 and AD 475, but no evidence of military destruction has come to light in the Straits area. From a historical point of view, Phase 6 opens with the Graeco-Gothic War, which affected an area already in a state of considerable decadence. It might be expected that on both sides of the Straits this would be marked by the progressive abandonment of the coast in favour of mountainous inland areas. This occurs only to a small extent in the area of the Monti Peloritani, where there is evidence of a migration from the coast toward inland areas, with only a few vigorous sites surviving. The transition to Phase 6 is even more striking in Calabria. Here, with the exception of the city of Reggio Calabria itself, there is a total disappearance of all sites, including those that had survived through more than one phase and were structurally and functionally stronger. The process of settlement contraction had already begun in Imperial times, but in Phase 6 it is subject to massive acceleration. The economic model offered so far to account for this phenomenon (von Falkenhausen 1982) is unsatisfactory because even productive and receptive structures which centralized activities in late antique times were abandoned. Structures capable of fulfilling similar functions had been created (e.g. the Pellaro-Lume kilns), but their life had been short. Furthermore, the Phase 5 map (Fig. 2e) clearly shows that there is no direct causal relationship between the abandonment of coastal sites 'in a state of crisis' and the opening up of new potential development areas. Factor (2) is the substantial contraction of the Messina settlement, already documented in Phase 5 and even more marked in Phase 6. On the basis of available data, it is probably no exaggeration to describe this as an almost total abandonment of the urban area. There are indications of this phenomenon at Reggio Calabria as well, although probably to a lesser extent than in Messina. Factor (3) is the striking incidence of a phenomenon that we might call 'migration' of inscriptions. There is evidence for six thirdand fourth-century AD inscriptions at Reggio
AIMS AND METHODS IN TERRITORIAL ARCHAEOLOGY Calabria being reused for building materials or found as erratic elements. At two sites (38 and 61), third- and fourth-century public inscriptions (Fiorelli 1888; Orsi 1922) were reused in fairly close chronological contexts. We interpret such a reuse for private contexts of public inscriptions after only a brief time interval as evidence of an abrupt and substantial decay of both the urban landscape and its administrative structures. Factor (4), the presence of tombs within the urban area of Reggio Calabria in Phase 5 also indicates a substantial contraction of the urban area. An inscription of AD 374, which mentions the collapse of the Reggio Calabria baths as a result of an earthquake (factor (5)), was first published by Putorti (1912) and was used as evidence for earthquake damage in Reggio Calabria (Guidoboni 1989; Guidoboni et al 1994). On the basis of this inscription, Baratta (1936) attributed the collapse of the baths to the earthquake of 21 July AD 365. The last two significant factors are the collapses found at Reggio Lido (factor (6)), dated to about the middle of the fourth century AD, and the associated evidence for a tsunami (factor (7)), which was identified from the sand deposits overlying the collapse.
Reappraisal of literary sources At this point it seemed reasonable to ask: how is it that surviving written records from the late antique period do not contain any explicit reference to this hypothesized seismic disaster? To answer that question and attempt to explain the written sources, we need to recall the main features of the cultural and political situation at the time. The second half of the fourth century was steeped in a fairly extreme climate of ideological conflict brought about by the emperor Julian. He was strongly opposed to Christianity and tried to revive ancient pagan culture. Julian came to power in November 361 and died on 26 June AD 363. His death dashed the hopes of many pagan intellectuals, including the famous rhetorician Libanius, and initiated a difficult period for the empire. Libanius is the author of a famous epitaph dedicated to Julian, probably written immediately after his death in June 363 (in the view of Henry (1985), and Jacques & Bousquet (1984)). A literal translation of the Greek text reads as follows: 'Earth, at least, was duly aware of her loss and has honoured our hero [Julian] with fitting mourning. Like a horse tossing its riders, she has destroyed a great number of
61
cities - many in Palestine, and all those in Libya. The greatest cities of Sicily lie in ruins, as does every city in Greece except one [according to Henry's (1985, p. 48) hypothesis, it is the city of Athens]; Nicaea the lovely is laid low, and our loveliest of cities [Nicomedia] is shaken and can have no confidence in the future' (Lib. Or. 18. 291-293). Libanius appears to have used for his own ideological and rhetorical purposes various earthquakes that had indeed struck the Mediterranean area but he did not bother about the exact chronology of events or the places struck, referring to the latter only in a poetic and metaphorical way. Nevertheless, some of the cities and of the associated earthquakes are easily identified because they are also recorded in other sources: Nicomedia, Macedonia and Pontus (AD 358), Nicaea and Nicomedia (2 December AD 362), Corinth and other Greek cities (between AD 361 and AD 363), Palestine (363) and Libya (mid-fourth century AD) (Guidoboni et al. 1994). All of these earthquakes occurred before Julian's death, and therefore before the AD 365 earthquake, and seem to have been used by Libanius as a prediction. Also, it must not be forgotten that in Libanius' earlier Monodia, mention was made of earthquakes that had shaken the world (Lib. Or. 17. 30). A strong earthquake may therefore have involved Sicily and particularly Messina, which was indeed one of the major cities on the island, in the second half of the fourth century AD, probably between AD 350 and 363 (the year of Julian's death). Libanius' reference to the destruction of Sicilian cities in earthquakes is not the only evidence of Sicily's involvement in strong seismic activity. A passage of the Chronicon of Jerome (c. AD 347-419), who continues that of Eusebius of Caesaria, mentions Sicily as having been struck by the great tsunami of 21 July AD 365. It must be pointed out that Jerome was probably about 18 at the time and living in Rome, though he later lived in Antioch and Constantinople; hence he may well have been able to gain indirect oral evidence of what happened. The date of the tsunami was treated as epoch-making by Christian rhetoricians, who, for strictly ideological purposes, intended to stress the crisis that struck the empire after the death of Julian. It is therefore reasonable to suggest that this date may have attracted other local events to the general description of 'earthquake'. Given the present state of knowledge of the earthquake of 21 July AD 365 we find it very difficult to identify effects on the coast of Sicily that could be seen as even vaguely fitting Jerome's text: There was an earthquake
E. GUIDOBONI, A. MUGGIA & G. VALENSISE
62
throughout the world, and the sea flowed over the shore, causing suffering to countless peoples in Sicily and many other islands.' We must point out that the literary expression 'throughout the world' seems to have been used by Jerome more to bring together various seismic events that could be placed under the general heading 'earthquake' rather than to indicate a real geographical horizon (even to the world as it was known at that time) within which a single phenomenon occurred. This same literary device is also found in the monastic annals of early and mid-medieval culture, where the earthquake as a sign becomes universal, in spite of its appearance as a local event and a phenomenon circumscribed by the natural world. We think that the universalistic aspect of late antique Mediterranean culture, permeated as it was with values and expectations tending to accentuate moments of ideological crisis, provides a key to these enigmatic references in surviving contemporary sources to a seismic disaster involving Sicily. As for the failure to mention Calabria, on the other side of the Straits, this may well be a case of synecdoche; even today, the 1908 earthquake is often referred to as 'the Messina earthquake', although the number of damaged sites on the Calabrian side is larger than the number of Sicilian sites (see Table 4). It is worth mentioning that earlier research did find possible traces of earthquake activity in mid-fourth-century Sicily and Calabria by simple analysis of written sources (Guidoboni 1989; Guidoboni et al 1994), but these studies concluded nothing about the quality of the phenomena, and the effects were portrayed as separate regional events. For a number of reasons, ancient written sources can indeed preserve the record of even very small earthquakes. What our archaeological research has now done is to provide new and convergent evidence supporting the hypothesis that a strong earthquake, large enough to produce destruction on both sides of the Straits, struck the region between AD 350 and 363. Table 4. Number of sites of Calabria and Sicily that were damaged by the 28 December 1908 earthquake (intensity VII-XI)
MCS intensity
Calabrian sites
Sicilian sites
XI X and X-XI IX and IX-X VIII and VIII-IX VII and VII-VIII Total
10 57 20 110 85 282
1 13 15 61 70 160
Conclusions Previous sections have described in detail and discussed several lines of evidence converging towards a scenario that is typical of a large earthquake. Such evidence includes a substantial contraction of the settlements on both sides of the Straits between the fourth and the beginning of the fifth centuries; the presence of tombs of about the same age within the urban area of Reggio Calabria further supporting the contraction of the city; the existence of third- and fourth-century inscriptions reused for building materials only a few decades after their first emplacement; the presence of an AD 373 inscription referring to the collapse of the Reggio Calabria baths and of archaeological findings suggesting collapses at Reggio Lido around the middle of the fourth century; and the presence of anomalous sandy deposits overlying collapsed fourth-century buildings and suggesting the action of a tsunami wave. Can the 1908 earthquake in the Straits of Messina be taken as the 'modern analogue' of this fourth century disaster? Or, in other words, could the two earthquakes be representative of the activity of the very same seismogenic source? We believe the answer is 'yes'. Modern seismological wisdom suggests that the largest seismogenic sources tend to rupture repeatedly in successive similar-sized or 'characteristic' earthquakes occurring at relatively regular time intervals (Schwartz & Coppersmith 1984). Although the regularity in time of even the largest earthquakes has been since questioned and thoroughly debated, the hypothesis of similarity among subsequent ruptures along the same fault has been successfully tested in many areas of the world. For the specific case of the Straits of Messina, Valensise & Pantosti (1992) have shown that the main landscape features of the Straits (including the sharpness of the Peloritani range, the steepness of the fiumare and the existence of several elevated plateaux) are controlled by tectonic processes that are well explained by the superimposition of continuous land uplift and the pattern of crustal deformation observed following the 1908 earthquake. In other words, the present structure of the Straits is compatible with the repetition of earthquakes similar to that in 1908 as the main local tectonic agent. This finding has two fundamental implications. The first is that if the 1908 earthquake was a 'characteristic earthquake', one can expect any of its predecessors to have produced roughly the same pattern of damage and land modification as that seen at the beginning of this century (see Fig. 1). This is especially important in view
AIMS AND METHODS IN TERRITORIAL ARCHAEOLOGY of the fact that the 1908 event also generated a tsunami that was especially intense around Reggio Calabria. The second implication is that if 1908-type earthquakes dominate the tectonic scene of the Messina Straits, that is to say, if they relieve most of the tectonic stress periodically accumulated in the region, there should be little room left for additional significant sources unless they are located at the edges of the 1908 fault. As discussed in the Introduction, none of the earthquakes known to have occurred within the Straits produced widespread collapses in Reggio Calabria or Messina, and only two of those that occurred outside the Straits (the 5 and 6 February 1783 Calabrian earthquakes) induced limited building collapses around the Straits. The combination of archaeological and seismological evidence therefore suggests that the AD 350-363 earthquake can be taken as a 1500 years older ancestor of the 1908 catastrophe. Such a long interval falls at the upper limit of an existing palaeoseismological estimate (an average repeat time of 1000 (+500, -300) years for repeated 1908-type earthquakes was postulated by Valensise & Pantosti (1992)) and agrees well with average repeat times estimated for other large Italian earthquakes (for a summary see, e.g. Valensise & Pantosti (2000)). Explaining the profound changes in the Straits' territorial dynamics around AD 350-363 as the result of a catastrophic earthquake throws new light on the
63
history of occupation of this important site, supports the hypothesis that the Straits will not experience another catastrophic earthquake at least for several centuries, and highlights the value of archaeoseismology for a better understanding of seismic activity in areas where the candid assessment of the true earthquake potential has become an issue of national importance. Appendix 1: Territorial dynamics in the Straits of Messina area
Extra-urban sites The following is a brief phase-by-phase summary of the territorial scenarios obtained from the archaeological record for extra-urban settlements. The analysis focuses on the areas that would more clearly record the effects of a possible 1908-type earthquake. The two cities of Messina and Reggio Calabria are dealt with separately (see below). A general overview of the development of the settlement network is shown in Fig. 5 for the Sicilian sites and Fig. 6 for the Calabrian sites. Phase 1 (510-406 BC). 59 sites were identified for Phase 1 (Fig. 2a). Although this should be a stable and well-defined horizon in the political and economic organization of the Greek poleis and their respective territories, the associated evidence is not homogeneous on the two sides of the Straits. The only evidence on the Sicilian side is in Messina. The situation is more developed in Calabria, as most of the identified sites
Fig. 5. Trend of settlement distribution for the Sicilian side of the Straits.
64
E. GUIDOBONI, A. MUGGIA & G. VALENSISE
Fig. 6. Trend of settlement distribution for the Calabrian side of the Straits. (e.g. Fiumara di Muro 16, Scilla 53, Gallina 21, Pellaro 46, Gioia Tauro 22 and 23, Melito di Porto Salvo 32, Condofuri 13, Condofuri Marina 14 and 15, Palizzi 44) are spread along the Tyrrhenian and Ionian coast in a fairly regular way. The sites in question are settlements, cemeteries, uninterpretable contexts and/or sporadic materials, which are nevertheless evidence of more than occasional occupation. Sometimes they are associated with structures of a productive type (clay quarries at Calanna 5; a kiln and a landing place at Occhio di Pellaro 40). There is also evidence of farms as a form of settlement (Scilla 53). Penetration inland takes place along the river valleys up to a distance of about 10 km from the coast. Inland sites tend to be situated on the edges of the plateaux to control access. A fortified site in the chora of Reggio Calabria has also been documented (Serro di Tavola 54). Phase 2 (406-2731212 BC). A total of 89 finds were identified for Phase 2 (Fig. 2b), a period for which evidence exists on both sides of the Straits but with a very disproportionate number of sites. The only documented settlement on the Sicilian side is the city of Messina. Inland we find a hoard (Monforte S. Giorgio 48) and a cemetery (Rometta 68). Conversely, on the Calabrian side we find a significant increase in the number of sites, both on the coast and inland. Compared with the previous phase, we find 23 sites showing continuity of settlement but also at least 20 new sites. The occupation pattern on the Calabrian coast continues to be regular and rather spaced out rather than close-knit, and it now takes on a decidedly more stable form with the continuing presence of settlements (Occhio di Pellaro 55) and cemeteries (Melito di Porto
Salvo 43, Reggio S. Caterina 65 and Reggio Piani di Modena 66) as well as the appearance of sacred places (Saline 74). Inland occupation is denser (Fiumara di Muro 22, Calanna 8 and 9 Vito Superiore 89, S. Salvatore Cataforio 73, Motta S. Giovanni 49 and 50, Condofuri 18 and Gallina 28), as one can tell from the presence of sporadic materials and uninterpretable contexts. The quantitative relationship between the birth, death and survival of sites confirms that the general picture for the fourth century BC is one of developing territorial structures: 44 sites show continuity with Phase 1 15 sites disappear, and 45 new sites are recorded for the whole area. Phase 3 (273/212-27BC). A total of 92 finds were identified for Phase 3 (Fig. 2c). On the Sicilian side evidence exists along the northeastern coast of Sicily at Messina, at Spadafora 80 in the form of a kiln, and at Capo Peloro 10, a settlement with monumental features. The main inland evidence is the cemetery at Monforte S. Giorgio 50. The number of sites increases on the Calabrian side, where 25 new sites appear, 30 survive from the previous phase and 21 have disappeared, making a total of 55 sites. The numerical ratio between birth, death and survival of sites seems to suggest a substantial continuity of territorial structures during the transition from the Greek age to the domination of Rome, which is partially confirmed by the survival of Occhio di Pellaro 60 as a port and production centre. A careful examination of the Fig. 2c, however, shows that the habitational centre of gravity seems to be slipping towards the extreme southern tip of Calabria.
AIMS AND METHODS IN TERRITORIAL ARCHAEOLOGY Territorial structures seem to be developing principally on the coast, where we find new villas (Gioia Tauro 28, Motta S. Giovanni 55) and less prestigious settlements, possibly involved with production (Campo Calabro 9 and Condofuri Marina 20). Inland occupation is substantially unchanged in form and shows signs of continuity, whereas reported new sites are cemeteries and uninterpretable contexts (Motta S. Giovanni 51 and 53), largely situated in the more southerly stretches of the mountain area. Roman domination appears to have given priority to selective restructuring over the development of territorial structures. The overall distribution of cemeteries, however, might suggest that particular aspects of the Greek landscape were being deliberately modified. As far as religious and cultural structures are concerned, for example, there seems to be evidence of this in the abandonment of sacred places at Saline (Phase 2, 74). The general picture for this phase is rather unclear because a reduction in the number of sites was expected, whereas in fact there is a slight increase in the context of a fairly highly structured territory. This reveals a situation of transition, with possible evidence of a nascent negative trend, which, taken together with the selective nature of settlement and the shifting of its centre of gravity, can be related to the already noted reconfiguration of the settlement patterns. Imperial Phase 4 (27BC-313 AD). A total of 87 finds were identified for Phase 4 (Fig. 2d), for which evidence exists in Messina and other sites along both sides of the Straits. The most significant of these (Capo Peloro 16) survives from the preceding phase. Two particular new sites appear: the coastal settlement at Messina-Pistunina 50, which can be identified as a mansio or vicus with agricultural and pastoral structures, and the area of Fiumedinisi 28, situated slightly inland in the middle of a mining area characterized by the presence of remains of late antique farms. The site at Ganzirri 87, where there exist traces of Roman age occupation, may also have come into being during this phase. In contrast, Calabria shows a slight decrease in the overall number of sites to 52. Twenty-one sites from the previous phase have disappeared, but the persistence of 29 sites indicates stable settlement. These longlasting sites, the most outstanding being the villas of Motta S. Giovanni 56 and Gioia Tauro 35, and the functional centre of Occhio di Pellaro 59, are mostly located along the coast. The persistence of inland occupation is deduced from uninterpretable contexts. The appearance of new data-points on the map is related to finds of a settlement type: either villae rusticae with associated cemeteries, mostly situated along the coast (Pellaro 68, Cannitello 15, Taureana 80 and Gioia Tauro 36), or areas typified by small rural settlements. Also, this phase suggests a reduction in inland occupation and an increase of coastal sites, with the exception of a sudden reduction of occupation between Reggio Calabria and Villa S. Giovanni. Late antique Phase 5 (AD313-535). A total of 46 finds were identified for Phase 5 (Fig. 2e), which is characterized by a marked reduction in the number of
65
sites. On the Sicilian side and apart from Messina there is evidence of the survival of the three coastal sites that had come into being in the previous phase (Fiumedinisi 13, Pistunina 27 and Ganzirri 46). In particular, the mansio or vicus of Pistunina enjoys its period of greatest expansion. Its functional importance within local dynamics probably led to the restoration of buildings destroyed by a catastrophic flood around AD 425-450. (Bacci Spigo 1993-1994). In contrast, signs of occupation in the hinterland and on the northern coast now disappear. A similar situation is seen in Calabria, where the sampling of sites is more reliable. Out of 25 sites, 21 show continuity of occupation with the previous phase whereas four are new. Among the new sites, the most significant are the cemetery at Palmi 33 and the kiln at Pellaro—Lume 35, which, however, shows a very short period of activity (until about the middle of the fourth century AD). Inland sites are found to have survived, whereas there is a substantial reduction in coastal settlement involving the whole coastline north of Reggio Calabria, as far as Palmi 33 and Taureana 41. At the same time, however, the basic settlement fabric of villae rusticae and functional centres survives along the southern coast, most of them spanning more than one phase (Melito di Porto Salvo 26, Motta S. Giovanni 30, Taureana 41, Gioia Tauro 21 and Occhio di Pellaro 32: the last in use from Phase 1 onwards). Sometimes, as in the case of Taureana 41, survival is the result of new functions acquired in the course of time: there, the Roman villa, which is associated in the Early Christian period with the cult of S. Fantino, becomes a basilica with cemetery, and is hence an example of how topographic continuity is maintained within a totally different cultural system. It is worth pointing out that there are no traces of earthquake activity in the literature at any of the sites recorded in this phase, not even when the record is reinterpreted, but for four sites (Occhio di Pellaro 32, Pellaro 35, Gioia Tauro 21 and Gallina 17) we do record evidence for abandonment or change of activity. This evidence leads us to hypothesize that an anomalous perturbation of the territorial structure may have occurred in the second half, or perhaps at the end, of the fourth century AD (the archaeological dating system does not allow better accuracy). Phase 6 (AD535 to the end of the sixth century). A total of 27 finds were identified for Phase 6 (Fig. 2f), during which we find significant changes in the dynamics of the territorial system. On the Sicilian side there is an increase in the number of sites in the mountainous area of Rometta Messinese (21, 22 and 23) and Monforte S. Giorgio (15 and 16), which dominates the northern coast and provided no finds at all in previous phases. There is evidence of continuing occupation in the usual sites spanning more than one phase (Pistunina 14, Fiumedinisi 8 and Ganzirri 27). There is a decline in occupation, however, in the agricultural and mining area of Fiumedinisi, as one can tell from the way the archaeological record changes from settlement finds to uninterpretable contexts. Pistunina is confirmed as a site with stable, if'poor', occupation by the contextual presence of settlement and cemetery
66
E. GUIDOBONI, A. MUGGIA & G. VALENSISE
structures. There is indisputable evidence of stable structures at Ganzirri. By contrast, the Calabrian coast shows a strong contraction of the population: only seven sites are recorded, and 20 disappear from the previous phase. In the Straits area, all we find are a hoard (18) and a new fortification (19), both in the hinterland of Motta S. Giovanni. The ecclesiastical structures of Taureana 26 also survive.
The city of Messina From a historical point of view, the Zankle-Messana site is characterized by uninterrupted occupation from the eithth century BC until today, although no urban structures have survived. Topographical data for the Greek and Roman periods, therefore, have to be drawn from a variety of scattered archaeological evidence. As we already pointed out, archaeological research at Messina is hampered by the geological nature of the site, with its characteristic abundance of thick sandy and gravelly deposits associated with rapid and catastrophic alluviation phenomena. Little research was carried out in the nineteenth century. The 1908 earthquake was, in a way, a missed opportunity for collecting new data, but at least ample and good quality evidence has been collected in recent decades. Topographic pattern and evidence of its diachronic dynamics. The city of Messina is bounded by the Torrente Portalegni to the south and the Camaro to the north, and stands a few metres above sea level on a 1 km long (north to south) sloping platform having a maximum width of 500m (east to west). It is within this morphological unit that the most significant cemetery and settlement finds relating to the period of maximum urban development of Zankle (sixth to early fifth century BC) have been made. As no defensive structures from archaic times have survived, the size of the settlement has to be deduced from materials (mostly sporadic) found within the city fabric. The settlement occupied the alluvial plain to the south of the port and of the S. Ranieri peninsula. It is not possible to establish whether the settlement finds from Via S. Cecilia relate to urban or suburban dwellings; but there may be some evidence as to the southern limit of the archaic city in the huge cemetery along the S. Cosimo stream, 1300m to the south of Via S. Cecilia. Establishing the northern limit of the city is even more difficult. There are some fifth-century materials from isolato 327 in the modern town plan (the site of a Hellenistic cemetery), which may well belong to a funerary context. If so, the archaic city extended no further north than the Hellenistic and Roman cities. At the time of its maximum expansion, probably at the end of the sixth century BC, the site of Zankle covered a semicircular area facing the port and having a diameter of 1500m. In this phase (Phase 1) the city appears to have been made up of nearly east-west oriented rectangular blocks with the main streets parallel to the watercourses. Inhabited areas alternated with free areas, as excavation evidence clearly shows, and sanctuaries were arranged along the edges of the settlement. At this time the urban area probably covered about 80 ha.
In the mid-fifth century BC, the urban area was reduced by half in the east-west direction, as we deduced from the position in present-day Piazza Cairoli of the cemetery, which was to remain in use until late Hellenistic times. The expansion of the cemetery area is followed by the abandonment of a large part of the urban area so far considered, but there are also signs in the Hellenistic city (Phases 2 and 3) that the centre of gravity of the settlement shifted from the southern plain in a northerly and northwesterly direction. The city centre continues to gravitate around the port, though there is no evidence of buildings on the S. Ranieri peninsula. A substantial group of finds indicates that the heights of Montepiselli and the Tirone foothills, which face the southern plain, were inhabited from the end of the fourth or beginning of the third century BC. In late Hellenistic times, however, there is evidence (kilns) that the area between Viale S. Martino and the coast was used for production activities. The Hellenistic cemeteries together follow a sort of curve forming the city's southern boundary, thus allowing us to establish its size at the beginning of the third century BC; but we have no evidence to suggest how its size may have changed during the following period. There is evidence that in classical and Hellenistic times, the western edge of the inhabited area was near isolato 187 (where the Scuola Galatti is situated today) and 108 (Cinema Garden) near Via Martino. Subsurface surveys at the first of these sites have shown the presence of Hellenistic deposits below the level of the Roman inhabited area. At the second site, the modest remains of fourthcentury dwellings looking onto a broad open space suggest that this was the periphery of the ancient inhabited area, and that there was little urbanization. The city thus shrinks to inside the Torrente Portalegni to the south and the Torrente Bocetta to the north, lying squeezed between the sea and the hill slopes. Its area decreases appreciably, perhaps to as little as 40 ha. The reduction in size in Hellenistic times has been related to the Carthaginian destruction of the city in 396 BC; but the process of contraction had in fact already begun, as the Via Cairoli cemetery indicates, and must have occurred for reasons of a more economic nature and because of changes in the functional use of space. Remains of structures are very scanty, and in general equally scanty are finds relating to Imperial and late antique times (Phases 4 and 5), which seem to be concentrated between the Torrente Portalegni and the Teatro Vittorio Emanuele to its north and extend no further than the Torrente Bocetta cemetery. The structure of the Roman city must have been homogeneous, and seems to have been concentrated around the port, with a possible expansion westwards. Late antique (Phase 5) finds prove to be concentrated in the small coastal strip to the west of the port and indicate a further decrease in the size of the inhabited area (which shrinks to about 30 ha), whereas evidence for the Byzantine period (Phase 6) is scattered and insubstantial. During these two phases the number of mapped sites drops from eight to three. Although it is still punctuated by gaps, the archaeological description of urban Messina appears to be reasonably representative of the site's topographical dynamics (Vallet 1958; Scibona 1992, 1993). There is
AIMS AND METHODS IN TERRITORIAL ARCHAEOLOGY
67
Fig. 7. Population trends for the towns of Messina and Reggio Calabria.
therefore particular significance in the almost total lack of evidence for the Byzantine period, which, unless we postulate drastic changes in location strategies (although such changes have not been identified by archaeologists), implies that the contraction of the city must have been extreme. The port. We have little information about coastline changes along the S. Ranieri peninsula. Votive finds indicate sacred places that were frequented over a long period in areas that are still usable. Soundings in Piazza Cavallotti show that in Greek and Roman times the area was covered by a marshy beach without buildings. The rise of the coastline is thus a recent and entirely artificial phenomenon, which began in the Middle Ages, when earth was brought in to reclaim new land and subsequently to build on it (Bacci Spigo 1993-1994). Population: long-term estimates. We have made estimates of the resident population of ancient urban Messina using the criteria set out above. The results are shown in Fig. 7.
The city of Reggio Calabria Unlike Messina, the archaeological inheritance of Reggio Calabria was actively protected in the nineteenth century, and since the 1908 earthquake activity has been systematically and well documented. Recent activity by the Superintendency, however, was concentrated on a small number of excavations, for the museum sector has been given preferential treatment. The net result is an increase in the number of archaeo-
logical finds that are easy to situate topographically but often insufficiently informative for dating purposes. Topographic pattern and evidence of its diachronic dynamics. Thanks to the survival of significant fragments of its late classical defensive works, a reconstruction of the topography of Reggio Calabria can be of much greater help than is the case with Messina. Archaic and classical age (Phase 1) finds of a habitational and religious kind seem to be located mostly in the northern part of the city, near the Rada dei Giunchi, as well as in what may have been an area outside the walls (keeping in mind that the inhabited area in archaic times may have been 22-25 ha). Like the modern city, the Reggio Calabria of Hellenistic times (Phase 2) extended along the coast for about 1 km, and was bounded by the Torrente S. Lucia to the north and by the Torrente Calopinace to the south. If we accept Tropea Barbara's convincing reconstruction (1967), the late fifth-century BC city walls stretched about 750m up into the eastern heights bounding an area of 75 ha, although only about 30 ha of the coastal strip were settled and the rest was an open space. There is some debate, however, about the archaeological evidence (Vallet 1958; Turano 1966). The fact is that we have comparatively few settlement finds for Reggio Calabria, but a very large number of finds relating to religious practices and infrastructures, especially wells and cisterns for the local water supply. The conic section cisterns are a useful clue to the boundaries of the urban area, for they are arranged both within the street system and on the hill slopes around it. They belong to classical times and remained in use until the first century BC, when aqueducts were built. In Roman times (Phases 3 and 4) we notice an increase in settlement structures, i.e. houses and baths
68
E. GUIDOBONI, A. MUGGIA & G. VALENSISE
(Orsi 1922). There is evidence for expansion beyond the walls both to the north and to the south, probably in the form of large suburban villas rather than as a continuous fabric, and the size of the urban area became about the same as that of the Hellenistic city (about 30 ha). We cannot say that there was any contraction of the settled area in late antique times (Phase 5), for there is homogeneous evidence of villas and baths right along the urban strip, but the appearance of some tombs within the city (15, 39) suggests that the settlement fabric had become less compact. With the arrival of the Byzantine age, the number of finds decreases sharply from 19 to nine. Unlike the case of Messina, evidence available in this phase (Phase 6) enables us to identify an area of denser settlement in Reggio Calabria. The close proximity of a building (61), some sacred places (67, 69 and 52) and some bronze inscriptions (32) suggests that this may have been a public area, perhaps with an archivum with funerary contexts around it. This pole of attraction shows topographical continuity with structures belonging to the late antique phase, and occupies an area of 8 ha. At the northern end of the city, near the Stazione Lido, there are some Byzantine structures of an artisan type (49: fish preparation plant and cemetery), which may indicate the existence of a new type of settlement, organized in separate but neighbouring nuclei. To sum up: the settlement dynamics at Reggio Calabria is different from that of Messina, for we find in the former an archaic town of moderate dimensions (25 ha) which spreads to 30 ha in accordance with an organized town plan. From Hellenistic times to the late antique period the settlement remains constant at least in area (30 ha), if not in terms of the homogeneity of the urban fabric. In the Byzantine phase, from the mid-fourth century onward, the settlement becomes much smaller and acquires a more scattered form. The port. There has been a lively debate over the identification of the Greek and Roman port (a synthesis and bibliography have been given by Tropea Barbaro (1967, pp. 66-83)). The present-day port, to the north of the mouth of the Torrente dell'Annunziata, is an artificial basin dating to 1873. Sediments carried by the Torrente delPAnnunziata cannot have had much effect on the morphology of the Rada dei Giunchi, though they must have produced a moderate advancement of the coastline. On the basis of Thucydides (VI 44, 2-3), many scholars have identified the Rada dei Giunchi as the place where the Athenians disembarked in 415BC. It must be stressed, however, that the present coastal morphology lacks an important point of reference, Punta Calamizzi, a small promontory located to the south of the city, which disappeared into the sea in 1562 (Tegani 1873). If it is true that Punta Calamizzi was cultivated, as local historians maintain, it presumably stretched some way out into the sea, forming a large sheltered inlet together with Rada dei Giunchi. Population: long-term estimates. We have made estimates of the resident population of ancient urban Reggio Calabria using the criteria set out above. The results are shown in Fig. 7.
References ARDOVINO, A. M. 1978, Edifici ellenistici e romani ed assetto territoriale a nord-ovest delle mura di Reggio. Klearchos, 20(77-80), 75-112. BACCI, M. G. & Rizzo, C. 1993-1994. Attivita della Soprintendenza: Taormina. Kokalos, 39-40, 945-951. BACCI SPIGO, M. 1993-1994. Attivita della sezione ai Beni Archeologici della Soprintendenza, B.C.A. di Messina negli anni 1989-1993. Kokalos, 39-40, 923-943. BARATTA, M. 1901.1 ten-emoti d'Italia. Saggio di storia geografia e bibliografia sismica italiana. Bocca Torino (anastatic reprint, Forni Sala Bolognese, 1979). 1936, / terremoti in Italia. R. Pubblicazioni della commissione italiana per lo studio delle grandi calamita, Vol. VI. Accademia Nazionale dei Lincei, Le Monnier, Florence. BEJOR, G. 1986. Gli insediamenti nella Sicilia romana: distribuzione, tipologie e sviluppo da un primo inventario dei dati archeologici. In: GIARDINA, A. (ed.) Societd romana e impero tardoantico, HI. Le merci e gli insediamenti. La terza, Bari, 463-519. BERNABO BREA, L. 1964-1965. Due secoli di studi, scavi e restauri del teatro greco di Tindari. Rivista dell'Istituto Nazionale di Archeologia e Storia dell'Arte, 13-14, 99-144. 1997. Note sul terremoto del 365 d.C. a Lipari e nella Sicilia nord orientale. In: GIARRIZZO, G. (ed.) La Sicilia dei terremoti. Lunga durata e dinamiche sociali. Maimore Catania, 87-97. BONANNO, C. 1993-1994. Scavi e ricerche a Caronia e a S. Marco d'Alunzio, Kokalos, 39-40, 953-986. BOSHI, E., GUIDOBONI, E., FERRARI, G., VALENSISE, G. & GASPERINI, P. (eds) 1995, 1997. Catalogo dei forti terremoti in Italia dal 461 a.C. al 1990. Istituto Nazionale di Geofisica and Storia Geofisica Ambiente (SGA) Bologna and Rome (also on CD-ROM and web site http://storing.ingrm.it/cft/ index.htm). BOSCHI, E., PANTOSTI, D. & VALENSISE, G. 1994. L'identificazione geologica delle faglie sismogenetiche. Le Scienze, 310, 36-46. BUTZER, K. W. 1982. Archaeology as Human Ecology. CUP, Cambridge. CAMBI, F. & TERRENATO, N. 1994. Introduzione all'archeologia dei paesaggi. Nuova Italia Scientifica, Rome. CAVALIER, M. 1994. Panarea. Bibliografia Topografica della Colonizzazione Greca in Italia e nelle hole Tirreniche, 13, 321-329. & BERNABO BREA, L. 1993-1994. Attivita della Soprintendenza: isole Eolie. Kokalos, 39-40, 987-1000. CHERRY, J. 1983. Frogs around the pond: perspectives on current archaeological survey in the Mediterranean region. In: KELLER, D. R. & RUPP, D. W. (eds) Archaeological Survey in the Mediterranean Region. British Archaeological Report (BAR) International Series, 155, 375-416. COARELLI, F. 1981. La Sicilia tra la fine della guerra annibalica e Cicerone. In: GIARDINA, A. &
AIMS AND METHODS IN TERRITORIAL ARCHAEOLOGY SCHIAVONE, A. (eds) Societd romana e produzione schiavistica I. L'Italia: insediamenti e forme economiche. Laterza, Bari, 2-18. FERRI, S. 1926. Gioiosa lonica (Marina). Teatro romano e rinvenimenti varii. Notizie degli Scavi di Antichitd, 1926, 332-338. FIORELLI, G. 18860. Reggio di Calabria. Nota del can., A. M. Di Lorenzo Vicedirettore del Museo di Reggio. Notizie degli Scavi di Antichitd, 1886, 59-64. 1886Z?. Reggio di Calabria. Note del vice direttore del Museo can., A. M. Di Lorenzo. Notizie degli Scavi di Antichitd, 1886, 436-441. 1888. Reggio di Calabria. Avanzi di edificio termale ed epigrafi onorarie latine scoperte in Reggio. Rapporto del vice-direttore del Museo civico can. A. Di Lorenzo. Notizie degli Scavi di Antichitd, 1888, 715-717. FOWLER, P. J. 1990. Site, landscape and context. In: FRANCOVICH, R. & MANACORDA, D. (eds) Lo scavo archeologico dalla diagnosi all'edizione, III ciclo di lezioni sulla ricerca applicata in archeologia (Certosa di Pontignano 1988). All'ln segna del Giglia, Florence, 121-131. GRECO, E. 1992. Archeologia della Magna Grecia. Laterza, Bari. GUIDOBONI, E. 1984. 3 Janvier 1117: le tremblement de terre du Moyen Age roman, aspects des sources. In: HELLY, B. & POLLING, A. (eds) Tremblements de terre histoire et archeologie, IVemes rencontres Internationales d'archeologie et d'histoire d'Antibes. Actes du colloque, 2-4 novembre 1983, Association pour la Promotion et la Diffusion des Connoissance Archeologiques Valbonne, 119-139. & BOSCHI, E. 1989. I grandi terremoti medievali in Italia. Le Scienze, 249, 22-35. & TRAINA, G. 1996. Earthquakes in medieval Sicily. A historical revision (VII-XIII century). Annali di Geofisica, 39, 1201-1225. , COMASTRI, A. & TRAINA, G. 1994. Catalogue of ancient earthquakes in the Mediterranean area up to the 10th century. Istituto Nazionale di Geofisica and Storia Geofisica Ambiente (SGA), Bologna. Guzzo, P. G. 1981. II territorio dei Brutii. In: GIARDINA, A. & SCHIAVONE, A. (eds) Societd romana e produzione schiavistica. I. L'Italia: insediamenti e forme economiche. Academic Press, Bari, 115-136. HASSAN, F. A. 1981. Demographic Archaeology. New York. LATTANZI, E. 1982. Attivita archeologica della Soprintendenza Archeologica della Calabria. In: Megale Hellas. Nome e immagine. Atti del XXI convegno di studi sulla Magna Grecia, Toronto 1981, Istituto per la Storia e 1'archeologia della Magna Grecia, Taranto, 217-236. 1986. L'attivita archeologica in Calabria nel 1985. In: Neapolis, Atti del XXV convegno di studi sulla Magna Grecia, Taranto 1985, 417-430. LEONARDI, G. 19920. Assunzione e analisi dei dati territoriali in funzione della valutazione della diacronia e delle modalita del popolamento. In: BERNARDI, M. (ed.) Archeologia del paesaggio,
69
IV ciclo di lezioni sulla ricerca applicata in archeologia (Certosa di Pontignano 1992). All'ln segna del Griglio, Florence, 25-66. \992b. II deposito archeologico: bacini, processi formativi e trasformativi. In: LEONARDI, G. (ed.) Processi formativi della stratificazione archeologica, Atti del seminario internazionale. 'Formation processes and excavation methods in archaeology: perspectives', Departmento di scienze dell'antictifa Universita degli studi, Padua, 13-47. & BALISTA, C. 1992. Linee di approccio al deposito archeologico. In: LEONARDI, G. (ed.) Processi formativi della stratificazione archeologica. Atti del seminario internazionale 'Formation processes and excavation methods in archaeology: perspectives', Departmento di scienze dell'antictifa Universita degli studi, Padua, 75-99. MUGGIA, A. 1997. L'area di rispetto nelle colonie magno-greche e siceliote. Studio di antropologia della forma urbana. Sellerio, Palermo. ORSI, P. 1899. Buscemi. Sacri spechi con iscrizioni greche, scoperti presso Akrai. Notizie degli Scavi di Antichitd, 24,452-471. \9l6a. Messana. La necropoli romana di S. Placido e di altre scoperte avvenute nel 19101915. Monumenti Antichi pubblicati per cura della Reale Accademia dei Lincei, 24, 121-218. 19166. Nocera Tirinese. Ricerche al Piano della Tirena sede dell'antica Nuceria. Notizie degli Scavi di Antichitd, 1916, 335-362. 1922. Reggio Calabria. Scoperte negli anni dal 1911 al 1921. Notizie degli Scavi di Antichitd, 1922, 151-186. PANTOSTI, D., SCHWARTZ, D. P. & VALENSISE, G. 1993. Paleoseismology along the 1980 Irpinia earthquake fault and implications for earthquake recurrence in the southern Apennines. Journal of Geophysical Research, 98, 6561-6577. PUTORTI, N. 1912. Di un titolo termale scoperto in Reggio Calabria. Rendiconti della Reale Accademia dei Lincei. Classe di scienze morali, storiche e filologiche, serie V, 21, 791-802. SCHIFFER, M. B. 1987. Formation Processes of the Archaeological Record, University of Utah Press, Albuquerque. SCHWARTZ, D. P. & COPPERSMITH, K. J. 1984. Fault behavior and characteristic earthquakes: examples from the Wasatch and San Andreas fault zones, Journal of Geophysical Research, 89, 5681-5698. SCIBONA, M. 1984-1985. Messina: notizia preliminare sulla necropoli romana e sul giacimento preistorico del torrente Boccetta. Kokalos, 30-31, 855-861. 1992. Messina. Bibliografia Topografica della Colonizzazione Greco in Italia e nelle Isole Tineniche, 10, 16-36. 1993. Punti fermi e problemi di topografia antica a Messina: 1966-1986. In: Lo stretto crocevia di culture, Atti del XXVI convegno di studi sulla Magna Grecia, Taranto 1986, 433-458. SPADEA, R. 1993. Le citta dello stretto e il loro territorio. Reggio Calabria. In: Lo stretto crocevia di culture, Atti del XXVI Convegno di Studi sulla Magna Grecia, Taranto 1986, 459-474.
70
E. GUIDOBONI, A. MUGGIA & G. VALENSISE
SPIGO, U. 1993-1994. Capo d'Orlando: il complesso termale di eta imperiale romana di Bagnoli S. Gregorio. Scavi 1987-1992. Kokalos, 39-40, 1027-1037. STORIA GEOFISICA AMBIENTE (SGA) 1996. Territorial archaeology in the Straits of Messina area between 500 BC and AD 500. Bologna, Rome, Report 158/ 1996. [In Italian.] TEGANI, A. (ed.) 1873. Cronaca del cantore Antonio Tegani [e dei suoi continuatori (1480-1625)], a cura di, A. M. De Lorenzo. Memorie da servire alia storia sacra e civile di Reggio e delle Calabrie, vol. I (parte I fasc. 3). Tipografia Siclari Reggio Calabria, 9-60. TROPEA BARBARO, E. 1967. II muro di cinta occidentale e la topografia di Reggio ellenica, Klearchos, 9, 7-130. TURANO, C. 1966. Carta archeologica di Reggio Calabria del XIX secolo. Klearchos, 8(29-32), 159-180. VALENSISE, G. & GUIDOBONI, E. 1995. Verso nuove strategic di ricerca: zone sismogenetiche silenti o silenzio delle fonti? In: BOSCHI, E., FERRARI, G., GASPERINI, P., GUIDOBONI, E., SMRIGLIO, G. & VALENSISE, G. (eds) Catalogo deiforti terremoti in Italia dal 461 a. C. al 1980. Istituto Nazionale di Geofisica and Storia Geofisica Ambiente (SGA), Bologna and Rome, 112-127. & PANTOSTI, D. 1992. A 125 Kyr long geological record of seismic source repeatability: the Messina Straits (southern Italy) and the 1908 earthquake (Ms 7±). Terra Nova, 4, 472-483. & 2000. Seismogenic faulting, moment release patterns and seismic hazard along the central and southern Apennines and the Calabrian Arc. In: VAI, G. B. & MARTINI, I. P. (eds) Anatomy of a Mountain Chain: the Apennines and Adjacent Mediterranean Basins. Kluwer, Dordrecht.
VALLET, G. 1958. Rhegion et Zancle. Histoire, commerce et civilisation des cites chalcidiennes du detroit de Messine. De Boccard, Paris. VON FALKENHAUSEN, V. 1982.1 Bizantini in Italia. In: PUGLIESE CARRATELLI, G. (ed.) / Bizantini in Italia. Scheiwiller, Milan, 1-136. VOZA, G. 1980-1981. L'attivita della Soprintendenza alle Antichita della Sicilia Orientale. Parte I. Kokalos, 26-27, 674-693. 1984-1985. Attivita nel territorio della Soprintendenza alle Antichita di Siracusa nel quadriennio 1980-1984. Kokalos, 30-31, 657-678. WILSON, R. J. A. 1990. Sicily under the Roman empire. The archaeology of a Roman province. 36BC-AD535. Aris & Phillips, Warminster. WOOD, W. R. & JOHNSON, D. L. 1978. A survey of disturbance processes in archaeological site formation. In: SCHIFFER, M. B. (ed.) Advances in Archaeological Method and Theory I. Academic Press, New York, 315-381.
Archaeological journals subjected to systematic scrutiny Archivio Storico Messinese (1969-1994) Archivio Storico per la Calabria e la Lucania (1931— 1994) Archivio Storico per la Sicilia Orientale (1904-1994) Archivio Storico Siciliano, IV serie (1975-1992) Atti dei Convegni di Studio sulla Magna Grecia (Taranto, 1960-1992) Bibliografia Topografica della Colonizzazione Greca in Italia e nelle hole Tirreniche (1977-1995) Klearchos (1959-1984) Kokalos (1955-1994) Notizie degli Scavi di Antichita (1876-1960) Rivista Storica Calabrese (1893-1994) Sicilia Archeologica (1968-1994)
Santorini (Greece) before the Minoan eruption: a reconstruction of the ring-island, natural resources and clay deposits from the Akrotiri excavation WALTER L. FRIEDRICH, MARIT-SOLVEIG SEIDENKRANTZ & OLE BJ0RSLEV NIELSEN
Department of Earth Sciences, University of Aarhus, DK-8000 Arhus C, Denmark Abstract: Before the catastrophic eruption around 1640BC, Thera, Therasia and Aspronisi formed a ring-shaped island with a sea-flooded caldera in the middle. The so-called PreKameni Island was situated in the centre of the caldera. This reconstruction is based on the study of stromatolites found in eruption products as well as other geological observations. The location of pre-eruption settlements or sites on the present rim of the Santorini caldera seems to support this reconstruction. Many of the rocks and minerals used in the Bronze Age culture are of local origin. Foraminiferal and mineralogical studies enable us to trace the source areas of a clay deposit found in a grave chamber in the Akrotiri excavation. This clay can be used for pottery making. The foraminiferal and mineralogical studies also help identify the natural drainage system and thus the freshwater supply, which may have been an important factor deciding the location of the Bronze Age settlement.
The volcanic island of Santorini (36.40°N, 25.40°E) (Fig. 1) in the Aegean Sea is of considerable interest for both geology and archaeology, as a flourishing Late Bronze Age settlement was buried under the volcanic products of the Minoan eruption. According to a new ice-core dating (Clausen et al. 1997), this eruption, which caused the end of the Minoan settlement on Santorini, occurred around 1640BC. A large buried settlement, located on the Akrotiri Peninsula (Fig. 2), is currently under excavation. The source areas of some of the natural resources (especially clay and water supply) from this excavation will be discussed in this paper. The geography of Santorini before the Minoan eruption
Geological evidence Today Santorini consists of the three older islands (Thera, Therasia and Aspronisi), as well as the two young Kamenis (Palea and Nea Kameni) (Fig. 2). The formation of the Kameni Islands started at 192BC and has continued until modern times. Geological evidence shows, however, that Santorini formed a ring-shaped island
before the Minoan eruption (Fig. 3). Especially significant for this reconstruction were the findings of stromatolitic blocks in the third layer of the products from the Minoan eruption (Fig. 3). Stromatolites are calcareous, globular structures, which were formed by the interactions of algae and bacteria in shallow-marine conditions. They were radiocarbon dated to about 13000 years BC, and their distribution in the northern part of Thera and Therasia made it possible to trace them to their growth area (Friedrich et al 1988; Eriksen et al 1990; Friedrich 1994). Together with other geological observations, this helped answer one of the questions that geologists and archaeologists have asked themselves for a long period: what did Santorini look like before the eruption? It was concluded that the central part of the caldera that existed before the Minoan eruption contained an island, the so-called Pre-Kameni Island (Friedrich et al 1988) (Fig. 3). The location of this island was deduced from the location of the vent of the Minoan eruption and the fact that the first phase of the eruption (the Plinian phase) was not influenced by contact with water. This reconstruction has been confirmed by additional geological observations by Druitt & Francaviglia (1990 1992), who found pumice products of the Minoan eruption plastered on
From: McGuiRE, W. G., GRIFFITHS, D. R., HANCOCK, P. L. & STEWART, I. S. (eds) The Archaeology of Geological Catastrophes. Geological Society, London, Special Publications, 171, 71-80. 1-86239-062-2/00/S15.00 © The Geological Society of London 2000.
72
W. L. FRIEDRICH, M.-S. SEIDENKRANTZ & O. B. NIELSEN
Fig. 1. Location of Santorini (marked in black) in the Aegean Sea.
the inner side of the caldera wall underneath the town of Fira (Fig. 2) on Thera and at other localities. Archaeological observations The new idea of the shape of Santorini before the Minoan eruption has also had an influence on the concept of archaeological thinking, especially the question of whether there was access to local occurrence of minerals and ores. Furthermore, the distribution of settlements or sites on the present-day shape of the island could be explained much better using the new reconstruction (Friedrich & Doumas 1990).
tracing natural resources to local areas on Santorini are the following: (1) the talc used to produce white decorations on pottery (Aloupi & Maniatis 1990) occurs at Plaka (Fig. 3) (Friedrich & Doumas 1990; Friedrich 1994); (2) reddish marble used in the production of a stone vase found in the Akrotiri excavation occurs at Echendra, close to Cape Exomiti on Thera (Fig. 3); (3) pigments used in the production of wall paintings are still under investigation, but the sources for the pigments might be the phyllites that occur between Thermia and Plaka (Fig. 3). These localities were presumably accessible by boat before the Minoan eruption.
Clay from the Akrotiri excavation Source areas of materials found in the Akrotiri excavation In some cases, rocks used for the building of houses, as well as minerals, and other natural resources in the Akrotiri excavation could be traced back to their source areas (Fig. 3). This investigation was carried out mainly by Einfalt (19780,&) and supplemented by further observations by Friedrich (1994). Three examples of
During the construction of a Dextion roof in the Akrotiri excavation (Figs 3 and 4), the archaeologists dug a hole, where they placed a pillar for the roof construction. They eventually found a clay deposit on the site of Pillar 17 (Marinatos 1976) (Fig 4). This specific clay has been used for making modern pottery, as demonstrated in 1989 at the opening ceremony at the Third Conference on Thera and the Aegean World', when the leader of the Akrotiri excavation, Professor
SANTORINI BEFORE THE MINOAN ERUPTION
73
Fig. 2. Present geography of the Santorini island group, showing the whole volcanic edifice, including the Kolumbo volcano to the NE.
Christos Doumas, gave a piece of pottery to the organizer, Peter Nomikos. The clay deposit at Pillar 17 is found in the immediate surroundings of the pillar, where it has a thickness of 0.3-2.5m. The clay was also found in a small cave dug out under an ignimbrite (hard, volcanic rock) by the inhabitants of the settlement (Fig. 5). They closed the cave by a stone wall, and archaeological evidence indicates that the cave had the function of a grave chamber (Doumas 1995). Further excavations (Doumas 1996) revealed that the grave was from the Early Cycladic period and it contained seven
marble idols, stone settings and bones. It was not, however, possible to determine whether the bones were of human or animal origin. The clay is found behind the stone wall inside the grave chamber. The stratification and increasing thickness of the clay deposit towards this wall lead to the conclusion that the clay filled the cave at a later time through an opening in the wall. The cause of the infill could be either natural processes, e.g. the influence of wind and water, or human activity. The layering of the sediment, however, suggests that this deposit was formed naturally. This does not conflict with
74
W. L. FRIEDRICH, M.-S. SEIDENKRANTZ & O. B. NIELSEN
Fig. 3. Geography of Santorini before the Minoan eruption. The reconstruction is based on geological and archaeological evidence (after Friedrich 1994). Black squares mark the location of Bronze Age archaeological sites. Circles of different size show the present-day occurrences of ejected stromatolitic blocks (the larger the circle, the more common the stromatolitic blocks). White dots north of the Pre-Kamenia Island represent the assumed place of growth for stromatolites before the Minoan eruption. Possible sources of rocks and minerals found in the Akrotiri excavation are shown in this reconstructed shape of the Bronze Age island. Some of the information on the resources is from Einfalt (1978&). The interval between the isohypses is 100m.
the observation of the archaeologists (Doumas 1995), who found that the immediate surroundings of the area had been spared in later building, indicating that the Therans knew of the
existence of their ancestor's grave and respected the area as a holy site. The clay samples, which we analysed, contained a piece of charcoal, which has been
SANTORINI BEFORE THE MINOAN ERUPTION
75
Fig. 4. Plan of the Akrotiri excavation showing the location of Pillar 17-
radiocarbon dated using the accelerator mass spectrometry (AMS) method at the AMS Laboratory, University of Aarhus (sample no. AAR-1566). The sample gave an age of 4050 ± 6014C years BP, corresponding to 2570-2510BC
in calendar years (calibrated after Stuiver & Renner (1993)) (see also the discussion below). Geological setting around Pillar 77. Two of us (O.B.N. and W.L.F.) examined the site in 1993.
76
W. L. FRIEDRICH, M.-S. SEIDENKRANTZ & O. B. NIELSEN
Fig. 5. Sketch of the grave chamber with the clay at Pillar 17.
According to our observations, the roof of the grave chamber is formed by the so-called Cape Riva Ignimbrite, which can be traced from the caldera wall (Friedrich et al. 1977) to the excavation by observations in the field, i.e. in the Potamos Valley. The situation in the Akrotiri excavation is similar to that in the Potamos Valley, where the ignimbrite sheet ends abruptly parallel to the present-day coastline, as a result of coastal and fluviatile erosion. Furthermore, this ignimbrite crops out directly on the south coast, roughly halfway between the Akrotiri excavation and Cape Exomiti (Figs 3 and 6a, b). The source area of the clay. Clay samples taken from inside the grave chamber at Pillar 17 (Fig. 4) in the Akrotiri excavation have been identified as an erosional product of local sediments from the Archangelos-Loumaravi Complex (Figs 3 and 6a, b). The study was based on its content of foraminiferal and volcanic components. The clay sample contains a poor foraminiferal assemblage of shallow- and deep-water species (Fig. 6a), indicating that the clay deposit at Akrotiri is derived from more than one source area. Marine Plio-Pleistocene sediments from the Archangelos-Loumaravi Complex on the
Akrotiri Peninsula have previously been studied by Seidenkrantz & Friedrich (1993). The Foraminifera (Figs 6a and 7) revealed that both shallow- and deep-water marine sediments occur in this particular area. The sediments of the Archangelos-Loumaravi Complex are a mixture of clay, marl and pumice, and were deposited before or during the formation of a volcanic dome. The highest clay and marl content is found in the deeper water (upper epibathyal) deposits at Cape Loumaravi (Fig. 3), whereas the shallower (littoral to inner neritic) deposits at Mt Archangelos and Mt Loumaravi (Fig. 3) have a larger content of pumice. The foraminiferal assemblage from the Pillar 17 clay shows the closest affinity to the assemblages found at Cape Loumaravi (see Seidenkrantz & Friedrich 1993) (Figs 6a and 7) but with an element of the shallow-water or epiphytic species that dominate the assemblage from Mt Archangelos. We can thus partly confirm the idea of Fouque (1879), who investigated prehistoric pottery shards from the Akrotiri area that contain Foraminifera and other marine organisms. He concluded that the clay was of local origin, possibly derived from a shallowwater site between Aspronisi and the Akrotiri Peninsula, which might have been accessible before the Minoan eruption. Vaughan (1990), who studied Early Cycladic wares from the Akrotiri excavation, concluded that part of the material was of local origin. The mineralogical investigations, especially the contents of smectite, cristobalite and zeolites of the clinoptilolite-heulandite series, also indicate that the clay is an erosional product, derived from a volcanic source area. The compositions of the sediments around Pillar 17 and in the cave (Fig. 6b) are rather similar to the marine sediments of Akrotiri Village, which outcrop at the church of Agios Epiphanios. They differ from the deposits below the ignimbrite in the Potamos Valley, in which smectitic clay minerals are almost absent (Fig. 6b). We thus conclude that the Pillar 17 clay is an erosional product of the marine sediments from the Archangelos-Loumaravi updomed area on the Akrotiri Peninsula. The clay was washed out and deposited in depressions, such as observed at Pillar 17. Before the Minoan eruption, the area of outcropping marine Plio-Pleistocene sediments must have been larger, as today major parts of the Akrotiri Peninsula are still covered by products of the eruption. The erosional clay products from these sediments were presumably accessible to the prehistoric population at several localities. The clay of Pillar 17 thus gives us an
Fig. 6. The Akrotiri peninsula. Source area of the clay from Pillar 17. The extent of the Cape Riva Ignimbrite and the freshwater drainage system during the Bronze Age is marked as a white hatched area. Ak, Akrotiri Village; Pot, Potamos Valley. (A) Foraminiferal data. The Foraminifera are divided into categories according in part to wall structure (porcellaneous, agglutinated, hyaline; all benthic habitat) and in part according to ecological requirements (deep- and shallow-water benthic species (only for the hyaline forms), and planktonic species), de, deep-water hyaline, benthic Foraminifera; sh, shallow-water and epiphytic hyaline, benthic Foraminifera; po, porcellaneous benthic Foraminifera; ag, agglutinated benthic Foraminifera; ot, other benthic Foraminifera; pi, planktonic Foraminifera. (B) Mineralogical data. Ca, carbonates; Cl, clinoptilolite; Cr, cristobalite; Fe, feldspar; Qu, quartz; It, illite; Ka, kaolinite; Sm, smectite.
78
W. L. FRIEDRICH, M.-S. SEIDENKRANTZ & O. B. NIELSEN
Fig. 7. Some typical Foraminifera. Foraminifera are marine organisms that commonly have a size between 0.1 and 1.0mm. Some forms live on the sea floor (benthic); other species live in the upper water masses (planktonic). The illustration shows scanning electron microscope photographs of benthic (upper row) and planktonic (lower row) forms from the Plio-Pleistocene sediments of the Akrotiri Peninsula.
idea of how clay for pottery making could have that, at the time of their investigations, ruins in been obtained in prehistoric times. the Akrotiri valley were visible in several places. There are several possibilities as to how the This means that erosion had already removed prehistoric Therans could have obtained their much of the covering pumice material. The pottery clay. One or several natural outcrops of general erosion by wind and water may have the clay might have existed in the hills around the been intensified by tsunami activity, which has Akrotiri settlement, or the clay may have been hit the Akrotiri area at least twice since the collected in the lowlands from human-made cav- Minoan eruption: first, the tsunami generated ities, such as freshwater cisterns. Cisterns made from the collapse of the roof of the Minoan for animals in rural district act today as clay magma chamber, producing the 'fourth phase' traps and have to be cleaned from time to time. deposits; and, second, the tsunami generated in Our experiments have shown that the bottom connection with the eruption in AD 1650 (Ross sediment from a cistern for animal-use at Plaka 1840; Friedrich 1994) of the Kolumbo volcano, (Fig. 2) resulted in a good pottery clay. located in the northeastern part of the Santorini volcanic edifice (Fig. 2). Contemporaneous eyeErosion (wind, water and tsunamis). The clay witness reports state that water entered, among at Pillar 17 itself may have been deposited when others, the Akrotiri region (Ross 1840). Tsunami the settlement was still inhabited. However, it sediments of this event are also seen in the may also have been deposited after abandon- Potamos Valley, where rounded pebbles with ment of the settlement because of the Minoan marine, calcareous worm tubes Spirorbis were eruption, as erosion has cut its way through the found (Friedrich 1994). During this long period covering pumice deposits and reached the ruins of erosion, water carrying clay material might of the settlement. Reports of the first discovery have reached cavities in the buried settlement of the Akrotiri site by Mamet & Gorceix and deposited the clay. The dating of the above(18700,6) as well as Fouque (1879) mentioned mentioned charcoal found at Pillar 17 (sample
SANTORINI BEFORE THE MINOAN ERUPTION no. AAR-1566) does not necessarily give the age of the sedimentation process, as the charcoal could have been redeposited. The freshwater drainage system near the Akrotiri settlement before the Minoan eruption The south-facing bay with a shallow beach, where boats could be pulled ashore, was an excellent choice for a settlement for the Bronze Age population. The site also took advantage of the red Cape Riva Ignimbrite. This ignimbrite has been radiocarbon dated to approximately 18000 years BP on the basis of charcoal found underneath and within this rock formation (Eriksen et al 1990; Friedrich 1994). The red ignimbrite was deposited in a preexisting erosional valley. It can be traced from the caldera rim to the Akrotiri excavation (Fig. 6a and b) following the outer down-slope of the volcanic edifice with an angle of c. 8° to the south. Before the Minoan eruption, the houses were built both on top of the ignimbrite and below its distal erosional edge. The grave at Pillar 17 was located at this very edge, dug into the soft alluvial material underneath the ignimbrite. The valley in which the ignimbrite is deposited is now the main route for the flow of rainwater, as the ignimbrite prevents the water from penetrating into the ground. The valley was also the route for transport of rainwater prior to the Minoan eruption; it was part of the main drainage system of the Archangelos-Loumaravi Complex, transporting the erosional products, including the clay, from the hills to the Akrotiri settlement (Fig. 6a and b). The Cape Riva Ignimbrite thus had a very significant influence on the course of the drainage system and thus the freshwater supply. It was presumably, together with the natural harbour, among the important factors deciding the location of the Bronze Age settlement. We would like to thank C. G. Doumas, the leader of the Akrotiri excavation, University of Athens, for the kind permission to take samples in the Akrotiri excavation, and M. Arvanitis, Santorini, for his help in the field. We would also like to express our gratitude to B. Hallager. H. Sigala and A. Frang for their help in literature search, and to S. M. Christiansen, J. G. Nielsen, J. G. Petersen, U. Bjerring, T. K. Rasmussen, B. Winsl0v, M. Dybdahl and U. Viskum for technical assistance. The radiocarbon dating performed by J. Heinemeier and N. Rud is gratefully acknowledged. R. Wilson kindly helped us improve the English of the text. The Carlsberg Foundation (WLF) and the Danish Natural Science Foundation (MSS) funded this study.
79
References ALOUPI, E. & MANIATIS, Y. 1990. Investigation of the technology of manufacture of the local LBA Theran pottery: the body and pigment analysis. In: HARDY, D. A. (ed.) Thera and the Aegean World III, Vol. 2. Thera Foundation, London, 459-469. CLAUSEN, H. B., HAMMER, C. U., HVIDBERG, C. S., DAHL-JENSEN, D. & STEFFENSEN, J. P. 1997. A comparison of the volcanic records over the past 4000 years from the GRIP and DYE 3 Greenland ice cores. Journal of Geophysical Research, 102(C12), 25707-26723. DOUMAS, C. G. 1995. Excavation on Thera (1992). Praktika Archaiologikis Etaireias, 176-188 [in Greek]. 1996. Excavation on Thera (1993). Praktika Archaiologikis Etaireias, 176-178 [in Greek]. DRUITT, T. H. & FRANCAVIGLIA, V. 1990. An ancient caldera cliff line at Phira, and its significance for the topography and geology of Pre-Minoan Santorini. In: HARDY, D. A. (ed.) Thera and the Aegean World HI, Vol. 2. Thera Foundation, London, 362-369. & 1992. Caldera formation on Santorini and the physiography of the islands in the late Bronze Age. Bulletin of Volcanology, 54, 484-493. EINFALT, H.-C. 19780. Chemical and mineralogical investigations of sherds from the Akrotiri excavations. In: DOUMAS, C. (ed.) Thera and the Aegean World I. Thera Foundation, London, 459-469. 1978&. Stone materials in ancient Akrotiri a short compilation. In: DOUMAS, C. (ed.) Thera and the Aegean World I. Thera Foundation, London, 529-527. ERIKSEN, U., FRIEDRICH, W. L., BUCHARDT, B., TAUBER, H. & THOMSEN, M. S. 1990. The Stronghyle Caldera: geological, palaeontological and stable isotope evidence from radiocarbon dated stromatolites from Santorini. In: HARDY, D. A. (ed.) Thera and the Aegean World III, Vol. 2. Thera Foundation, London, 139-150. FOUQUE, F. 1879. Santorin et ses Eruptions. Masson, Paris. FRIEDRICH, W. L. 1994. Feuer im Meer - Vulkanismus und die Naturgeschichte der Insel Santorin. Spektrum, Heidelberg. & DOUMAS, C. G. 1990. Was there local access to certain ores/minerals for the Theran people before the Minoan eruption? An addendum. In: HARDY, D. A. (ed.) Thera and the Aegean World III, Vol 1. Thera Foundation, London, 502-503. , ERIKSEN, U., TAUBER, H., HEINEMEIER, J., RUD, N., THOMSEN, M. S. & BUCHARDT, B. 1988. Existence of a water-filled caldera prior to the Minoan eruption of Santorini, Greece. Naturwissenschaften, 75, 567-569. , PICHLER, H. & KUSSMAUL, S. 1977. Quaternary pyroclastics from Santorini, Greece and their significance for the Mediterranean palaeoclimate. Bulletin of the Geological Society of Denmark, 26, 27-39.
80
W. L. FRIEDRICH, M.-S. SEIDENKRANTZ & O. B. NIELSEN
MAMET, H. & GORCEIX, C. 18700. Archeologie et geologic. Recherches et fouilles. Bulletin de I'Ecole Francaise d'Athenes, 9, 183-191. & 1870&. Archeologie et geologic. Fouilles a Santorin. Bulletin de I'Ecole Francaise d'Athenes, 10, 199-203. MARINATOS, S. 1976. Excavations at Thera VII. Bibliotheke tes en Athenais archaiologikes hetaireios, Athens Ross, L. 1840. Reisen auf den griechischen Inseln des dgdischen Meeres. J. G. Cotta, Stuttgart und Tubingen.
SEIDENKRANTZ, M. S. & FRIEDRICH, W. L. 1993. Santorini, part of the Hellenic Arc: age relationship of its earliest volcanism. Bulletin of the Geological Society of Greece, 28(3), 99-115. STUIVER, M. & RENNER, P. J. 1993. Extended 14C data base and revised CALIB 3.014C age calibration. Radiocarbon, 35, 215-230. VAUGHAN, S. J. 1990. Petrographic analysis of the Early Cycladic wares from Akrotiri, Thera. In: HARDY, D. A. (ed.) Thera and the Aegean World HI, Vol. 1. Thera Foundation, London, 470-487.
The eruption of the Santorini volcano and its effects on Minoan Crete JAN DRIESSEN1 & COLIN F. MACDONALD2 1
Department d'archeologie, Universite Catholique de Louvain, B-1348-Louvain-la-Neuve, Belgium (e-mail:
[email protected]) 2 British School at Athens, 52, Souedias Str., Athens 106 76 Abstract: Sometime in the course of the second millennium BC, an earthquake appears to have triggered a massive eruption of the Santorini volcano. The immediate consequences of the earthquake closely followed by the eruption for Cretan society during the Late Minoan I period are rather difficult to characterize, although physical evidence in the form of Theran ash has shown up at an increasing number of sites. Certain features of the archaeological record, taken in isolation, have hardly been noticed in the past. The long-term effects of the eruption, however, have recently become more comprehensible thanks to a reconsideration of old and new archaeological evidence. The combined picture gives the impression of a period of societal stress following these events. Changes in architecture, storage and food production, artisan output, the distribution of prestige items, administrative patterns and ritual manifestations can be pinpointed archaeologically. These may and should be interpreted as disturbances in the political, economic, cult and security-related domains. It is argued that the inability of the Minoan palatial centres to adapt to changing circumstances caused by a double disaster, an earthquake followed by the eruption of Santorini, led to an increase in crisis-related situations, culminating in the widespread fire destructions which brought this palatial phase of Minoan civilization to an end and opened the way for mainland Mycenaean domination of the Aegean.
Iconographic or literary evidence for natural catastrophes during prehistory is rather scarce. A notable exception is the eruption of the Hasan Dag on the east fringe of the Konya plain, depicted on a seventh-millennium BC mural of £atal Huyiik. Even in historical times, catastrophes seldom seem directly to have influenced artistic creation; another exception is provided by some relief sculpture from a house at Pompeii. These show the effects of the AD 62 earthquake, the traces of which were still very visible in the cities destroyed by the eruption of Vesuvius 17 years later; the fatal mountain itself rarely inspired Roman artists. Sometime during the beginning of the Late Bronze Age, the Aegean witnessed one of the largest eruptions in the history of mankind when the volcanic island of Santorini erupted (see Driessen & Macdonald (1997) for references and details). This caused, amongst other things, the disappearance of the flourishing Cycladic town of Akrotiri beneath metres of ash (Fig. 2) We are not entirely sure of the absolute date of this eruption, as both the second half of the 17th century BC and the middle of the 16th century BC have been suggested. Recent dendrochronological evidence from Porsuk in Turkey, for instance, has been used by the supporters of the high date
(Kuniholm et al. 1996), whereas pumice found in the Nile Delta would rather be suggestive of a 16th-century date (Bietak 1996). The discussion on the absolute date of this eruption is still not closed and new evidence is presented regularly; hence we will probably hear more about the Nisyros eruption, closer to the Turkish coast, which is claimed to have occurred in more or less the same period (Liritzis et al. 1996), and during the London conference the Avellino eruption of Vesuvius was also dated to the 18th or 17th century BC. Moreover, the 1996 campaign at Palaikastro, a Minoan site in East Crete, came upon an extensive, 15 cm thick silt layer suggestive of a water event in an archaeological period dating to the end of the Middle Bronze Age, which would comfortably agree with the high date, suggested by dendrochronology, but is clearly earlier than the Santorini eruption (MacGillivray et al. 1998). At present, there appears to be a stalemate or rather an impasse with regard to the absolute date, and people agree to disagree. In the present paper we are concerned only with the relative date of this eruption and with its archaeological effects. Here we are on firmer ground. The settlement of Akrotiri was clearly destroyed when so-called Late Minoan (LM) IA pottery was in use, a Cretan pottery
From\ McGuiRE, W. G., GRIFFITHS, D. R., HANCOCK, P. L. & STEWART, I. S. (eds) The Archaeology of Geological Catastrophes. Geological Society, London, Special Publications, 171, 81-93. 1-86239-062-2/00/S15.00 © The Geological Society of London 2000.
Fig. 1. Map of Crete with sites mentioned in text.
SANTORINI VOLCANO AND MINOAN CRETE
Fig. 2. Minoanized Cycladic settlement of Akrotiri, Thera.
style characterized by a popularity of floral motifs. The style of the imported pottery found at Akrotiri indicates that this event may have happened rather late in the LM IA period. For many years, this eruption was also blamed for a broad horizon of destructions by fire on Crete (Fig. 1) and for the demise of Minoan palatial civilization. In this scenario, the combined effect of earthquakes, volcanic ash and tsunamis on the island should have caused the downfall of the Cretan palace societies and turned them into easy prey for marauding Mycenaeans from the Greek mainland. On Crete, however, the pottery found in the destruction layers is distinctly later than that found at Akrotiri, and is of a style called Late Minoan IB, the main characteristic of which is the preponderance of marine elements such as octopuses, starfish and argonauts. This implies that the annihilation of Akrotiri on Santorini and the destructions on Crete differed by at least a generation or two. Although some researchers have tried to harmonize the chronology of these two events, excavations of the last decade on Crete and the Dodecanese now make this an untenable sce-
83
nario. The sequence has been clarified and can fall in line with the Akrotiri evidence. First, a serious earthquake occurred sometime before the eruption, as there are signs of earthquake damage followed by clearing and rebuilding at Akrotiri. Then followed the eruption with massive ash fall and the destruction of the Akrotiri settlement. On the island of Rhodes, extensive Theran ash layers have been found up to a metre or more thick (Fig. 3). In East Crete, less substantial but still considerable ash layers have been identified, especially at the sites of Mochlos (see, e.g. Soles et al 1995) and Palaikastro. In these settlements, tephra was found associated with LM IA features and stratified below LM IB architectural features. The ash layers usually vary between 5 and 12cm in thickness, which, allowing for compression, dispersal and bioturbation, would imply an original ash carpet of more than 15 cm, sufficient to cause substantial damage to crops, livestock, buildings and water supplies. In most cases, rebuilding activities or repairs follow the ash fall, and many argue that Crete really witnessed its greatest days after the eruption, during the LM IB period.The eruption, at any rate, cannot have been the direct cause of the demise of the Minoan polities at the end of the LM IB period. What, then, was the cause of Minoan downfall and decline? Civil war, mainland Mycenaean conquest and earthquake, or a combination, are some of the most popular explanations. We suggest that the combined effect of an earthquake and the Santorini eruption in LM IA did indeed cause the eventual demise of Minoan civilization in the sense that these natural catastrophes caused serious crises on the island early in the succeeding Late Minoan IB period, producing a snowball effect that culminated in the destruction of the Minoan palace states. Some earth scientists have shown that there is a causal link between intermediate-depth earthquakes and volcanic eruptions and, in the case of Santorini, a delay of 2-5 years between an earthquake and the eruption has been suggested. There is good evidence for a destructive earthquake sometime before the tephra fell at Trianda on Rhodes, and at Mochlos and Palaikastro on Crete. Indeed, most of the Cretan sites, irrespective of the presence or absence of tephra, illustrate minor or major destructions or abandonments during the LM IA period (Fig. 4). This is why we regard it as highly likely that a major earthquake also triggered the Minoan eruption of Santorini, a hypothesis surely capable of scientific judgement in the near future. Unfortunately, there is still very little geomorphological evidence published that deals with the
84
J. DRIESSEN & C. F. MACDONALD
Fig. 3. Map of the south Aegean indicating general direction of the ash fall.
Fig. 4. Late Minoan IA destructions on Crete.
effects of the eruption on Crete. Minuscule quantities of tephra have been identified at about a dozen Cretan sites as well as in the Eastern Aegean, illustrating the path of the ash cloud. However, except for Mochlos, Palaikastro and the Dodecanesian sites, no good ash layers have yet been identified or published, although we expect this only to be a matter of time. Cores, for instance, in swamp areas close to the shore at Malia, have apparently failed to produce evidence for tephra. No data have been published
to our knowledge corroborating the 1981 study by Karsten and Cita indicating a major tsunami in approximately this period, on the basis of sea-bed anomalies, although at the London conference Marinos informed us that such evidence had been identified in 1996 on the coasts of Asia Minor, where tsunami deposits were overlaid by tephra. Indeed, recent tsunami research seems more concerned with elaborate simulation models than with actual field observations (Monaghan et al. 1994). The evidence of Theran
SANTORINI VOLCANO AND MINOAN CRETE pumice distribution is also better left aside as its taphonomy is debatable. However, as it is beyond doubt that a serious eruption did happen, it can be assumed that a tsunami destroyed some ships, damaged northfacing harbour installations (e.g. Poros, Amnissos and Nirou Chani in Central Crete) and salinated lowlying northern coastal areas, rendering them useless for agriculture for some years. That a major tidal wave did strike the coast is proved to us by Macdonald's personal observations of pumice on the hill west of the Villa of the Lilies. He has found pumice at the 15m contour level on the north side of the hill but no higher. Given erosion since the eruption, we might say that the tsunami that lifted the pumice to that height was over 15m in height but possibly less than 32m, i.e. the height of the hill. It should be noted that a tsunami does not carry pumice from the volcano; pumice ejected in the early phases of the eruption would have been carried, floating on the sea, south by currents and winds so that when the tsunami occurred at the Cretan coast, it was already there and was lifted by the wave(s) and deposited on the Amnissos hill. Tephra will also have affected crops and killed some animals, if not humans. Volcanic tremors may have caused yet more damage to buildings, although we should not expect evidence for a pan-Cretan volcanic earthquake. The abandonment and/or partial destruction of many Late Minoan IA sites could be an immediate consequence of the Santorini eruption, although it seems to us preferable to link this to the earthquake before the eruption. It may also be expected that climatological anomalies ensued, for example a 'volcanic winter', again with disastrous effects on agriculture and the economy in general. The so-called 'Storm' stele of pharaoh Ahmose (mid-late 16th century BC) has been cited in connection with this phenomenon, as it recounts a series of devastating storms early in the XVIIIth Dynasty. If the climatic anomalies referred to in the inscription are indeed a result of the eruption, the situation on Crete may have been devastating although it may have left few immediate traces in the archaeological record. Olive trees, vines and other crops may have suffered, a situation aggravated by any ash fall, which would also have been extremely dangerous for animals as it would have abraded their teeth and clogged their digestive system. Water pollution is also likely to have occurred, again with disastrous consequences, and it should be noted in connection with this that several Late Minoan I sites illustrate how some wells went out of use at this time and other new ones were dug, often protected from the elements
85
in some way. It may be recalled that the Tambora eruption in Indonesia in 1815 caused what has been called the 'last great subsistence crisis in European history' (Arnold 1988), when the cooler and wetter summer led to crop failure, disease, social unrest and famine in Western Europe. Lord Byron even recalled these events in his poem entitled Darkness (McGann 1986). It may be wise at this point to recall some anthropological features of 'disasters'. Essentially, a 'disaster is an event that involves a combination of a potentially destructive agent from the natural or technological environment and a population in a socially and technologically produced condition of vulnerability' (Oliver-Smith 1996). This combination leads to damage of the major social organizational elements and physical facilities of a community to such a degree that the essential functions of the society are interrupted or destroyed. This results in individual and group stress and social disorganization of varying degrees of severity. Disasters, therefore, tend to affect most aspects of community life. There can then be little doubt that on the psychological level, the eruption, from its first rumblings to the final explosive event, must have had an affect on the population of Crete. It is interesting, therefore, to note that that one of the effects of the AD 725 eruption of Santorini was to encourage the Emperor Leo III of Byzantium to remove icons from churches, as he assumed the eruption was a sign of divine wrath. It was the final not the only, reason for the onset of the iconoclastic period in the Byzantine world. A more recent eruption is also worthy of note. In the seven months after the eruption of Mount St Helens in 1980, there were increases of 18% in death rate, 21% in emergency room visits, 200% in stress-aggravated illnesses, 235% in mental illnesses, 45% in domestic violence and 37% in aggression, all mainly caused by the ash fall up to 40 miles away from the volcano (Adams & Adams 1984). If we turn to the purely archaeological data of Late Minoan I Crete, it can be argued that there are several features that may be interpreted as disaster induced and that, combined, may have caused Minoan society to begin to disintegrate. Such features include the following: abandonment of settlements; reduction of occupied space within settlements; decrease or absence of new construction and of new settlement; occurrence of crisis architecture (which implies changes of plan and function as well as construction with spolia and the erection of protective features); abandonment of old wells and the digging of new ones; evidence for an increase in conspicuous consumption by elites
86
J. DRIESSEN & C. F. MACDONALD
(visible in architecture and Palatial Style pottery); hoarding of precious metals and objects; concentration of economic resources (food production, artisan items and their storage become closely guarded in the hands of the elite); ritual excesses, victimization and shifts within religious beliefs; a tendency towards political fragmentation and the emergence of new alliances. Several of these features can be described as posttraumatic stress symptoms and some of these are briefly highlighted below. Of particular importance is the fact that, although many new settlements were founded in
LM IA, none can be said to have been established first in LM IB. Moreover, although there is evidence for reconstruction and repair after an LM IA destruction, very few new buildings can be said to have been constructed during LM IB. Almost all of the Cretan settlements show signs of disruption in LM IA, some not necessarily at its very end (e.g. Vathypetro, Galatas). Several were destroyed, mostly by earthquake, some by fire, or else deserted at this point; others were destroyed or damaged again either still within LM IA (e.g. Galatas and Petras) or in the course of LM IB (e.g. Palaikastro), and yet others only
Fig. 5. Vathypetro settlement of LM IA with later additions (hatched).
SANTORINI VOLCANO AND MINOAN CRETE at the very end of this period (e.g Zakros). Of the 54 settlements occupied in LM IA, only 32 definitely remained occupied in LM IB (c. 60%), with this number dropping to about 10 in LM II, after which time it rises again. Hence, there was a gradual reduction in the number of settlements, albeit with a sudden rise in the number of destructions during LM IB. This process of abandonment must be seen against the general background of settlement contraction and a reduction in the area occupied within settlements during the LM I period. What this drop both in the number of settlements and in the area occupied within settlements actually means in terms of population decline has rarely been addressed. This decline can only have been caused by 'prime movers' such as famine, exile, emigration, warfare, disease and natural catastrophe, as these are more likely to account for abrupt and drastic declines in population and settlement numbers. The abandonment or destruction of settlements during the LM I period surely also affected the overland communication on the island (e.g. abandonment of the Palace at Galatas) and must have had serious consequences with respect to internal and international trade and transport (e.g. destruction of Building J/T at Kommos). Santorini, of course, was at the very least a stepping stone for Minoan trade, and the importance of the elimination of the settlement of Akrotiri should not be underestimated. The hypothetical 'Western String' trade route, connecting Crete with the Argolid (via Melos) and Attica (via Kea), received a serious blow and the void may have been eagerly filled by Mycenaeans. Those settlements that were not destroyed or abandoned in LM IA witness a series of changes during the LM IB period, and it is these changes that are particularly informative. Either because of earthquake damage, or for other reasons, many fine mansions lost their former prestigious appearances. Broad entrances were either blocked or made narrower; large dwellings were subdivided; cheaper materials were used for repairs and fine rooms lost their original appeal. Most surprising perhaps is the fact that the access to many structures became much more difficult during LM IB. This loss of prestige and the restriction of access should be translated in terms of socio-economic changes. A related feature is the fact that during the mature LM I period a series of enclosures were constructed within towns or around countryhouses. This was either for defensive reasons and/or to keep animals close. In addition, several old wells were given up and new ones, some of monumental size, were dug. Most peculiar, however, was the change of room functions.
87
Large public rooms were subdivided and modified into areas for storage or food production. The clearest examples may be briefly mentioned. The fine mansion of Vathypetro (Fig. 5) suffered earthquake damage in LM IA and was then patched up, given an enclosure wall and most of its rooms were modified into storage and agricultural production areas. The mansions at Tylissos seem to have had their entrance systems entirely modified before their LM IB destruction and gave pride of place to storage; the same occurred at Nirou Chani. The fine mansion at Xeri Kara suffered earthquake destruction and was changed into a storage and production unit with the loss of its original architectural prestige; the same happened at Nerokourou and Amnissos. After an LM IA earthquake, the centre of Minoan Palaikastro was not rebuilt but reserved for an large enclosed space in which two new wells were dug, one of which seems to have been for public use. Part of the central area of the Gournia town was also abandoned after its LM IA destruction and other houses were subdivided afterwards to house more people. The major building (J/T) at Kommos with its fine colonnades on either side of a large court was largely abandoned during LM IA, but 'squatters' used the ruins for metal and pottery production thereafter, perhaps still within LM IA. The mansion at Zou, easy of access in LM IA, was enclosured on all sides and access made very difficult. The enclosure probably served to protect animals and a pottery production unit; the same appears to have happened at the Makryghialos mansion. The Villa Reale at Ayia Triada was subdivided some time before its LM IB destruction and its final functions appear especially to have included storage and metal and pottery production. Several of its finer rooms were changed into storage areas. There are many more examples, which continue to stress the same phenomena. The Minoan palaces show a similar pattern. The newly discovered palace at Galatas suffered some damage in LM IA, followed by a 'squatter'-like occupation concentrating on foodprocessing; the palace was then abandoned and later destroyed by an earthquake. LM IB is so far only represented by a destruction in a building outside the palace to the north. Another, smaller palace, recently found at Petras (Fig. 6) had a large open central court and state rooms in LM IA, which were then reduced in size and the rooms used as storage areas. The extent of disruption and change in LM IA at the palaces of Knossos, Phaistos and Malia has not been sufficiently understood, nor the fact that there is
88
J. DRIESSEN & C. F. MACDONALD
Fig. 6. Palatial building or palace at Petras, Siteia.
little evidence for them to continue as fully functional units in LM IB. They have in common that their access systems were modified (Fig. 7), probably after a LM IA destruction; thereafter, occupation appears in a limited scale only, on the basis of the small amount of LM IB ceramic evidence. Access is also restricted at the palaces of Zakros and Gournia during LM IB. In addition, in the Phaistos palace, a pottery production unit was installed on the east and a shrine placed in the West Wing, which was only accessible from outside the palace. The Malia palace also seems to have been largely abandoned before its destruction, but here too a shrine was placed in the South Wing, only accessible from outside. It is possible that Malia's so-called 'silos' to the southwest reflect a storage
area added to the palace only at a very late stage and with the specific purpose of serving the community rather than the palace. The palaces at Malia, Zakros and Phaistos have several blocked doorways, changed circulation patterns and fine rooms changed into storage areas. There are other related features such as the use of cheap building materials, special water provisions, more workshops and enclosed areas. We believe that all these examples illustrate some kind of crisis architecture, an adaptation forced by changing socio-economic conditions. The restrictions at exits and entrances imply a safety concern, and the other changes all seem to relate to making room for either agricultural or artisan production units in existing structures and adding storage rooms. The changes in room
SANTORINI VOLCANO AND MINOAN CRETE
89
Fig. 7. Restricted access after LM IA at Knossos, Gournia, Phaistos, Zakros and Malia.
function illustrate this most eloquently. The increase of storage is thus not a feature accompanying the simultaneous reduction of storage in the palaces at the beginning of the Neopalatial period, as some have claimed (Moody 1987), but a later feature, as some of the palace storage areas were created by modifying the original plan of the buildings. We may ask, therefore, whether local authorities were forced to increase storage capacities because the palaces were unable to allieviate the problems of food shortage by themselves? If so, it not only tells us something about the conditions of the palaces but also illustrates a new phenomenon in LM IB, namely decentralization of power. Indeed, the construction of community storage systems such as the 'silos' at Malia, the Bastione and other areas at Ayia Triada and elsewhere obviously increased the power basis of the local elites. In cult also there are many new features in the mature LM IA and LM IB period, which may perhaps be interpreted as disaster related and, therefore, represent forms of 'crisis cult'. Several, century-old cult forms, such as peak sanctuaries and the so-called lustral basins, were almost entirely given up in the course of this LM IA period. The main 'peak sanctuary' of Central Crete on Mount luktas is the only sanctuary of this kind to yield any evidence of use during LM IB, and so slight is that evidence that it may merely have been visited a few times; it was certainly not functioning as a major cult centre at this time, a phenomenon even clearer
at the other peak sanctuaries of Crete, which are entirely lacking in evidence for use at this time. The formally planned sanctuary of Kato Symi (Fig. 8) appears to have suffered earthquake destruction in LM I; thereafter, although it was rebuilt, its earlier grandeur with processional roadway and central podium was replaced with new constructions of a less impressive and formal nature. This change of plan may well reflect a radical change in worship at the site and a fall-off in palatial visitations. One of the new forms of cult in LM IB is the community shrine. We have mentioned the shrines in the palaces of Phaistos and Malia, which were only accessible from outside the palace and can, therefore, be described as public. At Palaikastro, some rooms of Building 5 (Fig. 9) were blocked off from the rest of the mansion and this part of the building was given an ashlar facade and a separate imposing entrance. The great statue of a young Minoan, made of ivory, gold and other materials (chryselephantine) was found in pieces scattered here and in the public area outside the house. Similar places of community worship can be found at other settlements: House B2 on Mochlos with its two Pillar Crypts, the Shrine on Pseira with its monumental fresco of a seated goddess in relief and its court, and the Gournia Shrine forming the end of a blind alley. That cults were brought into the settlements is further illustrated by the discovery of bronze figurines and inscribed Linear A stone vessels, votive objects previously found only in rural shrines.
90
J. DRIESSEN & C. F. MACDONALD
Fig. 8. Formal arrangement of the Kato Symi sanctuary with later modifications (hatched).
Fig. 9. Part of Building 5 at Palaikastro converted into a community shrine in Late Minoan IB.
We may add here that there are examples in which volcanic pumice was included in ritual offerings; for example, in a foundation deposit for architectural alterations beneath a threshold of the mansion at Nirou Chani. Horns of consecration of this period have been found at Palaikastro, Zakros, Amnissos and Petras.
New forms of cult appear making greater use of sacred symbols and here the new Marine Style, the ceramic hallmark of the LM IB period, should be stressed. Its sudden appearance demands a rational explanation. Marine style pottery is definitely a distinct elite style, a 'palatial' style. This is demonstrated not only
SANTORINI VOLCANO AND MINOAN CRETE by the high quality of the product but also by its specific find places. It appears to be the demonstration par excellence of elevated status and group affiliation. We suggest that its production also filled a sudden gap within the corpus of propaganda-laden objets d'art in other media such as stone and ivory. However, its cultic associations, we feel, almost certainly derive directly from the 'Santorini experience', displaying a new awareness of the power of the sea, as both the eruption, visible from Crete, and the ensuing tsunami will have had a profound effect on the Minoan psyche, following so closely on the heels of the earthquake that caused the eruption in the first place. There may even have been a certain feeling that the true origins of earthquakes lay within the sea itself. Palatial ceramic styles, of which the Marine Style is one aspect, bring us to changes in arts and crafts during the LM I period. It is largely in this area that it is possible to argue for & floruit of Neopalatial Crete in the LM IB period, as the greatest corpus of fine, valuable objects comes from LM IB destruction levels. A priori, the link between fine objets d'art and prosperity is often a tenuous one. For example, using very broad and crude comparisons, Switzerland is a very prosperous country yet, except for cuckoo clocks, it has rarely produced major works of art, whereas the Italian renaissance was an unstable and war-stricken period during which numerous masterpieces were produced. Minoan workshops had been scattered throughout the settlements in LM I A, whereas, in LM IB, they become prominent features of palatial and major buildings or replace fine houses; for example, the kilns west of the Stratigraphical Museum at Knossos. It seems that specialization in arts and crafts begins to occur in previously prestigious surroundings and may be under tighter control by the elites. Under this heading, we return to Palatial LM IB styles of pottery, which were not only
91
under the control of the palatial elite but also served their purposes, perhaps even in the dissemination of new religious aspects to proclaim that religious control remained a palatial preserve. That said, we must still ask to which period should the manufacture of the finest and most valuable objects be assigned. It is possible to argue that the floruit of production of elite objects, with the notable exception of ceramics, was during LM IA, on the basis of the amazing Minoan wealth of the LM IA-LH I Shaft Graves at Mycenae and the finest objects from Akrotiri on Thera. It is interesting to note that not a single stone vase rhyton was complete at the time of the LM IB destructions. This is not to say that all production ceased in LM IB, but rather there was a decline in the production of valuable objects, heirlooms became more valued, even when damaged, and Palatial Style ceramics were brought in to fill the gap and press changes in cult emphasis. The great fresco works of Knossos belong to the 'Great Rebuilding' in LM IA and not to LM IB. Certain LM IB exceptions appear to challenge our hypothesis, notably, the monumental frescoes from Room 14 in the Villa Reale at Ayia Triada and the relief fresco of a seated goddess in the Pseira Shrine. However, these appear to us to reinforce the decentralization of Crete during LM IB, when local elites sought to bolster their power base through palatial emulation. Abandonment and destruction deposits of the mature LM IA and LM IB periods are also informative on a different level. It has been noted before (Georgiou 1979) (Fig. 10) that there are a surprising number of bronze hoards or hidden collections of valuables in Late Minoan I contexts compared with earlier and later periods on Crete: daggers hidden beneath pithoi, copper ingots walled-up at Ayia Triada, three bronze hoards of tools and weapons at Gournia,
Fig. 10. Distribution of bronzes in Late Minoan I contexts on Crete.
92
J. DRIESSEN & C. F. MACDONALD
Fig. 11. Distribution of LM I administrative documents on Crete. bronze basins hidden in several places at Mochlos, weapons hidden both in the Malia palace and some of the houses, weapons and bronzes hidden at Palaikastro and Zakros, and bronze vases and tools hidden in several houses destroyed in mature LM IA at Knossos. These hoards seem to reflect an unsettled period, which had already begun before the end of LM IA. We have mentioned that many settlements increased their storage and artisan-production capacities by LM IB. One of the consequences of this decentralization would have been an increase of local administration. It is not surprising, therefore, that there is a proliferation of administrative documents throughout Crete in LM IB, both of Linear A tablets and sealings (Fig. 11). Surprisingly, the settlements with palaces yielded few or no administrative documents. This contrasts with the later picture of Linear B and argues for a notable degree of decentralization during LM I instead of the later single integrated economic system with the palaces as capitals. We argue that the local centralization of storage systems, industrial units and administrative records are all part of a single phenomenon, namely, the rise of powerful local eonomic systems because of failure within the earlier central, palatial administrations. On a wider level, LM IB is also the period during which the Minoans lose their influence in the Cycladic and Dodecanesian islands. All of this began with the eruption of Santorini and the destruction of the Minoan Cycladic base, Akrotiri. Conclusion The following historical reconstruction seems to us to be the most appropriate model, based on our interpretation of the evidence. The archaeological data suggest that there was a severe
economic dislocation in Crete, triggered by the Santorini eruption, and that this dislocation gradually worsened as LM I progressed. Moreover, a combination of a general feeling of uncertainty caused by the eruption and its accompanying effects, the earlier destructions caused by earthquake, and the need to rebuild and re-establish normal economic life, may have presaged the end of the LM IA centralized sociopolitical and economic system and created the suspicion that the existing palatial elites had lost their divine support. Because of problems with food production and distribution, the existing network disintegrated, resulting in a decentralization or fragmentation of the political landscape, which went hand in hand with an increase in elitist power and competition. In all probability, the latent tension between this and the demonstrable tendency to exclude those not of the group (by hoarding, segregation measures and storage) would have eventually led to conflict. That this did occur is indicated by the enormous number of conflagrations and selective destructions of houses, villas and palaces in the LM IB period. In such crises, there is a widespread tendency to explain disasters in terms of the sins of the people and one often looks for scapegoats. Although disasters can be somewhat even-handed in distributive destruction, affecting all levels of society in various ways, the recovery process is not at all egalitarian. Conflict in the post-emergency or rehabilitation period of a disaster tends to arise in two areas, namely, the allocation of blame and the allocation of resources for rehabilitation. The changes in architecture, production, storage, cult, and arts and crafts discussed above may well all have been triggered by the situation of conflict and tension existing after the eruption. Famine would have led to strife and years of internal unrest, population movement and religious
SANTORINI VOLCANO AND MINOAN CRETE upheaval. This is why we feel justified in assuming that the Santorini eruption affected almost every aspect of society, producing responses that, in all probability, created a chain reaction leading to fundamental changes. Many of the features highlighted resulted from post-eruption stress. During LM IB, Crete's main preoccupation seems to have become the protection of food production, storage, livestock and industrial production through local centralization. Indeed, it appears that the old nodal points, the palaces of LM IA, were not capable of offering an adequate response to the immediate problems. This inadequacy resulted in secondary centres being allowed (or, perhaps assuming) much greater regional control, which eventually led to a political fragmentation. The Santorini eruption is here given the role of a precipitant or catalyst, triggering an entire series of changes that culminated in Crete being absorbed to a greater or lesser extent into the Mycenaean, and thereafter, the Greek world. References ADAMS, P. R. & ADAMS, G. R. 1984. Mount Saint Helens ash fall. Evidence for disaster stress reaction. American Psychologist, 39, 252-263. ARNOLD, D. 1988. Famine, Social Crisis and Historical Change. Oxford, 30. BIETAK, M. 1996. Avaris. The Capital of the Hyksos. Recent Excavations at Tell el-Dab'a. British Museum Publications, London, 77-78.
93
DRIESSEN, J. & MACDONALD, C. F. 1997. The Troubled Island. Minoan Crete before and after the Santorini Eruption. Aegaeum, 17. GEORGIOU, H. 1979. The Late Minoan I Destruction of Crete. Metal Groups and Stratigraphic Considerations. Los Angeles. KUNIHOLM, P. I., KROMER, B., MANNING, S. W., NEWTON, M., LATINI, C. E. & BRUCE, M. J. 1996. Anatolian tree rings and the absolute chronology of the Eastern Mediterranean, 2220-71 SBC. Nature, 381, 780-783. LIRITZIS, I., MICHAEL, C. & GALLOWAY, R. B. 1996. A significant Aegean volcanic eruption during the second millennium BC revealed by theromoluminescence dating. Geoarchaeology, 11, 361-371. MACGlLLIVRAY, J. A., SACKETT, L. H. & DRIESSEN, J.
1998. Excavations at Palaikastro, 1994 and 1996. Annual of the British School at Athens, 93,221-268. McGANN, J. J. (ed.) 1986. Lord Byron. The Complete Poetical Works IV. Oxford, 41. MONAGHAN, J. J., BlCKNESS, P. J. & HUMBLE, R. J.
1994. Volcanoes, tsunamis and the demise of the Minoans. Physica D, 77, 217-228. MOODY, J. 1987. The Minoan palaces as a prestige artifact. In: HAGG, R. & MARINATOS, N. (eds) The Function of the Minoan Palaces. Svenska Institutet i Athen, Stockholm, 240-241. OLIVER-SMITH, A. 1996. Anthropological research on hazards and disasters. Annual Review of Anthropology, 25, 303-328. SOLES, J. S., TAYLOR, S. R. & VITALIANO, C. J. 1995. Tephra samples from Mochlos and their chronological implications for Neopalatial Crete. Archaeometry, 37, 385-393.
This page intentionally left blank
Late Minoan IB marine ware, the marine environment of the Aegean, and the Bronze Age eruption of the Thera volcano PETER BICKNELL Department of Classics and Archaeology, Monash University, Clayton, Vic. 3168, Australia Abstract: Late Minoan IB fine ware pottery includes a number of decorative styles. The most spectacular of these is characterized by motifs, hitherto only rarely deployed by Cretan vase painters, drawn from the marine world. Late Minoan IB marine ware turns up in ritual contexts, which include human sacrifice. The pottery style is likely to reflect, then, not simply a vagary of secular fashion, but a circumstance or circumstances requiring far-reaching religious attention. It is proposed that Late Minoan IB marine ware and the cult activities in which it was deployed were a response to negative effects of the Late Bronze Age Thera eruption on the marine environment of the Aegean.
At a point during the Late Bronze Age falling within the long reign, c. 1500-1450 BC, of the Egyptian Pharaoh Thutmose III (Wachsmann 1987, pp. 127-130) Mycenean invaders from the Greek mainland established control over Minoan Crete. Their appearance in the island is reflected in various ways, not least by the decorative style, motifs and shapes (in particular, the so-called Ephyraean goblet and squat alabastron) of Late Minoan II pottery (Betancourt 1985, p. 159). Late Minoan IPs ceramic predecessors, Late Minoan IA and IB, are still essentially Minoan in spirit and design. Although the content of the former is relatively dull and circumscribed, the artistic quality of some vessels, of which a group found in room C58 at Gournia (for location of sites mentioned in this paper, see Fig. 1) is a case especially in point, is extremely high. The Late Minoan IA vocabulary includes heraldic motifs such as the double axe and bull's head, others drawn from the plant world, and various abstract spirals, wheels, ripples and meanders (Popham 1967; Betancourt 1985: pp. 128-133). Late Minoan IB, the last ware produced while the Minoans were still masters of Crete, is far more complex. There are two overarching categories (Betancourt 1985, pp. 137-158). The first is the so-called standard tradition, relatively unexciting and uninnovative, which continues at the outset to exploit Late Minoan IA motifs. The second comprises stunning fine ware vessels that come in multiple concurrent varieties, each dominated by a different decorative mode. The principal modes are the floral, marine
and geometric-abstract. To judge from frequent incorporation of marine co-motifs and submotifs into other modes (in the case of one such hybrid, from Phaistos, for example, the base features a marine motif whereas on the body abstract rosettes and zigzags alternate), the marine mode was the most significant (Mountjoy 1984). Exemplars are certainly more abundant than those of its floral and geometric-abstract counterparts. The period, no more at most than a single generation (Popham 1990), during which Late Minoan IB pottery was in vogue turned out to be one of major disaster for the Minoans. Almost every significant site on Crete was terminally or temporarily destroyed, with violent conflagrations the normal accompaniment of devastation. Palace complexes such as those at Phaistos, Mallia and Kato Zakro, prosperous townships such as Gournia and Palaikastro, and rural mansions, with those at Sklavokambos, Nirou Khani and Makrygialos typical examples, were equally affected (Page 1970, pp. 1-12; Betancourt 1985, pp. 133-139). Almost, but not quite, exceptionally the largest palace complex, that at Knossos, survived relatively unscathed to become, in due course, the administrative centre of the Myceneans. Many of the structures around it, however, were incinerated like entire sites elsewhere (Wall et al 1986). Convincing explanation of the Late Minoan IB holocaust in Crete remains the greatest of the challenges confronting the historian of the Bronze Age Aegean. Significant involvement of the Late Bronze Age eruption of the Thera
From: McGuiRE, W. G., GRIFFITHS, D. R., HANCOCK, P. L. & STEWART, I. S. (eds) The Archaeology of Geological Catastrophes. Geological Society, London, Special Publications, 171, 95-103. 1-86239-062-2/OO/S 15.00 © The Geological Society of London 2000.
96
P. BICKNELL
Fig. 1. Location of sites mentioned in text.
volcano has been routinely discounted (Doumas 1983, pp. 144-147; Barber 1987, pp. 221-222; Dickinson 1994, p. 304), as the stratification on Crete of distal ashfall from the paroxysm put it beyond doubt that the event occurred while Late Minoan IA pottery was still in use (Warren 1991; Soles et al 1995; Soles & Davaras 1996). At first sight attractive attempts to inculpate rampaging Mycenean invaders have been devastatingly criticized on various grounds by Doro Levi (see Page 1970, p. 12) and others (Dickinson 1994, p. 304), and are not comfortably compatible with the fact that at least a decade may separate the Late Minoan IB destructions at Phaistos and peripheral Knossos from those at Palaikastro and Kato Zakro (Downey & Tarling 1984). It is as likely as not that multiple causes contributed to Minoan decline. In what follows, a new entry point into a labyrinth of problems is attempted by way of the motifs and context of Late Minoan IB marine style pottery. Although not immediately and overtly implicated, as once envisaged (Marinatos 1939), in the Late Minoan IB vicissitude of Crete, the Thera volcano, it will emerge, may have made at least one significant indirect contribution to conditions that paved the way for Mycenean takeover.
Late Minoan IB marine ware; motifs and context By 1984, when Elizabeth Mountjoy published a corpus of all Late Minoan IB marine ware vessels, together with marine-floral and marinegeometric-abstract hybrids, known to her, 19 sites in Crete were represented. Subsequent discoveries have brought to light further exemplars at both the same and additional sites, such as Mochlos and Poros 'Knossos' port. Possibly manufactured in a travelling workshop based on the Knossos palace (Betancourt 1985, p. 140), marine style vessels were not only distributed throughout Crete but found their way to Minoan outliers such as Kastri, Kythera and Rhodian Trianda and beyond to Egypt and the Levant. Several of the Late Minoan IB motifs were imitated or adapted in an often extravagant and baroque manner in the Cylades islands and on the Greek mainland, and, to return to Crete, an artistically degenerating selection is part of the vocabulary of Late Minoan II ware and its successors, all produced during the period of Mycenean domination. Late Minoan IB marine ware comes in a wide range of pleasing shapes and features sea scenes usually, but always underwater. Main
MINOAN MARINE WARE AND THE THERA ERUPTION and sub-motifs are often densely crowded provoking diagnosis of horror vacui on the part of artists concerned (Betancourt 1985, p. 145). An initial appearance of naturalism turns out on closer inspection to be a mirage. Counterfactual stylization of some motifs is the rule and by way of more fundamental departure from realism, the core Minoan heraldic emblem of the double axe and an enigmatic stellate object (see Fig. 2) occasionally intrude amongst marine fauna, flora and rocks. Also, it may be necessary to acknowledge instances of disregard of scale. Two classes of marine motif proper can be distinguished, main and background filling. The latter includes rocks, sometimes covered with a minute cellular pattern that may be intended to evoke coral (Fig. 2), water bubbles, seaweed of different types and a species of sea urchin (Fig. 2, 4 and 5) probably to be identified as the rockdwelling Paracentrotus lividus, whose 3cm long spines protrude from a test some 6cm in diameter. If, with Mountjoy (1984), we exclude fish depictions on isolated sherds as post Late Minoan IB innovations, there are four main motifs. First, and less common than its counterparts, is the common dolphin, Delphinus delphis (Fig. 2, 3). Much more frequent are two cephalopods and a mollusc of the gastropod class with whorled shell. Despite the single rather than double row of suckers on the tentacles, the octopus that appears, for example, on vessels from Gournia and Palaikastro (Fig. 2, 4 and 5) must be identified as Octopus vulgaris. Configuration in general and absence of webbing between the tentacles at their base in particular rule out Eledone moschata and Eledone cirrosa. The other cephalopod (Fig. 2, 1) is the so-called paper nautilus, Argonautica argo. The standard representation of this pelagic, nocturnal and usually submarine creature, with three arms emerging from a closed region of a distorted shell, is remote indeed from realism. The painters concerned either failed to understand, or, prompted by aesthetic considerations, chose to ignore, the true relationship between the cephalopod and its shell, and also the manner of its motion. The gastropod (Fig. 2, 4 and 5) has been variously identified. Influenced by the spiny protrusions, Marinates (Marinates & Hirmer 1960, p. 76) confidently identified it as one of the purple dye, producing murices, either Murex trunculus or Murex brandaris, both of which attain a maximum height of 8-9 cm. If Marinatos is right, the sharply demarcated and tapering secondary whorls, not a feature of either of the two species, must be regarded as yet another example of departure from realism.
97
Strikingly, and remarkably, the late fifth century die-cutters of Tyre, eventual eastern Mediterranean chief centre of the purple industry, depicted one of the two dye-yielding murices (probably M. trunculus, given that vast dumps of expended shells of this species lie close to Tyre's remains) similarly modified on the reverse of the shekel. Taking considerations of relative scale more seriously, Mountjoy (1984) pronounced the gastropod a triton without discussion. Gill (1985) conceived of insinuation of the trumpet triton, either Charonia variegata or Charonia nodifera, both of which are capable of reaching a length well above 40 cm, but actual representation, for aesthetic reasons, of the more distinctive, if smaller (up to 18cm in length), ranella, Ranella gigantea. Much earlier, Bosanquet (1904) and Sir Arthur Evans (1928, pp.306 and 316) tried to have it both ways by opting for a deliberately contrived conflation of murex and triton. In different ways both gastropods have a high Minoan profile. Evidence is abundant that the triton played a prominent role in religious ritual. On a lentoid seal from the Idaian cave a priestess is depicted holding a trumpet shell aloft with her left hand (Evans 1935, p. 344). Actual shells have been found in shrines and other cult areas at Knossos, Mallia, Palaikastro, Phaistos, Pseira, Pyrgos and Kato Zakro (Reese 1990). Minoan exploitation of the murex for dye production had a long history. Large-scale production during the protopalatial period is attested by substantial dumps of processed shells on the offshore island of Kouphonisi (Bosanquet 1904; Stieglitz 1994). Eventually, the centre of gravity of the industry shifted to the neighbourhood of Palaikastro, whose again impressive shell dumps represent a time-span extending from the protopalatial period through to the neopalatial Late Minoan IB period (Hood 1971, p. 94; Reese 1987). Neopalatial purple manufacture on a smaller scale is reflected, for example, by plentiful crushed Murex trunculus fragments in the remains of the coastal mansion at Makrygialos (Reese 1987). Final choice between triton and murex for identification of the Late Minoan IB marine ware gastropod motif is not easy, but the absence of extended pointed protrusions from Charonia nodifera and Ranella gigantea, and the crowded multiple molluscs depicted on some rhyta from Palaikastro tip the scales, to my mind, in favour of the murex. As Aristotle (History of Animals, 5. 15) was first to record, murices gather together in large numbers in spring. Tritons are consistently more solitary. Before the neopalatial period, the context of Minoan representations of marine creatures and
98
P. BICKNELL
Fig. 2. Late Minoan IB vessels: 1, provenance unknown; 2 and 5, Palaikastro; 3, Pseira; 4, Gournia.
of marine paraphernalia in general is unmistakably cultic. The religious association of the triton shell, noted above, is one case in point. In a religious sanctuary uncovered at Anemospilia pebbles from the seashore were placed upon a stepped altar (Sakellarakis & SapounaSakellaraki 1991). Protopalatial faience models of paper nautilus shells and other sea creatures, together with hundreds of actual seashells, colourfully painted, were found in repositories in the Knossos palace in the company of representations of a female deity (Evans 1921,
p. 520). Expectation of a similar setting for Late Minoan IB vessels decorated with marine motifs is born out by the results of a seminal study by Mountjoy published in 1985. In the paper concerned, Mountjoy noted four separate associations at Knossos of Late Minoan IB marine ware with areas clearly used for ritual purposes. She went on to record two further such juxtapositions at Palaikastro and single ones at Archanes close to Knossos, Gournia, Nirou Khani, Pseira, Pyrgos, Tylissos and Kato Zakro. To these sites we can probably
MINOAN MARINE WARE AND THE THERA ERUPTION add building B2 recently excavated at Mochlos (Soles & Davaras 1996). At Knossos one of the associations is especially dramatic. Remains of five Late Minoan IB marine ware vessels were discovered in the 1980s in the debris of an incinerated structure, one of the casualties of the Late Minoan IB period of disaster, that has been labelled the House of the Children's Bones. Here, on what is by far the most coherent and plausible interpretation of the evidence overall (Wall et al 1986), after ritual sacrifice and consumption by priests and votaries of portions of their flesh, stripped, disarticulated bones of three or four young children were deposited in a small basement room inaccessible to predators. The proximity of the isolated skull of a child to ritual appurtenances, including Late Minoan IB marine style vases, in the west wing of the palace complex at Kato Zakro (Platon 1971, p. 120) and a similar concatenation (here the skull is that of a young woman) in house B2 at Mochlos (Soles & Davaras, 1996) suggest that human sacrifice during the Late Minoan IB period was not confined to a single site. If Nilsson (1950, pp. 194-235) was right in insisting upon a sacrificial ambience for the double axe that pervades Minoan iconography, its occasional appearance amid the underwater scenery depicted on Late Minoan IB marine ware is readily explained. The provenance of two examples is the House of the Children's Bones (Mountjoy 1984). The Late Bronze Age Thera eruption and the marine environment of the Aegean In the case of the relatively mass representational medium of vase painting, Minoan deployment of motifs drawn from the marine world is for long both rare and sporadic. Occasional vessels featuring tunny, stylized octopuses or dolphins turn up in Middle Minoan II and III ware, the first entirely protopalatial, the latter extending into the new palace period. Within the decorative repertoire of the subsequent Late Minoan IA pottery, produced, in the opinion of Popham (1990) over a time-span approaching 75 years, we encounter not a single instance of a motif drawn from the coastal or open sea. Such prolonged dearth makes the proliferation and profusion of marine motifs in the dominant variety of Late Minoan IB all the more startling. The most conspicuous valency of Late Minoan IB marine ware, to reiterate, is religious. Rhyta and other types of vessel displaying sea creatures and marine environmental features were employed in, or in association with, ritual acts
99
that included human sacrifice. This particular conjunction is as startling as the sudden prominence of marine motifs. Provided one follows Hughes (1991, pp. 13-17) in discounting premature claims (Sakellarakis & Sapouna-Sakellaraki 1981, 1991) about discoveries in the early 1980s at Anemospilia, there is no evidence for earlier Minoan recourse to this kind of offering. Other things being equal, one would expect the decorative motifs of vessels chosen for ritual purposes to reflect the ambience and purpose of the rites concerned. Late Minoan IB marine ware, it ought to follow, was employed in ceremonies somehow connected with marine creatures and their habitat. These ceremonies, performed with the intent of influencing the deities who presided over the sea and all within it, included the sacrifice of human beings. To account for unprecedentedly intense cultic preoccupation with the sea, preoccupation sufficiently urgent to dictate recourse to the ultimate expedient in the ancient world for placating supposedly affronted gods, we are compelled to envisage at the outset of the period during which Late Minoan IB pottery was produced some profound and catastrophic disturbance of the local marine environment. The context and stratigraphy of ash fall-out at the Mirabello Bay site of Mochlos put it beyond reasonable doubt that the Late Bronze Age eruption of the Thera volcano took place at the very end of the Late Minoan IA pottery phase. Having made this point, Soles and his coinvestigators went on to note without further elaboration that it was not uncommon for natural catastrophes in the Aegean to be associated with changes in pottery style (Soles et al. 1995). From the religiously conditioned perspective of early peoples, natural catastrophes reflect extreme displeasure of the god or gods who preside over the sphere in which disaster takes place. Given the close juxtaposition that Soleg and his colleagues emphasized, it is difficult to resist the temptation to bring into close connection the Thera paroxysm and the marine upheaval to which Late IB marine style pottery and human sacrifice were religious responses. The first explicit attempt, as far as I am aware, to associate Late Minoan IB marine ware with the Thera eruption is that of Wilson in a popular book published in 1985. After observing that addition of marine motifs to the floral and heraldic ones that were the staple of Late Minoan IA pottery is striking, Wilson (1985, pp. 139-141) proceeded to draw his readers' attention to the description of the Late Roman imperial historian Ammianus Marcellinus (c. AD 325-395) of the great Mediterranean
100
P. BICKNELL
seismogenic tsunami of AD 365. In Ammianus' words (26. 10. 15; I have modified the translation Wilson uses): The sea with its swelling waves was driven back and withdrew from the land, so that ... many kinds of marine creatures were revealed in the mud and numerous ships were stranded Many people roamed about without qualms in the shallow remainder of the water in order to gather fish and similar creatures with their bare hands. Then the roaring sea ... gathered up and crossing the now seething former shallows, hurled itself on islands and broad stretches of mainland and levelled countless structures — The great volume of water killed many thousands of people.' Have presented this account, Wilson suggested that the Minoans introduced depictions of sea life on to their pottery in the wake of devastation of Crete's northern coast by devastating tsunamis propagated in the course of the Thera eruption. The giant waves were interpreted by their victims as punishment for some offence and the motifs of Late Minoan IB marine ware were a contribution to expiation. The main problem with this scenario is that there is considerable doubt as to whether any significant Thera-generated tsunamis were directed towards Crete. Formidably deterrent to belief in tsunami intrusion anywhere along the island's northern coast are a seemingly ubiquitous absence of marine-bed boulders deposited above storm tide levels (Dominey-Howes 1996, pp. 45-93) and lack of any convincing indication in the archaeological record of any Minoan site of any effects that could be ascribed unequivocally, or even with a reasonable degree of plausibility, to tsunami damage. It is true that Francaviglia (1990) reported recovery of tsunami-emplaced pumice at Amnisos from the hillslope, 10-15m high, behind a villa excavated by Marinates, but caution, I believe is called for. Francaviglia's claim appears to be in conflict with the representations of Marinates himself, who reported finding pumice only in the villa itself close to the sea (Dominey-Howes 1996, p. 87), and early in 1997 I failed to detect any trace of pumice on the hill despite persistent searching. In a nutshell, although it is just conceivable that evidence of large-scale Late Bronze Age tsunami incursion into Crete is still to emerge (by way, for example, of telltale sediment in cores extracted from coastal locales), current indications are that Minoan observation of a temporarily denuded sea bed, demolition by inundation of major sites in northern Crete, and longer-term detrimental effects, such as salination, of huge volcanogenic waves are unlikely to have provided the stimulus for Late Minoan IB pottery's marine motifs.
Wilson's approach ruled out, it becomes necessary to look for different connections between the Thera eruption and adverse conditions of the Aegean marine environment that dictated a dramatic change of Minoan pottery style. Further consideration of the motifs of Late Minoan IB marine ware may be a helpful preliminary to establishing what these might have been. Commenting on marine motifs in general in Minoan representational art and recognizing their basic cultic valency, Gill (1985) emphasized that they are likely to have been selected for practical rather than aesthetic reasons. The purpose of choice will have been to ensure continuity of harvests from the sea by depiction of those creatures most useful to human beings, and of the animal attendants or symbols of the god or gods who presided over that aspect of Minoan life. All four main sea creature motifs of Late Minoan IB marine ware together with the background filling sea urchin are eminently compatible in one way or another with their profile. A curious idea, strangely shared by many, that the Minoans avoided culinary use of products from the sea is comprehensively refuted by data assembled by Guest-Papamanoli (1983) and Powell (1996). There is little reason to doubt, it emerges, that the common octopus and the sea urchin Paracentrotus lividus, would have been protein sources as significant for the Minoans as they were for Crete's Greek occupants of the Classical period (Thompson 1947, pp.72 and 208). Although the Classical Greeks by and large refrained from hunting and eating cetaceans, a cylinder seal depicting dolphins hung from poles, like fish, to dry, indicates that Delphinus delphis contributed to the diet of their Minoan predecessors (Gill 1985). The murex provided food in Classical times (Thompson 1947, p. 218), and in this respect too the Minoans certainly anticipated those who came after them. Detritus from the so-called unexplored mansion at Knossos, a purely residential structure certainly unconnected with dye processing, included a few shells of Murex trunculus alongside those of other edible species and Paracentrotus tests (Popham 1984, pp. 246-256). Any dietary significance of the murex, however, would have been outweighed for the Minoans by its major importance in the purple industry. Given the evident extent of operations at Kouphonisi and then in the Palaikastro area, the dye produced ought to have been well in excess of local requirements. In all probability, purple was exported by the Minoans in large quantities. To conclude this conspectus with the paper
MINOAN MARINE WARE AND THE THERA ERUPTION nautilus, although it certainly could have been eaten by the Minoans, its comparative rarity rules out its having been a dietary staple. The pseudo-shell's function as cult object emerged in the previous section and it is possible, in addition, that specimens were used locally and exported elsewhere as decorative items. All but one of the sea creatures depicted on Late Minoan IB marine vessels, it can be said with confidence, are likely to have contributed significantly, or at least not negligibly to the Minoans' food supply. In particular, as well as being edible the murex was the source of a luxury product that may have been a key export item. Given such ambience, it is logical to seek to bring into connection with the pottery style and the conditions that determined its motifs indisputable effects or by-products of the Late Bronze Age Thera eruption inevitably deleterious to marine fauna. Prime candidates are floating pumice and distal ash fall. To begin with the former, the first paroxysmic phase of the Thera eruption resulted in the emplacement of pumice metres thick over the whole of the convulsed island. As in similar circumstances at Krakatoa in 1883, massive amounts will have been deposited into the surrounding sea. Again as in the case of the Krakatoa eruption, both discrete lumps of pumice of various sizes and rafts of accumulated pieces, some coherent and thick enough to support persons prepared to walk on them, will have drifted in all directions, ending up in coastal areas in some instances and remaining afloat in open water for years in others (Bullard 1976, p. 46; Simkin & Fiske 1983, p. 15, 91-95, 200-201 and 211). As well as presenting a major obstruction and hazard to fishing vessels, such seaborne tephra can be directly detrimental to a variety of marine life. In the wake of the Alaskan Katmai-Novarupta eruption of 1912, for example, floating mats of pumice frightened halibut from fishing grounds around Kodiak Island (Erskine 1962, pp. 204-208). Pumice projected into the sea during the 1952 Barcena (Mexico) eruption had more extensive effects on the marine environment, with intertidal fishes and marine invertebrates seriously depleted as a result of the material's abrasive action (Brattstrom 1963). Copious Thera-derived pumice should have been similarly deterrent to fish in the open sea near Crete and injurious to denizens, vertebrate and invertebrate, of coastal waters, especially, but far from only, to the north of the island. Given a natural temptation, stressed above, to treat projections as facts, it would be welcome to be presented with direct indication of Minoan
101
awareness of and concern about floating Thera pumice. It is just possible that such evidence exists. The third vessel depicted in the illustration is one of a pair of identical pear-shaped, Late Minoan IB marine ware rhyta discovered in the remains of a Minoan township on the offshore island of Pseira in the Mirabello Bay area of northern Crete. The dolphins, the main motif, are surrounded by a reticulate pattern which, as it does not cover the cetaceans themselves, cannot represent an entrapping net. In fact, the pattern, as Morgan (1988, p. 35) and others have explained, is a Minoan artistic convention for representation of the surface of the sea. Normally the modules of the sea connoting mesh are either empty, or contain a single central circle, a central or sub-central curved line, or a couple of unobtrusive dashes. Net elements packed with dots are unique for Minoan Crete and, as far as I am aware, paralleled only in parts of the complex, rococo ornamentation of vessels from Ayia Irini and Phylakopi in the Cyclades, and Vaphio on the Greek mainland, all three of which reflect imitation of a variety of Minoan models (Bosanquet 1904; Mountjoy 1984). The dot-clogged sea cells of the Pseira vases, I hesitantly suggest, are not an aberrant symptom of a painter's abhorrence of empty space, but by way of illustration of or allusion to a sea cluttered in the immediate aftermath of the Thera eruption with floating pumice, a circumstance leading to an absence of dolphins, an important Minoan food source. If such is the case, the representations on the two rhyta are apotropaic: a condition is depicted that it is hoped the deity or deities concerned will consent to bring to an end. In contrast, the majority of Late Minoan IB vases display underwater scenes in which cepholopods, gastropods and echinoderms flourish in evidently clear water. Here, by way of converse sympathetic magic, conditions are portrayed that have ceased to be the case and that the gods are solicited to restore. Distal Thera ash fall into the sea, as well as pumice floating on its surface, would also have been capable of a substantial contribution to marine environmental degradation sufficient to evoke extreme religious countermeasures. As Blong (1984, p. 338) observed, tephra falls of 3cm or less into water have been associated with the mortality of its inhabitants, with various specific circumstances such as particle size, heavy metal constituents or attached aerosols influencing the extent and nature of effects, which may be subtle. Only one link in a food chain, Blong went on to emphasize, need be affected in order for there to be detrimental consequences for an economically significant
102
P. BICKNELL
population. Relatively insignificant ash fall into the Bay of Naples after the 1906 Vesuvius eruption caused massive destruction of invertebrates and temporary disappearance of fish from the area (Bianco 1906). Minor ash fall again, from Hekla in 1947, was sufficient to drive fish away from coastal waters affected (Blong 1984, p. 337). At Mochlos in three places distal Thera ash up to 10cm in depth has been exposed during recent excavations. In all cases the tephra represents airborne deposit rather than being the result of human clearance or construction activity (Soles et al 1995). Before settling, the ash blanket may have been deeper than at present. It follows that at least 10cm of ash, probably more, fell into the sea to Crete's north. Dolphins, it is worth stressing, given that they were part of the Minoan diet, are especially sensitive and vulnerable to marine pollutants. Their immune system suppressed, they become sick, unable to feed and highly susceptible to virus infection. In the second half of 1990 an area of the Mediterranean between North Africa and Spain was severely contaminated as a result of industrial pollution. Poisoned by heavy metals and virally infected, close to 6000 dolphins died in the space of 3 months. No other species was palpably affected (Pastor 1991, pp. 60 and 128-129). Any significant deoxygenation of water stemming from the presence of pumice or ash would be especially deleterious to life in the sea. A possibility to bear in mind is that of accumulation of nutrients in Cretan and other Aegean coastal waters as a result of progressive mineral release. Given eventual action of sunlight on the nutrient-saturated sea, the stage is set for the syndrome (now familiar in the Adriatic as a result of deposition into the sea of artificial fertilizers leached from coastal soil) of runaway phytoplankton blooms, subsequent crash of the inflated algal population, and mass marine mortality as a result of oxygen-reducing bacterial decomposition. To proceed a little further, and more adventurously, down this track, there are some episodes of bloom formation, so-called red tides, for example, that kill fish and other sea creatures not only by eventual deoxygenation of their immediate environment, but also, earlier and more directly, through the release of virulent neurotoxins. Toxin-permeated phytoplankton is consumed in the first instance by bivalves and gastropods, whose metabolic and feeding rates are adversely affected so that large numbers die prematurely. The poisons simultaneously spread upward through the marine food chain in increasing concentration as shellfish are ingested by higher-order molluscs such as cepha-
lopods, which are in turn eaten by vertebrate predators including cetaceans and fish. The earliest description of a red tide (Exodus 8. 19-25) relates to an outbreak in the Nile Delta in the Late Bronze Age. The local fish population was exterminated by a combination, presumably, of poisoning and oxygen starvation. Land-based casualties of toxic blooms include, of course, human consumers of shellfish and other tainted fauna. Fatal poisoning of some Minoans by tainted molluscs, dolphin or fish would have dictated temporary abstention from sea food, whatever the shortage of other foodstuffs, and provided further stimulus to gruesome religious expiation. Conclusion Drawing to a close, I hasten to underline the speculative nature of the content of the final paragraph of the previous section. One is not compelled to think in terms of exotic algal blooms and deadly toxins, any more than to insist that huge tsunamis ravaged northern Crete, to bring the Late Bronze Age eruption of the Thera volcano into connection with a serious disturbance of the marine environment of the eastern Aegean. Prolonged adverse conditions as a result of profusion of floating pumice and copious distal ash fall will have been more than sufficient to elicit an extreme religious response on the part of the Minoans and to play a significant role in the enigmatic collapse of their civilization, which set the scene for eventual Mycenean domination of the whole of Late Bronze Age Crete. References BARBER, R. L. N. 1987. The Cyclades in the Bronze Age. Duckworth, London. BETANCOURT, P. P. 1985. The History of Minoan Pottery. Princeton University Press, Princeton, NJ. BIANCO, S. L. 1906. Azione della pioggia di cenere caduta durante 1'eruzione del Vesuvio dell' Aprile 1906 sugli animali marini. Mitteilungen Zool. Stat. Neapel. 18, 73-104. BLONG, R. J. 1984. Volcanic Hazards. Academic Press, Sydney, NSW. BOSANQUET, R. C. 1904. Some 'Late Minoan' vases found in Greece. Journal of Hellenic Studies, 24, 317-329. BRATTSTROM, B. H. 1963. Barcena Volcano 1952 - its effect on the fauna and flora of San Benedicto Island, Mexico. Proceedings, 10th Pacific Science Congress 1961. Bishop Museum Press, 499-524. BULLARD, F. M. 1976. Volcanoes of the Earth. University of Texas, Austin.
MINOAN MARINE WARE AND THE THERA ERUPTION DICKINSON, O. 1994. The Aegean Bronze Age. Cambridge University Press, Cambridge. DOMINEY-HOWES, D. 1996. The geomorphology and sedimentology of five tsunamis in the Aegean Sea region, Greece. PhD thesis, Coventry University. DOUMAS, C. G. 1983. Thera; Pompeii of the Ancient Aegean. Thames and Hudson, London. DOWNEY, W. S. & TARLING, D. H. 1984. Archaeomagnetic dating of Santorini volcanic eruptions and fired destruction levels of Late Minoan civilisation. Nature, 309, 519-523. ERSKINE, W. F. 1962. Katmai. Abelard-Schuman, London. EVANS, A. J. 1921 1928 and 1935. The Palace of Minos, Vols. I II and IV. Macmillan, London. FRANCAVIGLIA, V. 1990. Sea-borne pumice deposits of archaeological interest on Aegean and eastern Mediterranean beaches. In: HARDY, D. A. & RENFREW, A. C. (eds) Thera and the Aegean World HI, Vol. 3. Thera Foundation, London, 127-134. GILL, M. A. V. 1985. Some observations on representations of marine animals in Minoan art, and their identification. Bulletin de Correspondance Hellenique, Supplement XI, 64-81. GUEST-PAPAMANOLI, A. 1983. Peche et pecheurs minoens. In: KRZYSZKOWSKA, O. & NIXON, L. (eds) Minoan Society. Bristol Classical Press, Bristol, 101-108. HOOD, M. S. 1971. The Minoans. Thames and Hudson, London. HUGHES, D. D. 1991. Human Sacrifice in Greece. Routledge, London. MARINATOS, S. 1939. The volcanic destruction of Minoan Crete. Antiquity, 13, 425-439. & HIRMER, M. 1960. Crete and Mycenae. Thames and Hudson, London. MORGAN, L. 1988. The Miniature Wall Paintings of Thera. Cambridge University Press, Cambridge. MOUNTJOY, P. A. 1984. The marine style pottery of LM I B/LH II A: towards a corpus. Annual of the British School at Athens, 79, 161-218. 1985. Ritual associations for LM I B marine style vessels. Bulletin de Correspondance Hellenique, Supplement XI, 231-242. NILSSON, M. P. 1950. The Minoan-Mycenean Religion and its Survival in Greek Religion. Kungl. Humanistiska Vetenskapssamfundet, Lund. PAGE, D. L. 1970. The Santorini Volcano and the Desolation of Minoan Crete. Society for the Promotion of Hellenic Studies, London. PASTOR, X. 1991. The Mediterranean. Greenpeace Books, Collins and Brown, London.
103
PECK, R. (trans.) 1970. Aristole: The History of Animals, Harvard University Press, Cambridge, Mass. PLATON, N. 1971. Zahros: the Discovery of a Lost Palace of Ancient Crete. Scribner, New York. POPHAM, M. R. 1967. Late Minoan pottery, a summary. Annual of the British School at Athens, 62, 337-351. 1984. The Minoan Unexplored Mansion at Knossos. Thames and Hudson, London. 1990. Pottery styles and chronology. In: HARDY, D. A. & RENFREW, A. C. (eds) Thera and the Aegean World III, Vol. 3. Thera Foundation, London, 27-28. POWELL, J. 1996. Fishing in the Prehistoric Aegean. Paul Astroms Forlag, Jonsered. REESE, D. S. 1987 Palaikastro shells and Bronze Age purple dye production in the Mediterranean basin. Annual of the British School of Athens, 82, 201-206. 1990. Triton shells from east Mediterranean sanctuaries and graves. Journal of Prehistoric Religion, 3-4, 7-14. ROLFE, J. C. (trans.) 1940. Ammianus Marcellinus: The History, Harvard University Press, Cambridge, Mass. SAKELLARAKIS, J. A. & SAPOUNA-SAKELLARAKI, E. 1981. Drama of death in a Minoan temple. National Geographic, 166, 205-222. & 1991. Archanes. Ekdotike Athenan, Athens. SIMKIN, T. & FISKE, R. S. 1983. Krahatau 1883. Smithsonian Institution Press, Washington, DC. SOLES, J. S. & DAVARAS, C. 1996. Excavations at Mochlos 1992-1993. Hesperia, 65, 175-230. , TAYLOR, S. R. & VITALIANO, C. J. 1995. Tephra samples from Mochlos and their chronological implication for neopalatial Crete. Archaeometry, 37, 385-393. STIEGLITZ, R. R. 1994. The Minoan origin of purple. Biblical Archaeologist, 57, 47-54. THOMPSON, D'A. W. 1947. A Glossary of Greek Fishes. Oxford University Press, Oxford. WACHSMANN, S. 1987. Aegeans in the Theban Tombs. Uitgeverij Peeters, Leuven. WALL, S. M., MUSGRAVE, J. H. & WARREN, P. M. 1986. Human bones from a Late Minoan I B house at Knossos. Annual of the British School at Athens, 81, 333-388. WARREN, P. M. 1991. The Minoan civilisation of Crete and the volcano of Thera. Journal of the Ancient Chronological Forum, 4, 29-39. WILSON, I. 1985. The Exodus Enigma. Weidenfeld and Nicolson, London.
This page intentionally left blank
Ground-penetrating radar mapping of Minoan volcanic deposits and the Late Bronze Age palaeotopography, Thera, Greece JAMES K. RUSSELL1 & MARK V. STASIUK2 1
Igneous Petrology Laboratory, Department of Earth and Ocean Sciences, University of British Columbia, Vancouver, B.C. V6T 1Z4, Canada (e-mail:
[email protected]) 2 Environmental Science Division, Institute of Environmental and Biological Sciences, Lancaster University, Lancaster LAI 4YQ, UK Abstract: The Late Bronze Age (LBA) eruption of Santorini volcano deposited ash over most of the eastern Mediterranean, distributed thick deposits of pyroclastic material over the local landscape, and instantly buried the Minoan-aged living surface of these islands. Ground-penetrating radar (GPR) studies of the LBA volcanic deposits on Thera have allowed us to establish the thickness of individual pyroclastic units, to trace units laterally, and to establish facies variations in areas where the deposits are unexposed. GPR data are presented for two sites: Site A is a survey over LBA volcanic deposits exposed in the Phira quarry, immediately south of the town of Phira, and Site B is a 550m survey of the LBA deposits underlying the Akrotiri peninsula immediately north and south of the Akrotiri archaeological excavation. These traverses show that GPR can define structures as deep as 18-20m (velocity 0.1 mns" 1 ) and can accurately map the thicknesses of the LBA volcanic deposits from the caldera wall to Thera's southern coast. Furthermore, our best datasets suggest that the Phase 1 fall deposits can be differentiated from the Phase 2-4 deposits, and that GPR can clearly image the interface between the volcanic deposits and the underlying LBA living surface. Future GPR surveys could be used to delineate palaeotopographic lows and valleys associated with LBA streams or drainages or, used in combination with geological mapping, could refine the position and nature of LBA shorelines.
The present-day landscape of the Santorini islands results directly from the Late Bronze Age (LBA) or Minoan eruption of Santorini volcano. The eruption enlarged the central caldera to its present 70-80 km2 and produced a series of rhyodacitic pyroclastic deposits that blanket most of the islands (Bond & Sparks 1976; Heiken & McCoy 1984; Druitt 1990). The deposits bury the Minoan living surface to depths of tens of metres; in several instances, the pyroclastic deposits have buried several important Minoan settlements (e.g. Marinatos 1939; Sparks 1979; McCoy & Heiken 1994), the most notable of which is that near Akrotiri on the south coast of Thera. In this paper we explore the potential of ground-penetrating radar (GPR) for subsurface mapping of the LBA pyroclastic deposits on Thera. GPR data are presented for two sites (Fig. 1). Site A is a calibration survey over LBA volcanic deposits exposed in the Phira quarry, immediately south of the town of Phira, and Site
B is a regional 550m survey of the LBA deposits underlying the Akrotiri peninsula immediately north and south of the Akrotiri archaeological excavation. Our original intent was to use GPR to map volcanic stratigraphy. Specifically, we aimed to map volcanic deposits in regions that had little vertical exposure and to trace individual pyroclastic units within the subsurface, This endeavour is relevant in that the Minoan deposits are relatively undissected and, except along the crater wall or the coastlines, are poorly exposed. Stratigraphic measurements such as these can provide hard data with which to constrain ideas on the eruptive and depositional processes attending the LBA eruption. Our results also directly benefit archaeological investigations. With GPR, we are clearly able to image the interface between the LBA volcanic deposits and the underlying Minoan living surface (e.g. Vaughan 1986; McCoy et al. 1992; Camerlynck et al. 1994; Goodman 1994). Radar surveys, therefore, can show the total thickness
From: McGuiRE, W. G., GRIFFITHS, D. R., HANCOCK, P. L. & STEWART, I. S. (eds) The Archaeology of Geological Catastrophes. Geological Society, London, Special Publications, 171, 105-121. 1-86239-062-2/00/ $15.00 © The Geological Society of London 2000.
106
J. K. RUSSELL & M. V. STASIUK
Fig. 1. Physiographic map of Santorini showing the locations of surveys in relation to major geographical features. Survey sites include the Phira quarry (Site A) and a traverse across the Akrotiri peninsula (Site B). Digital elevation model cartography is courtesy of Tim Druitt.
of volcanic material overlying specific sites or horizons, can map palaeotopography in the subsurface, and can map the position and nature of the ancient Minoan shorelines (e.g. Heiken et al. 1990; Druitt & Francaviglia 1992; Forsyth 1996; Friedrich et al. 2000). The Minoan volcanic deposits The Santorini islands include Thera, Therassia, Aspronisi and the younger Palaea Kameni and Nea Kameni (Fig. 1). Thera, Therassia and Aspronisi are remnants of the LBA island that was partially destroyed and heavily modified by the cataclysmic 3600 years BP (e.g. Aitken 1988) Plinian eruption of Santorini volcano. These islands form a broken ring around the nowflooded caldera, produced by a series of collapse events associated with the Minoan and previous voluminous eruptions (Heiken & McCoy 1984; Druitt & Francaviglia 1992; Forsyth 1996). The Kameni islands are substantially younger than the LBA eruption (e.g. Fytikas et al. 1990) and are, in fact, the emergent portions of a large (2km 3 ) submarine dacitic intracaldera shield volcano. There have been at least nine subaerial eruptions between 197BC and 1950 (Fytikas et al. 1990; Druitt et al. 1996).
The Minoan eruption of Santorini produced over 36 km3 of volcanic material. Previous workers have established a stratigraphy of four mappable pyroclastic units (Phases 1-4). Below is a summary of the eruption sequence and a description of the resulting pyroclastic deposits based on the work of Bond & Sparks (1976), Heiken & McCoy (1984) and Druitt (1990), and summarized by Druitt et al. (1999). The eruption began with phreatic and phreatomagmatic explosions from a vent near present-day Nea Kameni and showered SE Thera with about 106m3 of ash (Heiken & McCoy 1984; Doumas et al. 1997). The Plinian phase of the main eruption (Phase 1) commenced at the same vent and the column attained a height of 36km (Sigurdsson et al. 1990). Phase 1 fall deposits represent c. 2 km3 of magma and occur over all of the Santorini islands to a maximum depth of 6 m. Several hours into the Plinian eruption, sea water gained access to the vent, causing violent phreatomagmatic explosions and the production of surge deposits (Phase 2). The surge deposits are laminated, cross-bedded and poorly sorted, and are themselves overlain by massive, poorly sorted deposits up to 35m thick (Phase 3). The Phase 3 deposits are thought to represent lowtemperature, fluid-rich pyroclastic flows caused by high-intensity phreatomagmatic explosions. Phase 4 deposits are dominantly fine-grained ignimbrite and are thought to derive from hightemperature pyroclastic flows; they are distributed mainly around the outer coasts of Thera, Therassia and Aspronisi. The Phase 4 deposits are fan-shaped in cross-section, in that they thicken with distance from the caldera wall (up to 40 m). One of the most notable features of the Phase 4 deposits is the abundance of lenses and layers of lithic breccias enclosed within the fine-grained ignimbrite. Lastly, overlying the entire eruption sequence, there are alluvial deposits associated with flash floods presumed to have occurred shortly after the Plinian eruption. In the vicinity of the Akrotiri excavation site, these alluvial gravel and boulder deposits have eroded and covered the Phase 4 ignimbrite. Ground-penetrating radar GPR is used extensively for imaging the shallow subsurface; the technique uses electromagnetic (EM) waves in the Megahertz (MHz) frequency range to image subsurface variations in electrical properties. Conceptually, it is the EM analogue of reflection seismology (e.g. Ardon 1985); the fundamental principles of the technique have been well described by Annan & Davis (1977) and Davis & Annan (1989).
GPR STUDIES OF VOLCANIC DEPOSITS ON THERA GPR has been used effectively in the fields of glaciology (e.g. Clarke & Cross 1989), geotechnical engineering (e.g. Ardon 1985; Holloway et al. 1986), environmental geophysics (Knoll et al 1991; Rea et al 1994) and, recently, in archaeology (e.g. Vaughan 1986; Goodman 1994; Camerlynck et al 1994; Marco et al 1997) as a complement to other geophysical surveys (e.g. Williams & Cronkite-Price 1995). The development of higher-power transmitters (e.g. 1000 V), lower-frequency antennae (e.g. <50MHz) and more focused beams has improved the effectiveness of the technique for geological studies. For example, GPR has been used on a wide variety of lithified geological deposits such as clastic sedimentary rocks (e.g. Jol & Smith 1992; Smith & Jol 1992), Palaeozoic limestone reefs (Pratt & Miall 1993), and more structurally complex rocks (e.g. Holloway et al 1986; Liner & Liner 1995). Volcanic stratigraphy is particularly appropriate for study by GPR because the deposits are commonly thin and compositionally homogeneous, and they can be electrically resistive (Stasiuk & Russell 1994; Russell & Stasiuk 1997). The deposits on Thera are also particularly dry and situated well above the water table, which means that there will be little attenuation of the radar signal by pore or surface water. Previous studies by McCoy et al (1992), Stasiuk & Russell (1994), Gilbert et al (1996), Russell & Stasiuk (1997) and Russell et al (1998) have shown that GPR has tremendous potential for quantifying distributions, thicknesses and volumes of volcanic deposits, and perhaps can even be used to study facies variations within individual units. Radar survey conditions All GPR data were collected with a Sensors and Software 'PulseEKKO' 100 instrument. The radar was controlled with a Toshiba 486 PC laptop. During the course of the field programme
107
we experimented with different configurations of antennae (50 and 100 MHz) and transmitters (400 and 1000 V). On the basis of these experiments, we elected to use a 1000 V source and the 50 MHz antennae for our surveys. Transmitter and console computer were powered by a rechargeable, 13 A, 12V gelsel battery. Table 1 summarizes the survey variables (e.g. acquisition time windows, etc.) for all traverses presented in this paper. Apart from a single common-mid-point (CMP) survey, all surveys were run in common-offset reflection mode. We employed a 3m separation between transmitter and receiver antennae to reduce the effects of ringing on surveys with longer listening times (e.g. MK4A). Data were collected with a fixed gain but, for interpretation and presentation purposes, were analysed with automatic gain control (AGC) or spreading and exponential compensation (SEC) gain control. Both gain functions are standard options within the PulseEKKO 4.2 software. Gain artificially increases the amplitudes of signals recorded in the traces and can be used to visually compensate for decreased amplitudes reflected from greater depths. Increasing gain to elucidate deeper structures is effective only where reflected amplitudes are significantly larger than background noise amplitudes (large signal to noise ratio). Situated at the top of Mount Profitis Ilias (Fig. 1) is a large radar installation that appears to serve civilian and military air traffic control. The installation has line-of-sight over most of Thera and broadcasts a signal continuously. This signal was received by our antenna with the result that each trace contains two components of energy: (a) the portion of our original EM pulse that was reflected from subsurface features, and (b) energy derived from the ambient cultural radar. The second component contributes highfrequency noise to each trace. For shallow reflectors this noise represents little more than a nuisance because the signal (primary reflection) to noise (ambient cultural radar) ratio is high.
Table 1. Survey variables used for each traverse segment, including listening time (TW), stacking, number of traces collected and sample interval (SI) Site
Location
Mode
Files
TW(ns}>
Stacks
Traces
SI(m)
Length (m)
A A B B B B
Phira quarry Phira quarry Akrotiri Akrotiri Akrotiri Akrotiri
CMP COR COR COR COR COR
CMP1 MK1E
512 512 900 512 512 512
64 256 256 128 128 128
54 283 80 200 160 192
0.2 0.5 1.0 1.0 1.0 1.0
141 79 199 159 191
MK4A MK4B MK4C MK4D
CMP, common-mid-point survey; COR, common-offset reflection survey.
108
J. K. RUSSELL & M. V. STASIUK
For deeper reflectors, however, our ability to image is severely impeded because primary signals generally become weaker with depth, resulting in a low signal to noise ratio. In such instances, gain cannot be used to enhance the
primary signals without increasing the noise. We found that when we operated within lineof-sight of Profitis Ilias the background radar signal masked all reflectors at depths greater than c. 8 m.
Fig. 2. Volcanic stratigraphy of LBA deposits exposed in Phira quarry, (a) Field photograph of quarry bench that was used for 141 m GPR survey. Numbers denote positions of measured stratigraphic columns. (b) Schematic cross-section of volcanic stratigraphy shown in field photograph (after Milner 1997). (c) Stratigraphic columns for measured sections of LBA volcanic deposits (see text and Fig. 4).
GPR STUDIES OF VOLCANIC DEPOSITS ON THERA To address this interference we applied a lowpass filter to selectively eliminate the highfrequency noise associated with the background radar signal. Specifically, we chose a cutoff of 10% of the Nyquist frequency or 62.5 MHz and passed only lower-frequency information. All radar data shown in this paper had a lowpass filter applied.
Site A: Phira Quarry Immediately south of the town of Phira is an abandoned quarry situated on the eastern lip of the caldera. The quarry exposes several hundred metres of Minoan volcanic deposits (Bond & Sparks 1976; Druitt & Francaviglia 1992) lying on top of older volcanic rocks that comprise the ancient Late Bronze Age (Minoan) living surface. The LBA volcanic deposits are all pyroclastic in origin and, in this location, are proximal in nature. We chose to run a GPR survey for calibration purposes along the top of one of the quarry benches (Fig. 2a). Stratigraphic sections were measured at six locations along the front face of the bench. The locations of the sections are marked in each of Fig. 2a, 2b, and 2c. Detailed, and more general, descriptions of these deposits are available from a number of sources (Bond & Sparks 1976; Heiken & McCoy 1984; Druitt et al. 1999). The top of the bench is cut into Phase 3 pyroclastic deposits and the survey line is underlain by pyroclastic flow (Phase 3), pyroclastic surge (Phase 2), and fall deposit (Phase 1). The pyroclastic deposits are themselves underlain by well-indurated pre-Minoan volcanic rocks, including a basaltic-andesite lava breccia and well-stratified, finer-grained tuffs. Figure 2b is a sketch of the field photograph shown in Fig. 2a, and shows the distribution of the individual pyroclastic units in cross-section. The six measured stratigraphic sections are plotted in Fig. 2c. The main features of this stratigraphic section include the following: (1) the contact between the LBA volcanic deposits and the basement dips slightly to the east and lies between 9 and 14m beneath the surface of the bench. (2) Overlying the pre-Minoan volcanic deposits and the Minoan living surface is a wellsorted, massive, 6m thick fall deposit (Phase 1) comprising pumice lapilli of 1-20 cm. The largest blocks are c. 0.2m. The only structure observable in this unit is a crude coarsening upwards of pumice clasts.
109
(3) Phase 2 pyroclastic surge is 4-6 m thick, poorly sorted, moderately well laminated, and comprises ash and lapilli with rare block-sized lithic fragments. Laminations dip c. 15° to the east and merge asymptotically with the base of the unit. Many layers show prominent cross-laminations. (4) Phase 3 pyroclastic flow is massive, poorly sorted, and contains large (0.5m to >2m) blocks of lithic fragments. Phase 3 contains weak interlayers of pumice beds and has an upper lithic-rich zone. The apparent dip of this crude stratification is c. 5-10° east. (5) As a package, the pyroclastic flow and surge are crudely bedded, form sets of prograding lobes or beds, and dip slightly to the east.
Common-mid-point (CMP) analysis Two-way travel times are easily converted to depths wherever the velocity of the radar in the deposit is known. The radar velocities or the dielectric properties of these deposits are generally not known a priori (e.g. Russell & Stasiuk 1997) and, therefore, an important first task is to estimate the radar velocity of the target deposit. For a well-exposed section, as found in the Phira quarry (Fig. 2a), there are two ways to do so. A relatively simple way to constrain the velocity is by selecting a range of velocities and finding the one that gives the best 'depth' match between stratigraphic features in the deposit and corresponding reflections in the GPR profile. This method works well where the deposits are homogeneous and contain prominent stratigraphic markers that are also geophysically distinct. In the Phira quarry section, we found that a velocity of O.lmns" 1 produced a good correspondence between the measured depth to prominent reflectors (e.g. Minoan surface; Fig. 2) and the apparent depth in the radargram (e.g. Fig. 4, see below). The average radar velocity for these deposits has also been estimated with a field experiment called a CMP survey (Table 1), the results of which are summarized in Fig. 3. Figure 3a shows the GPR data plotted as distance between antennae and energy returned as a function of time. These data are displayed after application of a lowpass filter and with a moderate gain (SEC). Figure 3b shows the results of the CMP analysis in which numerous repeated traces are averaged (stacked) to produce single traces and the results are shown as a function of model velocities (0.01-0.3 mns" 1 ). The output record is time v. velocity. The appropriate velocity of the deposit is recognized by the trace with the largest
110
J. K. RUSSELL & M. V. STASIUK
Fig. 3. Results of common-midpoint (CMP) survey run in Phira quarry, used to estimate a value for average radar velocity for LBA volcanic deposits. The CMP data are shown in (a) and the analysis of CMP data is plotted in (b) (see text).
amplitudes at depth. In this case (ignoring the direct ground- and air-wave data), the velocity appears to be constrained to between 0.09 and O.llmns^ 1 . We elected to use a mean value of 0.1 mns" 1 based on this survey, which is consistent with the 'stratigraphic' estimate of average radar velocity.
GPR survey results from Phira quarry Our survey of LBA volcanic deposits in the Phira quarry was 141m long, used a 0.5m sample spacing, and comprised 283 traces. The main
purpose of this survey was to calibrate the geophysical results against a well-characterized section of the Minoan pyroclastic deposits. Figure 4 shows the survey results and compares them with the stratigraphic sections prepared for the face of the quarry bench (e.g. Fig. 2a-c). Figure 4a is a radargram for the survey shown at 2.5 times vertical exaggeration. The radar data are shown after application of a dewow and the lowpass (10%) filter, and have been enhanced with a moderate SEC gain. Figure 4b is a line diagram drawn for the radargram and showing the position and shape of the
Fig. 4. Results of GPR survey of Phira quarry section, (a) Radargram (looking south); caldera rim is on the western end of the traverse, (b) Line drawing of same section showing prominent reflectors within upper (Phase 2 and 3) pyroclastic flows (bold lines), structures within underlying pre-LBA units (dashed lines labelled B) and a reflection that possibly derives from the quarry wall or other out-of-plane features (R). The shaded region denotes the upper and lower surfaces of the Plinian fall deposit as interpreted from the geophysical data.
GPR STUDIES OF VOLCANIC DEPOSITS ON THERA 111
112
J. K. RUSSELL & M. V. STASIUK
prominent geophysical reflectors. Superimposed on the line diagram are lithological columns representing the measured stratigraphic sections as well as a shaded region representing the distribution of the Phase 1 fall deposit as interpreted from the geophysical data. The radar profile (Fig. 4a) shows information collected for two-way travel times of 0-400 ns; this range corresponds to a maximum depth of just over 20m (v^O.l mns" 1 ). The radargram shows strong reflectors down to 200-250 ns and weaker, but distinct, reflectors down to 350400ns. Furthermore, there is no apparent ringing (echoing of EM waves) at these depths, corroborating our decision to use a 3 m separation. Our interpretation of geological features is based on identifying continuous reflectors or horizons of reflectors that show the same polarity (black v. white). Reflectors that represent the same geological interface should show the same polarity. There appear to be three distinct packages or regions in the radargram. At times between 200 and 350ns, the radargram shows a number of continuous but weak reflectors. The reflectors (dashed lines labelled B in Fig. 4b) are variable in attitude: they dip to the east (0-40 m), or are concave upwards (40-100m), or dip to the west (110-140 m). These are 'basement' reflectors and represent geological structures in the pre-Minoan stratigraphy. The concave-upwards structures are particularly evident in Fig. 2a and b. The signals are weak relative to the more shallow parts of the radargram, but they are constant in character and on this basis are reliable. The upper portion of the radargram (0-150 ns) also has a very distinctive pattern. This stratigraphic package shows abundant strong discontinuous dipping reflectors (see Fig. 4a and b). The package, taken as a whole, represents the Phase 2 and 3 pyroclastic surge and flow deposits overlying the Phase 1 fall deposit. The most distinctive feature of this part of the radar profile is the abundant choppy, eastward-dipping reflectors seen right across the panel. This feature mirrors the characteristic laminated structure of the surge and flow deposits (Fig. 2a-c). The last zone shown in Fig. 4a is best displayed between horizontal positions 10 and 110m and at times of 130 and 190ns. This part of the radargram is characterized by little or no internal reflection; amplitudes are generally low (little black) and the reflectors that exist tend to be flat and broken-up. The pattern, where viewed from the side, is somewhat reminiscent of cross-hatching. We interpret this zone as indicative of the pyroclastic fall. The fall is homogeneous, massive, and has virtually no
internal structure. These properties are consistent with the relatively featureless character of the corresponding portions of the radar section. Our interpretation of the distribution of the Phase 1 fall, based on the radar data, is shown as a shaded region in Fig. 4b. The correlation between our interpretation and the actual distribution (see observed lithological columns) is not perfect but there are pronounced similarities in thickness, dip and position. Our interpretation of the radargram is somewhat hampered by the presence of a very strong horizontal reflector, which occurs on the eastern end of the traverse at times of 190-210ns. This reflection, labelled R in Fig. 4b, results from interaction of the EM waves with either the face of the cliff to the quarry bench or the surface of a road located at the base of the cliff-face. The GPR survey was run within a few metres of the face of the bench to optimize the correlations between the radar data and the geological features measured on the face of the quarry bench. The radar beam, however, appears not to be sufficiently focused to avoid generating these artefacts.
Site B: Akrotiri Peninsula Akrotiri peninsula comprises the southern end of Thera (Fig. 1); at its narrowest the peninsula is just over 1 km in width (Fig. 5). The LBA pyroclastic deposits form a continuous sheet across the peninsula and vertical exposures are more or less limited to the caldera wall or the southern coastline. We ran a regional-scale GPR survey across the peninsula (Table 1) in two segments as shown in Fig. 5. The traverse is oriented radially to the vent region of the 3600 years BP eruption. The position and orientation of the traverse were chosen so that the survey could sample more proximal deposits near the caldera wall (Fig. 5, no. 1), intersect the Akrotiri archaeological site (Fig. 5, no. 2), and finish at the southern shoreline, where there are more distal pyroclastic deposits (Fig. 5, no. 3). The primary aim of the regional survey was to map the distribution of LBA volcanic deposits in the subsurface between points of known stratigraphy (Fig. 5, nos 1 and 3). Mapping of young volcanic deposits commonly relies on trust: stratigraphy derived from a few well-exposed sections is used to extrapolate across regions with little or no stratigraphic exposure. Volcanic stratigraphy is notoriously complex, however, and this practice can be very
GPR STUDIES OF VOLCANIC DEPOSITS ON THERA
113
Fig. 5. Detail location map for GPR survey of Akrotiri peninsula showing locations of individual segments of survey line with respect to: (1) stratigraphic section measured on rim of caldera, (2) archaeological excavation site, and (3) stratigraphic column measured on south coast of peninsula. dangerous without substantial information from areas between sections. Regional-scale GPR surveys can be used to test and strengthen these extrapolations. In addition to elucidating the distribution of the LBA volcanic deposits, our survey has the ability to define and trace the palaeotopography of the Minoan living surface. In this survey, the results should be of import to researchers working on the Akrotiri excavation site because our results define the thickness of the LBA volcanic deposits and provide physical descriptions of the hinterland to the Minoan townsite. Potentially, the survey can also recover the ancient coastline.
Volcanic stratigraphy of Akrotiri peninsula Our geophysical survey was constrained at two locations where we measured the stratigraphy of LBA volcanic deposits. We measured a complete section through the LBA deposits exposed in the caldera wall on the northern margin of Akrotiri peninsula. The location of section II-1 is shown in Fig. 5 (no. 1). Another section (ML-2) was measured on the southern margin of the peninsula just south of the archaeological excavation (Fig. 5, no. 3). The southern section comprises over 30m of LBA pyroclastic material and
clearly represents more distal deposits that were accreted to Thera during the 3600BP eruption. Both stratigraphic sections are summarized in Fig. 6. At the caldera wall there are 12m of LBA volcanic deposits resting unconformably on the pre-Minoan surface. In this location the basement comprises Cape Riva ignimbrite (Druitt 1985; Druitt & Francaviglia 1992). Immediately above the Cape Riva formation is 2 m of Phase 1 fall deposit. Phase 2 surge overlies the fall; it is strongly laminated and contains abundant redoxidized lapilli. The upper 7 m of the section is made of poorly laminated Phase 3 pyroclastic flow (c. 5 m) capped by just over 2 m of Phase 4 ignimbrite. The Phase 4 ignimbrite comprises a tan-coloured ash matrix containing abundant lithic fragments. The unit shows a crude stratification defined by concentrations of block-sized dense lava lithic fragments. The concentrations of lithic fragments form discrete horizons, which show substantial topography and are commonly channel-like in form. The top of the section is more or less even in elevation with the north end of the GPR traverse (e.g. MK4D, Table 1). The stratigraphic sequence measured at the southern coastline (ML-2; Fig. 6) is substantially different. There is a depth of over 30m of exposed LBA volcanic deposits and, although the section is continuous to sea level, the
114
J. K. RUSSELL & M. V. STASIUK
Fig. 6. A comparison of LBA volcanic stratigraphy measured for the northern edge of the Akrotiri peninsula and situated on the south rim of the caldera (no. 1: Fig. 5), and for the southern shoreline of the Akrotiri peninsula and immediately south of the archaeological excavation site (no. 3: Fig. 5). Sections compare thicknesses of pyroclastic fall and flow (Pf) deposits and show distribution of lag breccias (Bx).
GPR STUDIES OF VOLCANIC DEPOSITS ON THERA pre-Minoan surface is not exposed. Phase 1 and 2 deposits are absent from the section. The base of the section is made up of slightly over 3 m of Phase 3 pyroclastic flow; the unit is massive and made up of lithic and pumice lapilli in a fine-grained ash matrix. Overlying Phase 3 and making up most of the section (>24m) is Phase 4 ignimbrite. The majority of the unit is massive, fine-grained, poorly sorted material with lenses and lithic-rich layers or domains. These layers mainly comprise dense lava clasts, which range in size from lapilli to blocks. The layers can be continuous and define a slight (<2°) dip towards the coast. Conversely, they can occur as discontinuous and trough-shaped lenses. The layers vary from centimetre scale to several metres in thickness. Finer-grained lithic lenses sometimes are graded. In the region of the Akrotiri archaeological site, Phase 4 deposits are capped by fluvial gravel and boulder deposits. These deposits are generally erosive into the Phase 4 ignimbrite and form discontinuous channels or sheets. At this locale (ML-2) there are between 2 and 4m of this deposit sitting on top of Phase 4. Also shown in Fig. 6 is the elevation of the GPR traverse (MK4A) relative to the stratigraphic column. The southern end of our GPR survey ends at an elevation below the flash flood gravel deposits.
GPR results from Akrotiri peninsula Our survey across the Akrotiri peninsula was completed in two segments: a northern segment that ran south from a point near the caldera wall towards the northern edge of the Akrotiri excavation site, and a short segment that extended from the southern boundary of the same excavation site towards the coast (Fig. 5). The survey was over 0.5km long and used a 1.0m sample spacing (Table 1). The survey results are shown in two separate radargrams at c. 2X vertical exaggeration (Figs 7 and 8). Figure 7 shows the survey results for the northern 552m segment from the caldera side of the peninsula (N) to the edge of the Akrotiri excavation (S). The three panels are continuous, and oriented as if the viewer is looking westward, and are best read from right (N) to left (S). The radar data reveal a simple consistent stratigraphy for the first 350m of the traverse (upper & middle panels, Fig. 7). Again, the radargram for 0-350 m can be broken down into three packages on the basis of their distinctive geophysical character. In the middle of the section between 100 and 200ns (4.5-9m) there
115
is a zone of low-amplitude signal (featureless grey). The lower half of the section between 200 and 400ns is characterized by weaker, discontinuous reflectors overprinted by occasional stronger, concave-down reflectors. The upper portion of the radargram between 0 and 100ns shows a third distinctive pattern characterized by strong reflections. The radar data from the distinctive lowamplitude zone suggest a deposit that is uniform in thickness (2-3 m), continuous, and homogeneous without internal structure (e.g. no internal reflectors). We interpret this zone (Fig. 7; P^ as corresponding to the Phase 1 fall; the lower contact is against the pre-Minoan surface of Cape Riva ignimbrite (Fig. 7; CR) and the upper contact is with Phase 2 and 3 LBA pyroclastic deposits (Fig. 7; ^2,3,4)- The upper contact is marked by a prominent, flat-lying, and more or less continuous reflector at 4.0-5.5m depth, which maintains its polarity across the radargram; this is evident from the consistent redblue-red striping. The lower contact between LBA fall and Cape Riva ignimbrite is marked by a set of weaker, flat-lying, somewhat discontinuous reflectors at about 9 m depth. The fact that the polarity and orientation of these reflections is also maintained across the section is strong support for our interpretation. The radar data from 200ns and below represent structure within the pre-Minoan basement. Most reflectors are weak and horizontal; however, between 200 and 340m several deeper (12-16m) and more continuous curving reflectors appear in the unmigrated data. The upper 100ns of the radargram comprises strong, continuous and segmented, horizontal reflectors, which are interwoven with strong, discontinuous, southward-dipping reflectors. The data are consistent with a series of strongly layered deposits, which are locally crosslaminated or have abundant internal structure. We interpret these as LBA pyroclastic surge and flow units (e.g. Phases 2-4) sitting on top of the Phase 1 fall deposit. In particular, the dipping structures seen at positions 133m, 150m, 230m, and between 270 and 330m are entirely consistent with the overall lobate structure of these pyroclastic deposits and their strong internal structure (see Fig. 6). The lower panel shows survey results from immediately north of the excavation site. Electromagnetic interference from a metal fence can be seen in traces from the last 5 m of the survey (Fig. 7; F). The radar data in the lower panel show the same stratigraphic packages but with some important differences. First, the contact between the Phase 1 fall and the Cape Riva
116
J. K. RUSSELL & M. V. STASIUK
GPR STUDIES OF VOLCANIC DEPOSITS ON THERA ignimbrite begins at about 9-1 Om depth but shallows substantially to the south as the basement rises. This rise is the first topographic irregularity seen on the pre-Minoan surface over this traverse (Fig. 7; T); to this point the topography has been uniformly shallow dipping and flat. Attending this rise in topography is a distinctive thinning of the Phase 1 fall deposit relative to the flats. The upper package of pyroclastic flows (0-100ns) has a character consistent with that observed in the other two panels and it also appears to thin over the basement high. A second difference is seen in the lower half of the radargram. Many more continuous reflectors are seen within the pre-Minoan rocks (Fig. 7; B). Indeed, there are prominent structures seen at depths of 20-22 m. The continuous concave-downwards reflectors seen in the unmigrated section may represent the edges of small cliffs or steps in the basement topography. In summary, the northern segment of this GPR survey establishes several important elements. First, we are clearly able to see the interface between the LBA volcanic deposits and the Minoan living surface as represented by the Cape Riva ignimbrite. Second, we can identify the Phase 1 fall deposit based on its geophysical character. Because of its unique character, it is possible to trace the fall deposit in the subsurface for the full length of this segment of the survey. The unit is between 2 and 3m thick and clearly mantles the underlying topography. The fall appears to be of uniform thickness and character except where it mantles a relative high in the palaeotopography. Third, the fall is capped by LBA pyroclastic surge and flow (Phases 2-4). These units are more or less constant in thickness (c. 5 m), although there is a hint that the total package thickens very slightly from north to south from 4.5 to 6m respectively. In some instances our survey elucidates structures as deep as 20-22 m. The southern segment of the traverse across the Akrotiri peninsula is 30m long and begins on the southern edge of the archaeological excavation site. Figure 8 shows radar results for this segment of the traverse (left panel) and, for comparison, repeats (right panel) the radar data for the last 90m of the traverse shown in Fig. 7. The two radar panels are shown at the same scale
117
and vertical exaggeration, although longer listening times (Table 1) were used for the southern leg, to image deeper structures. The interval between the two panels represents the fenced area (400-450 m) of the archeological site. Within the radargram for the southern segment (Fig. 8; left panel) is a single 'nongeological' reflection; the artefact is an EM reflection from an overhead powerline situated a short distance south of the end of the section (Fig. 8; A). This reflection is manifested in the radargram by the apparent reflecting interface dipping to the north with a characteristic slope corresponding to the speed of light in air (OJmns- 1 ) at depths of 16-24m (Fig. 8). The remaining reflectors (all south dipping or flat) are interpreted as geological features in the subsurface. The subsurface structure for the southern segment is distinct from that found for the northern part of the traverse (Fig. 7; Fig. 8, right panel). First, the Phase 1 fall unit, characterized by a continuous, mantling zone of low amplitudes is absent. Second, there is no clear evidence of an interface that could be interpreted as the Minoan living surface. Furthermore, the interfaces that can be seen define a package of deposits dipping southward with dips shallowing with decreasing depth. In the near surface (top 6m) the structures are similar in both parts of the survey (compare the left and right panels, Fig. 8). The southern portion of the survey is short and our coverage in this area somewhat sparse, hence only a preliminary interpretation is offered here. Our interpretation relies heavily on geological observations made at the coastal cliffs 50-60 m south of the endpoint of the GPR survey. The stratigraphic section (ML-2) for the south coast (Fig. 6) is repeated in Fig. 8, and has been positioned to reflect its proper elevation relative to the GPR survey. At this location on the coast, Cape Riva ignimbrite is not exposed and neither is the Phase 1 fall, nor Phase 2 surge deposits. Phase 3 and 4 pyroclastic flow deposits are thicker than at more proximal locations (e.g. IL-1; Figs 5 and 6), having a total combined exposed thickness of about 25m. It is evident from this and other distal locations on the coast that the pyroclastic flow deposits have considerably extended the coastline of the island.
Fig 7. Radargrams for the northern portion of the GPR survey across the Akrotiri peninsula with the viewer looking west. The 552m traverse from the caldera side of the peninsula (N) to the edge of the Akrotiri excavation (S) is portrayed as three panels, each of which should be read continuously and from right (N) to left (S). Symbols in upper panel denote Cape Riva (CR) basement overlain by Phase 1 fall (Pi) and a combination of Phase 2, 3 and 4 deposits (P2,3,4). Symbols in lower panel illustrate prominent reflector in basement (B), thinning of Phase 1 as a result of topographic rise (T), and interference because of metal fence enclosing the Akrotiri site (F).
Fig. 8. Radar results for the southern termination of the traverse across the Akrotiri peninsula. The segment is 30 m long and ends against the southern edge of the Akrotiri archaeological excavation site (left panel). The radar data for the last 90m of the traverse shown in Fig. 7 are repeated for comparison. The interval between the two panels represents the fenced area of the archaeological site. Symbol (A) indicates a non-geological reflection because of above-ground powerline (see text).
GPR STUDIES OF VOLCANIC DEPOSITS ON THERA We would argue, on the basis of the GPR data shown in Fig. 8 (left panel) and the geological observations summarized in Fig. 6 (ML-2), that the Minoan shoreline was situated somewhere between the southern leg of our survey (MK4a) and the current ancient Akrotiri excavation site. The deposits that we have imaged south of the excavation site represent, in our opinion, a pyroclastic apron extending into the Late Bronze Age sea. The signal character, depth and upwardshallowing dips shown in the radargram (Fig. 8; left panel) are consistent with the internal structure of a prograding pyroclastic apron (Jol & Smith 1991) formed from numerous flow units or pulses of pyroclastic flows. Notably, at times below 420 ns there is an absence of reflectors; this depth corresponds to sea level and is the probable location of the local water table. If the location of the southern leg of the GPR survey is at a position that was offshore before the LBA eruption, can we also infer the nature of this portion of the LBA coastline? The total thickness of these deposits (>25m), combined with their proximity to the partially excavated Minoan settlement (which establishes a relative elevation for the Minoan living surface), suggests a significant elevation drop (20-30 m) over a narrow horizontal distance (<50-75 m). We imagine that the Minoan living surface must have dropped away to the sea very rapidly from the southern limit of the settlement, and suggest that this part of the settlement was built out onto sea cliffs that were situated slightly north of the left panel in Fig. 8. This argues for the presence of a bay or harbour serving the ancient town, which must have been located east or west of the survey section. It is clear that further work is needed to establish the exact nature of this portion of the ancient Minoan coastline, and that GPR could be fruitfully used for this purpose. Conclusions The results of our GPR surveys of LBA pyroclastic deposits on Thera indicate that GPR can consistently image LBA deposits to depths of 15 m and can return information on reflectors at depths of 20-25 m. The method can distinguish between the LBA deposits and the underlying Minoan living surface, as well as distinguish between different types of pyroclastic deposits such as air fall vs. pyroclastic flow. The consequences of these points are as follows. First, the distributions and thicknesses of the different components of the LBA deposits can be traced across the island. This has
119
implications for understanding both the volcanic history of the island (e.g. Bond & Sparks 1976; Druitt 1990; Doumas et al 1997), the destruction of the island (Sparks 1979; Sigurdsson et al. 1990; McCoy & Heiken 1994) and nature of the pre-eruption surface (e.g. Rackham 1990; Forsyth 1996; Friedrich et al 2000). For example, the area immediately north of the ancient Akrotiri townsite is underlain by a uniform 10-12m blanket of LBA deposits including pyroclastic fall, surge and flow deposits. The stratigraphic package changes profoundly in nature immediately south of the town site. Second, the interface between the LBA volcanic deposits and the Minoan living surface can be defined and traced even where not exposed, allowing mapping of the palaeotopography (e.g. Heiken et al. 1990; Druitt & Francaviglia 1992; Forsyth 1996). For example, our survey shows that immediately north of the Akrotiri town site lay flat, probably uninhabited land which would have been appropriate for agriculture (e.g. Rackham 1990), and that the town site was probably built out onto sea cliffs. Furthermore, although our surveys did not uncover any direct evidence in the subsurface of man-made structures, these data suggest that future regional surveys could identify walls and buildings belonging to the Minoan civilization. Funding for this research derives from a number of sources. Purchase of the PulseEKKO 100 instrument from Sensors and Software was made possible through an NSERC-Industry partnership grant between the University of British Columbia and Golders Associates. We are indebted to M. Maxwell and J. Schmok for this partnership and for their technical advice. Operational funds for the fieldwork came from NSERC Research Grant A0820 (J.K.R.) and from the University of Toronto (M.V.S.). We are indebted to T. Druitt for his introduction to the geology and volcanology of Thera, and for his critical analysis of results from our field surveys. We also were aided by the efforts and support of G. Vougioukalakis from the Institute of Geology and Mineral Exploration. Lastly, our fieldwork was greatly facilitated with the assistance of P. L. Cowlishaw, L. Stasiuk and L. Zeppos.
References AITKEN, M. J. 1988. The Thera eruption: continuing discussion of the dating. Archaeometry, 30, 165-182. ANNAN, A. P. & DAVIS, J. L. 1977. Radar Range Analysis for Geological Materials. Geological Survey of Canada, Paper, 77-1B, 117-124. ARDON, O. F. P. 1985. Comparison among seismic refraction, electrical resistivity and ground probing radar methods for shallow underground
120
J. K, RUSSELL & M. V. STASIUK
structure investigation. Individual Studies by Participants at the International Institute of Seismology and Earthquake Engineering, 21, 83-98. BOND, A. & SPARKS, R. S. J. 1976. The Minoan eruption of Santorini, Greece. Journal of the Geological Society, London, 132, 1-16. CAMERLYNCK, C, DABAS, M. & PANISSOD, C. 1994. Comparison between GPR and four electromagnetic methods for stone features characterization: an example. Archaeological Prospection, 1, 5-17. CLARKE, K. C. & CROSS, G. M. 1989. Radar imaging of glaciovolcanic stratigraphy, Mount Wrangell caldera, Alaska: interpretation, model and results. Journal of Geophysical Research, 94, 7237-7249. DAVIS, J. L. & ANNAN, A. P. 1989. Ground-penetrating radar for high-resolution mapping of soil and rock stratigraphy. Geophysical Prospecting, 37, 531-551. DOUMAS, C., GURIOLI, L., SBRANA, A. & VOUGIOUKALAKIS, G. 1997. Stratigraphy of the 1628 BC (Minoan) Plinian deposits in the Akrotiri settlement: inferences on precursory phenomena and eruptive scenario of the Minoan event and comparisons with Pompeii and Ercolano archeological settlements. In: Volcanoes, Earthquakes and Archaeology, Volcanic & Magmatic Studies Meeting, April 1997, 17. DRUITT, T. H. 1985. Vent evolution and lag breccia formation during the Cape Riva eruption of Santorini, Greece. Journal of Geology, 93, 439-454. 1990. The pyroclastic stratigraphy and volcanology of Santorini. In: HARDY, D. A. (ed.) Thera and the Aegean World III, Vol. 2. Thera Foundation, London, 27-28. & FRANCAVIGLIA, V. 1992. Caldera formation on Santorini and the physiography of the islands in the Late Bronze Age. Bulletin of Volcanology, 54, 484-493. , EDWARDS, L., MELLORS, R. M., PYLE, D. M., SPARKS, R. S. J., LANPHERE, M., DAVIES, M. S. & BARREIRIO, B. 1999. Santorini Volcano. The Geological Society, London, Memoir, 19. FORSYTH, P. Y. 1996. The pre-eruption shape of Bronze Age Thera: a new model. Ancient History Bulletin, 10.1, 1-10. FRIEDRICH, W. L., SEIDENKRANTZ, M.-S. & NIELSEN, O. B. 2000. Santorini (Greece) before the Minoan eruption; a reconstruction of the ring-island, natural resources and clay deposits from the Akrotiri excavation. This volume. FYTIKAS, M., KOLIOS, N. & VOUGIOUKALAKIS, G. 1990. Post-Minoan volcanic activity of the Santorini volcano. Volcanic hazard and risk, forecasting possibilities. In: HARDY, D. A. (ed.) Thera and the Aegean World III, Vol. 2. Thera Foundation, London, 241-249. GILBERT, J. S., STASIUK, M. V., LANE, S. J., ADAM, C. R., MURPHY, M. D., SPARKS, R. S. J. & NARANJO, J. A. 1996. Non-explosive, constructional evolution of the ice-filled caldera at Volcan Sollipulli, Chile. Bulletin of Volcanology, 58, 67-83. GOODMAN, D. 1994. Ground-penetrating radar simulation in engineering and archeology. Geophysics, 59, 224-232
HEIKEN, G. & McCoY, F. 1984. Caldera development during the Minoan eruption, Thira, Cyclades, Greece. Journal of Geophysical Research, 89, 8441-8462. , & SHERIDAN, M. 1990. Palaeotopographic and palaeogeologic reconstruction of Minan Thera. In: HARDY, D. A. (ed.) Thera and the Aegean World III, Vol. 2. Thera Foundation, London, 370-376. HOLLOWAY, A. L., SOONAWALA, N. M. & COLLETT, L. S. 1986. Three-dimensional fracture mapping in granite excavations using ground-penetrating radar. Canadian Institute of Mining Bulletin, 79, 54-59. JOL, H. M. & SMITH, D. G. 1991. Ground penetrating radar of northern lacustrine deltas. Canadian Journal of Earth Sciences, 28, 1939-1947. & 1992. Geometry and Structure of Deltas in Large Lakes: a Ground-Penetrating Radar Overview. Geological Survey Finland, Special Paper, 16, 159-168. KNOLL, M. D., HAENI, F. P. & KNIGHT, R. J. 1991. Characterization of a sand and gravel aquifer using ground-penetrating radar, Cape Cod, Massachusetts. US Geological Survey, Water Resources, Investigation Report 91-4034, 29-35. LINER, C. L. & LINER, J. L. 1995. Ground-penetrating radar: a near-face experience from Washington County, Arkansas. Leading Edge, 18, 17-21. MARCO, S., AGNON, A., ELLENBLUM, R., EIDELMAN, A., BASSON, U. & BOAS, A. 1997. 817-year-old walls offset sinistrally 2.1 m by the Dead Sea transform, Israel. Journal of Geodynamics, 24, 11-20. MARINATOS, S. 1939. The volcanic destruction of Minoan Crete. Antiquity, 13, 425-439. McCoY, F. W. & HEIKEN, G. 1994. The Minoan eruption; a Late Bronze Age volcanic disaster in Greece. Geological Society of America, 1994 Annual Meeting, Abstracts with Programs, 26, 263. , PAPMARINOPOULOS, S., DOUMAS, C. & PALYvou, C. 1992. Probing the Minoan eruption on Thera with ground-penetrating radar: buried extent of Akrotiri, tephra stratigraphy and lateBronze Age volcanic hazards. Geological Society of America, 1992 Annual Meeting, Abstracts with Programs, 24, 26. MILNER, W. 1997. Forward modelling GPR response over dry volcanic fades: testing the suitability of GPR mapping at Santorini volcano. BASc thesis, University of British Columbia, Vancouver. PRATT, B. R. & MIALL, A. D. 1993. Anatomy of a bioclastic grainstone megashoal (Middle Silurian, southern Ontario, revealed by ground-penetrating radar. Geology, 21, 223-226 RACKHAM, O. 1990. Observations on the historical ecology of Santorini. In: HARDY, D. A. (ed.) Thera and the Aegean World HI, Vol. 2. Thera Foundation, London, 384-391. REA, J., KNIGHT, R. & RICKETTS, B. D. 1994. GroundPenetration Radar Survey of the Brookswood Aquifer, Fraser Valley, British Columbia. Current Research, Geological Survey of Canada, 1994-A, 211-216.
GPR STUDIES OF VOLCANIC DEPOSITS ON THERA RUSSELL, J. K. & STASIUK, M. V. 1997. Characterization of volcanic deposits with ground penetrating radar. Bulletin of Volcanology, 58, 515-527. , SCHMOK, J., NICHOLLS, J. et al. 1998. The Ice Cap of Hoodoo Mountain Volcano, Northwestern British Columbia: Estimates of Shape and Thickness from Surface Radar Surveys. Geological Survey of Canada, Current Research, 1998-A, 55-63. SMITH, D. G. & JOL, H. M. 1992. Ground-penetrating radar investigation of a Lake Bonneville delta, Provo level, Brigham City, Utah. Geology, 20, 1083-1086. SIGURDSSON, H., CAREY, S. & DEVINE, J. D. 1990. Assessment of the mass, dynamics and environmental effects of the Minoan eruption of Santorini volcano. In: HARDY, D. A. (ed.) Thera and the
121
Aegean World HI, Vol. 2. Thera Foundation, London, 100-112. SPARKS, R. S. J. 1979. The Santorini eruption and its consequences. Endeavor, New Series, 3-1, 27-31. STASIUK, M. V. & RUSSELL, J. K. 1994. Preliminary Studies of Recent Volcanic Deposits in Southwestern British Columbia using Ground Penetrating Radar. Current Research, Geological Survey of Canada, 1994-A, 151-157. VAUGHAN, C. J. 1986. Ground-penetrating radar survey used in archeological investigations. Geophysics, 51, 595-604. WILLIAMS, H. & CRONKITE-PRICE, S. M. 1995. Excavations at Stymphalos 1994. Echos du Monde Classique - Classical Views, XXXIX (n.s. 14), 1-22.
This page intentionally left blank
Precursory phenomena and destructive events related to the Late Bronze Age Minoan (Thera, Greece) and AD 79 (Vesuvius, Italy) Plinian eruptions; inferences from the stratigraphy in the archaeological areas RAFFAELLO CIONI1'3, LUCIA GURIOLI1, ALESSANDRO SBRANA1 & GEORGES VOUGIOUKALAKIS2 1
Earth Science Department, University of Pisa, via S. Maria 53, 1-56126, Pisa, Italy Institute of Geology and Mineral Exploration, 70 Messogion, 11527, Athens, Greece 3 Now at: Earth Science Department, University of Cagliari, via Trentino 51, 109124, Cagliari, Italy Abstract: Volcanological studies in the Bronze Age settlement of Akrotiri (Santorini, Greece) and in the Roman towns of Pompeii and Herculaneum (Vesuvius, Italy) have provided information about the precursory phenomena preceding the Minoan and AD 79 Plinian eruptions and the impact of the eruptive products on the human settlements. The Akrotiri settlement was badly damaged by earthquakes before the onset of the eruption. A building debris layer, related to these earthquakes, covers the Minoan soil. The fallout pumice bed, mantling the ruins, freezes a state of partial destruction of the settlement. The deposition of the following pyroclastic flows completed the covering of the site. Strong seismicity also occurred during the opening and the Plinian phases. At the Herculaneum and Pompeii excavations clear evidence of strong pre- and syn-eruptive earthquakes is absent. Herculaneum, just 7 km west of the crater of Vesuvius, was destroyed by several pyroclastic flows, which buried the town under 20m of deposits. Pompeii was covered by a 3 m thick blanket of pumice fall deposit. Distal dilute and turbulent ash clouds reached the town toward the end of the Plinian phase, killing all remaining inhabitants. The following turbulent cloud related to the onset of the caldera collapse completely destroyed the town, which was successively covered by the final phreatomagmatic products of the eruption.
Archaeological sites in volcanic areas represent a sory phenomena and of the impact of volcanic precious and sometimes unique source of data products on the human settlements. The interest relative to the precursors of catastrophic erup- in this type of study is enhanced by the location tions and to the impact of volcanic deposits on of these two volcanoes in moderately to densely human settlements. Two of the most famous populated regions exposed to volcanic hazard. Plinian eruptions in the history of volcanology, Knowledge of the precursors of these large magthe Santorini Island (Greece) Minoan and the nitude events represents an important aspect for Vesuvius (Italy) AD 79 eruptions, struck the Med- the volcanic hazard evaluation of the two areas, iterranean area in different times during the life In the following we will use the term pyroof very advanced cultures. They resulted in the clastic flow to describe any type of primary, gasdestruction of important settlements, greatly rich, pyroclastic gravity flow regardless of its affecting life in the surrounding areas. The transport and depositional mechanism, somestratigraphy and eruptive scenario of these erup- times specifying their concentration (dilute, tions are well known, as a result of earlier dense) or flow regime (turbulent). The term studies. In this paper, we present data from pyroclastic-surge deposits refers to dune-bedded volcanological studies carried out in the archae- deposits deriving from a sequence of pulses, each ological excavations of the Bronze Age settle- related to a turbulent, dilute pyroclastic flow, ment of Akrotiri (Thera Island, Greece), and from the Roman towns of Pompeii and Herculaneum (Vesuvius, Italy), which were destroyed ^he Minoan eruption and buried by these two Plinian eruptions. The aim of this paper is to compare the two events, The Akrotiri settlement is located in the southern mainly stressing differences in terms of precur- part of the island of Thera (Fig. 1). The village, From: McGuiRE, W. G., GRIFFITHS, D. R., HANCOCK, P. L. & STEWART, I. S. (eds) The Archaeology of Geological Catastrophes. Geological Society, London, Special Publications, 171, 123-141. 1-86239-062-2/00/ $15.00 © The Geological Society of London 2000.
124
R. CIONI ET AL.
Fig. 1. Stratigraphy of the deposits of the Minoan eruption cropping out around the Akrotiri excavations. The location of the studied sections is also reported.
inhabited from the middle Early Bronze Age (Doumas 1980), was totally buried by pumices and ashes of the Minoan eruption dated around 1626-1700 BC (Baillie 1990; Friedrich et al. 1990; Hammer & Clausen 1990; Hubberten et al. 1990; Kuniholm 1990; Pyle 1990). The excavations preserve the evidence for damage caused by several earthquakes before
and during the destructive volcanic event. A discussion on this aspect has been reported by Doumas (1980, 1983, 1990). The stratigraphy of the Minoan eruption deposits was described in detail by Bond & Sparks (1976), Pichler & Friedrich (1980), Heiken & McCoy (1984) and Druitt et al. (1989). Sparks & Wilson (1990) revised the stratigraphy
PRECURSORY PHENOMENA AND DESTRUCTIVE EVENTS and the general reconstruction of the eruptive sequence, accepting the general division into four main phases already adopted in the preceding papers. We refer to these workers for the nomenclature and subdivision of the deposits. To correlate the deposits inside the excavations with the general sequence of the eruption, we studied the whole area of the Akrotiri valley, from the more proximal deposits on the caldera wall, to the deposits on the valley bottom and on Akrotiri beach (Figs 1 and 2). Phase 1 deposits represent the Plinian phase of the climactic eruption. The deposits are represented by a pumice fallout sheet, with an ESE dispersal (Bond & Sparks 1976). Fieldwork on the Akrotiri peninsula revealed the existence of three sub-units in the fallout deposit. The basal subunit (SU1) has not been described previously, probably because of its restricted southerly dispersal, contrasting with the direction of
Fig. 2. Schematic geological map of the Akrotiri peninsula.
125
the fan of the main fallout (Fig. 3). It is a crudely stratified, reversely to normally graded bed, characterized by the presence of very peculiar, highly porphyritic, dacitic pumice lapilli. This sub-unit is absent at the base of the reference section for the fall deposit of the Phira Quarry (see, e.g. fig 2-3 of Wilson & Houghton (1990)). The main fallout is formed by two massive, reversely graded sub-units (SU2 and SU3), separated by an ashy surge bed ('phreatomagmatic break' of Heiken & McCoy (1984)). Phase 2 deposits are represented by pyroclastic surge deposits related to a clear phreatomagmatic activity. These deposits consist of a stratified and cross-bedded ashy sequence. Several units were identified, sometimes separated by wet fallout ashes and swarms of ballistic ejecta. At least five depositional units cover the southern sector of Thera; only three of these are present in the sequences inside the archaeological area. During phase 3 massive, wet, phreatomagmatic pyroclastic flow units were emplaced, crudely mantling the topography. They were related to cold pyroclastic flows (McClelland & Thomas 1990) determining the formation of a giant tuff ring (Sparks & Wilson 1990). These flow deposits are missing in the preserved outcrop of the Akrotiri excavations. The Phase 4 deposits are pyroclastic flow deposits that cover the whole island, and reach thicknesses of tens of metres. They generally show drastic vertical and lateral facies variations. On the southern sector of the island, the main facies is represented by lithic-rich, massive
Fig. 3. Isopach map of SU1 of Minoan fallout (continuous lines; in cm). The dashed lines represent the isopachs (cm) of the main fallout, according to Bond & Sparks (1976).
126
R. CIONI ET AL.
to stratified, ash and pumice flow deposits, thickening toward the external coastal plains. Lithic-enriched veneer facies, a few metres thick, drape the caldera border. Coarse, breccia-like facies are exposed in the cliffs of Akrotiri beach, and along the pre-existing valleys (Fig. 2). Precursory activity or opening phase? The climactic eruption was preceded by an opening phase, which corresponds to the precursory activity described by Heiken & McCoy (1990; hereafter HM). We studied these deposits throughout the southern sector of the island, to define the effects of this eruptive phase on the island and on its inhabitants, and also to clarify its role as a real precursor or as a very early opening phase of the eruption. HM distinguished four sub-units in the basal portion of the Minoan deposits (Fig. 4), representing a sequence of events, with two distinct precursory phreatic (sub-unit 1) and phreatomagmatic (sub-unit 2) explosions, and the onset of the climactic eruption (sub-units 3 and 4). In our reconstruction we distinguish a first, phre-
atomagmatic to magmatic event (beds AI and Ap; sub-units 1 and 2 of HM; Fig. 4), followed by a new phreatomagmatic pulse (bed B; subunit 3 of HM). Bed B immediately preceded the onset of the climactic eruption, marked by the lithic-rich, fine lapilli bed of sub-unit 4 of HM. We consider sub-unit 4 not as a real precursory or opening phase of the eruption, but simply as the lithic-rich, fine-grained bed marking the clearing of the conduit and the first phases of raising of the sustained Plinian column. Bed AI is a lithic-rich coarse ash, showing an upward transition to the overlying Bed Ap. Bed AI is characterized by the presence of white rhyolitic pumices. Coatings of the glasses and bubble fillings of fine material reveal the phreatomagmatic character of the bed, in agreement with its high lithic content (about 60%, HM). The bed is spread in a narrow fan, and reaches a maximum thickness of 2cm (Fig. 4). The vent area is difficult to trace, but there are no indications for a different position from that proposed by Heiken & McCoy (1984) for the Plinian Phase 1.
Fig. 4. Stratigraphy of the deposits of the opening phase of the Minoan eruption at Section J5 of Fig. 1. We distinguish a first sub-Plinian event (Beds AI and Ap, I and II of Heiken & McCoy (1990) (HM)) followed by a new pulse (Bed B III of HM). Also shown are the dispersal area of the Minoan opening phase (OP) fallout beds (thickness in mm), and the grain-size analyses of Beds Ap, AI and B.
PRECURSORY PHENOMENA AND DESTRUCTIVE EVENTS Bed Ap is a whitish, pumice-rich, lapilli layer, up to 5cm thick (Fig. 4). Vesicularity of pumices is generally higher than in Bed Ap and the pumices are characterized by fresh-appearing surfaces. These morphological and textural features suggest magmatic fragmentation mechanisms. The larger dispersal area with respect to the underlying Bed Aj (Fig. 4) can be related to an increase in the magma discharge rate, in agreement with the coarser grain size of the deposit. Bed B (Fig. 4) is a yellow-orange, fine to coarse ash layer, often indurated. It is formed by lithic, hydrothermalized lava fragments and pumice lumps. Scanning electron microscopy morphology of pumices always revealed ashy coated glassy surfaces and bubble fillings, typical of phreatomagmatic activity. Bed B is widely distributed on the southern and eastern part of Thera (Fig. 4), whereas in the more lateral outcrops it occurs as a discontinuous, dusty yellow veneer on the pumices of Bed Ap. The magma discharge rate (MDR), calculated for Bed Ap (around 10 6 kgs~ 1 , equivalent to a maximum column height between 7 and 10km, according to the model of Wilson & Walker (1987); Fig. 5) and the volume of the deposit (at least 3 x 10 6 m 3 ) suggest that this eruptive phase lasted not less than 30-40 min. These calculations, even if affected by huge uncertainties, suggest that the settling rate of the pyroclasts was very low. If we assume the pyroclastic shower lasted at least for the entire duration of the event a maximum settling rate of less than lOcmhr" 1 can be calculated. Baxter et al (1998)
Fig. 5. Isopleth map of Bed Ap of the opening phase (average of the largest dimension of the five maximum pumices).
127
suggested that the minimum concentration of inhalable dust (<100/^m) capable of causing asphyxia is 0.1 kgm~ 3 . As only 10wt% (Fig. 4) of the fragments in the deposits is smaller than 100/mi, we can estimate that this threshold was never exceeded on the basis of terminal velocities of the pyroclasts and minimum duration of the event. In this context, it seems unlikely that forced evacuation of the island occurred because of breathing difficulties, as suggested by HM. According to HM, no clear evidence of erosional contact between the different beds was noted, suggesting very short periods of quiescence between their deposition. The gradual nature of the passage from Bed Aj to Ap suggests deposition from a single eruptive pulse, with a character gradually shifting from phreatomagmatic to magmatic. No purely phreatic explosions are recorded by the deposits. Bed B could herald the onset of the climactic phase. The
Fig. 6. Location of the stratigraphic sections studied inside the Akrotiri excavations.
128
R. CIONI ET AL.
interpretation of these levels as a real precursory activity or simply as an opening phase of the eruption is a matter of debate. We think that, apart from terminology, the deposition of Beds AI and Ap, shortly preceded the onset of the main eruption, and the main precursor was the strong seismic activity.
The Late Bronze age settlement of Akrotiri Akrotiri Minoan village is located at the outlet of the Akrotiri valley, near the present-day shoreline. Before the Minoan eruption, the geography was represented by an inclined plateau on the densely welded Cape Riva ignimbrite, gently dipping toward the south in this area (Fig. 2).
Several stratigraphic logs were measured at the Akrotiri excavations (Fig. 6) to reconstruct the eruptive sequence and to observe the variations induced by the buildings in the emplacement of the products. The general stratigraphy is well represented in Section L4 (Fig. 7). The pre-emptive state of the settlement. In the Akrotiri village most of the streets and squares are irregularly covered by a building debris layer made up of metre-sized wall blocks and mortar pieces (Fig. 8a). The presence of this layer is related to the partial destruction of the village caused by one or more earthquake(s) preceding the Minoan eruption (Doumas 1990). The occurrence of clearance and restoration works after the initial earthquake(s) and before the eruption
Fig. 7. General stratigraphy of the Minoan deposits cropping out in the archaeological area of Akrotiri, with a summary of the damage caused by the earthquakes and by the products of the Minoan eruption at Akrotiri.
PRECURSORY PHENOMENA AND DESTRUCTIVE EVENTS
129
is suggested by the presence of piled boulders and large blocks at the sides of the roads and buildings (Doumas 1983). The rubble is covered by the deposits of the opening phase, previously described, and Plinian fallout (Figs 7 and 8a). In many sites a layer of dust mixed with mortar and charcoal (Davidson 1980), up to 10cm thick, is interbedded between the building debris layer and the tephra of the opening phase (Fig. 7). This layer could be related to the deposition and/or following reworking of the dusty material generated during the earthquake-induced collapse of the buildings. Its polymodal grain size rules out the possibility of an aeolian origin for this layer. The lack of erosional features cutting the upper surface of this very soft material suggests that the time interval separating its deposition from the onset of the eruption was short (probably less than a rainy season). The start of the eruption. The first deposits are represented by the fallout Beds A] and Ap of the opening phase (Fig. 7). On the whole, they show a laterally variable thickness, up to 10cm, as a result of the presence of strong irregularities of the substratum. No local reworking of material, for example, because of efforts at clearing up by Minoan inhabitants, has been observed. In several sites (Sections LI, L2, LI7, Fig. 6) the fallout beds of the opening phase are covered by metre-sized blocks resulting from the collapse of the buildings. The limited thickness of this pumice-rich, dry deposit rules out the possibility of collapses caused by overloading of the buildings, so suggesting a strong seismic activity before the onset of the climactic phase. The yellow ashes of Bed B (up to 1 cm in thickness) lie on these blocks (Section LI) and are followed by the main pumice fall deposit (Fig. 8b). The Plinian fallout. The buildings, partially destroyed by the pre and syn-eruptive earthquakes, are buried under a pumice layer with a minimum thickness of 120cm. The internal stratification of the pumice fall deposit is formed by the three main sub-units described Fig. 8. Major damage caused by pre- and syn-eruptive earthquakes at Akrotiri. The building debris layer (BDL) related to the pre-eruptive earthquake(s) covers most of the streets and the squares, and is followed by the opening phase (OP) and the pumice fallout deposits at Section L18. The white bar is 70cm (A). Megablocks of the Minoan buildings, probably related to collapses during syn-eruptive earthquakes, are embedded between Beds A and B of the OP at Section L2 (B), and included in the Plinian fallout deposits at Section LI8 in (C).
130
R. CIONI ET AL.
above. Sub-units 2 and 3 are separated by an ashy surge deposit (Fig. 7). Local thickenings of up to 170-200 cm are essentially due to preferential accumulations close to the buildings. At Section L7 (Fig. 6), in the centre of the village, a wall that has fallen down in the direction of flow is embedded in this surge deposit. This is the only preserved evidence of a wall presumed to have been knocked down by pyroclastic surges in the excavated area. In the southeastern sector of the village building blocks are covered by and included in the main Plinian fall deposit (Fig. 8c). The occurrence of this rubble scattered throughout the sequence of the pumice fallout is strong evidence of continuous, intense, syn-eruptive seismic activity.
The surge deposits. In the Akrotiri area, Phase 2 deposits comprise three depositional units. The first two are constituted by a lower, fine-grained, low-angle cross-stratified layer followed by an upper, coarser, dune-bedded deposit (Fig. 9). They are separated by a vesiculated ash layer. Some impact sags, up to 10cm in diameter, are present. The third unit crops out continuously only in the eastern sector of the Akrotiri excavations. It is a coarse-grained, dune-bedded, lithic-rich, reddish layer, up to 3m in thickness. Ballistic blocks, up to 15cm in diameter, are present. The dry character of the deposits is evidenced by the lack of cohesive plastering of the ashes against the walls, which is shown only by the vesiculated ash separating the two first units
Fig. 9. Section dug parallel to the flow direction (from N to S traced from a photograph) at Section L5. The general coarsening upward of the deposit is mainly related to an increase in the proportion of lapilli layers in the upper unit. In both flow units the wavelengths of the basal and finer-grained dunes (about 4m with amplitudes of several decimetres) are generally shorter than those of the upper, coarser-grained, dunes. The association of poorly sorted progressive dunes topped by fines-rich and vesiculated ash layers indicates tractional transport beneath turbulent, relatively low concentration, unsteady pyroclastic clouds. The reverse grading of the deposits and the increase in dune wavelengths suggest an aggradational deposition from flows of increasing energy. Grain-size analyses of individual laminae were performed at 0.5 intervals by classic dry sieving method for the size down to 40. The analysis of the finest portion was performed using a Cilas 164 laser counter in wet mode at 0.50 intervals. (See the hammer for scale).
PRECURSORY PHENOMENA AND DESTRUCTIVE EVENTS (Section L15, Fig. 6). The finer-grained, laminated ash deposits of the first unit drape the upwind sides of obstacles (essentially the building walls) forming with them angles up to 70°. Downwind of obstacles, they develop low angle (10), lensoid, fines-poor, pumiceous beds (Section L6), suggesting enhanced turbulence. Building debris is scarce in these deposits, which did not scour the substratum. The destructive effects of the pyroclastic surges on the buildings can be inferred in Section L9. An outside wall (60cm thick) is broken in correspondence to the upper, coarser part of the first unit. The finer portion of the pyroclastic surge mantles the buildings. We tried to make an estimation of the minimum velocities of the flow required to suspend the abundant pumice lapilli in these flow units through the grain-suspension criterion (Middleton & Southard 1984; Komar 1985; Lajoie et al 1989). Brissette & Lajoie (1990) suggested that the settling velocity of a clast with mean grain size (Mz) represents the best approximation to the minimum shear velocity of the transporting suspension. Referring to this work, we estimated the mean velocities ( U f ) of the flow, with the assumption of a flow thickness greater than 100m (the height of the pre-existing caldera wall), substratum roughness of 1 m (according to an average height of the walls emerging from the fall deposits) and pure air as transporting medium at 100-200° C (according to the palaeomagnetic data of McClelland & Thomas (1990)). These calculations suggest that Akrotiri was swept by hot, dilute, turbulent currents with velocities of around 10-15 ms" 1 with peaks up to 80 ms^ 1 (Table 1). The destructive effects are related to the first, faster flows (upper portion of the Unit 1). These data are in agreement with the description of Taylor (1958) relative to the 1951 Lamington eruption; in that case dilute pyroclastic flows with velocities of about 33ms" 1
131
completely destroyed the buildings at Higaturu, at 10km from the crater. The ignimbrite. Massive, lithic-rich pyroclastic flow deposits of Phase 4 unconformably cover the underlying pyroclastic units, above an erosive and sometimes channelized contact with Phase 1 and Phase 2 deposits (Fig. 9). Some fine grained sedimentation units, up to 2-3 m thick, are visible, sometimes developing lithic-rich ground layers. The breccia-like units, largely cropping out down-valley from the excavations (Fig. 2), are here totally missing. Probably the presence of the village, sited on a plateau separating two valleys, represented a significant relief for the densest flows, which were deviated by and channelled round the settlement into the topographic lows.
The AD 79 eruption The stratigraphy of the AD 79 'Pompeii' eruption of Vesuvius (Italy) has been studied by several researchers. Sigurdsson and coworkers (Sigurdsson et al. 1982, 1985; Carey & Sigurdsson 1987)
Table 1. Estimates of shear velocities (U*) and mean velocities (Uf) for several surge deposits, at Akrotiri, of phase 2 Section Sample Mz L5 (0)
P
U*
Uf
Unit I
Kol226 3.106 Kol227 3.02 Kol228 2.947 Kol229 -0.262
1700 1700 1700 1200
1 1 1 4
10 11 11
Unit II Kol230 3.286 Kol231 2.516 Kol232 -1.795
1800 1500 750
1 1 6
(kgm- 3 ) (ms- 1 ) (ms- 1 )
51 ±2
9
14
84 ±3
T 100-200°C;/ (substratum roughness) 1m; D (flow thickness) 100-200 m.
Fig. 10. Isopach map (in cm) of Pompei pumice fall deposits (continuous and dashed lines represent white and grey fallout; after Sigurdsson et al. (1985)). The whole area shown in the figure was invaded by pyroclastic flows.
132
R. CIONI ET AL.
revised the general stratigraphy previously proposed by Lirer et al. (1973) and Sheridan et al. (1981), and also described the stratigraphy inside several archaeological sites. Cioni et al, (1991, 1992) proposed a new stratigraphic and chronological scheme for the eruption, describing the effects of the main eruptive phases on the human settlements. The new data reported in the present paper define more precisely the chronology of the destruction of the two main towns of Herculaneum and Pompeii (Fig. 10), and the effects of precursory and syn-eruptive earthquakes. General stratigraphy and timing of the eruption. Cioni et al. (1992) subdivided the AD 79 deposits into eight Eruption Units (EU) (Fig. 11); these were grouped into an opening phreatomagmatic phase (EU1), a Plinian magmatic phase (EU2 and EU3), and a final phreatomagmatic phase (EU4-EU8). They were inserted in the chronological scheme of the eruption
proposed by Sigurdsson et al. (1982, 1985) on the basis of the accounts made by Pliny the Younger in his famous letters to Tacitus on the death of his uncle Pliny the Elder. According to the reconstruction of Sigurdsson et al. (1985) the eruption started after noon on 24 August, AD 79. The first EU is mainly represented by a pisolitic fallout ash layer with an eastern dispersal (Sigurdsson et al. 1985). On the western slopes of the volcano, EU1 is sometimes associated with fine-grained pyroclastic-surge deposits. According to Sigurdsson et al. (1985) and Barberi et al. (1989a), it records a first, transient phreatomagmatic phase marking the clearing of the Plinian conduit. The exact timing of EU1 deposition is not very clear, and this event could be the cause of the first plea for help to Pliny the Elder from the Vesuvian area (Sigurdsson et al. 1985), made around noon. The climactic Plinian phase of the eruption started shortly after EU1 deposition. A sustained, 30km
Fig. 11. Sequence of events during the AD 79 eruption as reconstructed by Cioni et al. (1992). Timing after Sigurdsson et al. (1985) and Cioni et al. (1991). On the left the nomenclature proposed by Sigurdsson et al. (1985) is reported. Sigurdsson et al. (1985) interpreted the eruptive sequence in terms of a prolonged phase of a Plinian sustained column which partially collapsed several times, generating pyroclastic flows and associated ground surges. According to those workers the first pyroclastic flow, which Cioni et al. (1992) located at the passage between white and grey pumices, is within the grey pumice fall deposit, and the phreatomagmatic phase is confined to the final stages of the eruption.
PRECURSORY PHENOMENA AND DESTRUCTIVE EVENTS high (Carey & Sigurdsson 1987), eruptive column rose above the crater, depositing a thick (up to a maximum of 250 cm) pumice fallout blanket to the southeast (see Fig. 17, below). The lower portion of this deposit is formed by a white pumice layer (EU2f), capped in the proximal areas by a thin whitish ash flow deposit (EU23pf). This is followed by a grey pumice fallout (EU3f) interrupted by at least four thin interbeds of ash flow deposits (EU3pf), deriving from partial collapses of the eruptive column (Sigurdsson et al. 1985). Pumice and ash flow deposits close the magmatic phase in the proximal sections of the southern sector. An approximate interval of 7 h (from 1 p.m. to 8 p.m., 24 August) for the white pumice fallout, and of 12 h (from 8p.m., 24 August to 8a.m., 25 August) was estimated by Sigurdsson et al. (1985), on the assumption of a constant sedimentation rate of ^cmhr" 1 at Pompeii. According to these estimates, the first pyroclastic flow occurring during grey pumice deposition (EU3pfl) was emplaced during the night, around 1 a.m., of 25 August. The end of the Plinian phase was followed by a drastic change in the eruptive dynamics, marked by the destabilization of the volcano feeding system and the onset of the caldera collapse. Grain-size and component data (Barberi et al. I9%9a,b; Cioni et al. 1992) suggest an important role of magma-water interaction, reflecting the ingression of phreatic fluids into the feeding system. The beginning of the phreatomagmatic phase is everywhere marked by the radial emplacement of a highly energetic turbulent pyroclastic flow (EU4), following a short-lived restoration of a Plinian column after the end of the magmatic phase of the eruption (Cioni et al. 1992, 1996). Some minor pyroclastic flow deposits (EU5), overlying EU4 in the proximal sectors, are channelled in the main valleys and are followed by a very coarse, lithic rich, breccia flow deposit (EU6). This is overlain by a widespread pyroclastic flow unit (EU7). The waning stages of the eruption are represented by wet, phreatomagmatic, pisolitebearing ash deposits (EU8). The Herculaneum excavations The stratigraphic scheme of Cioni et al. (1992) was applied to the successions in the Herculaneum excavations. Several stratigraphic sections were measured and described (Fig. 12). The proposed correlation and timing of the main eruptive events are significantly different from those proposed by Sigurdsson et al. (1985), who
133
Fig. 12. Map of the Herculaneum archaeological excavations with the locations of the studied sections. interpreted the whole succession of flow units as a sequence of massive, magmatic, pyroclastic flow deposits (F) associated with their ground surge layers (S) at the base (Fig. 13). The deposits (23m thick) in front of the Suburban Thermae of Herculaneum (Fig. 13) are the thickest exposure of the AD 79 pyroclastic flows that buried the town. Several flow units can be recognized. The deposits lie on a platform, along the Herculaneum shoreline, of yellow tuff related to the final phases of the preceding 'Avellino' Plinian eruption of Vesuvius. In some sites the yellow tuff is separated from the overlying pyroclastic deposits by the black sand of a palaeoshore. In the Palestra area (Fig. 12), at the base of the sequence, Sigurdsson et al. (1985) described local remnants of an ash fall (7 mm thick) of the opening phase (EU1).
The deposits of the magmatic phase EU2-3pf. All the skeletons of the human victims are embedded in the deposits of the first unconsolidated flow unit (50cm thick). It is a normally graded ash layer massive at the base and with low-angle cross-stratification at the top (SI of Sigurdsson et al. (1985)). In some sections the lower portion of the deposit is very rich in building materials (columns, roof tiles, rafters). Archaeological evidence suggests a very short transport for these building materials, according to Sigurdsson et al. (1985). Degassing pipes
134
R. CIONI ET AL.
Fig. 13. Stratigraphy of the AD 79 deposits in front of the Suburban Thermae at Herculaneum. The white bar is 100cm. On the right the grain-size and component analyses (from - 4 to cj> - 1) are reported. J/L is the ratio between juvenile fragments (white) and lithic fragments (dark). S/D is the ratio between 'shallow' lithic fragments (lavas and tuffs; dark grey) and deep lithic fragments (cumulites, skarns and marbles; light grey).
develop along the margins of this debris, probably as a result of a channelling effect on the escaping gases. At Villa dei Papiri, inside the Triclini room, the EU2-3pf deposit covers pieces of mortar lying directly on the floor, probably detached by low-magnitude, precursory earthquakes or syn-eruptive tremors. In a section uphill from Herculaneum (Case Sarcinello section, Figs 10 and 14) a similar flow deposit is interlayered between the white (EU2) and grey (EU3) pumice fall layers. The chemical composition of the ash fragments from this deposit is identical to that of the shards from the first flow unit at Herculaneum (Fig. 15), and to the composition found by Cioni et al. (1995) for the deposits at the passage between white and grey pumice in other outcrops. We suggest the EU2-3pf deposit records the white-grey transition during the Plinian fallout phase and was emplaced by a turbulent dilute flow generated by partial collapse of the unstable Plinian column.
The correlation of this layer with the transition phase between EU2 and EU3 is important, because it places the first destructive event that occurred at Herculaneum in the evening of 24 August and not during the early morning of the following day as suggested by Sigurdsson et al 1985. EUSpfl. The first layer is overlain by a lithified massive pumice flow deposit up to 1m thick (EU3pfl). This bed crops out discontinuously in lensoid bodies (Fig. 13). It contains some charcoal, but fragments of tiles and other building materials are generally scarce. Sigurdsson et al. (1985) related this layer (Fl) to the settling from the body of the same cloud that deposited the underlying units (SI, EU2EU3pf). EU3pf2. This deposit is represented by two main beds. The upper, massive, pumice-rich bed
PRECURSORY PHENOMENA AND DESTRUCTIVE EVENTS
Fig. 14. EU2-EU3pf deposits interlayered between white (EU2) and grey (EU3) pumice fallout layers in a section uphill of Herculaneum (Case Sarcinello).
(up to 5m in thickness) has a regular distribution over the town, maintaining a horizontal upper surface, and thickening along the coastline (EU3pf2b, F2 of Sigurdsson et al (1985)). This bed contains some building fragments close only to the base. The basal bed (EU3pf 2a; S2 of Sigurdsson et al. (1985)) is an incoherent, finesdepleted (Fig 13, sample 3), coarse-grained,
135
massive, pumice-rich layer, up to 200cm thick, carrying abundant building material. The matrix is represented by a lithic- and crystal-rich coarse ash. It generally grades into the overlying bed, with a progressive fining upward and increase of the matrix. This layer occurs downwind of obstacles, and thins and fines away from them. It could represent the sedimentation from the head portion of the pyroclastic flow. This 'ground layer' was deposited where obstacles created a high ground roughness, which enhanced turbulence. The strong fines depletion is related to the turbulence and is enhanced by the vaporization of water (sea and swimming-pool), and wood ignition. The high content of building materials and the presence of lithified pieces of the underlying units suggest a highly erosive capability of the cloud. Inside the Triclini room at Villa dei Papiri (Section 11) the destructive effects of this pyroclastic flow are very clear. Tiles, vault fragments and a large wooden girder lie at the base of EU3pf 2, recording the collapse of the roofs. In the same section a fines-poor lens of angular, light grey pumice fragments, with rare lithics lies on the EU3pf2 deposit. This lens could be related to the grey fall deposit.
The deposits of the phreatomagmatic phase EU4-EU8. A fines-poor, discontinuous bed of pumice lapilli and lithic fragments (skarn, marbles and cumulitic rocks) represents the most important discontinuity visible in this area (S3 on the Herculaneum beach and S4 on the
Fig. 15. FeOtot, CaO, and A12O3 frequency histograms showing the composition of the glassy matrix in pumices from EU2-EU3pf deposits, sampled in the Herculaneum excavations and at Case Sarcinello.
136
R. CIONI ET AL.
Palestra wall, according to Sigurdsson et al. (1985)). Its scalloped bottom is clearly related to the impact of ballistic blocks. Impact sags are filled by coarse lapilli with minor ash, probably deriving from the combination of fallout and deposition from the following flow. On the basis of its components and sedimentological features, we correlate this bed with the basal layer present at the bottom of EU4 in the southern sector of the volcano (EU4bl). The 'nube atra' (dark cloud) and the strong earthquakes described by Pliny the Younger at daybreak of 25 August could correspond to this event. The presence of several impact sags at the base of the bed suggests the underlying flow units were already deflated when the fragments settled, possibly indicating a brief time break (less than 5 hr, according to the proposed timing of the eruption; Fig. 11) between the emplacement of the pyroclastic flow and the onset of the phreatomagmatic phase. If this was the case, we should relate the EUSpfl and EU3pf2 at Herculaneum to the ash flow deposits interlayered in the grey pumice fallout in the southern sector (occurring during the night of 24-25 August; Fig. 11), and not to the deposits of the pyroclastic flows marking the closure of the Plinian phase (Fig. 11). The occurrence of a lens of grey fallout deposit at the top of EU3pf2 at Section 11 supports this interpretation. This bed (EU4bl in Fig. 13) marks the transition to the thick (up to 15m), lithic-rich flow deposits, related to the phreatomagmatic
phase of the eruption (F3, F4, F5 and F6 of Sigurdsson et al. (1985)). These deposits thin toward the Palestra wall, where they are topped by the EU8 deposits (Cl of Sigurdsson et al. (1985)), formed by lowangle, cross-stratified, basal ash layers followed by massive, pisolite-rich, fine ash beds. These last pyroclastic flows caused the demolition of the higher part of the buildings and the total burial of the town. The Pompeii excavations Some new excavations have provided at Pompeii (Fig. 16) a good opportunity to make a more precise definition of chronology of the events and assessment of damage produced by the eruption products inside the town.
The deposits of the magmatic phase EU2f-EU3f. The deposits of the Plinian fallout phase crop out extensively in this sector, where they reach their maximum thickness (150cm for the EU2 deposits, 125cm for the EU3). Their deposition caused the partial burial of the town, with anomalously thick accumulations (up to 4m) close to the buildings, as a result of discharge from the inclined roofs. Most of the roofs collapsed under this heavy load, and several victims were found inside the houses under these collapses (Blong, 1984).
Fig. 16. The Pompeii archaeological area with the locations of the studied sections (numbered black dots). Contour lines (in metres) reveal that Pompei was founded on a hill.
PRECURSORY PHENOMENA AND DESTRUCTIVE EVENTS
137
have swept the western part of the plain, south of Vesuvius, for it is always absent in all the outcrops east of the town. The presence of the ash coatings on pumices of the topmost part of the EU3 deposit in Pompeii suggest that that the last tens of minutes of pumice fallout occurred through a very ash-rich atmosphere. The presence of this ash inside the town, although it did not deposit a well-identified layer, could have provoked the death from asphyxia of the outdoors surviving inhabitants, as a result of both the temperature of gases (well above the temperature of condensation, according to the features of the deposits) and the increased fine ash concentration of the air (Baxter et al. 1998).
Fig. 17. Stratigraphic relationships at Villa dei Misteri section, Pompeii (Section 1 Fig. 16). The photo shows a detail of the pyroclastic flow deposits interbedded with and topping the fallout sequence.
EUSpf. The stratigraphy of the pyroclastic flow deposits in the archaeological area is particularly interesting. The first pyroclastic flow that reached the town, just before the end of the Plinian fallout phase, is recorded in a section on its western outskirts, near the Villa dei Misteri excavations (Section 1; Figs 16 and 17). The deposit comprises a 2-3 cm thick ash layer, with rounded pumice lapilli and rare lithic fragments. Following Sigurdsson et al. (1985), we correlate this deposit with the third ash bed (S3 of Sigurdsson et al. (1985)) interlayered in the grey pumice fallout at Villa Regina and Oplontis (Fig. 16). This deposit loses its identity inside the town, where it occurs only as a thin veneer of fine ash on the pumices of the upper 5 cm of the fallout deposit. This very sharp lateral variation is clearly related to a topographical effect. Pompeii was founded on a low hill formed by the deposits of a 19 000-year-old parasitic scoria cone (Di Vito et al. 1998), with a scarp (a palaeocliff) about 10m high on its western edge. The presence of this cliff acted as a barrier for this first flow, as testified by the absence of this deposit in the excavations of Villa dei Misteri (Section Ib, Fig. 16), only 200m from Section 1 and about 10m higher. The topographical effect on the deposition from the cloud suggests that only the basal portion of the cloud had a particle concentration sufficient enough to give a sedimentation rate greater than that of the contemporaneous pumice fallout (estimated at 10-15cmh"1 for this site). This first flow must
EU3pf final deposits. The stratigraphy of the deposits overlying the Plinian fallout is very well exposed in several outcrops along the external perimeter of the excavations and inside the town (Figs 16 and 17). The deposits of at least three more ash clouds are visible (EU3pfa-c in Fig. 17). This very peculiar sequence of flow units is dispersed throughout the plain south of Vesuvius, always closing the fallout sequence. These units correspond to the distal facies of the thick pumice and ash flow deposits covering the Plinian fallout in the proximal sites (Cioni et al. 1992). The destructive potential of these final ash clouds is demonstrated by their effects on the buildings recorded in the deposits just outside Porta Marina and Porta Nola (Sections 4 and 8 Fig. 16). In these locations, a few decimetresized tiles and bricks occur in the deposits of the second flow unit (EU3pfb of Fig. 17). Inside the town (Section 9 in Villa dei Casti Amanti), parts of human skeletons were found at this stratigraphic height. Observations made during the excavation suggest that people were killed by these EU3pf clouds, and the bodies were swept away and buried by the following EU4 deposit (Dal Masco et al. 1999). We conclude that the pyroclastic flows generated by the total collapse of the column at the end of the Plinian phase reached the town and killed the survivors as a result of their high ash concentration and temperature. This probably occurred near the daybreak of 25 August. In that moment, the fate of Herculaneum was sealed.
The deposits of the phreatomagmatic phase The stratigraphy of the more important EUs of the phreatomagmatic phase (EU4, EU7, EUS; Cioni et al. 1992) is extremely clear throughout the plain south of Vesuvius.
138
R. CIONI ET AL.
EU4-EU8. The products of the magmatic phase are always followed by the deposits of EU4, represented by a clear bedset with a basal (EU4bl) coarse, structureless, grain-supported lapilli layer followed by a fining upward sequence of massive to laminated ashes. The bedset is closed by a pisolite-bearing ash bed. The pyroclastic cloud recorded by the EU4 deposits, always well exposed at Pompeii, was clearly responsible for the total destruction of the buildings, and it had an impressive shattering effect even on the stately external walls of the town. Most of the victims were recovered in the ashy portion of this bedset, and the famous moulds of the victims were all poured into this bed. A group of nine victims was found in the new excavations of Via Nucera (courtesy of A. Varone) in the massive basal portion of the bedset, covered by the cross-laminated upper ashes. These bodies were probably rolled and grouped at an embayment of the flow front, as suggested by some wooden poles aligned in the flow direction, testifying to the ability to lift and transport heavy 'obstacles'. EU4 is always overlain by an alternation of two grain-supported, lithic-rich lapilli beds separated by a cohesive ash layer (3cm thick). This sequence is capped by a plane-parallel, laminated, coarse ash layer topped by massive, pisolite-bearing fine ashes (EU7). The sequence is closed by some pisolite-bearing layers (EU8) that mantle the completely ravaged town.
Evidence for seismic activity Historical accounts describe a great earthquake on 5 February, AD 62. It partially damaged buildings in Pompeii, Herculaneum, Nucera and Neapolis, an area not farther than 15km from Vesuvius (Marturano & Rinaldis 1995). The limited extent of the damaged area and the magnitude of the earthquake, estimated at less than five by Marturano & Rinaldis (1995), led those researchers to ascribe the earthquake to a volcanic origin. This contrasts with the conclusions of Sigurdsson et al. (1985), who suggested a tectonic origin for the event. Several indications exist of seismic activity during the decades preceding the eruption: (1) Tacitus (Annals, 15, 33, 2-34, 1 in Ling 1995) reported a tremor causing the collapse of the theatre in Neapolis 2 years after the earthquake of AD 62; (2) Pliny the Younger wrote that there had been tremors for several days before the onset of the eruption (Epistles, 6, 20, 3 in Ling 1995); (3) much restoration work inside houses and on the sewerage system was in progress at Pompeii (Allison 1995; Ling 1995; Varone 1995),
suggesting the possibility of a second earthquake shortly before the eruption; (4) some Roman farms in the neighbourhood of Pompeii were abandoned before the eruption (De Spagnolis Conticello 1995). However, this seismic activity caused no serious damage, and provoked only little alarm. The social economic system of the region was in fact flourishing just before the eruption. Evidence of syn-eruptive seismic activity is very scarce. The presence of pieces of mortar lying directly on the floor at Villa dei Papiri, Herculaneum, suggests the occurrence of lowamplitude tremors before or during the Plinian phase. According to Pliny the Younger, tremors occurred during the night of 24 August, and strong shocks at daybreak on 25 August. We relate this renewal of intense seismic activity to the onset of magma chamber collapse and to the beginning of the phreatomagmatic phase of the eruption. Summary and conclusions The two large Plinian eruptions described show some similarities and several differences with regard to the effects on the human settlements.
Seismic precursors The Minoan eruption was heralded by strong earthquake(s), forcing an early evacuation of the island and causing the destruction of Akrotiri before the onset of the eruption. The fallout pumice bed, mantling the ruins, freezes this state of destruction inside Akrotiri. Archaeological and stratigraphical evidence suggests a short time lapse (probably not longer than a rainy season) separating the paroxysmal seismic activity and the onset of the eruption. No clear evidence exists of intense, destructive seismic activity immediately preceding the AD 79 eruption of Vesuvius. Only some indications of low magnitude earthquakes are reported. Both the eruptions were characterized by precursory seismic activity, with very different magnitudes and consequent effects on the inhabitants. The different precursory activity in the two eruptions could be related to different preemptive conditions of the shallow magmatic system and/or to a different seismic response of the volcanic area. The presence of an outcropping crystalline basement on Thera could have enhanced and channelled seismic waves. In contrast, the thick Pleistocene sedimentary and volcaniclastic pile under Vesuvius could have partially damped volcanic earthquakes.
PRECURSORY PHENOMENA AND DESTRUCTIVE EVENTS
Volcanic precursors No clear evidence of real precursory volcanic phenomena (both magmatic and phreatic) is recorded in the deposits of the two eruptions. The onset of the eruption in both cases was represented by phreatomagmatic pulses of low energy, shortly followed by the Plinian phase. The opening of the Minoan eruption was marked by the establishment of a short-lived, convective column. The Vesuvius AD 79 event was opened by an explosion associated with the deposition of fallout phreatomagmatic ash and poorly dispersed pyroclastic surges. The involvement in the opening phases of the two eruptions of a hydrothermal system is suggestive of pre-eruptive conditions characterized by hydrothermal activity. Likewise, as in several modern eruptions (Mount St Helens), phreatic blasts could have occurred. However, their limited dispersal, restricted to the vent over an area that foundered during the syn-eruptive caldera collapses, prevents their being found in the pyroclastic sequences.
Syn-eruptive seismic activity At Santorini, strong syn-eruptive earthquakes occurred during the opening phase of the eruption, after the deposition of the first pumice bed (Bed A). This strong seismic activity could coincide with the phreatomagmatic explosion (Bed B) involving a shallow hydrothermal system, which immediately preceded the paroxysmal Plinian phase. Strong earthquakes occurred also during the Plinian phase of the eruption, as suggested by the abundant building blocks embedded in the Plinian fallout deposits. The occurrence of high-magnitude events is also suggested by the intense fracturing of the stillstanding, huge building blocks of the houses in Akrotiri. Apart from the strong shocks that occurred at daybreak on 25 August, only low magnitude earthquakes or tremors occurred during the AD 79 eruption. The intensity of the tremors was limited, and poor evidence of destructive effects is recorded in the deposits.
139
the pumice fallout had left wall remains. At Herculaneum, at the foot of the volcano, the first ash flow killed all the inhabitants, pulling down columns, and lifting and transporting tiles. The following flow units knocked down walls, sometimes eroding the underlying deposits and forming channels inside the town. At Pompeii, 10km south of the vent, distal ash clouds enveloped the town at the end of the Plinian phase, leaving very sparse deposits but killing inhabitants. The flow units of the phreatomagmatic phase still had destructive effects at this distance from the vent. A complete destruction of the settlements on the volcano slopes and surroundings was caused by the several pyroclastic flows of the magmatic phase (Cioni et al. 1992). The proposed reconstruction induces us to revise the inferred times of death of the inhabitants; in both cities, most of the inhabitants were killed during the magmatic phase, around 8p.m. on 24 August AD 79 in Herculaneum, only 7 h after the onset of the eruption, and before daybreak, around 17 h after eruption began, on 25 August in Pompeii. At Akrotiri no roofs, or very few of them, were present at the moment of the eruption, because of the destructive effects of pre-eruption earthquakes. No major damage or erosion was caused by pyroclastic flows in Akrotiri. The surging of Phase 2 flows against the village caused mantling of the ruins, and only minor collapses of the walls occurred at the passage of the more energetic flows. The pre-eruptive topography and the village itself channelled the main Phase 4 flows away from the village, avoiding further destruction. We are grateful to C. Doumas (Director of the Akrotiri excavations) and to P. G. Guzzo, M. Pagano and A. Varone (Soprintendenza Archeologica di Pompei) for permission to carry out fieldwork in the Akrotiri excavations and Vesuvius archaeological areas. The suggestions of D. Pyle and anonymous referees contributed to improve the manuscript. Thanks are due to M. Gini for some photographs of Herculaneum. We gratefully acknowledge the financial assistance of the Gruppo Nazionale per la Vulcanologia (GNV) of CNR (Italy), and of the EC.
Effects of eruptive phenomena
References
Extensive roof collapses occurred at Pompeii because of the high mass loading of pumices during the Plinian phase ((1.5-2) x 103 kgm~ 2 ). The impact of pyroclastic flows on towns and farms (Villae Rusticae) of the Vesuvian area was tremendous. They completely knocked down the buildings, even where a first collapse because of
ALLISON, P. 1995. On-going seismic activity and its effects on the living conditions in Pompeii in the last decades. In: Archaologie und Seismologie. Biering & Brinkmann, Munchen, 183-190. BAILLIE, M. G. L. 1990. Irish tree rings and an event in 1628 BC. In: HARDY, D. (ed.) Thera and the Aegean World III, Vol. 3. The Thera Foundation, London, 160-166.
140
R. CIONI ET AL.
BARBERI, F., CIONI, R., Rosi, M., SANTACROCE, R. SBRANA, A. & VECCI, R. 19890. Magmatic and phreatomagmatic phases in explosive eruptions of Vesuvius as deduced by grain-size and compositional analysis of pyroclastic deposits. Journal of Volcanology and Geothermal Research, 38, 287-307. , , SANTACROCE, R., SBRANA, A. & VECCI, R. 1989&. Variazioni laterali nella distribuzione dei componenti nei depositi dell'eruzioni pliniane. Bollettino Gruppo Nazionale per la Vulcanologia, 2, 595-609. BAXTER, P. T., NERI, A. & TODESCO, M. 1998. Physical modelling and human survival in pyroclastic flows. Natural Hazard, 17, 163-176. BLONG, R. J. 1984. Volcanic Hazard. Academic Press, Sydney, Australia. BOND, A. & SPARKS, S. J. 1976. The Minoan eruption of Santorini, Greece. Journal of the Geological Society, London, 132, 1-16. BRISSETTE, F. C. & LAJOIE, J. 1990. Depositional mechanics of turbulent nuees ardentes (surges) from their grain sizes. Bulletin of Volcanology, 53, 60-66. CAREY, S. & SIGURDSSON, H. 1987. Temporal variations in column high and magma discharge rate during the AD 79 eruption of Vesuvius. Geological Society of America Bulletin, 99, 303-314. CIONI, R., CIVETTA, L., MARIANELLI, P., METRICH, N., SANTACROCE, R. & SBRANA, A. 1995. Compositional layering and syneruptive mixing of a periodically refilled shallow magma chamber: the AD 79 Plinian eruption of Vesuvius. Journal of Petrology, 36(3), 739-776. , MARIANELLI, P. & SBRANA, A. 1991. L'eruzione del 79 d.C.: stratigrafia dei depositi ed impatto sugli insediamenti romani nel settore orientale e meridionale del Somma Vesuvio. Rivista di studi Pompeiani, 4, 179-198. , & 1992. Dynamics of the AD 79 eruption: stratigraphic, sedimentologic and geochemical data on the successions of the SommaVesuvius southern and eastern sector. Acta Vulcanologica, Marinelli, 2, 109-124. , SBRANA, A. & GURIOLI, L. 1996. The deposits of AD 79 eruption. In: SANTACROCE, R., Rosi, M., SBRANA, A., CIONI, R. & CIVETTA, L. (eds) Vesuvius Decade Volcano Workshop Handbook. International CEV-CMVD Workshop on Vesuvius, Pisa, September 1996. Consiglio Nazionale delle Ricerche, Italy. DAL MASCO, C., MARTURANO, A., VARCOME, A. (1999). Pampei, il caccomto dell'eruzionie. Le Scienze, 371, 58-65. DAVIDSON, D. A. 1980. Aegean soils during the second millennium BC with reference to Thera. In: DOUMAS, C. (ed.) Thera and the Aegean World, Vol. 1. The Thera Foundation, London, 725-739. DE SPAGNOLIS CONTICELLO, M. 1995. Osservazioni sulle fasi edilizie di alcune ville rustiche di Scafati, suburbio orientale di Pompei seppellite dalla eruzione del 79 d.C. In: Archaologie und Seismologie. Biering & Brinkmann, Munich, 93-102.
Di VITO, M., SULPIZIO, R., ZANCHETTA, G. & CALDERONI, G. 1998. Geo-volcanological studies on densely inhabited areas: an example from the south western slopes of Somma-Vesuvius, Italy. Acta volcanologica, 10, 383-393. DOUMAS, C. 1980. The stratigraphy of Akrotiri. In: DOUMAS, C. (ed.) Thera and the Aegean World, Vol. L The Thera Foundation, London, 777-782. 1983. Thera: Pompeii of the Ancient Aegean, London. 1990. Archaeological observations at Akrotiri relating to the volcanic destruction. In: HARDY, D. (ed.) Thera and the Aegean World III, Vol. 3. The Thera Foundation, London, 48-50. DRUITT, T. H., MELLORS, R. A., PYLE, D. M. & SPARKS, R. S. J. 1989. Explosive volcanism on Santorini, Greece. Geological Magazine, 126, 95-126. FRIEDRICH, W. L., WAGNER, P. & TAUBER, H. 1990. Radiocarbon dated plant remains from the Akrotiri excavation on Santorini, Greece. In: HARDY, D. (ed.) Thera and the Aegean World III, Vol. 3. Thera Foundation, London, 188-196. HAMMER, C. U. & CLAUSEN, H. B. 1990. The precision of the Ice-Core dating. In: HARDY, D. (ed.) Thera and the Aegean World III, Vol. 3. Thera Foundation, London, 174-178. HEIKEN, G. & McCoY, F. 1984. Caldera development, during the Minoan eruption, Thera, Cyclades, Greece. Journal of Geophysical Research, 89, 8441-8462 & 1990. Precursory activity to the Minoan eruption, Thera, Greece. In: HARDY, D. (ed.) Thera and the Aegean World III, Vo. 2. Thera Foundation, London, 79-88. HUBBERTEN, C. U., BRUNS, M., CALAMIOTOU, M.,
APOSTOLAKIS, C., FILIPPAKIS, S. & GRIMANIS, A. 1990. Radiocarbon dated from the Akrotiri excavations. In: HARDY, D. (ed.) Thera and the Aegean World III, Vol. 3. The Thera Foundation, London, 179-187. KOMAR, P. D. 1985. The hydraulic interpretation of turbidites from their grain sizes and sedimentary structures. Sedimentology, 32, 395-407. KUNIHOLM, P. I. 1990. Overview and assessment of the evidence for the date of the eruption of Thera. In: HARDY, D. (ed.) Thera and the Aegean World III, Vol. 3. The Thera Foundation, London, 13-18. LAJOIE, J., BOUDON, G. & BOURDIER, J. L. 1989. Depositional mechanics of the 1902 pyroclastic nuees ardente deposits of Mt. Pelee, Martinique. Journal of Volcanology and Geothermal Research, 38, 131-142. LING, R. 1995. Earthquake damage in Pompeii I 10: one earthquakes or two? In: Archaologie und Seismologie. Biering & Brinkmann, Munich, 201-209. LIRER, L., PESCATORE, T., BOOTH, B. & WALKER, G. P. L. 1973. Two Plinian pumice-fall deposits from Somma-Vesuvius, Italy. Geological Society of America Bulletin, 84, 759-772. MARTURANO, A. & RINALDIS, V. 1995. II terremoto del 62 d.C.: un evento carico di responsabilita. In: Archaologie und Seismologie. Biering & Brinkmann, Munich, 131-135.
PRECURSORY PHENOMENA AND DESTRUCTIVE EVENTS MCCLELLAND, E. & THOMAS, R. 1990. A palaeomagnetic study of Minoan age tephra from Thera. In: HARDY, D. (ed) Thera and the Aegean World III, Vol. 2. The Thera Foundation, London, 129-138. MIDDLETON, G. V. & SOUTHARD, J. B. (eds) 1984. Mechanics of Sediment Movement. Society of Economic Paleontologists and Mineralogists, Tulsa, Okla. PYLE, D. M. 1990. The application of tree-ring and icecore studies to the dating of the Minoan eruption. In: HARDY, D. (ed) Thera and the Aegean World III, Vol. 3. The Thera Foundation, London, 167-173. PICHLER, H. & FRIEDRICH, W. L. 1980. Mechanism of the Minoan eruption of Santorini. In: DOUMAS, C. (ed). Thera and the Aegean World, Vol. 1. The Thera Foundation, London, 15-30. SHERIDAN, M. F., BARBERI, F., Rosi, M. & SANTACROCE, R. 1981. A model for Plinian eruptions of Vesuvius. Nature, 289, 282-285. SIGURDSSON, H., CAREY, S., CORNELL, W. & PESCATORE, T. 1985. The eruption of Vesuvius in AD 79. National Geographic Research, 1, 332-387. , CASHDOLLAR, S. & SPARKS, R. S. J. 1982. The eruption of Vesuvius in AD 79: reconstruction from historical and volcanological evidence. American Journal of Archeology, 86, 39-51. SPARKS, R. S. J. & WILSON, C. J. N. 1990. The Minoan deposits: a review of their characteristics and
141
interpretation. In: HARDY, D. (ed.) Thera and the Aegean World III, Vol. 2. Thera Foundation, London, 89-99. TAYLOR, G. A. M. 1958. The 1951 eruption of Mount Lamington, Papua. Bureau of Mineral Resources of Australia. Bulletin, 38. 2nd edition (published 1983) Australian Government Publishing Service. VARONE, A. 1995. Piu terremoti a Pompei? I nuovi dati degli scavi in via dell'Abbondanza. In: Archaologie und Seismologie. Biering & Brinkmann, Munich, 29-35. & MARTURANO, A. 1999. Prosecuzione dello scavo lungo via della'Abbondanza: L'eruzione vesuviana del 24 agosto del 79d.C. attraverso le lettere di Plinio il Giovane e le nuove evidenze archeologiche. Rivista Studi Pompeiani (submitted). WILSON, C. J. N. & HOUGHTON, B. F. 1990. Eruptive mechanisms in the Minoan eruption: evidence from pumice vesicularity. In: HARDY, D. (ed.) Thera and the Aegean World III, Vol. 2. Thera Foundation, London, 122-128. WILSON, L. & WALKER, G. P. L. 1987. Explosive volcanic eruptions - VI. Ejecta dispersal in plinian eruptions: the control of eruption conditions and atmospheric properties. Geophysical Journal of the Royal Astronomic Society, 89, 657-679.
This page intentionally left blank
A geographical information system for the archaeological area of Pompeii M. T. PARESCHI1, G. STEFANI2, A. VARONE2, L. CAVARRA1,
F. GIANNINI1 & A. MERIGGI1
1
Centra di Studio per la Geologia Dinamica e Strutturale dell'Appennino, CNR, via S. Maria 53, 1-56100 Pisa, Italy 2 Soprintendenza Archeologica di Pompeii, Pompeii, Italy Abstract: The increasing degradation of archaeological evidence in the circumvesuvian area requires a suitable and exhaustive census of all the archaeological features of the area that are accessible now or that have come to light at some time in the past. The high population density of the area demands a conservation policy based on appropriately planned prevention, rather than on occasional intervention with the criterion of 'saving what can be saved'. Here we present a geographical information system (GIS) of the archaeological data in their territorial environment for the circumvesuvian area, with the aim of providing a useful tool for archaeological feature management. The database includes the following: (1) Current information on the circumvesuvian area (18 council districts with a total area coverage of about 250km2). The database consists mainly of vector data at the original scale of 1: 5000 (although raster images are also included), and it is updated to the period 19881995, depending on the zone. The digital information includes natural and man-made features: roads, buildings, railways, spot heights, etc. (2) Archaeological sites. The 'objects' are organized in three main 'layers': areas with building constraints, buried elements (of uncertain and certain location), visible constructions (e.g. aqueduct, amphitheatre, honorary arch, forum, harbour structures, theatre, temple, roads, villas, etc.). The census is at present under development. To date, about 300 structures have been included in the GIS in the area of interest.
On the morning of 24 August, AD 79, a tremen- of late of Satyrs set afoot their dances; this dous explosion shook the territory around [Pompeii] was the haunt of Venus, more pleasant Mount Vesuvius, which had been silent for to her than Lacedaemon; this spot was made many centuries. From the mouth of the volcano, glorious by the name of Hercules. All lies a column of gas and volcanic debris shot drowned in fire and melancholy ash; even the thousands of metres into the air with incredible High Gods could have wished this had been violence. A terrible rain of lapilli and ashes fell, permitted them.' Or again, Statius (see Slater submerging Pompeii and the surrounding area 1908), only a few years after the eruption, wrote: (Macedonio et al. 1988). The land was covered 'Will future generations believe, when the crops by a blanket two and a half metres thick, more in grow again and these deserts flower, that here some places, depending on the contours of the beneath their feet, beneath the weight of the terrain. At tremendous velocity, hot pyroclastic earth, lie cities and people and that the ancestral flows flowed down the slopes of Vesuvius fields were thus swallowed up.' (Sigurdsson et al. 1985). The violence of the The AD 79 eruption, burying ancient towns eruption, which led to the complete disruption of and localities under a thin covering of pyroclasthe cities around the volcano (Pompeii, Hercu- tic material, has passed on to us testimonies of laneum, Stabiae, etc.), left contemporary society that world, in terms of both the number and shocked and horrified. The poet Martial (see Ker great variety of man-made structures. Today the 1947) interpreted the common sentiment with his area has a population of around 600 000, with a verse: This is Vesbius [Vesuvius], green yester- population density that in some cases is greater day with viny shades; here had the noble grape than that of Hong Kong. The high demographic loaded the dripping vats; these ridges Bacchus pressure risks cancelling indefinitely or altering loved more than the hills of Nysa; on this mount these ancient testimonies. This is the reason why From: McGuiRE, W. G., GRIFFITHS, D. R., HANCOCK, P. L. & STEWART, I. S. (eds) The Archaeology of Geological Catastrophes. Geological Society, London, Special Publications, 171, 143-158. 1-86239-062-2/00/ $15.00 © The Geological Society of London 2000.
Fig. 1. The area around Vesuvius. A Landsat TM image is draped on a digital elevation model (DEM), to obtain a 3D perspective view.
Fig. 2. (a) The DEM of the area, (b) The area (c. 250 km2) around Vesuvius, where data in vector format are available. Black, buildings; red, roads; yellow, railways; blue, administrative boundaries; light blue, main contour lines and coastline.
A GIS FOR THE ARCHAEOLOICAL AREA OF POMPEII
145
Fig. 2. (continued)
there is an urgent need to catalogue and georeference the archaeological resources in the Vesuvian area with a view to their protection, maintenance and development. The need to locate precisely the archaeological sites investigated, either with excavation campaigns or following chance finds, so as to make an overall study of human presence in an area is a requirement that appeared only in the final decades of the nineteenth century. However, it was only at the beginning of the twentieth century that such ideas were formalized on a
national scale. The first proposal for the production of an archaeological map was by Cozza in November 1881, and was accepted by the Ministry of Education, which instituted an ad hoc scientific committee headed by Gamurrini et al (1972). This was followed by the creation of a specific Office for the Archaeological Map of Italy, with the same staff, to which Mengarelli was added (Bencivenni et al. 1992). It was the combination of the immense archaeological wealth of the Pompeiian area and the dedication of site archaeologists, such
146
M. T. PARESCHI ET AL.
as Sogliano and Cozza, which prompted the proposal for an archaeological map specific to the area. The map still presents a very modern methodological organization despite the fact that it dates back to the beginning of the century. Sogliano proposed the location of archaeological resources on an official cadastral map, with a scale of 1:2000. This would make it possible to show the perimeters of buildings, locate and align them exactly, and give essential information for the definition of the reconstruction of transport routes, residential districts and property boundaries, and illustrate the organization of the ancient urban infrastructure. However, the project put forward by Sogliano had no follow-up, apparently because of lack of personnel and means. After numerous out-oftown archaeological excavations, carried out for the most part by private enterprises and not always with great scientific rigour, which brought to light many interesting buildings throughout the area surrounding Pompeii, there were only occasional discoveries following new building operations. These finds were investigated merely with the aim of protecting the structures found. This great mass of data unfortunately has never had any supporting cartographic production. Therefore, in many cases it is not possible to trace the exact location of some important sites and it is not possible to have a legislative guarantee of conservation of the sites if they are not visible. In fact it is only recently that two local scholars, Bianco & Casale (1979), bridged this severe information gap. The survey was limited to the area to the north of Pompeii. It had the inevitable uncertainties and inaccuracies caused by difficulties of obtaining data. The data were often completely missing, and verification of information was possible only through frequently inadequate reports of the official authorities. However, the survey is still a precious research source, indispensable from a scientific and administrative point of view, and has been consulted time and time again, often uncritically, by numerous scholars working on the topography of the countryside around Pompeii. More recently, another attempt at an archaeological map was made by the Consorzio Neapolis supported by the Law on Cultural Resources (Neapolis 1994). The Territorial Archaeological Census produced will soon be made available again for consultation but requires careful and in-depth verification, correction and updating. The present work has the aim of the definitive production of a digital archaeological map of the circumvesuvian area (using a geographical information system (GIS)), involving collaboration
between the Office of Archaeological Resources of Pompeii (Soprintendenza Archeologica di Pompei) and the CSGSDA-CNR of Pisa. The project foresees computerized cartography of the area. It will be possible both to locate archaeological discoveries outside the built-up districts and to define topographically the urban infrastructure of the ancient settlements. There will also be analytical identification of the single components of this infrastructure. With this aim in mind there will be a thorough survey of the local administrative offices to obtain all the useful cartographic data and archive documentation. Particular attention will be paid to aerial photography documentation carried out on the territory in general and on the ancient sites in particular. This will identify preexisting sites now destroyed or no longer visible in areas that have undergone disruptive and extensive alteration in recent times. This paper presents the current status of a GIS of the circumvesuvian area (Fig. 1), aimed at the management of this exceptional archaeological heritage. GIS features Geographical information systems (GISs) are computer-based systems used to store and Table 1. Principal layers and sub-layers of the GIS of Vesuvius (natural and artificial elements) Natural features contours spot heights coastline rivers or channels lakes Roads motorways roads footpaths Railways Man-made structures buildings partition walls bridges greenhouses sports facilities sheds Green (vegetated) areas public green areas agricultural areas Administrative borders Toponymy street names, toponymy of first, second and third order of localities, rivers, etc.
Fig. 3. The data in vector format of the GIS, related to natural and man-made features. Output on video, (a) Upper Vesuvius: red, principal contour lines; blue, contour lines (at 5 m interval); green, pathway; white, spot heights, (b) The area southwest of Vesuvius (Herculaneum and Torre del Greco): white, buildings; blue and red, contour lines, (c) The area west of Vesuvius: white, buildings; red, roads, (d, e) The area of Torre Annunziata: blue, coastline; green, public green areas; white, spot heights; violet, greenhouses; light blue, rivers or channels; pink, sheds; light green, electric power lines, (f, g) Population density.
Fig. 3. (continued)
A GIS FOR THE ARCHAEOLOICAL AREA OF POMPEII
Fig. 3. (continued)
149
150
M. T. PARESCHI ET AL.
Fig. 3. (continued}
manipulate geographical information. They are designed to 'support the capture, management, manipulation, analysis, modelling and display of spatially referenced data for the solution of complex planning and management problems' (Aranoff 1989). In other words, GISs are designed for the collection, storage and analysis of objects and phenomena for which geographical location is an important characteristic or critical to the analysis itself. GISs are the result of linking parallel developments in many separate spatial data processing disciplines: digital cartography, computer-aided design and computer graphics, surveying and photogrammetry, spatial analysis using rasterized data from thematic maps, interpolation from point data and remote sensing technology (Burrough 1989). Because data from different sources can be accessed and transformed interactively, newly derived information can be easily and usefully generated. The manipulation of data extends from the simple overlay of different thematic maps for the identification of areas with specific required conditions, to the more sophisticated use of mathematical operators or integrated numerical models. Vector data and raster images, organized in layers, are the two main categories of objects managed by the GIS. A layer represents groups of similar elements of the database, and it constitutes the GIS's informative level. A layer is a complex structure consisting of various sub-layers that represent
different typologies of the same theme. For example, the layer 'buildings' has as sub-layers 'productive commercial building', 'generic public building', 'houses' and so on. Each sub-layer is univocally identified by a number code, for the retrieval of information. Different thematic layers can be compared or merged. This convenient system permits the user to maintain an extremely complex drawing in an unlimited number of constituent parts and to be able to simplify his or her view at any given moment by selecting which of those constituent parts will be visible and which will be suppressed. The layer allows identification, as on a transparent page in an anatomy atlas, of items of a similar type. Similarly, all the items of the same type or different types can be visualized, allowing subtractions or additions of the various typologies, to outline relationships between different items of information. The selection and combination of certain layers from among those present in the database is one of the most important characteristics of the GIS. In this way, it is possible to bring out only those elements that might be relevant to the enquiry, making the map produced clear and easily interpretable, and easily readable with reference to the analysis to be carried out. The objects can be represented on video following a dimensional criterion: only the objects with an area or a 'linear dimension' greater than a given threshold can be drawn. In practice, therefore, it is possible to decide what level of
Fig. 4. For each object, a file is available with alphanumeric information.
Fig. 5. The archaeological features in the GIS (pink dots).
152
M. T. PARESCHI ET AL.
Fig. 6. Ancient Pompeii (output on video). Different zoom views are used (a-d). Only generic buildings are represented.
detail is required by the image on the basis of the scale of the area to be drawn. The software used is ARCvmwS.O. In this framework, the spatial data are linked to attribute tables, containing alphanumeric additional information, within a DBMS (Data Base Management System) framework. The operations, interactively selected from a menu, can act on the spatial information (topological operators) or on the alphanumeric attributes, or on a combination of both. The basic functions are: management of point, linear and areal objects; selection of objects (according to spatial characteristics, or alphanumeric attributes); management of symbols; management of polygons; thematic analyses; plotting of the maps; management of different combined functions; etc.
In this way it is possible to carry out topological and logical operations: to display all the 'objects' near (e.g. at a distance less than a given value) a street or near a building, or inside a municipal district or near an industry at risk, etc.; to select and display all the man-made structures of a given age; to display all the protected areas in a given zone, etc. The data The present database refers to an area of about 250km2 around Vesuvius (Fig. 1). The available information consists of (1) a digital elevation model (DEM), (2) high-resolution satellite images, (3) a database on natural and man-made
A GIS FOR THE ARCHAEOLOICAL AREA OF POMPEII features in a vector format (roads, railway, contour lines, buildings, etc.), (4) details on archaeological structures.
The digital elevation model The available topography consists of contour lines defined by a set of x,y coordinates at constant elevations. These files contain the necessary information for obtaining a 3D representation. A suitable algorithm (Macedonio & Pareschi 1991) approximates terrain surfaces as a network of planar triangles (triangulated irregular network; TIN), where vertices are points of known elevation. Triangulation provides a means for using data in a manner that facilitates decisions regarding the site. For the area (Fig. 2a) there is a DEM at a scale of 1:25 000 (Istituto Geografies Militare source), within an area of 2500km2, and a DEM at a scale of 1: 5000, in an area of 250 km2 (Pareschi & Santacroce 1993).
Digital images from satellite or aircraft The sources are Landsat TM, SPOT, MIVIS, IRS-1C. These images are registered together and with digital terrain, to obtain coincident
Fig. 7. A detail of ancient Pompeii. Output on video.
153
maps for the same scene. A recent (1997) image (IRS-1C) with a spatial resolution of 6 m is available, and is used as support for the database.
Database on natural and man-made features The information, in a vector format, is available in the above-mentioned area of 250km2 (Fig. 2b). The original scale of the data was 1:5000. The current database is organized into seven main layers (each organized into sublayers), according to Table 1: natural features, roads, railways, man-made structures, green (vegetated) areas, administrative borders, toponymy (Fig. 3a-e). Information on the population densities (which reach peaks of 20000 inhabitants per square kilometre near the coast, Fig. 3f and g) are also available (source: 1991 ISTAT census). Each of the items mentioned above is memorized according to georeferenced lines and points. An important characteristic of the structure of the databank is the association of each object with a descriptive alphanumeric file: the latter, visualized by selecting the relevant object with the appropriate function, lists the different characteristics of the object itself (Fig. 4). These
154
M. T. PARESCHI ET AL.
Table 2. List of thematic items in the third layer (visible structures) for Archaeological resources 101100 Private buildings for living purposes 101101 house 101102 house with atrium (Italic type, with or without lateral areas, with or without viridarium or xystus, characterized by the presence of impluvia 101103 house with atrium and peristyle (or posticus) 101104 unified house or house with more than one atrium and/or peristilia (created by the merger of various living units) 101105 apartment on upper floor 101106 pergula (storerooms or galleries inside shops, accessible by means of wooden or brick staircase) 101107 unclassified house 101108 unidentified house type (partially excavated or reburied) 101109 postico (secondary entrance) 101110 muitima 101111 villa rustice 101112 farmhouse 101113 nymphaeum 101114 farm boundary wall 101115 destined for living or production 101116 destined for living or commercial 101117 destined for production or commercial 101118 destined for living or production or commercial 101200 101201 101202 101203 101204 101205 101207 101208 101209 101210 101211 101212 101213 101214 101215 101216
Commercial activities (retail or sale of services) veterarius: second-hand dealer caupona or taberna: tavern hospitium: inn (rooms for rent) bakery fishmonger taberna argentaria: banker's office, money exchange taberna lactaria: milk seller taberna pomaria: fruiterer thermopolium: drinks seller, 'bar' tonstrina: barber taberna asin ui: donkey driver's shop taberna dissigmatous: office for organization of entertainment taberna lusoria aleariorum: gambling house, dice games room school commercial activity uncertain
101300 Artisan activities (for working or transformation of products) 101301 fullonica: laundry 101302 officina coactiliaria: felt workshop 101303 officina lanificaria: wool workshop 101304 officina libraria or scriptoria: bookshop, copyist, writer of edicts and electoral manifestos 101305 officina lignaria or lignaria plostraria: carpentry workshop or for cart or chariot repairs 101306 officina musivaria: mosaicist's workshop
101307 officina sutoria: footware 101308 officina textoria: textiles workshop 101309 officina tinctoria: dyeworks 101310 building with artisan activity uncertain 101400 101401 101402 101403 101404 101405 101406 101407 101408 101409 101410 101411 101412 101413 101414 101415 101416 101417 101418 101419 101420
Activities of production and retail (shops or activities for working, transformation and retail of products) officina aeraria: foundry, workshop for bronze and copper working officina coriariorum: leather workshop officina gari: production offish sauces production of oil-lamps officina lucernarum lucerne: soap works officina eboraria: ivory workshop officina ferraria: smithy, blacksmith's workshop officina marmoraria or statuaria: workshop of marble worker or sculptor officina pigmentaria: workshop for production of colours officina and/or taberna olearia: workshop for production and/or retail of oil officina tegetaria: workshop for production of matting officina vasaria: workshop for production of pots or vases officina vestiaria: workshop for production of clothes officina and/or taberna vinaria: workshop and/or shop for production and/or retail of wine officina vitraria: glass workshop pistrinum: baker's shop, production of flour and bread pistrinum dolciarium: pastry shop, production of cakes perfumes workshop pharmacy building with production or retail uncertain
101500 Activities or shops of production 101501 warehouses attached to shops, workshops or other types of facilities and commercial activities 101502 cella vinaria: wine cellar 101503 backshop (deposits and/or living area attached to shops, workshops, and commercial activities) 101600 101601 101602 101603 101604 101605 101606 101607 101608 101609 101610 101611
Coverings and structural elements of buildings column splayed loop-hole window grating moenianum opaion upper floor door tiled roof reinforced concrete roof penthouse
A GIS FOR THE ARCHAEOLOICAL AREA OF POMPEII
155
Table 2. (continued) 101700 Public buildings for entertainment 101701 amphitheatre 101800 101801 101802 101803
Thermal baths thermal bath buildings private baths (in private houses) public baths
101900 101901 101902 101903 101904 101905 101906
Sacred buildings shrine sacellum votive statues deposit temple sanctuary generic sacred building
102000 Sporting facilities 102001 caserma: meeting place or training area for gladiators 102002 palaestra or gymnasium 102003 generic sports building 102100 102101 102102 102103 102104 102105 102106 102107 102108 102109 102110 102111 102112 102113 102114 102115 102116 102117 102118 102119 102120
Other public buildings honorary arch basilica comitium administrative building generic public building forum or porticus olitorium forum archway horrea port facilities latrinae macellum mensa ponderaria wall gate porticus post scaenam specula theatre tower public spaces or areas
102200 Road system 102201 pavement 102202 angiportus: pedestrian precinct (alleyways marked with a street number) 102203 bridge 102204 road 102205 cardo 102206 decumanus 102207 provincial road 102208 road connecting farm boundaries 102300 102301 102302 102303
Boundaries insula street number regio
102400 Quarters or areas of corporations 102401 sodalicium or collegium (reserved for meetings or guilds) 102402 static: post house 102500 102501 102502 102503 102504 102505 102506 102507 102508 102509
Water works aqueduct canal system castellum aquae cistern drainage system fountain well water tower piping
102600 102601 102602 102603 102604 102605 102606 102607 102608 102610 102611 102612 102613 102614 102615 102616 102617 102618 102619 102620
Funerary Buildings funerary monument necropolis funerary enclosure tomb tomb with altar tomb with stepped altar columbarium tomb with column on podium tomba with shrine barrel-vault tomb pit tomb Naiskos tomb tomb with niches tomb with enclosure tomb with schola or exedra tambour tomb funerary triclinium tomb of unidentified type ustrium: place destined for cremation of the dead
102700 102701 102702 102703
Brothels cella meretricia brothel, specific separate building brothel as annex of private house
102800
Unidentifiable building
102900 Building of uncertain use 103000 103001 103002 103003 103004 103005 010306
Green areas cultivated area garden garden or vegetable garden vegetable garden vineyard nursery
103100 103101 103102 103103 103104 103105
Orography-hydrography in AD 79 river lake lagoon coastline contour
(continued}
M. T. PARESCHI ET AL.
156
Table 2. (continued) 103106 marsh 013107 spot heights 103100 Structures of uncertain interpretation 103200 Miscellaneous material 103300 103301 103302 103303 103304
Land use arboretum wood quarry vegetable crops
mainly concern geographical position (longitude, latitude and altitude of the part of the object selected), the graphic characteristics (type of line, colour, etc.), the layer of the object, the real dimensions of the object, the scale of acquisition, the last updating, the seismic vulnerability (Angeletti et al. 1998), etc. The characteristics listed vary according to the type of selected object.
The archaeological information There are five principal information representations relating to the management of archaeological data: (1) maps, which illustrate the locations of these entities in relationship to geographical coordinates and land features (of artificial or natural origin); (2) drawings, which are generally more detailed representations of specific structures, sites or buildings; (3) photographs; (4) data describing the characteristics of the structure (age, kind of materials, status, etc.), in numerical or text format, which can be manipulated to produce summaries and reports; (5) documents containing procedures (of restoration, cleaning, etc.) and specifications. The GIS is designed to store and provide access to these information types in support of archaeological resource management, land planning and decision-making activities. The census of the archaeological features around Vesuvius includes about 300 items at present (Fig. 5), but it is expanding. For each object, information about drawing, photographs, etc., is being introducing into the GIS. The objects are organized into three main layers: (1) areas with building constraints, (2) buried structures (of uncertain and certain location), (3) visible structures (Figs 6, 7 and 8). Table 2 reports items related to layer (3). The chosen classification is based on the use of the structures. All private houses have been indicated by the code 101 Ixx. The code xx indicates
103305 103306 103307 103308 103309 103310 103311 103312 103313 103314 103315
beechwood orchard olive grove pasture oak wood saltpan cereal crops textile crops dune vegetation riverside vegetation vineyard
the typology of these private buildings; in particular, the code (1011)04 indicates complex houses, which have resulted from a prearranged development plan or by the merger of different pre-existing units as a result of the socialeconomic evolution of the quarter. In the class 1012xx, there are the commercial facilities, exclusively devoted to the retail of different products (dairies, bakers, fishmongers, fruiterers, etc.) or shops providing services, such as the cauponae, the hospitalia, the thermopolia, the banker's or exchange office, the barber shop, etc. In this class the school is also included, as it was managed by private teachers. Layers 1013xx and 1014xx distinguish between facilities that carry out a productive activity, with or without the sale of finished goods; often, it is difficult to distinguish between the cases. Layers 1015xx (warehouses, cellars, back-shops, etc.) and 1016xx (sheds) refer to buildings related to others, such as shops, commercial facilities, houses, workshops, etc. Layers 1017xx (buildings for entertainment), 101803 (public thermae), 1019xx (sacred buildings), 1020xx (sports buildings), 102Ixx (municipal buildings), 1022xx refer to public buildings. The places assigned to meetings or sodalicia activities or trade guilds are organized in a distinct layer (1024xx), including post-stages. Funerary buildings (1026xx) have been organized according to the classification introduced by Kockel (1983) for the necropolis of Porta Ercolano and that of D'Ambrosio and De Caro for the necropolis of Porta Nocera. The classification is inspired by Van der Poel's work (1983), and on Neapolis work (Varone 1988; Neapolis 1994). Conclusion and future developments The present GIS has three important advantages for the scientific authorities and public officials who manage archaeological information, especially in places where the structures are as
A GIS FOR THE ARCHAEOLOICAL AREA OF POMPEII
157
Fig. 8. Ancient Herculaneum. Output on plotter. exceptional as in the area around Vesuvius. First, the approach enhances the visualization of complex material through the 3D visualization of features and the use of layers. Second, a GIS creates a data file that maintains
accurate dimensions and locations without limits imposed by scale (the spatial information), and with the possibility of automatic drafting processes. Third, through the integrated DBMS technology, it is possible to store and query
158
M. T. PARESCHI ET AL.
a large amount of information related to the spatial data. The present GIS of the archaeological structures in the circumvesuvian area is in the process of updating and expansion. The fundamental lines on which these operations are articulated are the following: (1) the creation of a GIS at a scale of 1:2000 for inhabited areas, for the more precise location of archaeological resources. Special funding has been given by the Italian Civil Protection, and the numerical map, based on a flight carried out on 3-4 March 1997, should be ready in a year and a half. In this project, the number of layers will be further increased, with a detailing of particular typologies (to around 400, from the 20 available in the current database. For example, buildings will be further subdivided into schools, fire stations, police stations, factories, and types of factory, etc.). (2) The further updating of the archaeological emergencies and related information, carried out by officials of the Archaeological Office of Pompeii in collaboration, as regards aspects of computerization and cartography, with the CSGSDA-CNR. Pompeii and the other ancient localities in the Vesuvian area were unique in all their aspects, be they an entire town, a villa or a humble house, abruptly wiped out by the eruption in a pulsing moment of life on a single day and kept almost intact up to our day, with all that existed at that fatal moment. We hope that the uncontrolled and disorderly urban development, for which the Vesuvian area has been sadly well known since the early 1960s, will not jeopardize the archaeological resources of the area. The work carried out so far, and in constant progress, is a modest step in the direction of preserving the testimonies of a single moment and a world that are unique and far away. The work was financially supported by CNR-Progetto Finalizzato Beni Cultural! and by the Italian Civil Protection. References ANGELETTI, P., CHERUBIM, A. PETRAZZUOLI, S. M. & PETRNI, V. 1998. Quick evaluation of seismic
vulnerability: an application to seismic risk assessment of Vesuvian area. Cities on Volcanoes, Rome and Naples (Italy), 22. ARANOFF, S. 1989. Geographic Information Systems: a Management Perspective. WDL, Ottawa, Ont. BENCIVENNI, M., DALLA NEGRA, N. & GRIFONI, P. 1992. Monumenti e istituzioni, II. II decollo e le riforme del servizio di tutela dei monument! in Italia (1880-1915), Firenze. BIANCO, A. & CASALE, A. 1979. Primo contributo alia topografia del suburbio pompeiano. Antiqua, 15, 27-56. BURROUGH, P. A. 1989. Principles of Geographical Information Systems for Land Resources Assessment. Clarendon Press, Oxford. D'AMROSIO, A. & DE CARO, S. 1983. Fotopiano e documenta zoine della Meciopoli di Porta Noceia, Un Imfegno per Pompei, T.C.I., 23-43. GAMURRINI, G. F., COZZA, A., PASQUI, A. & MENGARELLI, R. 1972. Carta Archeologica a"Italia (1881-1897). Forma Italic, II, Firenze, 429-459. KER, W. C. A. 1947. Martial Epigrams. Harvard University Press, London, IV, 44, p. 261. KOCKEL, V. M. 1983. Die Grabbauten vor dem Herkulaner Tor in Pompeji. Mainz am Rhein. MACEDONIO, G. & PARESCHI, M. T. 1991. An algorithm for the triangulation of arbitrarily distributed points: applications to volume estimate and terrain fitting. Computers and Geosciences, 17, 859-874. , & SANTACROCE, R. 1988. A numerical simulation of the Plinian fall phase of 79 AD eruption of Vesuvius. Journal of Geophysical Research, 93, 14817-14827. NEAPOLIS 1994. Progetto-sistema per la valorizzazione integrale delle risorse ambientali e artistiche dell'area vesuviana, I. La valorizzazione dei Beni ambientali, Roma, 62-64. PARESCHI, M. T. & SANTACROCE, R. 1993. GIS applications in volcanic hazard. Proceedings of the International Workshop on Geographical Information Systems in Assessing Natural Hazard, September 1993, University for Foreign Students, Perugia, 112-118. SIGURDSSON, H., CAREY, S., CORNELL, W. & PESCATORE, T. 1985. The eruption of Vesuvius in AD 79. National Geographic Research, 1, 332-387. SLATER, D. A. 1908. Statins. The Silvae. Clarendon Press, Oxford, IV, 4, 81-84, p. 156. VAN DER POEL, H. B. 1983. Corpus Topographicum Pompeianum, Part II. Toponymy, Rome. VARONE, A. 1988. La struttura insediativa di Pompei: 1'avvio di un'indagine computerizzata per la conoscenza della realta economico e sociale di una citta campana della prima eta imperiale, Pompei. L'informatica al servizio di una citta antica, Roma, 25-48.
Apulian Bronze Age pottery as a long-distance indicator of the Avellino Pumice eruption (Vesuvius, Italy) RAFFAELLO CIONI1'3, SARA LEVI2 & ROBERTO SULPIZIO1 1
Dipartmento Scienze della Terra, Via S. Maria 53, 1-56126 Pisa, Italy Dipartmento Scienze Storiche Archeologiche e Antropologiche dell'Antichitd, 'La Sapienza', piazzale A. Mow 5 Rome, Italy 3 Now at: Dipartmento Scienze della Terra, V. Trentino 51, 09124 Cagliari, Italy
2
Abstract: During the Bronze Age, Vesuvius had a Plinian eruption whose deposits are known as the Avellino Pumice. The eruption spread a blanket of white and grey pumice across southern Italy, and there was a severe impact on proximal areas. Assessment of volcanological factors for the Plinian phase gives intensities of 5.7 x 107 kgs"1 for the white pumice phase and 1.7x 10 8 kgs~ 1 for the grey pumice phase, corresponding to column heights of 23 and 31 km, respectively. Volume (magnitude) calculations using the crystal concentration method (CCM) give respectively 0.32 and 1.25km3 of deposit, in a total minimum period of about 3 h. Archaeometric studies on Bronze Age domestic pottery from several settlements in Apulia (SE Italy) reveal the presence of pumice fragments mixed with the clay, and petrological and chemical criteria suggest that these pumices are from the Avellino eruption. This relationship allows us to fix precise correlations between different archaeological facies of the Italian Bronze Age. To explore the possibility of an extensive use of pumices in these distal regions (about 140km from Vesuvius), we calculated the possible thickness of the tephra blanket. We propose a method to extrapolate proximal data on the deposit to calculate its minimum distal thickness. Such a method could also be used in volcanic hazard studies to assess the distal impact of large past eruptions.
During the Bronze Age, the civilizations of the Mediterranean area were disrupted by two large Plinian eruptions. The Aegean was stricken by the massive Minoan eruption of Thera Island (Doumas 1980), which dispersed ash and pumice over thousands of square kilometres to the east, as far as Turkey (Sullivan 1988). At a similar time, Southern Italy suffered the tremendous impact of a Plinian eruption of Vesuvius, which resulted in the deposit known as the Avellino Pumice (Lirer et al 1973). This eruption, despite being much smaller than the Minoan, dispersed its fallout products across the Italian peninsula in a NE direction. Pyroclastic flows and ash clouds completely flooded and destroyed the plain north of Vesuvius, up to tens of kilometres from the volcano (Cioni et al. 19950; Di Vito 1999). In contrast to the highly evolved Minoan civilization (the towns of Knossos on Crete and Akrotiri on Thera were flourishing), Bronze Age settlements in Southern Italy were generally primitive. Notwithstanding this, scattered remains of these settlements have been found all
around the volcano, buried by the deposits of the Avellino eruption. Working on the petrographic composition of pottery from a long-frequented, Bronze Age settlement in Apulia (Southeastern Italy), archaeologists found that, during a specific period of development of that settlement, inhabitants used pumice fragments as temper for the pottery (Levi et al. 1995). Such a use of pumice was probably related to the lightness and thermal resistance of the resulting pottery. Through studying the composition and mineral chemistry of these pumice fragments, we have found that these pumices are the result of the Avellino eruption. This contribution to Bronze Age chronology in Italy allows new inferences on the correlation of Bronze Age civilizations in the Mediterranean area. In this paper, we present the data used to support this conclusion, and describe the composition, thickness and dispersal area of fall deposits in distal areas as inferred from extrapolation of proximal and medial data.
From: McGuiRE, W. G., GRIFFITHS, D. R., HANCOCK, P. L. & STEWART, I. S. (eds) The Archaeology of Geological Catastrophes. Geological Society, London, Special Publications, 171, 159-177. 1-86239-062-2/00/ $15.00 © The Geological Society of London 2000.
Fig. 1. Composite stratigraphic log of the Avellino eruption deposits. For each eruption unit a brief description of lithology, dispersal area and main effects on the surroundings are given.
AVELLINO PUMICE IN APULIAN BRONZE AGE POTTERY General outline of the eruption The Avellino eruption (3590 ±25 years BP; Andronico et al. 1995) was one of the most powerful, Plinian-type eruptions of SommaVesuvius. It occurred in the Early Bronze Age and dispersed its products over a narrow fan throughout southern Italy. Johnston-Lavis (1895) first recognized the products of the eruption in his pioneering paper on the geology of Vesuvius. Lirer et al. (1973) described the main petrographic and depositional features of the fall deposits. Rolandi et al (1993) and Cioni et al. (1999) published studies of the stratigraphic sequence of the eruption, which detailed the depositional and dispersal features of the associated pyroclastic flow units, and proposed an interpretation of the eruption sequence. Pescatore et al. (1987) studied the grain-size and dispersal features of the fall deposit, and collected information on the dynamics of the Plinian phase. The petrography and geochemistry of the products were detailed by Barberi & Leoni (1980), Joron et al (1987) and Civetta et al (1991). Cioni et al (1999) subdivided the stratigraphic sequence into six main Eruption Units (EU), as defined by Fisher & Schmincke (1984). Similar to the other Plinian eruptions of Vesuvius (Cioni et al 19956; Bertagnini et al 1998), three distinct stratigraphic packages can be recognized, which correlate with three phases of the eruption (Fig. 1): an opening phase (EU1), a Plinian phase (EU2-EU4) and a final, phreatomagmatic phase (EU5 and EU6). The onset of the eruption resulted in the emplacement of a double bedset (EUla and -b) of white pumiceous lapilli and brown fine ash, with abundant lithic fragments and loose crystals. The deposit is pyroclastic fall in nature, and is exposed on the northeastern slopes of the volcano. The deposits are interlayered with two coarse, topographically controlled, l-2m thick pumice flow units cropping out in the western sector. The following Plinian deposits of EU2 (phonolitic white pumice) and EU3 (phonolitictephritic grey pumice) mark the formation and establishment of a sustained Plinian column, from which fall products were dispersed over a wide area in a northeasterly direction. The deposits of one pyroclastic flow (EU3pf), deriving from partial collapse of the sustained column, are interbedded in the EU3 deposit (Fig. 1). A fine brown ash bed, only a few centimetres in thickness, separates the EU3 deposit from the following EU4. Unit EU4 is a fall deposit of grey pumice lapilli, with abundant lithic fragments and loose crystals, and a maximum
161
thickness of 6-7 cm. The dispersal axis for this deposit is directed towards the northeast, and thicknesses of 1-2 cm can be measured up to 20-25 km from the volcano (Fig. 1). The EU5 products, which mark the onset of the phreatomagmatic phase, are dune-bedded, thinly stratified, ash deposits, reaching thicknesses of 8-1 Om in the proximal western sector. This phase of the eruption severely affected the northwestern plain surrounding the' volcano. Ash cloud deposits related to this EU can be recognized as far as Mugnano di Napoli, more than 25 km from the volcano (Fig. 1; Cioni et al 1995a; Di Vito 1999). In this area several human settlements, covered by EU5 deposits, of the Early Bronze Age have been discovered. The following EU6 deposits have a much more limited dispersal, cropping out only on the slopes of the volcano, where they form a tuff ring bordering the Piano delle Ginestre area (Fig. 1). Lithic rich, massive, pisolitebearing, valley-ponded pyroclastic flow deposits, interlayered with lithic-rich fallout beds, form Unit EU6.
The Plinian fall deposits In the Plinian fall deposits, a lower bed (EU2) of phonolitic, highly porphyritic, sanidine-bearing white pumice is followed by tephri-phonolitic, sanidine- and pyroxene-bearing, grey pumice (EU3) (Table 1). The transition between the two EUs is gradual, through a layer of banded pumices.
Table 1. Variation of relative abundance of crystals or porphyricity at three stratigraphic heights for EU2 and EU3 deposits; the values are expressed as weight percent of the total mass (crystals plus vitric fraction)
EU3
EU2
Base Centre Top Base Centre Top Sanidine 10.3 15.9 Nepheline 0.1 0.2 Scapolite 0.1 0.2 Colourless cpx 0.1 0.2 Green cpx 0.1 0.2 Micas 0.1 Amphibole 0.1 0.2 Garnet 0.3 0.4 Porphyricity 11.1 17.4 Total sialic 10.8 16.7 0.3 0.7 Total femic
0.2 0.2' 0.2 0.2 1.3 0.2 1.0 0.1 0.3 0.2 0.2 0.5 0.4
7.0 0.1 0.2 1.7 1.3 0.4 0.3 0.5
2.8
3.7
14.9 10.1
0.2
16.5 13.7 11.5 15.8 10.9 7.8
0.7
9.3 0.1 0.2 2.1 1.7 0.5 0.4 0.7
15.0 10.3
4.7
162
R. CIONI ET AL.
Fig. 2. Isopach maps of the Avellino plinian deposits, (a) EU2 deposit; (b) EU3 deposit. Thicknesses in centimetres.
The fall deposit is widely dispersed over a narrow area through southern Italy, and the direction of the main dispersal axis is to the northeast (Fig. 2). The deposits of EU2 and EU3 have different dispersal axes, with a progressive shift from N65E for the EU2 to N55E for the EU3 deposits. Both white and grey deposits show a good exponential decay of thickness with distance from the vent (Fig. 3). Very thick, coarse, clast-supported massive
deposits, with abundant ballistic lithic blocks, crop out in the NW sector of the volcano. We consider them to be the proximal facies of the fall deposits, approximately marking the vent position, located about 3km west of the present crater in a flat area on the volcano slope, known as Piano delle Ginestre. Cioni et al. (19956) speculated that this flat area was formed after a major caldera collapse related to the Avellino eruption.
AVELLINO PUMICE IN APULIAN BRONZE AGE POTTERY
Fig. 3. Thickness v. distance (along the dispersal axis) diagram, showing exponential fit through the data for the EU2 (A) and EU3 deposits (Q).
The fall deposits were studied by sampling them at the same stratigraphic heights at several locations. For the study isochronous levels were considered. These included the white-grey transition, the ashy layer (EU3pf) interbedded with the EU3 fall deposit (Fig. 1), and a characteristic
163
light-coloured band toward the top of the EU3 deposit, related to the presence of abundant white and grey banded pumices. The white pumice deposit of EU2 shows reverse grading. Its median diameter (Md or 4>5Q percentile) increases from the base to the top (Fig. 4a). The same increase, very clear in the more distal sections, is shown by the 016 percentile (Fig. 4b). Lithic content decreases towards the top of the sequence, both in proximal and in distal sections (Fig. 4c), and the loose crystals fraction of the deposit is mainly represented by sanidine (Fig. 4d). The constant value of Md in the EU3 deposit together with the increase in lithic content (Fig. 4) suggest a continuous increase in eruption strength. The composition of the loose crystals fraction is fairly constant (Fig. 4d), with a slight decrease in the ratio of felsic (sanidine and rarer nepheline and scapolite) to mafic (essentially clinopyroxene) crystals in the basal portion of the deposit. This reflects a progressive shift towards more mafic compositions (Cioni et al 19956). Distance from the vent does not exert a strong control on the lithic content of the EU3 deposits (Fig. 4c), with few relative differences between proximal and medial sections.
Fig. 4. Variation with stratigraphic height and distance from the vent of grain size parameters (Md, (/>16), component data (% lithic fragments, felsic-mafic crystal ratio) and mean terminal velocity (TV) of EU2 and EU3 deposits. TV = —log2TV (ms"1).
164
R. CIONI ET AL.
Fig. 5. Variation of sorting and median diameter (acj) and Md, Inman, 1952) v. distance from the vent of EU2 (D) and EU3 (O) deposits. Vertical bars represent the standard deviation of the mean value of cr and Md. These were obtained by the arithmetic mean of acj) and Md values of the samples collected in the same site.
Variation of Md and sorting (crfy with increasing distance from the vent for the two EUs is shown in Fig. 5. Md values show an exponential decay with distance, according to the model of Sparks et al. (1992). A similar trend is shown by acj) values (Fig. 5), with a scattering of data
related to the sampling position. We generally observe that EU2 deposits from sites south of the dispersal axis are better sorted than those from sites closer to it. This effect might be due to a counterclockwise rotation of the dispersal axis during the progressive strengthening of the
Fig. 6. Isopleth maps of the maximum lithic fragments for EU2 and EU3 deposits. Values in centimetres. Each value is the arithmetic mean of the five largest clasts in the outcrop.
AVELLINO PUMICE IN APULIAN BRONZE AGE POTTERY eruption. Waitt et al. (1981) described similar behaviour for the post-climactic eruptions of Mount St Helens. Isopleth maps of the averaged longest axis of the five maximum lithic clasts (ML) are shown in Fig. 6. The more elliptical shape and restricted extent of the EU2 curves compared with that of the EU3 deposit reflect the lower eruptive energy of EU2. The asymmetry in the distribution of the deposits is also clear on this map, although less marked than for the isopachs (Figs 2 and 6).
Fig. 7. Assessment of peak column heights and wind speed during the EU2 and EU3 eruptive phases, using the method of Carey & Sparks (1986) for lithic clast diameters of 0.8, 1.6 and 3.2cm.
165
Eruption dynamics The coarsening of grain size with stratigraphic height, first shown by Pescatore et al. (1987), suggests a sustained eruptive column progressively increasing in height. The increase of the column height, and the consequent change in direction of the wind, explains the asymmetry of the isopach and isopleth maps, which is particularly evident for the EU2 deposit (Fig. 6). Eruption intensities (Carey & Sigurdsson 1989) were calculated according to the model of Carey & Sparks (1986). The calculations were performed only for the peak magma discharge rate (MDR), and gave averaged peak column heights of 23km for the EU2 eruptive phase and 31km for that of EU3, with a wind speed of around 30ms" 1 (Fig. 7). These results agree well with the data previously published by Pescatore et al. (1987). The peak column heights correspond to MDR values of 5.7 x K^kgs- 1 for EU2 and 1.7 x K^kgs" 1 for EU3 (Fig. 8). The theoretical exponential decay of Md with distance from the vent was used by Sparks et al. (1992) to calculate eruption column height on the basis of the grain-size features of the deposits. Despite the exponential decay shown by the measured Afd values with increasing distance from the vent (Fig. 5), they do not fit the theoretical curves (Fig. 9a). This could be due to the effect of strong winds governing plume dispersal, not accounted for by the model. A better fit of the data was obtained by
Fig. 8. Peak mass discharge rate during EU2 and EU3 eruptive phases, using the diagram of Sparks (1986). The numbers on the curves indicate the magma temperatures. Column heights from the diagrams of Fig. 7 magma temperature of 800°C and atmospheric profile for a temperate area.
166
R. CIONI ET AL.
Fig. 10. Log thickness v. square root of the area diagram for EU2 (A) and EU3 (O) deposits.
the distal features (in particular thickness) of the Avellino deposits. For the Avellino eruption the Plinian deposits account for more than 90% of the total volume of the erupted products Fig. 9. Variation of Md v. distance from the vent for (Cioni et al. 1999). EU2 (A) and EU3 (•) deposits, (a) windy conditions; Several methods of calculation of tephra (b) no wind conditions. Vertical bars summarize fallout volume have been described in the litervertical variability in each section analysed. Fine lines ature (Rose et al. 1973; Walker 1980; Froggat in both diagrams are the theoretical curves of Sparks 1982; Pyle 1989; Fierstein & Nathenson 1992). et al. (1992) with the corresponding column height For the Avellino Plinian deposits calculations values in kilometres. Bold line in (a) shows the made according to the method proposed by exponential fit through the data in the case of windy Rose et al. (1973) yielded volumes of 0.29km 3 conditions. Framework on the right side of (b) shows for EU2 and 1.64km3 for EU3, whereas assessan enlargement of the proximal data with horizontal axis 1.5 greater than the vertical. ment made with the method of Pyle gave volumes of 0.2km 3 for EU2 and 0.9km 3 for EU3 (Fig. 10). Both these methods contain inherent uncertainties in the evaluation of tephra replacing the distance from the vent with the value of the equivalent radius of the Md isopleth volumes, especially for ancient deposits with few isopach data. In particular, the method of Rose for each site (Fig. 9b). Such an assumption is similar to that made by Pyle (1989) in his is very sensitive to the number of data points method for volume calculations, and it could be and to the thickness of the most distal isopach, and this results in unreliability in the slope of the used to reduce the data of eruptions occurring in a strong wind field. Maximum heights of 17km regression lines (Fierstein & Nathenson 1992). for EU2 and 22 km for EU3 result from such an In turn, the method of Pyle (1989) provides good approximation. These are very close to the results only for isopach maps with thicknesses values estimated according to Carey & Sparks down to less than 5-10 cm. This is due to the (1986) for neutral buoyancy rather than total change in slope of the regression lines that often occurs below these values (Pyle 1990; Fierstein & column height. Nathenson 1992). An effective method of estimating volumes is the crystal concentration method (CCM) derived by Walker (1980). This method applies Volume of the Plinian deposit to eruptions of porphyritic magma and is based Assessment of the volume of ejecta can provide on four assumptions (Walker 1980; Fierstein & an estimation of several important eruption Nathenson 1992): (a) the measured porphyricity factors, such as magnitude and duration of (in wt%) of coarse pumices is representative the event. Furthermore, in this paper we pro- of the magma porphyricity; (b) the mass loadpose using the total volume assessment to infer ings of pumices, crystals and lithic fragments
AVELLINO PUMICE IN APULIAN BRONZE AGE POTTERY
167
Fig. 11. Isomass on land for the EU2 deposit.
all follow an exponential law of decay with distance; (c) below the mean crystal size, juvenile material is represented only by loose crystals and vitric fragments; (d) the total mass of loose crystals is completely dispersed inside the mappable deposit. If all these conditions are satisfied, the volume of the 'lost' vitric material can be assessed (by 'lost' is here intended the missing material, dispersed in a different way from the exponential thickness decrease resulting from PM (Pyle's Method). To calculate volumes with this method, we made maps of the mass loading of the total deposit, loose crystals, lithic fragments and vitric material (juvenile fragments smaller than 4mm; Figs 11 and 12). We used a maximum size of 4mm for the vitric material (instead of the 2mm diameter proposed by Walker (1980)). This was because of the larger mean size of crystals in the Avellino as opposed to the Taupo pumices. The total mass on land for the different components (Table 2) was obtained applying the method of Pyle to the isocurves of Figs 11 and 12. The volume of calculated 'lost' vitric material (Table 2) is a large fraction of the total volume (respectively 60% and 40% of the volume calculated with the method of Pyle (1989) for the EU2 and EU3
deposits). The resulting total volumes are 0.32km3 for EU2 and 1.25km3 for EU3. We must stress that the choice of a density value for the vitric fraction in the CCM is very important, because it substantially influences the value of the 'lost' volume (Fierstein & Nathenson 1992). The assessment of reliable values is not a simple exercise because it should take into account the relative proportion of the differently vesiculated clasts in the 'lost' material. The values chosen for the EU2 (1.3 gem- 3 ) and the EU3 (2.0gcm~ 3 ) products take into account the large amount of relatively coarse, vesicular juvenile material found in the Apulian pottery. Assuming the volume value as derived from the CCM and the estimated peak MDR, we can estimate a minimum length of the Plinian phase of the eruption of about 3 h.
Pyroclastic temper in Bronze Age pottery from Apulia (SE Italy) Archaeometrical analyses of Bronze Age pottery from Apulian (southern Italy) sites reveal a massive use of pyroclastic material as temper
R. CIONI ET AL.
168
Fig. 12. Isomass on land for the EU3 deposit. Table 2. Calculation of lost masses and volumes using the crystal concentration method (CCM) of Walker (1980) for the EU2 and EU3 deposits
Measured values (a) M/A total (1011 kg) (b) M/A juv <4mm (1011 kg) (c) M/A crystals (1011 kg) (d) M/A lithic fragments (1011 kg) (e) Magma porphyricity (f ) Glass/crystals ratio (g) Mean density of juv <4mm (gem-3) Calculation of lost mass (h) Total M/A juv <4mm = (f ) x (c) (10 n kg) (i) Lost M/A juv <4 mm = (h) - (b) (10 n kg) Calculation of total volume (j) Measured volume (Pyle, km3) (k) Lost volume = (h)/(g) (km3) (1) Total volume = ( j) + (k) (km3) M/A, mass on land.
EU2
EU3
1.50 0.41 0.36 0.16 15% 5.67 1.30
6.15 2.41 1.39 1.05 13% 6.69 2.00
2.04
9.30
1.60
6.90
0.20 0.12 0.32
0.90 0.35 1.25
(Amadori et al 1995; Levi et al 1995). Thinsection analyses identified several kinds of temper mixed with clay, including pumice and calcite. Grog tempering was also found to be common. The presence of abundant carbonates is clearly a reflection of the local geology, which is dominated by sedimentary calcareous formations. On the other hand, pyroclastic deposits are rare in northern Apulia and absent from present-day geological outcrops. Some archaeometrical studies, concentrating on neolithic pottery from the same area, have not found pumiceous temper. Rare volcanic materials (small augitic pyroxene crystals) have been interpreted as products of the Pleistocene activity of Vulture volcano (Mannoni 1980, 1983; Dell'Anna 1986; Di Lernia et al 1993; Lorenzoni et al. 1995), situated about 60km SW of northern Apulia. Several samples of pottery with different provenances were studied. The sherds come from sites in the province of Foggia (Fig. 13): Coppa Nevigata (Manfredonia), Terra di Corte (S. Ferdinando), Madonna di Loreto
AVELLINO PUMICE IN APULIAN BRONZE AGE POTTERY
169
Coppa Nevigata (Apulia)
Fig. 13. Locations of archaeological sites (•). (Trinitapoli), and Madonna di Ripalta (Cerignola). For comparison, very similar samples from La Starza (Ariano Irpino, Campania, about 80km SW of the others), were analysed. The sites of provenance and their archaeological context are briefly described below (for a more exhaustive description, see quoted references).
This site is very important becouse it was occupied for a long time. Settlement started in the Neolithic period and the site was continuously inhabited, with only minor gaps, through the Bronze Age up to the Early Iron Age (Cazzella & Moscoloni 1987, 1988). The Bronze Age occupation started in the protoappennine facies. About 200 pottery samples from this site were studied by optical microscopy, X-ray diffraction (XRD) and X-ray fluorescence (XRF) (Amadori et al 1995). Pottery tempered with pumice was first discovered at this site. The maximum use of volcanic temper appears to have been in the lowest Bronze Age levels of the site. During the following periods a sharp (although not constant) decrease is observed (Fig. 14). Pottery made with volcanic temper was of all types, but mainly bowls.
Terra di Corte (Apulia) The Terra di Corte rock-cut chamber 3 (Tunzi Sisto 1997) contained some anthropological remains and traces of activity that have been interpreted as of the cultic type. There are other similar chambers in the area, all with a grave or cult function. The pottery is of protoappennine age and various fragments contain pumices.
Madonna di Loreto (Apulia) The site is characterized by the presence of a rock-cut corridor tomb of apennine age, with some traces of former activity (Protoapennine?), which have been interpreted as cultic (Tunzi Sisto 19920, 1997). All the apennine pots were visually inspected and most of them contain pumices.
Madonna di Ripalta (Apulia) The site consist of a multistratified settlement with a sequence spanning at least from the Middle Bronze Age to Early Iron Age. The site overlooks the northern side of the river Ofanto, 30km from the sea (Tunzi Sisto 19926).
La Starza (Campania)
Fig. 14. Variations of the different types of temper with time in the Coppa Nevigata site. In the lower part of the diagram a division of the archaeological time scale is also shown.
The site of La Starza, near Ariano Irpino, in the Apennine Mountains (Fig. 13), has a multistratified human occupation, from the neolithic to the end of the Bronze Age (Albore Livadie 1990). A layer (about 10cm thick) of pumice from the Avellino eruption underlies remains of
R. CIONI ET AL.
170
the Bronze Age occupation (Vernet et al. 1999). Pottery for the analyses was chosen from the level immediately on top of the pumice layer. Some general considerations on the main features of the pottery are as follows: (1) pumice in the pottery is often associated with other kinds of temper, particularly with calcite. (2) XRF chemical analyses (bulk analyses of the pottery; Levi et al. 1995) show that all the Apulian samples have a similar composition, suggesting that they were regionally produced. There are some minor compositional differences between pottery, with different kinds of temper from the same site and between pottery from different sites. In contrast, samples from La Starza (Ariano Irpino), 60 km from the Apulian sites, show different petrographic and compositional features. (3) The mean grain size of the pumices used as temper is different from site to site, suggesting different areas of provenance and/or exploitation. (4) Volcanic material was used as temper also for daubs, ovens and other non-transportable objects. (5) Pumiceous temper is absent in neolithic pottery. Its appearance occurs between this period and the Middle Bronze Age. Petrographic features of the volcanic material Petrographic study of several samples of Bronze Age pottery from the sites described was carried out. Microanalysis was performed on pumice
chips and minerals in the matrix of pottery from several of these samples, using the microanalytical facilities at the Earth Science Department of the University of Pisa. A Philips 515 scanning electron microscope equipped with an energy dispersive spectrometry (EDS) microanalytical system EDAX PV 9900 was used (operating conditions 20 kV acceleration voltage, 100s live time, 200-500 nm beam diameter). Differences in the petrography of the pottery mainly reflect the grain size of pumice fragments and the amount and type of crystals scattered in the temper. In particular, pottery from La Starza is strongly enriched in sanidine crystals, with minor scapolite and nepheline as well as some fragments of leucititic lava. In turn, pottery from the Apulian sites has a lower content of sub-millimetre crystals. Pumice fragments and glass shards are generally unaltered and well preserved. Some samples, with strongly altered glass shards, were not analysed. Pumice fragments from Apulian pottery are sub-aphyric to aphyric, and show microphenocrysts of sanidine, salitic clinopyroxene, mica, potassic amphibole, with rarer plagioclase, scapolite, nepheline and garnet. The same phases also occur, with different proportions, in the groundmass of the pottery, together with spar calcite. Pottery of a similar age from the La Starza site shows the same features. In these samples pumice fragments generally have a coarser grain size and are more porphyritic, and the temper is strongly enriched in crystals of sanidine in comparison with the Apulian samples. Pumice mineralogy is uncommon in the coeval magmatic products of Italy and the Mediterranean region, with the contemporaneous occurrence of nepheline and scapolite of marialitic composition, together with potassic minerals
Table 3. EDS analyses of pumice from Apulian pottery and pumice from proximal deposits. Grey pumice [99]
SiO2 55.3 A1203 22.37 TiO2 0.23 FeO 2.41 MgO 0.88 CaO 3.48 Na20 7.63 K20 7.43 Cl 0.61
CN 6
CN7
MLO2
SD
[25]
SD
[10]
SD
1.2 0.94 0.13 0.79 0.36 1.42 2.14 1.08 0.12
55.86 22.13 0.25 2.74 0.53 3.98 6.78 7.14 0.55
0.67 0.43 0.05 0.46 0.21 0.65 0.8 1.17 0.08
55.03 21.87 0.26 2.96 0.77 3.98 7.28 7.32 0.55
0.63 55.15 0.61 0.21 22.58 0.52 0.11 0.22 0.05 0.56 2.47 0.46 0.46 0.66 0.13 0.7 3.37 0.63 0.77 7.05 0.29 0.29 7.68 0.48 0.06 0.68 0.05
[6]
SD
TDC4
White pumice
MLO 2
[5]
SD
[97]
SD
[3]
SD
55.39 22.76 0.18 2.72 0.89 3.79 6.39 7.14 0.6
0.58 0.17 0.05 0.3 0.24 0.51 0.49 0.27 0.07
56.56 23.89 BDL 1.47 0.6 1.65 7.56 7.75 0.73
1.27 0.68
57.31 24.33 BDL 1.34 0.42 1.43 7.91 6.38 0.71
0.92 0.73
0.46 0.28 0.47 2.06 1.42 0.15
0.1 0.12 0.09 0.42 0.25 0.07
TDC4 [2] 56.65 25.12 BDL 1.26 0.51 1.58 6.58 7.6 0.52
Number of analyses in brackets; SD, standard deviation of analyses; BDL, below detection limit. CN 6 and 7 from Coppa Nevigata; MLO2 from Madonna di Loreto; TDC4 from Terra di Corte.
AVELLINO PUMICE IN APULIAN BRONZE AGE POTTERY
171
Fig. 15. Comparison of EDS chemical analysis of juvenile material from proximal fall deposits of the Avellino eruption and fragments of pottery of Apulian archaeological sites.
(mica and amphibole). The same mineralogical assemblage was first described by Barberi et al. (1981) for the products of the Avellino Pumice. The glass composition is homogeneous and identical to that of the grey pumice (Table 3 and Fig. 15) measured on more than 200 juvenile fragments from the pyroclastic deposits of the Avellino Pumice (Cioni et al. 1999). Analyses of the glass from Madonna di Ripalta pottery and some analyses from Terra di Corte pottery (Table 3) are very similar to the mean corn-
Table 4. EDS analyses of scapolite crystals from the main deposit (EU2 and EU3) and temper of the pottery from Madonna di Loreto (MLO2) and Terra di Corte (TDC4) EU2
EU2
EU3
EU3
MLO 2 TDC 4
Si02 48.90 51.74 51.71 51.21 50.23 A1203 26.82 25.84 25.22 26.02 25.75 Fe2O3 0.22 0.25 0.20 0.19 0.30 CaO 12.43 12.06 11.85 11.14 11.75 Na2O 6.86 6.41 7.21 7.72 7.61 K2O 1.64 1.48 1.55 1.46 1.66 Cl 2.33 2.00 2.11 2.19 2.10
50.52 26.25 0.31 11.96 7.00 1.47 2.02
Recalculation on the basis of 12 cations Si 7.29 7.55 7.62 7.50 7.48 Al 4.71 4.45 4.38 4.50 4.52 0.01 0.01 0.01 0.01 0.02 Fe Ca 1.98 1.89 1.87 1.75 1.87 Na 1.98 1.81 2.06 2.19 2.20 K 0.31 0.28 0.29 0.27 0.32 Cl 0.59 0.49 0.53 0.54 0.53 % Me 46.50 47.62 44.44 41.71 42.95
7.44 4.56 0.02 1.89 2.00 0.28 0.50 45.57
position of the white pumices (Fig. 15). The magmatic crystals (mainly sanidine and salitic pyroxene) scattered in the temper of the pottery exactly fit the compositional range of the same mineral phases from the pumices. The scapolite microcrysts from the pumices in the pottery are also similar to those from the Avellino grey pumice (Table 4). In conclusion, the major element data on both glass and minerals strongly suggest that the pumices of the Apulian and La Starza samples are very homogeneous, and support the proposal that they are related to the Avellino Pumice eruption of Vesuvius. Glass composition and mineral chemistry of products from Mt Vulture, a Quaternary volcano in the Appennine chain (Guest et al. 1988), no more than 60km west of northern Apulia, were also considered as a possible source for pumices in pyroclastic temper. The different mineralogy of the major Mount Vulture phonolitic eruptions, always characterized by the presence of haliyne (De Fino et al. 1986), contrast with the very typical association of nepheline and scapolite occurring both in Avellino Pumice and in the fragments from Apulian pottery. Major element data are not usually considered useful to trace correlations between tephra layers (Sarna-Wojcicki et al. 1979). This general rule is not fully valid for Mediterranean alkaline magmas, where detectable differences often exist in the major element compositions of both glasses and minerals from the products of the different eruptions (see, e.g. the papers of Keller et al. (1978) and Paterae et al. (1988)). A Vesuvian origin for the pumice is also indirectly suggested by some leucitebearing lava fragments occurring in the matrix
172
R. CIONI ET AL.
of the pottery, related to lithic fragments eroded in the upper portion of the volcano. We cannot draw any definitive conclusion about the provenance of the different fragments of pottery. The finer grain size of pumices in pottery from Apulia (<50/mi) compared with the pumices (between 1 mm and 500 /mi) in the samples from La Starza (Fig. 13) suggest that pumices of the Apulian pottery were collected at more distal sites. This is a primary depositional feature in agreement with a general fining downwind of the deposit. The deposits of the eruption are still present at La Starza, where they show petrographic and grain-size features very similar to those of the juvenile fragments in the pottery.
correspondence to the extrapolated M& values of the fallout deposits at the same distances (Fig. 5). In the Madonna di Loreto chamber tomb, one bowl displays coarser pumices than the others, and this is regarded as an import. Deposits of the Avellino eruption are now not found in the Apulian region. This is not unusual, if we remember that the Apulian archaeological sites of this study are as far as 140km from Vesuvius. On the other hand, these sites lie exactly on the dispersal axis of the Avellino Pumice Plinian fall deposit. Some information on the main features of the pyroclastic fall products in these distal regions can be inferred from the previously discussed data on proximal and medial deposits. In particular, we can
The provenance of the pyroclastic temper: inferences on long-distance features of the Plinian fallout deposit The presence of pumice fragments related to the Avellino eruption in the Apulian Bronze Age pottery implies various interpretations. Below, we briefly discuss the possibilities, giving an interpretation on the basis of our data. (1) Pottery is allocthonous. The possibility that pottery was imported from elsewhere can probably be ruled out on the grounds of archaeological considerations (Levi et al. 1995). An extensive, long-distance exchange of pottery during the Bronze Age is unlikely, because of technological and social considerations. (2) Temper is allocthonous. We cannot completely rule out the possibility that the material for volcanic temper was imported. The differences in grain size and composition (relative proportions of pumice and crystals) of the temper of each site would imply, however, a complex, unlikely trading network for this 'humble' material. (3) Pottery and pyroclastic temper are autochthonous. The possibility of a local provenance of pottery and material for the temper is most likely, on the basis of archaeological and volcanological data. The observed decrease of both the mean grain size of pumice fragments in the temper and the amount of heavy material (such as loose crystals), passing from the La Starza to the Apulian sites suggests a local provenance of the pyroclastic material. The variation in mean grain size of pumice fragments in the pottery shows a good
Fig. 16. Accommodation of lost volumes assessed by the crystal concentration method in log of thickness v. square root of the area diagrams for EU2 and EU3 deposits. The accommodation is shown using three different points for the break in slope in both diagrams. The framework in the upper diagram shows either the area of Pyle's volume (assessed using a single straight-line segment) or the area of the lost volume (assessed by the crystal concentration method).
AVELLINO PUMICE IN APULIAN BRONZE AGE POTTERY extrapolate thickness values for EU2 and EU3 deposits in the Apulian area. Thickness data on proximal and medial sites do not allow a direct extrapolation to distal values. Pyle (1989) and Fierstein & Nathenson (1992) showed in particular that the exponential decrease of thickness (T) with distance is not regular. Most deposits show a thickness decrease following a two straight-line segments law on a log T v. distance (square root of the isopach area, SRA) diagram, with a slower decrease of thickness for distal than for proximal areas. This is probably due to a change in the settling law for the smaller grain sizes (Rose 1993). Our data on the fall deposits are within a maximum range of 35 km and they fit a single straight-line segment on a log T v. SRA diagram (Figs 3 and 10). The slope of the second straight-line segment should be fundamental for the assessment of distal thicknesses. In this section, we present a method for the calculation of the second straight-line segment parameters. To do this, the total volume assessed by the CCM (Walker 1980) represents the starting point. As a first approximation, we can assume that the total volume calculated with the CCM is equal to the area underlying the two straight-line segments in the log TV. SRA diagram (Fig. 16). The first segment is well constrained by our dispersal data. The slope of the second segment can be calculated to accommodate the CCM 'lost' volume, by the assumption that the break in slope (interception point of the two straightline segments) occurs at a thickness of between 10 and 1 cm (Fierstein & Nathenson 1992). To do this, we have to extract the value of the slope, —k\, of the second straight-line segment (beyond the point of interception) from the equation for the volume (Fierstein & Nathenson 1992):
where TQ is the thickness at source, A^ is the value of the square root of the area at the point of interception, k is the slope of the first straight-line segment. Hence, assuming k\ ^ 0, we have
173
Table 5. Parameters of the three straight-line segment used for the accommodation of lost volumes shown in Fig. 17 (see text for the symbols) 7* (cm)
Aip (km)
*i
£
2a (km)
2b (km)
128 125 88 212 221 217
36 35 24
EU2
5 3 1
60.22 58.4 41.06
-0.050 -0.041 -0.030
0.96 0.96 0.96
EU3
8 5 3
120.05 125.55 123.25
-0.029 -0.024 -0.019
0.91 0.91 0.91
87 91 89
IF
we have
which yields two real solutions, k\t\ and k\^The only acceptable solution must satisfy the condition
The coordinates of the point of interception being P\p = (A\^\ Jip), we are able to calculate the square root of the area, A^ , of each isopach line beyond the break in slope, simply by solving the following equation:
where Tis the thickness of the unknown isopach. Assuming the shape of the unknown isopach to be an ellipse with eccentricity e — ^/(a2 — b2)/a equal to the mean of the available isopachs, and area Ae\i — abir, we can calculate the two axes (2a and 2b) of the ellipse. Some examples of accommodation of lost volume to the isopach data for the Avellino products are shown in Fig. 16, choosing different positions for the break in slope. Table 5 shows the calculations for the 1 cm isopach. According to these calculations, the total thickness of the two fall deposits in the Apulian area around Coppa Nevigata was not more than 4-5 cm (Fig. 17). The very limited thickness estimated for the EU2 deposits in this area probably prevented their utilization in the temper of the pottery. The finding of few white pumices only in some potteries from Madonna di Ripalta and
174
R. CIONI ET AL.
The recognition of juvenile fragments related to the Avellino Plinian eruption of Vesuvius in pottery from Bronze Age Apulian settlements indicates the presence of an important tephra sheet in the area. Such an occurrence allows us to be confident of the assumption of a two straight-lines law for exponential thinning of the deposits. The extrapolation of proximal data, in fact, would not account for such a presence. The extensive utilization of pumice in the temper commonly used for autochthonous pottery, and the relatively low thickness of the primary deposits, suggest, however, the possibility of a large accumulation of tephra related to posteruptive mobilization toward the coastal, marshy area of Northern Apulia. As a general rule, the proposed method of extrapolation of distal thickness from proximal data and total volume assessment can be used to infer the area covered by a limited thickness of deposit (a few centimetres) in large explosive eruptions. This method would illustrate possible important perspectives for volcanic hazard assessment from incomplete field data on large, past eruptions. From the archaeological point of view, the presence of pumices related to the Avellino eruption in the Bronze Age pottery allows us to draw accurate correlations between geographically different cultural facies. The development of the Protoappennine phase followed the erupFig. 17. Reconstruction of the 1 cm isopach line using tion (Fig. 14). At Coppa Nevigata, pumiceous temper was used in pottery temper before the the 7^, value of 5cm (Table 2) for the EU2 and EU3 deposits. For the EU3 deposit the 3 cm isopach is also Middle Bronze Age. The decrease in its use drawn. Italic type indicates the archaeological sites of during the Bronze Age (Fig. 14) may reflect a Coppa Nevigata and Madonna di Ripalta. reduction in availability. Such considerations are very important for the relative chronology of the Italian Bronze Age, connecting directly regions Terra di Corte is in good agreement with the such as Campania and Apulia, independently position of this site, closer to the dispersal axis of from archaeological stylistic studies. The Avellino eruption covers many sites of the Early the EU2 deposit than the others (Fig. 17). Bronze Age Palma Campania facies (in the Campanian Plain; Albore Livadie et al. 1980), placing the eruption between the Early Bronze Conclusions Age and the beginning of the Middle Bronze Age. The medium-sized, short-lived Plinian eruption In terms of absolute chronology it may represent of the Avellino Pumice probably had a severe a central reference for Italian prehistory. The impact on the environment and human settle- calibrated ages of the eruption (3960-3780 BP; ments. Proximal areas were severely affected by Andronico et al. 1995) suggest an age older than tephra fall and pyroclastic flows of the phreato- that previously accepted for the Italian Middle magmatic phase, which caused some casualties Bronze Age. This is in agreement with the 'high' (Petrone & Fedele 1995). In distal sectors, the chronology proposed for the Aegean (Manning main effects are likely to have been damage to 1995), reinforced by the studies on the Thera vegetation (scorching and defoliation) and pol- 'Minoan' eruption (Kuniholm 1989). lution of water over a huge area (more than The extensive study of large prehistoric 10000km2). The deposition of a sheet of a few eruptions, if closely integrated with archaeolocentimetres of ash and pumice over such an area gical research, can yield information on the would have created ideal conditions for famine environmental impact of these fortunately rare and epidemic. events.
AVELLINO PUMICE IN APULIAN BRONZE AGE POTTERY C. Albore Livadie, A Cazzella, M. Cipolloni, M. Moscoloni and A. M. Tunzi made this work possible by providing samples for analyses and supporting us scientifically. Many thanks are due to many others who contributed to this work with discussions and technical help: M. L. Amadori, A. and R. Coeli, S. Conticelli, M. Di Pillo, C. Manganelli, R. Peroni and A. Vanzetti. A. Duncan and J. Gilbert greatly improved a first version of the manuscript. The financial support for this work was provided by the Italian National Group for Volcanology (GNV) of the National Council of Research (CNR). References ALBORE LIVADIE, C. 1980. Palma Campania (Napoli). Resti di un abitato dell'eta del Bronzo Antico. Atti dell'Accademia Nazionale del Lincei. Notizie degli scavi di antichitd, 105, 59-101. 1990. Nuovi scavi alia Starza di Ariano Irpino. Rassegna di Archeologia, 10, 481-491. AMADORI, M. L., Di PILLO, M., FRATINI, F., LEVI, S. T. & PECCHIONI, E. 1995. The Bronze Age pottery of Coppa Nevigata (FG-Italy): raw materials and production. In: VENDRELL-SAZ, M., PRADELL, T., MOLERA, J. & GARCIA , M. (eds) Proceedings of the European Meeting on Ancient Ceramics Archaeometrical and Archaeological Studies. Generalitat de Catawnya, Barcelona, 18-20 November 1993, 45-52. ANDRONICO, D., CALDERONI, G., CIONI, R., SBRANA, A., SULPIZIO, R. & SANTACROCE, R. 1995. Geological map of Somma-Vesuvius volcano. Periodico di Mineralogia, 64, 77—78. BARBERI, F. & LEONI, L. 1980. Metamorphic carbonate ejecta from Vesuvius plinian eruptions: evidence of the occurrence of shallow magma chambers. Bulletin of Volcanology, 43, 107-120. , BIZOUARD, H., CLOCCHIATTI, R., METRICH, N, SANTACROCE, R., & SBRANA, A. 1981. The Somma-Vesuvius magma chamber: a petrological and volcanological approach. Bulletin of Volcanology, 44, 294-315. BERTAGNINI, A., LANDI, P., Rosi, M. & VIGLIARGIO, A. 1998. The Pomici di Base plinian eruption of Somma-Vesuvius. Journal of Volcanology and Geothermal Research, B3, 219-239. CAREY, S. & SIGURDSSON, H. 1989. The intensity of plinian eruptions. Bulletin of Volcanology, 51, 28-40. & SPARKS, R. S. J. 1986. Quantitative models of the fallout and dispersal of tephra from volcanic eruption columns. Bulletin of Volcanology, 48, 109-125. CAZZELLA, A. & MOSCOLONI, M. 1987. Eta del Bronzo. La ricerca archeologica. In: CASSANO, S. M., CAZZELLA, A., MANFREDINI, A. & MOSCOLONI, M. (eds) Coppa Nevigata e il suo territorio: testimonianze archeologiche dal VII al II millennio, a.C. Quasar, Rome, 109-130. & 1988. La sequenza dell'eta del Bronzo di Coppa Nevigata. Atti deU'VIH Convegno Nazionale sulla Preistoria-Protostoria-Storia della Dau-
175
nia, San Severo 1986, Archeocust DTtalian, San Severe, 137-163. CIONI, R., CIVETTA, L., MARIANELLI, P., METRICH, N., SANTACROCE, R. & SBRANA, A. 19956. Compositional layering and syn-eruptive mixing of a periodically refilled shallow magma chamber: the AD 79 plinian eruption of Vesuvius. Journal of Petrology, 36, 739-776. , DE VITO, M. A., SBRANA, A. & SULPIZIO, R. 19950. The Avellino eruption (Vesuvius, Italy): volcanological aspects and implications on the palaeoenvironmental reconstruction. Periodico di Mineralogia, 64, 133-134. , MORANDI, D., SBRANA, A. & SULPIZIO, R. 1999. L'eruzione pliniana di Avellino del Somma— Vesuvio (ca. 3400 b.p.): aspetti stratigraficovulcanologici e dinamica eruttiva. In: LIVADIE, C. A. (ed.) L'eruzione vesuviana delle Pomici di Avellino e la fades di Palma Campana (Bronzo Antico), De Rosa, Malori (in press). CIVETTA, L, GALATI, R. & SANTACROCE, R. 1991. Magma mixing and convective compositional layering within the Vesuvius magma-chamber. Bulletin of Volcanology, 53, 287-300. DE FINO, M., LA VOLPE, L., PECCERILLO, A., PICCARRETA, G. & POLI, G. 1986. Petrogenesis of Monte Vulture volcano (Italy): inferences from mineral chemistry, major and trace element data. Contributions to Mineralogy and Petrology, 92, 135-145. DELL'ANNA, A. 1986. Applicazione dell'analisi mineralogica allo studio della ceramica impressa di C.no San Matteo-Chiantinelle (FG). Taras, VI, 1-2, 97-103. Di LERNIA, S., FRANCHI, R. & PALLECCHI, P. 1993. Manufacture characteristics. Provenance problems and content residues: an archaeometric approach to the Neolithic pottery of the Defensola Mina (Vieste, Italy). Quaternaria Nova, III, 151-175. Di VITO, M. A. 1999. Distribuzione dei depositi dell'eruzione delle 'Pomici di Avellino' (XVII-XVIII sec., a. C.) nell'area napoletana e ricostruzione del paleoambiente prima e dopo 1'eruzione. In: LIVADIE, C. A. (ed.) L'eruzione vesuviana delle Pomici di Avellino e la fades di Palma Campana (Bronzo Antico), De Rosa, Malori (in press). DOUMAS, C. 1980. The stratigraphy of Akrotiri. In: DOUMAS, C. (ed.) Them and the Aegean World, I, Thera and the Aegean World, London, 777-782. FIERSTEIN, J. & NATHENSON, M. 1992. Another look at the calculation of fallout tephra volumes. Bulletin of Volcanology, 54, 156-167. FISHER, R. V. & SCHMINKE, U. 1984. Pyroclastic Rocks. Springer, Berlin. FROGGAT, P. C. 1982. Review of methods of estimating rhyolitic tephra volumes; applications to the Taupo volcanic zone, New Zealand. Journal of Volcanology and Geothermal Research, 14, 301-318. GUEST, J. E., DUNCAN, A. M. & CHESTER, D. K. 1988. Monte Vulture volcano (Basilicata, Italy): an analysis of morphology and volcaniclastic facies. Bulletin of Volcanology, 50, 244-257.
176
R. CIONI ET AL.
INMAN, D. L. 1952. Measures for describing the size distribution of sediments. Journal of Sedimentary Petrology, 22, 125-145. JOHNSTON-LAvis, H. J. 1895. The geology of Mt. Somma and Vesuvius: being a study in volcanology. Quarterly Journal of the Geological Society of London, 40, 35-149. JORON, J. L., METRICH, N., Rosi, M., SANTACROCE, R. & SBRANA, A. 1987. Chemistry and petrography. In: SANTACROCE, R. (ed.) Somma-Vesuvius. Quaderni della Ricerca Scientifica, CNR, Roma, 8, 105-171. KELLER, J., RYAN, W. B. F., NINKOVICH, D. & ALTHERR, R. 1978. Explosive volcanic activity in the Mediterranean over the past 200,000 yr as recorded in deep-sea sediments. Geological Society of America Bulletin, 89, 591-604. KUNIHOLM, P. I. 1989. Overview and assessment of the evidence for the date of the eruption of Thera. In: HARDY, D. A., DOUMAS, C. G., SAUELARAVIS, J. A. & WARREN, D. M. (eds) Thera and the Aegean World III. Proceedings of the Third International Congress, Santorini, 3-9 September 1989. The Thera Foundation, London. LEVI, S. T., AMADORI, M. L., Di PILLO, M., FRATINI, F. & PECCHIONI, E. 1995. Archaeometric and archaeological research on the pottery of Coppa Nevigata (FG-Italy): production and provenance. In: VINCENZINI, P. (ed.). The Ceramics Cultural Heritage. Proceedings of the International Symposium: The Ceramics Heritage of the 8th CIMTECWorld Ceramic Congress and Forum on New Materials, Firenze, 28 June—2 July 1994. Gruppo editoriale Faenza, Faenza, 423-432. LIRER, L., PESCATORE, T., BOOTH, P. & WALKER, G. P. L. 1973. Two plinian pumice-fall deposits from Somma-Vesuvius, Italy. Geological Society of America Bulletin, 84, 759-772. LORENZONI, S., PALLARA, M. & ZANETTIN, E. 1995. The bricks of the so-called Tomba della Medusa' of the Arpi necropolis (Southern Italy). In: VENDRELL-SAZ, M., PRADELL, T., MOLERA, J. & GARCIA, M. (eds) Proceedings of the European Meeting on Ancient Ceramics-Archaeometrical and Archaeological Studies, Barcelona, 18-20 November 1993. Barcelona. MANNING, S. W. 1995. The Absolute Chronology of the Aegean Early Bronze Age. Sheffield Academic Press, Sheffield. MANNONI, T. 1980. Analisi minero-petrografiche in sezione sottile delle ceramiche del Guadone. Rivista di Scienze Preistoriche, XXXV, 69-70. 1983. Caratterizzazione mineralogico-petrografiche e tecniche di alcune ceramiche del Tavoliere. In: TINE, S. (ed.) Passo di Corvo e le civiltd neolitiche del Tavoliere. SAGEP, Genova, 94-97. PATERNE, M., GUICHARD, F. & LABEYRIE, J. 1988. Explosive activity of the south Italian volcanoes during the past 80,000 years as determined by marine tephrocronology. Journal of Volcanology and Geothermal Research, 34, 153-172. PESCATORE, T., SPARKS, R. S. J. & BRAZIER, S. 1987. Reverse grading in the Avellino plinian deposits
of Vesuvius. Bollettino della Societd Geologica Italiana, 106, 667-672. PETRONE, P. P. & FEDELE, F. 1995. Un'eruzione di 4000 anni fa. Le Scienze (Italian issue of Scientific American), 331, 22-23. PYLE, D. M. 1989. The thickness, volume and grain size of tephra fall deposits. Bulletin of Volcanology, 51, 1-15. ROLANDI, G., MASTROLORENZO, G., BARRELLA, A. N. & BORRELLI, A. 1993. The Avellino plinian eruption of Somma-Vesuvius (3760 y.B.p.): the progressive evolution from magmatic to hydromagmatic style. Journal of Volcanology and Geothermal Research, 58, 67-88. ROSE, W. I. 1993. Comment on 'Another look at the calculation of fallout tephra volumes' by Judy Fierstein and Manuel Nathenson. Bulletin of Volcanology, 55, 372-374. , BONIS, S., STOIBER, R. E., KELLER, M. & BICKFORD, T. 1973. Studies of volcanic ash from two recent Central American eruptions. Bulletin of Volcanology, 37, 338-364. SARNA-WOJCICKI, A. M., BOWMAN, H. W. & RUSSEL, P. C. 1979. Chemical correlation of some Late Cenozoic tuffs of Northern and Central California by neutron activation analysis of glass and comparison with X-ray fluorescence analysis. US Geological Survey Professional Paper, 1147, 1-14. SPARKS, R. S. J. 1986. The dimensions and dynamics of volcanic eruption columns. Bulletin of Volcanology, 48, 3-15. , BURSIK, M. I., ABLAY, G. J., THOMAS, R. M. E. & CAREY, S. N. 1992. Sedimentation of tephra by volcanic plumes. Part 2: controls on thickness and grain-size variation of tephra fall deposits. Bulletin of Volcanology, 54, 685-695. SULLIVAN, D. G. 1988. The discovery of Santorini Minoan tephra in western Turkey. Nature, 333, 552-554. TUNZI SISTO, A. M. I992a. L'ipogeo di Madonna di Loreto (Trinitapoli, Foggia). In: COCCHI GENICK, D. (ed.) L'etd del Bronzo in Italia nei secoli dal XVI al XIV a.C. Atti del Convegno, Viareggio, 26-30 ottobre 1989. Rassegna di Archeologia, 10, 545-552. 1992&. II villaggio preistorico di Madonna di Ripalta. In: COCCHI GENICK, D. (ed.) L'etd del Bronzo in Italia nei secoli dal XVI al XIV a.C. Atti del Convegno, Viareggio, 26-30 ottobre 1989. Rassegna di Archeologia, 10, 738-739. 1997. Ipogei della Daunia. Guida alia mostra archeologica (Manfredonia). Foggia. VERNET, G., RAYNAL, J. P. & ALBORE LIVADIE, C. 1999. La tephra d'Ariano, un aspect distal de 1'eruption plinienne d'Avellino du Monte Somma (Campanie, Italic). In: LIVADIE, C. A. (ed.) L'eruzione vesuviana delle Pomici di Avellino e la fades di Palma Campana (Bronzo Antico)', De Rosa, Malori (in press). VOGEL, J. S., CORNELL, W. NELSON, D. E. & SOUTHON, J. R. 1990. Vesuvius/Avellino, one possible source of seventeenth century BC climatic disturbance. Nature, 344, 534-537.
AVELLINO PUMICE IN APULIAN BRONZE AGE POTTERY WAITT, R. B., HANSEN, V. L., SARNA-WOJCICKI & WOOD, S. H. 1981. Proximal air-fall deposits of eruptions between May 24 and August 7, 1980. Stratigraphy and field sedimentology. In: LIPMAN, P. V. & MULLINEAUX, D. R. (eds) The 1980 Eruptions of Mount St. Helens, Washington. US Geological Survey Professional Paper, 1250, 617-628.
177
WALKER, G. P. L. 1971. Grain size characteristics of pyroclastic deposits. Journal of Geology, 79, 696-714. 1980. The Taupo pumice: product of the most powerful known (ultraplinian) eruption? Journal of Volcanology and Geothermal Research, 8, 69-94.
This page intentionally left blank
Human response to Etna volcano during the classical period D. K. CHESTER1, A. M. DUNCAN2, J. E. GUEST3, P. A. JOHNSTON4 & J. J. L. SMOLENAARS5 1
Department of Geography, University of Liverpool, Liverpool L69 3BX, UK 2 Centre for Volcanic Studies, University of Luton, Luton LU1 3JU, UK 3 Planetary Image Centre, University College London, Mill Hill Park, London NW7 2QS, UK 4 Department of Classical Studies, Brandeis University M.S. 016, Waltham, MA 02254-9110, USA 5 Klassiek Seminarium, Universiteit van Amsterdam, Oude Turfmarkt 129, 1012 GC Amsterdam, Netherlands Abstract: Volcanoes and eruptive activity played a part in the lives of many people in southern Italy during the classical era, no more so than on the flanks of Mount Etna (Sicily), a volcano that has been continually active throughout the historical period. Both the Romans and Greeks settled at the foot and on the lower flanks of the volcano and it seems likely that they were attracted to the region by its considerable agricultural potential, in particular its plentiful supplies of water. In this paper, literary sources are used to explore three aspects of human response to the activity of Etna during the classical period. First, the role of Etna as a stimulus to the development of myth and legend is considered, and is followed by a discussion of more 'scientific' explanations of the volcano's activity. To a large extent Etna's volcanic activity was ascribed by early writers to mythological figures, but other authors, such as Empedocles and Lucretius, stand apart from this tradition by seeking more rational explanations. The paper concludes with a discussion of what can and cannot be gleaned about the eruptive behaviour of Etna during the classical period by using literature-based and geological sources of information in combination. Records suggest that the city of Catania was partly destroyed by lava in c. 693 BC and in 425 BC. The eruption of 122sc was unusual for Etna in being explosive, and significant amounts of ash and lapilli were deposited on the southeastern flanks of the volcano, causing great distress in Catania, which required the provision of state aid.
Vesuvius and Etna played a prominent part in human perception of landscape during classical times. The volcanic origin of Mount Vesuvius, though having long been dormant, was recognized by the Greek geographer, Strabo (c. 58 BCAD 24), sometime before the terrible eruption of AD 79. So great was the impact of this eruption on contemporary consciousness that it became rapidly imprinted as a powerful symbol of divine wrath (Stauffer 1955, p. 147), influencing amongst others the author of the New Testament Book of Revelation (Bauckham 1977, p. 230). Etna, in contrast, is a much larger volcano, around 3000m in height (when compared with c. 1800m in the case of Vesuvius), and covers an area of some 1750km2 (when compared with c. 130km2 in the case of Vesuvius; Chester et al (1985)), and was almost continuously active during classical times, its
residents being only too familiar with its eruptive behaviour. Much of the activity of Etna, which is typically effusive in nature, takes place well away from inhabited areas, but occasionally larger flank eruptions occur, which cause more widespread destruction. Continuous records of eruptions are available from the 15th century AD and more fragmentary accounts stretch back to the seventh century BC. Accounts by authors of the classical period provide many valuable clues about the volcano, its activity and impact upon human consciousness over several hundred years of human settlement, In this paper, sources from classical literature will be used to better understand the relationships between Etna, the people who lived on its flanks and the intellectual community of the time. More specifically, classical source materials will be used to explore three aspects of human
From: McGuiRE, W. G., GRIFFITHS, D. R., HANCOCK, P. L. & STEWART, I. S. (eds) The Archaeology of Geological Catastrophes. Geological Society, London, Special Publications, 171, 179-188. 1-86239-062-2/00/ $15.00 © The Geological Society of London 2000.
180
D. K. CHESTER ET AL.
interchangeably with the Greek word for 'fire'. Handbooks of Greek literature and mythology emphasize the constructive aspect of this deity. He controls fire, by means of which he assisted the one-eyed Cyclopes to forge metals into Zeus' bolts of thunder and lightning; these weapons also enable Zeus to maintain his control over gods and mortals. So great is the importance of fire that Zeus conceals it from mortals, until it is stolen and given to mortals by Prometheus (Johnston 1996; pp. 55-57). For the Greeks, fire represents not only a powerful physical tool for mortals but is also equally important as the key to knowledge and foresight, and indeed, Mythological interpretations of the 'foresight' is the meaning of Pro-metis [pro = activity of Etna before; metis = thought], or Prometheus. In ProDominated as they were by the sheer size of metheus Bound by Aeschylus, Hephaistos directs Etna and confronted by its frequent activity, the giants Brontes, Bia and Steropesas as they since earliest times the inhabitants of eastern chain Prometheus to the Caucasian cliffs, from Sicily have invoked mythological interpretations where his eternally regenerating innards will be of Etna's volcanic phenomena. Even before the consumed by vultures. Whilst he is being colonization of Sicily by the Greeks in c. 740 BC chained to the cliff, Prometheus muses how his the Sicels had cults that associated volcanism theft has benefited mortals: 'I found them witless with subterranean processes. There was a temple and gave them the use of their wits and made to the fire god, Hadranus, near Adrano at the them masters of their minds.' After naming the foot of Etna (Fig. 1) (Freeman 1892, p. 34). benefits, particularly the intellectual benefits, These cults later became incorporated into that came through this gift, he concludes: 'All Greek myths. Volcanic eruptions were some- arts that mortals have come from Prometheus' times thought to be caused by a giant pinned (quoted by Johnston (1996), p. 57 and referdown beneath the volcano; according to Hesiod ences), that is, through his gift of fire. This and Pindar, for example, the Giant Typhoeus is constructive aspect of fire has tended to be said to be pinned down in Tartarus, the deepest highlighted in Roman literature as well. Handpart of the Underworld, which was located books on Greek and Roman mythology show beneath Mt Etna. In Book 3 of the Aeneid, Virgil that Roman Volcanus is equivalent to the Greek names the giant, Enceladus, as being in this Hephaestos. The name Volcanus, however, does position. Sometimes, however, volcanic activity not appear to have roots in either the Greek or is associated with the workshop of the Greek Latin languages; instead, it appears to come divine smith, Hephaestos, or his Latin equiva- from early Italic, probably Etruscan roots. lent, Vulcan, whose forges were believed to be Etruscologists derive the name from the located inside volcanoes. This god is usually seen Etruscan god Velkhan, who was primarily a god as a constructive craftsman, a cuckolded hus- of destructive fire, although Stoltenberg (1957), band (he is married to Aphrodite/Venus) and the following Altheim (1930), also calls him 'the father of creatures associated with fire. One of god of heavenly fire of the sun with its light and his sons is the monster Cacus, whom Hercules warmth', the god of productive fire, and god of encounters and kills at the future site of Rome. the hearth. The last aspect of Velkhan may have The cave of Cacus is likened to Tartarus, and some bearing on the scene in Aeneid 8.408-415 Cacus himself, when killed, is described as (throughout this paper, references to classical having a mouth scorched by the destructive texts are cited in the standard form according flames he exhaled. The 'constructive' side of to the Oxford Latin Dictionary. For examVulcan is very important: this metal-working ple, Aeneid 8.408-415 refers to book 8, pages deity and his assistants, the Cyclopes, whose 408-415) (Johnston 1996, footnote 16) where workshops were thought to be located inside Vulcan is compared with the housewife who gets volcanoes, craft the bolts of thunder and light- up at dawn to begin her chores; her first act is to ning with which Zeus controls the universe. resuscitate the fire on the hearth: she awakens For the Greeks Hephaestos was a divine (suscitat) the ashes and sopitos ... ignis, literally, craftsman, and the god of metal workers. His 'fire that has been lulled to sleep'; so, too, writes name, which is etymologically connected to the Virgil, 'the Lord of Fire (ignipotens) ... arose Greek verb flegein, 'to burn', is often used from his soft bed to forge the weapons'. But the
responses to the activity of Etna. First, the role of Etna as a stimulus to the development of myth and legend will be considered. This will be followed by a discussion of the more 'rational' explanations of both volcanic activity in general and Etnean activity in particular, which appear in the literature from the classical age. The paper will conclude with an account of what may be gleaned about the eruptive behaviour of Etna during the classical period, when textural source materials are combined with geological evidence.
ETNA VOLCANO DURING THE CLASSICAL PERIOD
181
Fig. 1. Mount Etna: isopach map (contour map showing the thickness of deposits, in cm) of pyroclastic fall deposits from the 122 BC eruptions (after Coltelli et al. (19950)) and the principal locations mentioned in the text.
Etruscan god Velkhan is primarily identified as the god of destructive fire (Stoltenberg 1957, pp. 68-69). Another Etruscan god of fire, whom Etruscologists more often equate with Roman Vulcan, is Sethlans, the Etruscan deity associated with the productive use of fire. There
was no generic name for volcanoes in antiquity, so individual volcanoes had to be referred to by their proper names or by their locations, or by a description of their behaviour: the dramatic, fiery eruptions, the streams of molten lava, the fumaroles and earthquakes, and so
182
D. K. CHESTER ET AL.
forth. The god of fire, the fire itself, and volcanic behaviour thus all seem susceptible to this distinction between constructiveness and destructiveness, simultaneously embodied in the ambiguous force of the god of fire. As the embodiment of one or both of these Etruscan fire gods, Vulcan seems to have been worshipped at Rome from the earliest known times; his temple, according to Vitruvius, stood outside the city, because of his association with destructive fire; the constructive side of Roman Vulcan seems also to have been accepted early. Pallotino (1975) has on several occasions pointed out an interesting reference in Pliny the Elder (Natural History 35.157) to a sixth century sculptor from Veii named Vulca, who was commissioned by Tarquinius Superbus to work on a statue of Capitoline Zeus. Capdeville (1995) has recently examined the religious practices concerning this god and has shown that fire and Vulcan is one of the most primitive and longstanding sources of religious worship in archaic and classical Rome. This god, then, appears to have been associated with metal-working and fire from a very early time, not only among the Greeks and Romans, but also among the preRoman inhabitants of Italy. His epithets shows that he has power over fire itself, which in Prometheus Bound becomes the key to intelligence and forethought: he is ignipotens (Lord of Fire), Anax Aetnaeos (Aetnaean Lord), and the weapons he forges are irresistible. Virgil appears to have attempted to match the myth of the destructive, angry Titan at the base of the volcano to the violent eruptions of contemporary volcanoes, and the more benign volcanic activities to the myth that the Lord of metalcraft housed his forges in volcanoes. More rational interpretations of the activity of Etna Philosophically oriented authors, such as Empedocles (mid-fourth century BC) and Lucretius (first century BC), sought more rational explanations of volcanic activity, but were not immune to the religious and mythological beliefs of their times. Empedocles of Acragas (modern Agrigento, situated on the southern coast of Sicily), dabbled in magic and mysticism, but was also highly respected for his scientific theories. Among the many accounts of the death of Empedocles, perhaps the most intriguing is the one that argues that he died by throwing himself into Mount Etna. His detractors said he did this in an attempt to destroy any trace of his body, and thus to defraud his followers by making
everyone believe he had become a god. One of the bronze sandals he wore, however, was ejected by the volcano and exposed what had really happened. A clue to the possible symbolism of Empedocles' singular bronze sandal has been found in some of the surviving 'magical papyri', a collection of handbooks and treatises kept by priests of mystery cults, professional astrologers and other practitioners of the occult (Luck 1985, pp. 15-17). It has been suggested that the bronze sandals were a substitute for leather, because Empedocles had a horror of animal sacrifice; others argued that it was an attempt to place himself, as it were, on a bronze pedestal while still alive, or at any rate to establish some sort of symbol of his divinity. In his recent book on Empedocles, Kingsley has identified a passage in a magical papyrus showing that 'a bronze sandal in antiquity was a symbol connected specifically with underworld ritual and magic' (Kingsley 1995, p. 238). This papyrus, which provides magical formulae, in this case makes it clear that it had to be one sandal, and only one sandal, that was worn. This single bronze sandal was apparently a symbol of Hekate, who granted access to the Underworld. Empedocles devotes considerable attention to establishing his abilities as a magician, but as Kingsley argues, the information that is essential to the practice of magic was interwoven with 'knowledge of concrete details about the inner workings of nature' (Kingsley (1995), p. 229; see also Bussanich (1996), p. 604). Empedocles was the first person in western literature to formulate the theory of four basic elements. Whereas his predecessor Heraclitus (Robinson 1987) names only three elements (earth, water, and aither or fire) Empedocles separates aither, the third element, into air and fire, and he calls these elements the four 'roots' of all things, which he identifies with the names of deities. One of the unresolved puzzles about Empedocles continues to be the question of which elements these deities represent. Whereas current orthodoxy identifies Zeus as representative of fire, Hera of air, Hades of earth, and Nestis of the sea, Kingsley (1995) returns to an earlier orthodoxy and argues forcibly that Empedocles identified Zeus as air, Hera as earth, Hades (Aidoneus) as fire and Nestis as water (sea). Kingsley's identification of fire with Hades is of particular relevance here. He argues that Platonists and Neoplatonists misconstrued Hades as air and interpreted the sub-lunar realm as 'the dark region of air in the earth's shadow, ... the terrestrial region ... which is alive with the suffering and wailing of earthbound souls' (Bussanich 1996, p. 2). As the
ETNA VOLCANO DURING THE CLASSICAL PERIOD realm of Hades is in fact below the earth, Bussanich suggests that we 'take Empedocles at his word'. Beginning with Empedocles' statement that 'there are many fires burning beneath the earth', Kingsley (1995) suggests links between the volcanism of Sicily, with its subterranean rivers of lava, hot mud and hot springs, and Hades. As Empedocles names Hephaestus twice, Kingsley referring to the element of fire, 'focuses on the role of this fire divinity in Sicilian cult and myth, and especially his association with Mount Etna . . . [and]... concludes that Hades manifests fire in its destructive aspect, whereas Hephaestus as craftsman and creative force ... displays the creative aspect of fire' (Bussanich (1996); see para. 3). For the Presocratic philosophers, fire was one of the basic elements constituting the universe. Heraclitus of Ephesus (fl. 504-501 BC), the predecessor of Empedocles, identifies earth, sea and aither (which is sometimes equated with air, and sometimes with fire) as the three basic elements. For Heraclitus, the entire cosmos is composed of masses of earth and sea, surrounded by fire or aither, all these elements are continuously being extinguished and rekindled in a regular measured manner, so that the total proportion of each remains always the same. Kirk & Raven (1966) interpret the cosmos of Heraclitus as consisting of masses of earth that were interpenetrated with secondary fire, as in a volcano. This compares with a similar argument by the later, Roman poet, Lucretius in De rerum natura, who explains Etna's eruptions as being caused by the heated air in its hollow caves (6.701). Empedocles shared with Heraclitus a belief in the constant extinction and rekindling of elements, as well as the idea that the fires of heaven, including the sun, had their origins in the depths of the earth. Heraclitus is reported by Aetius to have said that the 'that the stars are compressed portions of fire' (Aetius 2.13.8; see Diels (1954)), and 'stars are nourished by the exhalation that comes from the earth' (2.17.4; see Diels (1954)). Empedocles makes repeated use of 'the visible sun as his chief example of elemental fire in the world around us'. The sun, he maintains, is not in itself fire, per se, but rather a reflection of the elemental fire, thus implying that the fires of heaven, including the sun, 'had their origins in the depths of the earth'. Even Heraclitus' contemporary, Parmenides, whose teleogical, fixed notions are in many ways directly opposed to the theories of Heraclitus and Empedocles, describes how the daughters of the Sun took him down into the house of Night, thus joining these two opposites in the house of Hades,
183
where such things can be shown to happen repeatedly in classical literature and philosophy (Kingsley 1995, pp. 52-54). Empedocles indicates that his teachings enable his student to control the weather as well as restore the dead to life: 'And you'll stop the force of the tireless winds that chase over the earth and destroy the fields with their gusts and blasts; But then again, if you so wish, you'll stir up winds as requital. Out of a black rainstorm you'll create a timely drought for men, and out of a summer drought you'll create Tree-nurturing floods that will stream through the ether. And you will fetch back from Hades the lifeforce of a man who has died' (DK Bill; see Kingsley (1995), pp. 52-54). Diels (1954) assigns the above fragment to the epilogue of the poem, and along with other commentators, dismisses its contents as 'pure magic'; but if we take Empedocles at his word, it is possible that he is thinking of himself and his teachings as somehow partaking in the powers of those volcano-dwelling makers of thunder and lightning. Kingsley (1995, 218fT.) compares the fragment to Diodorus Siculus' description of the Telchines, the legendary smiths who were able to produce clouds, rainstorms, hail and snow when they chose. In addition, he considers the scene in Apollonius of Rhodes' epic, The Argonautica, and also the scene in Aeneid 8, where all the ingredients of storms are included in the thunder and lightning-bolts themselves. They are described as very fine works of art, but they are none the less storm weapons, with prepackaged destruction included. The anonymous poet of Aetna (probably written in the first century AD before 79; see Goodyear (1984)) concludes his profession of love for the volcano with a satire on modern tourists, who travel all over the world to visit the so-called miracles such as Thebes, Athens and Troy in ashes, and hasten to see the famous paintings and sculptures of the Greeks. We should abandon these ignoble cares and rather look upon 'the colossal work of the artist nature' (599), i.e. Mount Etna. Earlier in the poem the praises of studying science were sung extensively: 'not cattle-like to gaze on the world's marvels merely with the eye, not to lie outstretched upon the ground feeding a weight of flesh,
184
D. K. CHESTER ET AL.
but to grasp the proof of things and search into doubtful causes . . . all this is the mind's divine and grateful pleasure' (222-257; see Duff (19610,6) for text references). According to Duff (19616), this proclamation of the majesty of physical research is most exceptional in ancient doxography; more often science is only the means to much higher aims, such as to free humanity from fear of the gods or of punishment after death, and from believing that volcanic eruptions are the acts of a god. The variety of theories on the causes of earthquakes and volcanic eruptions in Greek and Latin doxography are all concerned with air or wind, (sea) water and (underground) fire. Since Thales of Miletus (sixth century BC), a great number of Greek and Latin philosophers have offered theories on volcanic activity, some based on direct observation and others being purely speculative. Plato may have seen Etna in eruption during his first visit to Sicily (388387 BC). In the Phaedo he suggests that an enormous river of fire (Pyriphlegethori) in the depths of the earth feeds the volcanic craters; he was the first to refer to a lava-flow in his description of the subterranean regions: 'many streams of liquid mud of varying density, like those who precede the lava-flow in the volcano of Sicily, or like the lava itself. Aristotle (384322 BC) explained the underground fire from particles of air that burst into flames from the friction of the wind. Others, considering the Earth a living creature, explain its convulsions (volcanic manifestations) as analogous to flatulence and spasms in the human body. The greater part of Greek theory on volcanism has been handed down to us by the Latin authors Lucretius, Seneca and Pliny the Elder (AD 23 or 24-79), writer of the influential Naturalis Historia, completed in AD 77. Following the principles of Epicurean philosophy, the poet Lucretius (94-55 BC) discusses a series of possible explanations in his didactic poem De rerum natura (6.535-702; see Bailey (1947)), among which are included the fall of masses of earth into a subterranean lake, and subterranean winds upsetting the balance of the earth. Lucretius proposed that the volcano was hollow and riddled with caverns. These caverns penetrated to the sea and allowed water to percolate to the depth of the volcano's throat. In the caverns wind and air were warmed, heating the surrounding rocks, and with burning fire and sulphurous flames fire rises and forces itself out of the volcano, scattering ashes far and wide with rolls of smoke and pitchy blackness.
The study of volcanic activity became very popular when earthquakes caused great destruction in Campania, the region around and to the north of Naples, in AD 62-64, and pyroclastic deposits from Vesuvius buried Pompeii and Herculaneum in AD 79. Seneca's (5BC-AD65) dissertation on the theories of his predecessors in Naturales Quaestiones book F/(Gummere 1925) was triggered by the frightful events between AD 62 and 64; he is the first to adduce the principle of gas pressure in explaining the force of eruptions. Like Strabo, the anonymous poet of Aetna unfortunately considered Vesuvius to be extinct (431-432; see Duff (19616)), from which we may safely conclude that this enthusiastic poem was composed before AD 63 (Bomer 1976). Pliny the Younger (AD 62-114), writing for the historian Tacitus, provides a detailed description of the AD 79 eruption, which caused the death of his uncle Pliny the Elder, who was in command of the fleet at Misenum at that time and combined a rescue expedition with his scientific exploration of the great eruption, in his Letters, VI.16 and VI.20 (Rackham 1949). In this account Pliny the Younger offers the first detailed eyewitness description of a cataclysmic eruption. This style of volcanic activity is now formally referred to as Plinian by volcanologists (Newhall & Self 1982). Latin epic poetry persisted in explaining volcanic activity as a manifestation of superhuman forces, such as giants punished and buried under volcanoes, and Vulcan producing weapons for Jupiter in his workshop on Sicily, or the island of Vulcano. Since Virgil (70-19 BC), however, similar mythological descriptions are often cleverly phrased in the language of science (Fairclough 1954). Likewise, in the emotional poetic responses to the AD 79 eruption of Vesuvius in works by Statius and Martial (Mozley 1928), any distinction between mythological and scientific belief seems deliberately blurred. These poets would certainly have smiled at the seriousness of the cautionary voice of the poet of Aetna: 'First, let none be deceived by the fictions poets tell - that Aetna is the home of a god, that the fire gushing from her swollen jaws is Vulcan's fire, and that the echo in that cavernous prison comes from his restless work. No task so paltry have the gods' (Duff 19616, pp. 29-33). The Roman poet Ovid (43BC-AD17) proposed that the activity of Etna could be explained by its many air passages, which breathed flame and will continue until its fuel
ETNA VOLCANO DURING THE CLASSICAL PERIOD is exhausted (see Chester et al. (1985), p. 25) and at the close of antiquity, this wedding of the language of science and poetry is completed by the poet Claudian (c. AD 400). In the grotesque mythological picture which he paints in the Rape of Proserpina, Claudian envisions the goddess Ceres climbing Mount Etna to fire her torches; her knowledge of scientific theory will enable the goddess to continue her mythological search for her daughter in the darkness of the night. It is clear that, though contemptuous of the myths surrounding volcanic phenomena, these philosophers based their interpretations on speculation and musing, and showed little interest in empirical observation. It is interesting that the nineteenth century Scottish geologist, Sir Archibald Geikie, in the Williams Memorial Lecture presented at Johns Hopkins University in 1896 stated: 'the speculations of the philosophers who began to observe the operation of natural processes and who, through their deductions were often about as unscientific as the myths for which they are substituted, may yet be claimed the earliest pioneers of geology' (Geikie 1905, p. 7).
Literary and geological accounts of the eruptive behaviour of Etna in the classical period As well as being concerned to record myths and to supply early rational explanations of volcanic activity, classical authors also provide much valuable information about actual eruptions (Table 1). Some of this information may, moreover, be cross-referenced and correlated with geological evidence and allows an unparalleled record of volcanic activity to be reconstructed over the past 3000 years. One of the earliest references to an eruption of Etna was by Pindar in his Pythian Odes, and this probably concerns an eruption that occurred between 474 and 479 BC (see Rodwell 1878; Chester et al. 1985). There is some ambiguity, however, in relating early records to specific eruptions. Thucydides refers to several early eruptions, the first of which is considered to have occurred c. 693 BC, with vents opening on the southern flank of the volcano and sending a flow of lava to the coast, partially destroyed Catania (see Tanguy 1980; Romano & Sturiale 1982; Stothers & Rampino 1983; Chester et al. 1985). During this eruption it is recorded that as lava entered Catania two youths, Anapias and Amphinomus (known as the fratelli pit}, carried their aged parents on their shoulders through
185
the flaming streets to safety, and a temple was subsequently erected in their honour. Diodorus Siculus (14.59.3; see Stothers & Rampino 1983) reports a lava flow that entered the sea to the north of Catania in 394-396 BC. The eruption formed a cone, Mt Gorna, near Trecastagni (see Fig. 1) and lava flowed east before entering the sea at Santa Maria la Scala (Fig. 1; see also Romano & Sturiale 1982; Chester et al. 1985). Lava cut the main coastal road of eastern Sicily, and this caused the Carthaginian general, Himileo, to detour to the west of the volcano on his march from Messina to Syracuse. Some sources imply that the 394-396 BC lavas were erupted from Mt Moio, a cone on the northern periphery of Etna (see Fig. 1), but this is not possible. The lavas from Mt Moio are not only prehistoric in age, but the Greek settlement of Naxos is also built on the distal portion of these flows (see Fig. 1). It appears that there was increased activity on Etna after 141 BC and this culminated in a major eruption in 122 BC (Rodwell 1878; Hyde 1916). There was explosive activity from the summit craters and ash was deposited along a southeasterly axis of dispersal. Thick accumulations of ash covered roofs in Catania, leading to the collapse of many buildings and considerable damage. In response, the Roman authorities granted the people of Catania immunity from taxes for 10 years (Rodwell 1878). Lava was erupted near Trecastagni (Fig. 1) and flowed towards the sea to the north of Catania. An eruption of Etna is reported at the start of the civil war between Pompey and Caesar in 49 BC, and this was followed by a major explosive event, which took place at the time of Caesar's assassination in 44 BC (Chester et al. 1985). Stothers & Rampino (1983) suggest the hazy conditions and subsequent crop failures may have been caused by volcanic aerosols injected into the atmosphere by this explosive event. Virgil (Georgia 1.466-473, from Stothers & Rampino 1983, p. 6359) states: 'After the death of Caesar . . . how often we saw Etna flooding out from her burst furnaces, boiling over the Cyclopean fields and whirling forth balls of flame and molten stones'. These near-contemporary accounts indicate that this was certainly a large eruption by Etna's standards. According to Pliny the Elder (see Stothers & Rampino 1983, p. 6359), 'Etna . . . is so hot that it belches out sands in a ball of flame over a space of 50 to 100 (Roman) miles'. Ash falls from eruptions of Etna are rarely deposited more than 20 km from the volcano, so a distance of up to 150 km (c. 100 Roman miles) would support the interpretation that this was a large eruption.
D. K. CHESTER ET AL.
186
Table 1. Summary of documented activity of Etna from c. 700 BC to the mid-fifth century AD Date
Documented eruptive activity
Comment
Reference
c. 693 BC
Eruption on south flank: main vent probably Mt Monpilieri, 1 km south of Nicolosi and other vents reported at Campo Pio, 6 km north of Catania; lava invaded Catania, which was established in 729 BC
Eruption famous for the story of the fratelli pit, the two brothers who carried their parents away from the advancing lava
1, I, II, III
479-475 BC
Vent area probably on SSE flank
1, I, II, III 1, I, II, III
1, 2, I, III
425 BC
Substantial eruption
Sixth year of Peloponnesian war
396 BC
Flank eruption building up a cone at Mt Gorna, 3.5km N of Trecastagni; lava flows entered the sea just N of Acireale
Lava flows prevented the Carthaginian general Himileo from marching along the coast from Messina to Syracuse
135BC
Lava and ashes erupted; eruptions also reported in 140 and 126BC
122BC
Explosive activity at summit craters, ash deposited to the SE; lava erupted from vent near Trecastagni (Mt Trigona)
Thick ash broke roofs in Catania; inhabitants granted immunity from taxes for 10 years
3, 4, I, II, III
49 and 44 BC
49 BC at time of war between Pompey and Caesar; 44 BC at time of Caesar's death
Dimming of the sun and haze reported in Italy
5, I, II, III
36 BC
Report of eruption
32 BC
Lava causing damage
AD?70
Reported eruption
AD 252-253
Flank eruption, lava nearly reached Catania
AD417
Darkening of sun described by Marcellinus may be same eruption as reported by Olympiodorus
III
6, I, III
During civil war
7,111 6, I, III
Flow stopped (allegedly) when veil of St Agatha brought to the flow front
8, I, II, III
9, 10, III, IV
C. AD 420
Classical references: 1, Thucydides History of the Peloponnesian War 3.116; 2, Pindar Pythian Odes; 3, Lucretius De rerum natura 6.639-646; 4, Orosius Adversus 5.18; 5, Virgil Georgics 1.466-473; 6, Appian Civil Wars 5.117; 7, Dio Roman History 50.8.3; 8, Suetonius Caligula; 9, Marcellinus Chronicon; 10, Olympiodorus in Photius Library. Modern summaries of activity in classical times: I, Chester et al. (1985); II, Romano & Sturiale (1982); III, Stothers & Rampino (1983); IV, Uchrin (1990).
In AD 40, Suetonius (Caligula 51, referred to by Stothers & Rampino (1983)) mentions that it was an explosion of Etna that caused Caligula to flee from the region, but the detailed nature of this event remains unclear. The next major phase of activity to be recorded is the eruption of AD 252-253. This was a substantial flank eruption during which a cone, Mt Peloso (Fig. 1), was formed 2km north of Nicolosi (Romano & Sturiale 1982), and lava flowed south, threatening Catania. Inhabitants brought out the veil of the recently martyred St Agatha, which was paraded in front of the advancing lava. This was the first recorded appearance of
St Agatha's veil in its role of protecting the local community from the destructive forces of Etna, and this custom of attempted divine propitiation has continued to the present day (Chester et al. 1985; Duncan et al. 1996). The last record of an eruption of Etna during Roman times comes from the early fifth century AD (see Table 1) and is a reference in a fragment of Olympiodorus's History (quoted by Photius). Uchrin (1990) suggests that this eruption may have been the cause of the darkness of AD 417 described by Marcellinus in Chronicon. Over the next 600 years there are few references to the eruptive activity. This probably
ETNA VOLCANO DURING THE CLASSICAL PERIOD reflects the political and cultural environment of the so-called 'dark ages' and a lack of literate observers, rather than a period of calm in the history of this most active of volcanoes. Data from the historical records of the classical period are insufficiently detailed to allow a detailed picture of the eruptive behaviour of the volcano to be established, as the Romans and Greeks did not make detailed observations of the volcano and its behaviour. Etna is a large volcano and persistent activity at the summit and, indeed, most flank eruptions are unlikely to have had much impact on the urban communities, which were and are concentrated on its southern and eastern flanks. Coltelli et al (19950), in a detailed study of the pyroclastic deposits on the southern and eastern flanks of Etna, correlate one of their youngest layers (FG) with the 122BC eruption. They quote a radiocarbon date of 2180 ±60 for a palaeosol immediately below the FG layer. A map (adapted from Coltelli et al. (1995^)), showing the distribution of the pyroclastic fall deposits from the 122BC eruption, is reproduced as Fig. 1 and, as the figure shows, more than 10cm of ash was deposited in Catania. During the 1991 eruption of Mt Pinatubo in the Philippines over one-third of the buildings in the town of Castillejos suffered partial or total collapse as a consequence of an ash-fall thickness of 15-20 cm. Such explosive eruptions are often accompanied by heavy rainfall and the ash may become waterlogged, so increasing its weight. It is not surprising, therefore, that ash fall from the 122BC eruption of Etna led to damage in Catania that required support from the Roman authorities. This eruption is classified as Plinian by Coltelli et al. (19950), and was one of the most explosive events to have occurred on Etna over the last 5000 years. It seems likely that the eruption column would have reached a height of more than 10km. By comparison, the 1979 eruption, one of the more explosive in recent years, despite producing a high eruptive column leading to ash dispersal in a southeasterly direction, deposited only some 3mm of ash on Catania. It caused minimal disruption, which included closure of the airport for a short period of time. A pyroclastic layer that is younger than deposits from the 122BC event is suggestive of a subsequent explosive eruption of some significance and magnitude (Coltelli et al. 19956). It is itself overlain by materials that are dated at 980 ±60 years BP and Coltelli et al. (19956) suggest that the pyroclastic layer may be of first century date and correlated with literary accounts of ash which fell on Taormina and
187
Catania (see Fig. 1) between 38 and AD 40. Alternatively, it may be that this pyroclastic layer resulted from a major explosive eruption in 44 BC around the time of Julius Caesar's death.
Discussion Lack of documents dating from the period between the collapse of the Roman Empire in the west and the re-emergence of scholastic activity following the Norman conquest in the llth century AD means that with only a few exceptions, little is known about Etna's eruptions from c. AD 400 to 1329. Following the Renaissance, information quickly improved, and from the start of the 16th century the record of eruptions becomes virtually complete. Records show that in the 17th and early 18th centuries Etna behaved in a more vigorous fashion than either before or since, with a mean lava effusion rate of c. 1.19m3 s"1 being maintained (Hughes et al. 1990). Large flank eruptions produced several lava flows that reached to the limit of the volcanic deposits (see Fig. 1), and in 1669 Catania was destroyed. After 1750 Etna reverted to its pre-17th century style of activity, showing a lower mean lava effusion rate of only c. 0.18m3 s"1 (Guest & Duncan 1981). Activity during this period has been characterized by persistent activity at the summit craters, at times involving the low effusion rate discharge of lavas, punctuated every few years by flank eruptions showing higher rates of emissions. Only once during this long period has lava caused significant damage to property, with the destruction of the town of Mascali in 1928 (Duncan et al. 1996). It is also apparent that during the past 200 years of almost continuous activity, only occasionally have eruptions posed a significant threat to communities. In the classical period it is likely that only major eruptions would have been recorded, with additional accounts being produced when famous figures were visiting the region. The literary and geological evidence suggests that for most of the classical period Etna operated in a fashion broadly similar to the steady state observed from c. AD 1750 to 1999, but the fact that substantial lava flows reached the foot of the volcano in c.693fiC(?), 396 BC, 122BC and AD 252-253 may imply that for at least some of the time Etna was behaving in a fashion comparable with that of the 17th century AD. We wish to place on record our thanks to M. Balmuth, Tufts University, USA, for introducing the authors to each other and for providing a stimulating academic
188
D. K. CHESTER ET AL.
environment at the conference 'The cultural response to the volcanic landscape', held at the Department of Classics, Tufts, in November 1996. A.M.D. gratefully acknowledges financial support from the Clough Fund of the Edinburgh Geological Society.
References ALTHEIM, F. 1930. Griechische Cotter im Alien Rom. A. Topelmann, Giessen. BAILEY, C. 1947. Lucretius: de Rerum Natura Libri Sex. Oxford University Press, Oxford. BAUCKHAM, R. 1977. The eschatological earthquake in the Apocalypse of John. Novum Testamentum, 14(3), 224-233. BOMER, F. 1976. Metamorphosen P. OvidiusNaso; Kommentar von Franz Bomer. C. Winter, Heidelburg. BUSSANICH, J. 1996. Review of: KINGSLEY, P. 1995, Ancient Philosophy, Mystery, and Magic: Empedocles and Pythagorean Tradition. Oxford University Press, Oxford. Bryn Mawr Review, 17(7), 602-607. CAPDEVILLE, G. 1995. Volcanus: Recherches Comparatistes sur les Origines de Culte de Vulcain. Ecole Fran9aise de Rome, Rome. CHESTER, D. K., DUNCAN, A. M., GUEST, J. E. & KILBURN, C. R. J. 1985. Mount Etna: the Anatomy of a Volcano. Chapman and Hall, London. COLTELLI, M., CARLO, P. & VEZZOLI, L. \995a. A plinian eruption of basaltic composition in the historical activity of Mount Etna. Periodico di Mineralogia Roma, 64, 145-146. , & 19956. Stratigraphy of the Holocene Mt. Etna explosive eruptions. Periodico di Mineralogia Roma, 64, 141-143. DIELS, H. 1954. Die Fragmente del Vorsokratiker, 5th edn. Weidmann, Berlin. DUFF, J. D. \96la. Silius Italicus. Heinemann, London. 19616. Minor Latin Poets. Heinemann, London. DUNCAN, A. M., DIBBEN, C., CHESTER, D. K. & GUEST, J. E. 1996. The 1928 eruption of Mount Etna Volcano, Sicily and the destruction of the town of Mascali. Disasters, 20, 1—21. FAIRCLOUGH, H. R. 1954. Vergil. Heinemann, London. FREEMAN, E. A. 1892. Sicily, Phoenician, Greek and Roman (The Story of Nations, Vol. 31). Fisher Unwin, London. GEIKIE, A. 1905. The Founders of Geology, 2nd edn. Macmillan, London. GOODYEAR, F. R. D. 1984. The 'Aetna': thought, antecedents and style. In: TEMPORINI, H. & HAASE, W. (eds) Aufstieg und Niedergang der Romischen Welt, de Gruyter, Berlin and New York, 32(1), 344-363.
GUEST, J. E. & DUNCAN, A. M. 1981. Internal plumbing of Mount Etna. Nature, 290, 584-586. GUMMERE, R. M. 1925. Seneca, ad Lucilium epistulae morales. Heinemann, London. HUGHES, J. W., GUEST, J. E. & DUNCAN, A. M. 1990. Changing styles of effusive eruption on Mount Etna since AD 1600. In: RYAN, M. P. (ed.) Magma Transport and Storage. Wiley, Chichester, 385-406. HYDE, W. W. 1916. The volcanic history of Etna. Geographical Review, 1, 401-418. JOHNSTON, P. A. 1996. Under the volcano: volcanic myth and metaphor in Vergil's Aeneid. Vergilius, 42, 55-65. KINGSLEY, P. 1995. Ancient Philosophy, Mystery, and Magic: Empedocles and Pythagorean Tradition. Oxford University Press, Oxford. KIRK, G. S. & RAVEN, J. E. 1966. The Presocratic Philosophers. Cambridge University Press, Cambridge. LUCK, G. 1985. Arcana Mundi: Magic and the Occult in the Greek and Roman World. Johns Hopkins Press, Baltimore, MD. MOZLEY, J. H. 1928. Statins. Heinemann, London. NEWHALL, C. G. & SELF, S. 1982. The volcanic explosivity index (VEI): an estimate of the explosive magnitude of historical volcanism. Journal of Geophysical Research, 87(C), 1231-1238. PALLOTINO, M. 1975. The Etruscans (translated J. Cremona, ed. D. Ridgeway). Lane, Bloomington, IN. RACKHAM, H. 1949. Pliny: Natural History. Heinemann, London. ROBINSON, T. M. 1987. Heraclitus: Fragments. University of Toronto Press, Toronto, Ont. RODWELL, G. F. 1878. Etna: a History of the Mountain and its Eruptions. Paul, London. ROMANO, R. & STURIALE, C. 1982. The historical eruptions of Mt. Etna (volcanological data). In: ROMANO, R. (ed.) Mount Etna Volcano, a Review of Recent Earth Science Studies. Memorie della Societa Geologica Italiana, 23, 75-97. STAUFFER, E. 1955. Christ and the Caesars. SCM, London. STOLTENBERG, H. L. 1957. Etruskische Gottnamen. Gottschalk, Leverkusen. STOTHERS, R. B. & RAMPINO, M. R. 1983. Volcanic eruptions in the Mediterranean before AD 630 from written and archaeological sources. Journal of Geophysical Research, 88, 6357-6371. TANGUY, J.-C. 1980. L'Etna etude petrologique et paleomagnetique implications volcanologique. These de Doctorat, Universite Pierre et Marie Curie, Paris. UCHRIN, G. 1990 Olympiodorus's eruption of Mount Etna. A possible date of, AD417. Eos, 71(11), 329 &334.
The Johnston-Lavis collection: a unique record of Italian volcanism W. L. KIRK1, R. SIDDALL1 & S. STEAD2 1
Department of Geological Sciences and 2Manuscripts and Rare Books, The Library, University College London, Gower Street, London WC1E 6BT, UK (e-mail:
[email protected],
[email protected],
[email protected]) Abstract: Housed at University College London, the Henry James Johnston-Lavis collection of rocks, minerals, photographs, gouaches and engravings is extremely important in that it provides a record of volcanism in southern Italy, especially in the latter part of the 19th century. The collection of the intrepid Dr Johnston-Lavis also contains literature and materials relevant to earlier eruptions of the Italian volcanoes, and sample collections of both rocks and minerals from other world-wide locations of mineralogical and volcanological interest. Although regrettably now much depleted, a substantial part remains, and this continues to be a valuable resource for volcanologists, historical geologists and archaeologists.
'I ran to inform Lavis, and, of course, a few minutes later we were both hurrying to the seat of activity. What a night we spent on the Vesuvian inferno! Having reached the lavaplain of the Atrio, we began ascending the eruptive cone, which looked like a huge pile of yellow corn-meal owing to sublimates of ferric perchloride which coated its surface materials. A considerable portion of the cone had already fallen in, and the widened crater measured about three hundred feet in its widest diameter. The broken edge was all fissured and dangerous to approach; other great rents scored the cone in every direction. Vast quantities of tumultuous white steam escaped from this great cauldron, which was no longer an eruptive crater, but a gaping fumarole. We struggled over the incoherent scoriae, rounding the cone in an easterly direction in order to reach the opening from which the lava was issuing, but great gusts of choking sulphurous acid prevented our progress, and although lower down we had dared to jump over the fissure where it was only three or four feet wide, we were obliged to make a hasty retreat to avoid being scalded or suffocated by the heated, stifling vapours. The Atrio was also fissured in many places and covered by great patches of variegated sublimates, which contrasted sharply with the gloomy grey of the old lavas and scoriae. At about 2.30a.m. the ground beneath our feet trembled violently and, amidst flashes of lightning, a new gap burst open at the very foot of the eruptive cone, pouring forth a great stream of molten lava, which seemed to bear down upon us precipitously. Terrified by
the sudden unexpected explosion, we retreated as fast as we could, but found our way barred by the earlier stream, which was still copiously gushing out. Believing ourselves entirely cut off, and fearing lest we should be overwhelmed by either the scorching lava or the scalding irrespirable vapours, we proceeded to clamber up the almost perpendicular face of the Somma escarpment. Fortunately, Lavis had climbed it on a former occasion, whilst studying its exposed lava dykes, and we managed to reach safety in a piteous condition of exhaustion. From the summit of the Somma, still gasping, we witnessed the grand spectacle of the eruption.' So did Woodward (1918) describe an incident in June 1891, involving his intrepid colleague Dr Henry James Johnston-Lavis (1856-1914) (Fig. 1). Johnston-Lavis' interest in geology had begun as a schoolboy. In 1873, whilst training for the medical profession, he came to University College London, where he was taught for a while by the Professor of Geology, John Morris. He joined the Geologists' Association in 1874, and became an under-age Fellow of the Geological Society a year later. His first paper, not on volcanology, but on the Lower London Tertiary deposits in Kent, was read before the Geologists' Association in 1876 (Johnston-Lavis 1876). He went on to become a world expert on south Italian volcanoes, particularly on Vesuvius (e.g. Johnston-Lavis 18810, 1882, 1895). He had moved to Naples in 1879, where he was able to combine his medical profession with his passion for volcanoes, eventually becoming Professor of Volcanology at the Royal University of
From: McGuiRE, W. G., GRIFFITHS, D. R., HANCOCK, P. L. & STEWART, I. S. (eds) The Archaeology of Geological Catastrophes. Geological Society, London, Special Publications, 171, 189-194. 1-86239-062-2/00/ $15.00 © The Geological Society of London 2000.
190
W. L. KIRK, R. SIDDALL & S. STEAD
Fig. 1. Henry James Johnston-Lavis, posing for the camera on Vesuvius (1906) (Woodward 1918).
Naples in 1893 (Anon. 1914). He stayed for 15 years before moving to the south of France, spending the winters at Beaulieu, near Nice, and the summers at Vittel in the Vosges. He published frequently, producing over 200 books, memoirs, etc., some of which were on medical matters, but most were on volcanic eruptions and their products, particularly those in southern Italy. His first major work was that on 'The geology of Monte Somma and Vesuvius, being a study in vulcanology', published in the Quarterly Journal of the Geological Society in 1884 (Johnston-Lavis 1884). When published, this was 84 pages long, having been considerably condensed from the original, and in what has been described as an unwise and unsympathetic fashion in Lavis's absence from England! He was subsequently appointed secretary by the British Association in order to investigate Vesuvius further, and reported annually from 1886 to 1896 (e.g. Johnston-Lavis 1895). In 1891, he published a geological map of Vesuvius at a scale of 1:10000 in six sheets, a culmination of his survey work between 1880 and 1888 (Johnston-Lavis 189la). The geological map, together with its pamphlet, was reviewed in an article in Nature (1891), in which Lavis was described as 'an equally indefatigable investigator and
historian' as Sir William Hamilton, who studied Vesuvius in the latter half of the 18th century (Anon. 1891). Other field studies included work in the Lipari Islands and on Etna, and a brief visit to the Ponza Islands in 1884, whence he had to depart in great haste at short notice. This occurred because at the time of his arrival there was a great outbreak of cholera, and the local people, who saw him breaking off rocks and wrapping them up, surmised that he had been sent by the Government to spread 'cholera powder'. Another major work was his monograph on the earthquakes at Ischia, published in 1885 (Johnston-Lavis 1885; see also JohnstonLavis 1880, 18830, £). In his reports on the earthquakes, he proposed that money would be better spent on establishing monitoring stations, with the possibility of understanding how the earthquake focus moved, than on rebuilding the badly damaged buildings. Other places he visited included Iceland in 1890 (accompanied by Dr Tempest Anderson), the Auvergne in France, the Eifel district of Germany, the British Tertiary Province and Edinburgh, and Lake Balaton in Hungary. Not one to keep his knowledge to himself, Lavis also gave the occasional lecture, and passed on his expertise to others as field trip leader, for example during the Geologists' Association visit to the south Italian volcanoes in 1889 (Johnston-Lavis 189\b). Over the years, Johnston-Lavis acquired a magnificent collection not only of volcanological specimens, but of books, paintings and engravings as well (Fig. 2). He was an avid photographer, and had a superb record of the late 19th century Vesuvian eruptions, which almost cost him his life on at least one occasion. He bequeathed his entire collection to the University of London, to form a geodynamical section of the Geology Museum at University College. He died in a car accident in 1914, but transference of his collection from Beaulieu was delayed by the war, and it did not arrive at UCL until 1921. The collection was opened to the public in July 1925, at which point it was housed in 134 Gower Street. It was visited by the Geologists' Association in 1927, and was described as 'probably the finest collection in existence (the Neapolitan Collections not excepted) of specimens, maps and literature relating to South Italian volcanic phenomena' (Earle 1928). It was subsequently housed in a variety of locations, including some very unsuitable storage, with resultant loss of material in the late 1950s and early 1960s. The situation was reviewed and recommendations to safeguard the collection were put forward by McKay (1970) before it
THE JOHNSTON-LAVIS COLLECTION
191
Fig 2. Forum Vulcani vulgo Solfatara. A late 18th century engraving of Solfatara from the Johnston-Lavis Collection.
was brought back to the main Department of Geology in 1975. The rock and mineral collection Johnston-Lavis was an avid collector of rock and mineral samples. The existing rock catalogue, a hand-written copy of Johnston-Lavis' original, was made by K. W. Earle, the Geology Department Curator in 1925, and records 5251 numbered entries. His separate mineral catalogue, again a duplicate of the original, lists c. 3500 entries, many of which recorded several specimens. When brought into the department in 1975, the collection was much depleted. The best specimens were singled out for display cabinets, and others were kept in teaching laboratories. An inventory and conservation report were made on the mineral collection in 1989 by an independent curator (Timberlake 1989). Work is continuing on the collection, which is now secure. Approximately half of the original mineral collection comprised specimens from the volcanoes around the Bay of Naples, particularly
Vesuvius and Solfatara (Campi Phlegrei), collected during Johnston-Lavis' employment in Pozzuoli and Naples during the period 18791894. The remainder is a general mineral collection with specimens from world-wide localities and a small, but important, collection from the Classical greek silver mines of Laurion in Attica. The present collection of minerals has been reduced by approximately half as a result of degradation and loss of samples. The collection from Vesuvius and the Bay of Naples contained many rare and interesting specimens of unstable minerals formed as sublimates in association with fumarolic activity. Many of the minerals were unstable hydrous compounds and have fallen into decay over the years, despite their storage in test-tubes and phials. This portion of the mineral collection has suffered the worse degradation, with only c. 300 specimens now accounted for from an original 1644. The 'Greek collection' is reduced to some 150 specimens from an original set of c. 200. These were collected from the spoil tips and galleries of the Laurion mines, an area now seriously overcollected. The specimens are representative of the ores and gangues from the mines (lead,
192
W. L. KIRK, R. SIDDALL & S. STEAD
zinc and copper minerals) as well as specimens of aragonite and smithsonite, for which the mines are justly famous to mineral collectors. The 'general collection' remains relatively intact, over a thousand specimens remaining. Many of these samples, of varied chemistry, derive from classic localities in the UK, continental Europe (mainly France and Italy), and South and North America. Correspondence remains at University College (Manuscripts and Rare Books Room) concerning the mineral 'thermokalite', recorded as a new species of mineral in the duplicate catalogue prepared by Earle. Earle sent some of the material to L. J. Spencer at the Geology Museum for testing. Spencer's preliminary tests indicated it to be a rhodium carbonate. Spencer's advice to Earle was never to throw any of the specimens away 'for you will never be able to delete the name "Thermokalite" from print'. Later correspondence with Bannister suggested that the error was probably a clerical one on Lavis' part, rather than a scientific one. The material was then (9 October 1928) considered to be nahcolite, trona, thermonatrite and thenardite. The collection of rocks has suffered a fate similar to that of the minerals, having been reduced to some 1500 specimens from the
original total of more than 5000. Most of the rock samples were collected by Johnston-Lavis himself, and not surprisingly, the majority of the collection are specimens from Italian volcanoes, particularly Vesuvius and other volcanic edifices surrounding the Bay of Naples: Campi Flegrei (specifically Monte Nuovo and Solfatara), Vesuvius, and the Islands of Ischia and Procida. Material from Vesuvius-Monte Somma is by far in the majority. Other Italian rock suites are derived from the volcanoes of Roccamonfina, Etna, and the Lipari (Aeolian) Islands. Lavis' interest in volcanology went far deeper than the currently active volcanoes of Italy. He also visited and sampled the Chaine-dePuys of the Auvergne, French Massif Central (1895), Iceland (1890), the British Tertiary Volcanic Province (Antrim, Northern Ireland and the Hebridean Isles of Arran and Mull) and the Carboniferous volcanic rocks of the Edinburgh District. During other trips, to Hungary (1913), the Eifel district of Germany and the Rhine Graben (1904), he gathered small suites of volcanic and intrusive rocks. Although dominated by the products of volcanic activity, the collection also contains specimens of fossils (Fig. 3), and sedimentary and crystalline rocks from the Western European Alps, the Scottish Highlands and the South of England. Early in
Fig. 3. Laurel leaves (Laurus cananiensis) preserved in volcanic ash, from Lipari (JLC R4492).
THE JOHNSTON-LAVIS COLLECTION his medical career, in 1883, Johnston-Lavis made a trip to the USA (Woodward 1918) and visited various geological localities in New York State, as rocks within the catalogue testify. It appears that he also acquired a 'readymade' collection from dealers or acquaintances in the USA, as a number of samples derive from widely varying localities in the North American continent. A little correspondence survives at the British Museum of Natural History (BMNH), dating between 1906 and 1908, concerning matters such as the transfer of specimens for analysis or sale. A letter of 11 November 1906 states that Lavis had collected in triplicate, for himself, the British Museum and for the Smithsonian Institution. The BMNH catalogue records 25 separate items, including chlornatrokalite and chlormanganokalite. The photograph collection The photograph collection includes a plate catalogue listing over 2000 items, which seem to be negative numbers, not all of which may have been printed. Corresponding to these numbers are three albums totalling about 300 photographs of early maps and paintings. According to McKay (1970), there were then five albums. The originals date from the second half of the 16th century and include copies from Hamilton's Campi Phlegraei. Observations on the volcanoes of the two Sicilies, with corresponding British Museum numbers that indicate the source for many of the photos. There is also a collection of perhaps 400 loose photographs, often mounted on card, which may well have been used to illustrate his reports to the British Association. These have their own numbering system, for which no catalogue now exists, although they are usually labelled or at least identifiable. They were taken mostly between 1880 and about 1906, and provide records of Lavis' numerous visits to, among others, Vesuvius, Elba, Stromboli, Lipari, Vulcano and Ischia. A number of the Vesuvian photographs, of which there are about 100, show an eruption in progress, and it is interesting that many are taken from the same viewpoint over a period of years. There are also about 50 photographs showing the devastation of Casamicciola after the Ischian earthquakes of 1881 and 1883. The literature collection Many of the works in this collection are extremely scarce and are not held by the British Library. The volcanological literature consists of
193
some 600 volumes (some containing more than one work), plus offprints and periodicals, and some maps. One hundred and twenty-nine volumes are pre-1700. The majority of the books are again concerned with Italian volcanoes and geology, particularly Vesuvius and Etna. The earliest works are by Censorinus, De die natali (1503), Beroaldus, Opusculum de terremotu et pestilentia (1505), and Elisius, De balneis (c. 1510). There are several descriptions of Naples, Pozzuoli, and the surrounding area, the earliest dating from 1538. In December 1631 there occurred the first serious eruption of Vesuvius since AD 79 and no less than 44 books, dated 1632-1635, deal with this eruption alone. Two particularly interesting books are Kircher's Mundus Subterraneus (1665), which deals with earthquakes, volcanoes, and geology in general, and Sir William Hamilton's Campi Phlegraei. Observations on the volcanoes of the two Sicilies (1776-1779), in three volumes with very fine hand-coloured plates. Later works include Charles Babbage's Observations on the temple of Serapis at Pozzuoli (1847), a 1912 Baedeker for southern Italy and Sicily, William Buckland's Geology and mineralogy considered with reference to natural theology (1837), several works on hot springs by Jacques Etienne Chevalley de Rivaz (1834-1859), Nathaniel Crouch's The general history of earthquakes (1694), several works on Vesuvius by Giovanni Maria Delia Torre (1755-1797), Heneage Finch's extremely scarce Relation of the late prodigious earthquake and eruption of Mount ALtna (1669), other works by Sir William Hamilton, several works on Naples and Pozzuoli by Andrea di Jorio (1817-1835), George Poulett Scrope's Considerations on volcanos (1825) and several early publications of the Academia del Scienze of Naples (1738-1788). The future The Johnston-Lavis collection in its entirety, including geological specimens, books, photographs and works of art, is an invaluable record for those researching the geology, geomorphology, history and archaeology of Italy, particularly the Bay of Naples region. As part of the Imperial Roman Province, this region has been a focus of study of multidisciplinary research, with important geological and archaeological sites. The history and prehistory of the region have been strongly influenced, if not dominated by, the presence of Vesuvius and other volcanic structures. Obvious influences are the catastrophic, and therefore justly infamous, destruction of sites (i.e. Pompeii and Herculaneum).
194
W. L. KIRK, R. SIDDALL & S. STEAD
In addition, the volcanically derived material not only provides fertile soils, thus influencing settlement of the region, but also material for building, both as masonry and, because of the unique properties of volcanic ashes, as a hydraulic set in the concretes used by the Romans. The prints, books and photographs are unique records of a landscape continually in a state of change. Lavis' own photographs taken over a period of years show this in a series of localized areas, and the older books and prints show the eruptive products and their geomorphological influences since the 16th century AD. Of interest, too, are the paintings and prints of the Temple of Serapis at Pozzuoli, where the Classical columns show borings of marine organisms, illustrating a concept that was to revolutionize the view of early geologists, particularly Charles Lyell: an appreciation of the rates of change in geological processes, as, in this case, a fall and then rise of ground level had occurred since the construction of the temple, thus within the previous 2000 years. The Johnston-Lavis collection is a unique and scientifically very important record of volcanism for southern Italy. It is still a most valuable and interesting collection, although considerably reduced. Conservation and research into the collection is continuing, and in due course further details of the collection will be published. It is the hope of the authors that the collection will attract renewed interest from current volcanological researchers.
References ANON. 1891. Our bookshelf. Geological map of Monte Somma and Vesuvius. Nature, 44, 271-272. 1914. Henry James Johnston-Lavis, F.G.S., etc. Geological Magazine, 51, 574-576. EARLE, K. W. 1928. Visit to the Johnston-Lavis geophysical collection and the geological collections at University College, London. Proceedings of the Geologists' Association, 39, 96—98 JOHNSTON-LAVIS, H. J. 1876. On the Triassic strata which are exposed in the cliff sections near Sidmouth, and a note on the occurrence of an
ossiferous zone containing bones of a Labyrinthodon. Quarterly Journal of the Geological Society, 32, 274-277. 1880. The earthquake in Ischia. Nature, 23, 497-498. 1881. The late changes in the Vesuvian cone. Nature, 25, 294-295. 1882. Diary of Vesuvius from January 1 to July 16 1882. Nature, 26, 455-456. 1883<3. The Ischian earthquake of July 28th, 1883. Nature, 28, 346-348. 1883/7. The Ischian earthquake of July 28th 1883. Nature, 28, 437-439. 1884. The geology of Monte Somma and Vesuvius, being a study in vulcanology. Quarterly Journal of the Geological Society, 40, 35-119. 1885. Monograph of the Earthquakes of Ischia, a Memoir dealing with the Seismic Disturbances in that Island from Remotest Times, with Special Observations on those of 1881 and 1883, and some calculations by Rev. Prof. Samuel Haughton. F. Furcheim, Naples, Dulau, London. 18910. Geological Map of Vesuvius and Monte Somma, with a short and concise account of the geology and eruptive phenomena to serve as an explanation to the map. Scale 1:10000. Philip and Son, London. 18916. The South Italian Volcanoes, being the account of an Excursion to them made by English and other Geologists in 1889 under the auspices of the Geologists' Association of London and the direction of the author, with papers on the different localities by Messrs. Johnston-Lavis, Platania, Sambon, Zezi and Madam de Antonia Lavis, including the Bibliography of the Volcanic Districts. Furcheim, Naples. 1895. [Tenth Report of the Committee appointed for the Investigation of] The Volcanic Phenomena of Vesuvius and its Neighbourhood. Report of the British Association (1894), 315-318. McKAY, J. 1970. Report on the Johnston-Lavis collection. Unpublished report to the Department of Geology, UCL. TIMBERLAKE, S. 1989. Conservation report on the Johnston-Lavis mineral collection within the Department of Geology, University College London. Internal Report to the Department of Geology, UCL. WOODWARD, B. B. 1918. A short sketch of the life of Henry James Johnston-Lavis. Second edition of Bibliography of the Volcanoes of Southern Italy, by Johnston-Lavis, compiled after the author's death by Miss B. M. Stanton. University of London Press.
The archaeology of a Plinian eruption of the Popocatepetl volcano PATRICIA PLUNKET & GABRIELA URUNUELA Department of Anthropology, Universidad de las Americas-Puebla, 72820 Cholula, Puebla, Mexico Abstract: The northeastern flank of the Popocatepetl volcano in western Puebla, Mexico, has been subject to intense and destructive volcanic activity since the Terminal Preclassic period (100BC-AD 100), and it is still one of the highest-risk sectors in the region. Between AD 50 and 100 the communities that dotted this slope, an area known locally as Tetimpa, were abruptly buried by a pumice-fall deposit that preserved the buildings, activity areas and agricultural fields but devastated the settlements and made the region uninhabitable for generations to come. It is probable that this violent eruption had an important social and ideological impact on the emerging urban centre at Cholula 15 km to the east. The Preclassic occupation of the region can be divided into the Early (700-200 BQ and Late (50 BC-AD 100) Tetimpa phases. These appear to be separated by a short period of abandonment that might reflect an intensification of volcanic activity. This paper deals primarily with the Late Tetimpa phase, the occupation that was destroyed by a Plinian eruption in the second half of the first century. Excavations by the Universidad de las Americas-Puebla have uncovered agricultural fields with hand-made furrows interspersed with households consisting of two or three rooms on low stone platforms around an open patio. At the centre of the patio is a small shrine or altar; the themes of these shrines vary but they include volcano effigies. Architecture is simple but household patterning reflects strict adherence to rules for generating the domestic environment. Houses show little or no modification or renovation, suggesting a very short occupation. This is further supported by the fact that no burials have been located in the nine domestic areas explored to date. The evacuation of the settlement appears to have been abrupt, as many household goods, particularly large or heavy items, were left behind. We have not found any human remains under the collapsed roofs of the houses, and it would seem that the majority of the population took refuge at nearby towns and villages on the valley floor.
Mountains were held to be sacred in the ancient cosmic vision of central Mexico. They were used as territorial markers, as fixed points to measure the movements of the Sun and the stars, and they even served as pictographs to indicate place names in the native manuscripts. Fray Bernardino de Sahagun (1981, pp. 71-72), a Franciscan who documented many elements of the Prehispanic cultures, tells us that: '[in] the thirteenth month ... named Tepeilhuitl ... they celebrated a feast in honor of the high mountains... where the large clouds pile up. They made images of each one of them in human form, from [amaranth dough] and they laid offerings before these images in veneration of these same mountains.' As dramatic and imposing features of the natural landscape, these were sites where important sacred and ceremonial events took place, The Great Temple of the Aztecs was built as a dual symbolic reference to two mythical mountains: the mountain of sustenance presided over by Tlaloc, the patron of rain and storms, and the
hill of Coatepec where the tribal god Huitzilopochtli had been born as an adult warrior ready for battle (Matos 1982). The New Fire Ceremony, used to initiate the incoming 52-year cycle, was celebrated on top of a hill in the southern Basin of Mexico (Broda 1991, p. 473), and the temple-platforms of the Maya area were conceived of as sacred mountains with caves that led to the underworld (Schele & Freidel 1990, pp. 71-72). Volcanoes are very special kinds of mountains, however, and their eruptions figure importantly in the creation myths recorded for highland Mexico. The Codex Chimalpopoca relates the story of the five mythical epochs of creation and destruction. The first creation, or sun, was destroyed by water and its people were transformed into fish; the second sun was inhabited by giants who were eaten by jaguars when the heavens collapsed; the third sun, ruled over by the rain god, was destroyed by a rain of fire. The author's description (Sullivan & Knab 1994,
From: McGuiRE, W. G., GRIFFITHS, D. R., HANCOCK, P. L. & STEWART, I. S. (eds) The Archaeology of Geological Catastrophes. Geological Society, London, Special Publications, 171, 195-203. 1-86239-062-2/00/ $15.00 © The Geological Society of London 2000.
196
P. PLUNKET & G. URUNUELA
p. 66) makes it clear that this 'rain of fire' refers to a volcanic eruption that must have been witnessed long ago: 'The third Sun was established. 4-Rain was its sign; it was called the Sun of Rain. In this sun it occurred that it rained fire and the people were consumed by fire . . . It rained stones. They now say that this was when the stones we now see fell, and the lava rock boiled up. And also, it was when the great rocks formed into masses, and became red.'
Popocatepetl's northeastern flank The Popocatepetl volcano in central Mexico (Fig. 1) is one of the largest and most impressive active volcanoes in the world, and perhaps the eruption recorded in the Legend of the Suns refers to one of the two sets of volcanic events we have documented on the northeastern flank of the mountain in a region known as Tetimpa. Because of its great height, over 5400m, the
mountain's explosive activity is visible from many parts of the central highlands. Tetimpa is a particularly apt designation for this area, because in the native Nahuatl language it can mean 'filled up with rock' or it may refer to the act of casting down. In either case, the name is appropriate because the two major eruptive sequences recorded here left the rolling hills filled up with rock that was cast down from the crater of the Popocatepetl volcano. The first eruptive sequence took place during the later part of the first century AD, at a time when Teotihuacan and Cholula were emerging as the major cities of the central plateau. This was a Plinian style event with an eruptive column that rose at least 25 km before it collapsed over the northeastern flank of the volcano, burying the towns and villages that dotted the slopes with up to 1m of pumice lapilli (Panfil 1996, p. 16). On a clear day this eruption would have been visible from all of the surrounding valleys, and we have suggested elsewhere (Plunket & Urunuela 19980) that it probably had a significant impact on the ideologies of the peoples of Puebla, Tlaxcala, Morelos and the Valley of Mexico. The consequences for the nearby civicceremonial centre of Cholula must have been both more immediate and more substantial, as
Fig. 1. Map of Mexico locating the state of Puebla and the Popocatepetl volcano.
Fig. 2. Isopach map of the western Puebla-Tlaxcala Valley showing the regional distribution of the volcanic deposits from the first Plinian event (after Panfil 1996, p. 18).
198
P. PLUNKET & G. URUNUELA
the lateral extension of the pumice can be traced to within 5km of the city (Fig. 2). Not only would Cholula have suffered the shock of the explosion, but the loss of arable land in the surrounding countryside and the social and economic demands of the refugees would have further exacerbated an already stressful situation. Shortly after the Plinian phase of the eruption, a massive lava flow covered both the ravaged remains of the villages and the pumice with between 30 and 100m of solid rock along the southwestern edge of the pumice-fall deposit (Panfil 1996, p. 16). It took several hundred years before any recolonization of the Tetimpa region was attempted, and for the most part this appears to be correlated with the rate of soil formation. A new soil began to develop on the surface of the pumice, but it seems to have taken about 400 years before this was sufficient to allow any reasonable agricultural activity. Organic accumulations and a sandy matrix define a weak A soil horizon in the upper 15-40 cm of the pumice (Panfil 1996, p. 21), and this appears to have been sufficient to permit the re-establishment of farming villages in the area. The density of these new settlements, however, was much less than it had been before the eruption, and although the slopes of the volcano once again became an attractive ecological niche, we suspect that the region was less productive than before. At some point between AD 700 and 850, a second sequence of volcanic events, including pyroclastic surges and extensive lahars, ended with the deposition of more lapilli. Covering this new pumice is the modern topsoil composed of sand, silt and organic materials; there is no evidence for major volcanic events on the northeastern flank of the volcano after AD 850 (Panfil et al 1996; Siebe et al 1996) (Fig. 3). In the 1960s local peasants began to mine the pumice deposits in the Tetimpa region, and in the process they uncovered and destroyed adobe and stone structures, ceramics and furrowed agricultural fields that were briefly described by researchers from the Fundacion Alemana para la Investigation Cientifica (Seele 1973). This mining activity has continued over recent decades, and although much has been lost to the extractive machinery there are still many intact fields that encase and protect the cultural remains. Experimental work with both ground-penetrating radar and d.c. resistivity has demonstrated these to be worthwhile research techniques that provide extremely clear imaging of the structures and furrowed fields beneath the ash (Fink 1994), but the pace of mining has accelerated in the last few years so that our exploration is dictated
primarily by the ability of the front-end loader drivers to negotiate deals with individual landowners for their pumice. The Late Tetimpa phase Although we have surveyed several square kilometres of mined fields and excavated houses from both the Early and Late Tetimpa phases in addition to the Classic period Nealtican phase, we will limit our discussion to the Late Tetimpa phase, the occupation that was buried under the pumice from the first Plinian event. Our rescue work has revealed a dispersed village with highly standardized house compounds separated from each other by small farm plots; the distance between houses is usually between 50 and 90m (Fig. 4). Each compound consists of two or three stone platforms placed at right angles around a courtyard. The platforms have a central staircase and use a decorative, and perhaps functional, architectural system, a sloping wall crowned with a horizontal panel (known as 'talud-tablero'), that is considered diagnostic of Classic period Teotihuacan (AD 200-650) (Fig. 5). The facades of some buildings were embellished with figures modelled in daub and painted in red and yellow. The rooms that surmounted the platforms were simple wattle-and-daub structures whose roofs collapsed under the weight of the pumice, pulling the rather insubstantial walls down at the same time; consequently, ceramic vessels that were left behind inside the rooms lay smashed on the floors. Fragments of burned daub from the room walls litter the surface of the mined fields, and on one hillside in particular our excavations have uncovered extensive evidence of fire at the three domestic units explored. The fact that some areas of the settlement burned whereas others did not is puzzling. We do not know whether the falling pumice itself was hot enough to ignite the building materials. The variability in the burning may depend on the trajectory of the volcanic blast and minor variations in the landscape; for example, the one house where the lower 30 cm of the wattle-and-daub walls were still intact was situated in a topographic depression that would have protected it from the force of the blast, although not from the accumulating pumice. At some houses the roofs may have collapsed on top of braziers or hearths filled with live coals that people neglected to extinguish. On the hillside known as Cruz Verde, where we have recently explored several burned structures, the fires appear to be localized: one corner of a building is highly fired whereas the
PLINIAN ERUPTION OF THE POPOCATEPETL VOLCANO
Fig. 3. Calibrated radiocarbon dates shown at the 2<j range for the Tetimpa region.
199
200
P. PLUNKET & G. URUNUELA
Fig. 4. Spatial distribution of house compounds and ancient agricultural fields in a mined field in the Tetimpa region.
others are not, one storage bin is burned whereas the one next to it is not. In general, there is little evidence of charred beams or thatch on the floors of the rooms; most of the remains consist of highly fired daub from the base of the room walls. This daub provides important information about the poles and lashing materials used in household construction. The lapilli that filled the rooms, storage pits, grain silos and ceramic vessels during the eruption generally have a smaller clast size and are more rounded than the material found in open spaces; in addition, the lapilli tend to be grey or white as opposed to the more oxidized pigmentation of the larger clasts in the exterior spaces. This occurs, for example, underneath collapsed columns on the porches of the buildings and inside water jars and other closed-mouthed
Fig. 5. House platform with collapsed room from the Late Tetimpa phase.
vessels. The whiter coloration may result from the reduced amount of oxygen in the enclosed spaces, but in any event the phenomenon is a useful indicator of material that fell into an enclosed space that has since disappeared. It is a particularly useful indicator of the presence of a particular type of storage facility that does not survive the eruption intact. These are small mud-and-thatch silo-like structures called cuexcomates, still used in the more remote villages around the Popocatepetl volcano. Their rounded bodies are made of wicker plastered with daub, much like a swallow's nest, and they are raised off the ground on four to six large stones. Neck fragments from ceramic jars are set into the side walls to provide ventilation holes, and the structure is finished with a thatch roof. During the 1996 season we excavated a cuexcomate that had been burned, thus preserving the structural details in the fired daub, but generally we find only the remains of collapsed bins: quadrangular patterns of large stones covered by small grey-white pumice and topped with a layer of crumbled yellow dirt. There are few organic remains inside the cuexcomates, but then there are almost no organic remains to be found anywhere in these households save charcoal and fibres of grass thatch from burned structures. The lack of organic materials is curious, as the surrounding soil and courtyard floors have an almost neutral pH (between 6.8 and 7.2). The daub surface is sometimes intact on both the interior and exterior walls, but there is merely an empty space where the cane should be; hardened mud casts of the corner posts occur on the floors
PLINIAN ERUPTION OF THE POPOCATEPETL VOLCANO
201
of some rooms, but the posts themselves have disappeared; there are no baskets, no ropes, no cloth, no wooden tools or furniture, no bone needles or awls, and no plants or seeds. The almost non-existent midden areas have yielded nothing more than fragments of charcoal. The central courtyards of the houses tend to be clean, as the work areas and storage spaces are located in the corners formed by the platforms and along the perimeter of the compound, close to the entrance. At the centre of each courtyard is a small household shrine. We have recovered shrines from seven courtyards and we have yet to find any exact duplications, although there are repetitive elements; a good deal of individual freedom of expression is evident. Most of the shrines consist of one or two carved stones (although uncarved stones are also used), including felines, snakes, and anthropomorphic heads. In yet another case, the shrine was a low rectangular altar that had a small, plain vertical flagstone set into one extreme; on top were stone recipients covered with fine grey ash from the burnt offerings. The most striking shrines, considering the ultimate fate of Tetimpa, were the effigies of the Popocatepetl volcano itself (Plunket & Urunuela 1998a). The first of these consisted of two tiny
stone-and-daub platforms, each with an interior chimney, leading to a chamber filled with ash and charcoal, and crowned by a carved stone figure (Fig. 6). Smoke would have puffed out behind each stone in imitation of the ash and vapour plumes that are expelled from the volcano's crater during periods of activity. The second volcano shrine was more elaborate, consisting of a low rectangular platform plastered with polished daub (Fig. 7). Set into the western end of the platform was a nicely carved stone head, and on top of the platform was a circular stone, crudely pecked to represent a human face, that rested on top of a small orifice with remnants of ash embedded in the polished surface around it. At the eastern end of the platform, worshippers had built a conical 'volcano' and set a stone serpent-effigy on top of it. Underneath this figure was a chimney formed from a jar neck that led to a specially made ceramic chamber, again filled with ash. The image created by these volcano shrines suggests that the people of Tetimpa witnessed many pulses of vapour and ash before their settlements were actually buried by the collapse of the Plinian column. But at what time of year did this eruption occur? This is a difficult question to answer, as plant remains that could provide data on the
Fig. 6. Volcano shrine uncovered at the centre of a residential courtyard at Tetimpa in 1994.
Fig. 7. Volcano shrine uncovered at the centre of a residential courtyard at Tetimpa in 1996.
202
P. PLUNKET & G. URUNUELA
season (Sheets 1992) are absent. In spite of this, however, there are several lines of evidence that bear upon this issue. One of the most notable aspects of the pre-eruptive landscape is the furrowed agricultural fields (Fig. 4). These furrows run across the slope, helping to control erosion, much as contour farming does today; and we assume that, like the furrows of modern fields, they also provided a weed-free environment and protection from wind and frost. We have inspected the surfaces of the mounds and troughs for indications of planting depressions, the seeds themselves or the holes left by the stalks of more mature plants, but we have yet to find unequivocal evidence that the fields were planted at the time of the eruption. The prevailing winds around the Popocatepetl volcano blow east-to-west during the rainy season (May-October) and west-to-east in the dry season (November-April) (Delgado et al. 1995), and this suggests that the most likely time for a pumice-fall deposit from the collapse of a Plinian eruptive column on the northeastern flank of the mountain would have been between the months of November and April, a time when the fields lie fallow. Another line of reasoning comes from the houses themselves. Hearths occur both inside the rooms and at the base of the house platforms along the edge of the courtyards. In the wintertime, or in those months when strong winds sweep through the region, or during the rainy season, cooking would be easier in a covered and enclosed space; however, at other times of the year the possibility of cooking outside would provide the advantage of not having to sit in a small, dark, smoky room, and at the same time would allow the household members to socialize, something that would not be possible in the tiny kitchens filled with jars and bowls. In the excavated compounds, we find that the hearths inside the kitchens are clean, with few traces of charcoal or ash, whereas the outdoor hearths provide solid evidence that they were in use. The existence of this seasonal cooking pattern coincides well with the wind studies that postulate a November-April placement for the eruption, and we would suggest that the most likely time is late March or April, as both winter and the strong winds have passed, the fields are not yet planted and the rains have yet to begin (Urunuela & Plunket 1998).
water jars, amphorae and grinding stones were left leaning against the house platforms or in the activity areas around the edge of the courtyard; the stone idols were forsaken on their altars. There are few small artefacts in the house compounds: a few small ceramic bowls left inside the rooms or under the storage silos; an occasional axe or projectile point; discarded obsidian blades outside the kitchens; and small braziers used to incense the gods. Items made from organic materials (straw mats, baskets, wooden bowls, leather bags, nets and fabric) would not have survived, and in any event, the families probably took as much as they could carry when they fled their homes. It seems logical that their baggage would consist of lightweight essentials, primarily clothing and small tools, that could be carried in baskets or net bags. Most of the population probably escaped the devastation. We have found no evidence of bodies inside the collapsed rooms where people might have sought refuge from the barrage of pumice; in fact, the only human remains located to date is a skull offering placed underneath one of the altars. Although plant material did not survive, a turtle-shell drum and a deer antler from one house provide evidence that bone and teeth would have been preserved above ground, albeit in poor condition, and we may yet find the remains of some stubborn individuals who decided to remain in their homes. The Tetimpenos probably fled to the relative safety of the communities on the valley floor that witnessed the dramatic event. It is difficult to estimate the number of refugees involved in this disaster, as many of the remains that could be used to calculate population are buried under the pumice or have been destroyed by the mining activity, but our survey work suggests that the number would have been between two and three thousand (Plunket & Urunuela 19986). Many of
The abandonment of Tetimpa The abandonment of Tetimpa appears to have been abrupt. Large and heavy items such as
Fig. 8. View of house compound from Late Tetimpa phase.
PLINIAN ERUPTION OF THE POPOCATEPETL VOLCANO them may have settled at the emerging urban centre of Cholula, although it will be almost impossible to assess the impact of this population influx, as Cholula's Prehispanic history is sealed beneath the 2000 years of continuous human occupation. The refugees did not return to salvage their household goods, as most of the buildings collapsed and were completely buried by the volcanic deposit. The landscape was left barren, covered with too much pumice to be of any use, and the volcano may still have provided reason to stay away. The timing of the reoccupation obviously has to do with the ecological recovery of the countryside, with the development of sufficient soil to make farming an economically viable enterprise, but we would not discount the possibility that a certain number of generations were required before the dramatic events of the first century could recede into the mythic past and human populations once again believed 'it will never happen to us'. The house compounds at Tetimpa were abandoned quickly and they provide an almost ethnographic vignette of Prehispanic life at one specific moment in time (Fig. 8). The scenes we have recovered, however, are static, whereas daily life is dynamic, and although it follows certain patterns it is also fraught with individual decisions and aberrant behaviour. The variability inherent in these residential compounds provides us with an opportunity to identify different moments in the domestic cycle as households move through time, to explore the spatial distribution of activities, to study the significance of differences in house compound size, to test demographic models and assumptions, to look at household ritual, and to investigate the rules used to generate domestic space. By documenting a large sample of houses at Tetimpa we hope eventually to distinguish between the idiosyncratic behaviour imprinted at each household and the broader social and economic patterns of village life in ancient Mexico. We would like to thank the Mesoamerican Research Foundation for its generous and unflagging support of the Tetimpa Project. The Institute de Investigation y Posgrado of the Universidad de las Americas-Puebla provided us with the infrastructure and scholarships. None of the research would have been possible without the dedication of our workmen from San Nicolas de los Ranches, and the foresight of Paulino Luna of San Buenaventura Nealtican.
203
References BRODA, J. 1991. Cosmovision y observation de la naturaleza: el ejemplo del culto de los cerros en Mesoamerica. In: BRODA, J., IWANISZEWSKI, S. & MAUPOME, L. (eds) Arqueoastronomia y Etnoastronomia en Mesoamerica. Institute de Investigaciones Historicas, Universidad Nacional Autonoma de Mexico, Mexico City, 461-500. DELGADO, H., CARRASCO, G., CERVANTES, P., CORTES, R. & MOLINERO, R. 1995. Patrones de viento en la region del Volcan Popocatepetl y Ciudad de Mexico. In: Volcan Popocatepetl, Estudios Realizados Durante la Crisis de 1994-1995. CENAPRED-UNAM, Mexico City, 295-324 FINK, J. 1994. Report on geophysical equipment evaluation at the Tetimpa Project in the State ofPuebla, Mexico. Manuscript on file, Tetimpa Project, Universidad de las Americas-Puebla, Cholula, Puebla. MATOS, E. 1982. El Templo Mayor: economia e ideologia. In: MATOS, E. (ed.) El Templo Mayor: Excavaciones y Estudios. Institute Nacional de Antropologia e Historia, Mexico City, 109-118. PANFIL, M. 1996. The Late Holocene volcanic stratigraphy of the Tetimpa area, northeast flank of Popocatepetl volcano, central Mexico. MS thesis, Department of Geosciences, Pennsylvania State University, College Park. PLUNKET, P. & URUNUELA, G. 19980. Appeasing the Volcano Gods. Archaeology, 51(4), 36-42. & 1998/7. Preclassic household patterns preserved under volcanic ash at Tetimpa, Puebla, Mexico. Latin American Antiquity, 9(2), 287-309. SAHAGUN, F. B. DE 1981. Florentine Codex: General History of the Things of New Spain. Book 2: The Ceremonies. Translated from the Aztec into English with notes and illustrations by A. O. Anderson and C. Dibble. Monographs of the School of American Research, Santa Fe, NM. SCHELE, L. & FREIDEL, D. 1990. A Forest of Kings: the Untold Story of the Ancient Maya. Quill, William Morrow, New York. SEELE, E. 1973. Restos de milpas y poblaciones prehispanicas cerca de San Buenaventura Nealtican, Puebla. Comunicaciones, 7, 77-86. SHEETS, P. 1992. The Ceren Site: a prehistoric village buried by volcanic ash in Central America. In: QUILTER, J. (ed.) Case Studies in Archaeology Series. Harcourt Brace, Fort Worth, TX. SIEBE, C., ABRAMS, M., MACIAS, J. & OBENHOLZNER, J. 1996. Repeated volcanic disasters in Prehispanic time at Popocatepetl, central Mexico: past key to the future? Geology, 24(5), 11-37. SULLIVAN, T. & KNAB, T. 1994. A Scattering of Jades. Simon and Schuster, New York. URUNUELA, G. & PLUNKET, P. 1998. Areas de actividad en unidades domesticas del Formativo terminal en Tetimpa, Puebla. Arqueologia, 20, 3-19.
This page intentionally left blank
Timing of the prehistoric eruption of Xitle Volcano and the abandonment of Cuicuilco Pyramid, Southern Basin of Mexico SILVIA GONZALEZ1, ALEJANDRO PASTRANA2, GLAUS SIEBE3 & GEOFF DULLER4 School of Biological and Earth Sciences, Liverpool John Moores University, Byrom Street, Liverpool L3 3AF, UK (e-mail:
[email protected]) DICPA, Instituto Nacional de Antrapologia e Historia, Seminario 8 Centro Historico, Mexico City, C.P. 06060, Mexico ^Institute de Geofisica, Universidad Nacional Autonoma de Mexico, Coyoacdn, Mexico City, C.P. 04510, Mexico 4 Aberystwyth Luminescence Laboratory, University of Wales, Aberystwyth, Dyfed, SY23 3DB, UK Abstract: The Cuicuilco pyramid was one of the first true urban centres in the Basin of Mexico. Its construction started a few centuries BC, during the Late Preclassic period. The pyramid is partially covered by a basaltic lava flow produced by the Xitle monogenetic volcano. New stratigraphic work around the pyramid and the volcano together with new radiocarbon dates indicate that the pyramid and nearby settlements were abandoned as a direct consequence of the volcanic activity of Xitle. The new dates, obtained from material which clearly is contemporaneous with the volcanic activity, suggest that the eruption took place around 1670 years BP, some 300 years later than previously thought.
The Cuicuilco circular pyramid was the most important ceremonial centre in the Basin of Mexico around 2000 years ago (Heizer & Bennyhoff 1958, 1972). Situated close to a freshwater lake (Xochimilco Lake), it represents one of the largest and oldest Preclassic centres in Mesoamerica. It should be evaluated in terms of the beginnings of urbanization and the origin of the city-state in the Basin of Mexico (Pina-Chan 1967; Parsons 1989). Cuicuilco was suddenly destroyed by the appearance and volcanic activity of Xitle volcano, 7 km SW of the pyramid (Figs 1 and 2a). This small volcano is one of more than 200 scoria and cinder cones that constitute the Quaternary Chichinautzin monogenetic volcanic field, to the south of the Basin of Mexico. Products of the cones are andesitic to basaltic in composition and mostly calcalkaline in character (Gonzalez-Huesca, 1992). The eruption of Xitle volcano started with the emission of a basaltic tephra, which covered the Cuicuilco circular pyramid with a 10-15 cm thick layer of ash. Initial ash fall was followed by several basaltic lava flows, which flowed downslope in a northerly direction, covering an area of c. 80km2. The pyramid was surrounded
by one of the lava flows with a thickness of 4-6 m (Fig. 2b). The pyramid was rediscovered in 1922 (Cummings 1933). Excavations started with the aid of explosives, in order to break the basaltic layer and uncover the structure. Further investigations showed that the pyramid consists of at least three superimposed structures, indicating occupation of the site for several hundreds of years. Additional excavations have shown that the Cuicuilco pyramid was in the centre of a large settlement, with several minor pyramids, mounds and habitation structures, pointing to the existence of a well-organized culture and the beginnings of the city-state in Mesoamerica. New excavations were started in summer 1996 by the Mexican National Institute of Anthropology and History (INAH) (Perez-Campa et al. 1995). Several trenches around the main structure were dug with the purpose of clarifying the relationships between the volcanic stratigraphy and the associated archaeological horizons, so as to understand how the population reacted to the activity of Xitle. Because of the volcanic activity the pyramid is well preserved. One of the important questions is: was the pyramid in use at the moment of the
From: McGuiRE, W. G., GRIFFITHS, D. R., HANCOCK, P. L. & STEWART, I. S. (eds) The Archaeology of Geological Catastrophes. Geological Society, London, Special Publications, 171, 205-224. 1-86239-062-2/00/ $15.00 © The Geological Society of London 2000.
Fig. 1. Location of the Cuicuilco pyramid (-jfc-) and Xitle volcano (A) in the southern part of the Basin of Mexico. Xitle volcano is a scoria cone that produced lavas that flowed towards the NE into the lacustrine Basin of Mexico, where human settlements were buried under the lava. Other important archaeological sites of the Preclassic period within the Basin are also indicated, together with the lakes in existence at that time.
THE CUICUILO PYRAMID
207
Fig. 2. (a) The circular pyramid of Cuicuilco was surrounded and partially buried by lava flows originated by Xitle. Today, this archaeological site lies within the urban area of Mexico City. Photograph taken April 1997. (b) Trench excavated with the help of explosives at Cuicuilco several decades ago by Cummings. The people are standing at point Bl where the lava is in direct contact with archaeological remains. Photograph taken in April 1996.
208
S. GONZALEZ ET AL.
eruption or was it already abandoned long before? Here, we present stratigraphic evidence from the new excavations, which indicates that the pyramid was still in use when the eruption started. For example, the sections excavated in the eastern ramp (see Fig. 13, below) show that it was mainly built of clay, which was intensively baked by the heat of the lava flow. The ramp presents a clean, flat surface, without evidence of charred vegetation. In addition we present stratigraphic sections located at 1.5km to the SW and 12km to the
NE of the crater, which include archaeological soils and the lava and tephra produced by the Xitle volcano. Composition, grain-size distribution and magnetic susceptibility of the various layers were determined and new radiocarbon dates from charcoal within the tephra layer are presented. The new data reported in this paper strongly suggest that the abandonment of Cuicuilco was the direct result of the birth and activity of Xitle. This volcano is one of the youngest monogenetic cones in the vicinity of the Basin
Fig. 3. Xitle pillow lavas, exposed during the construction of the Inbursa Tower in May-July 1997. The construction site is 365 m south of the Cuicuilco pyramid. Person is standing in a trench exposing the lacustrine sediments of a formerly existing lake.
THE CUICUILO PYRAMID of Mexico. The burial of Cuicuilco by young lava flows and ash has important implications with regard to volcanic hazards that could affect Mexico City, with a population of more than 20 million inhabitants: volcanic hazards that could affect Mexico City are not restricted to the large composite volcanoes such as Popocatepetl, but also include the activity of relatively small monogenetic volcanoes. Xitle volcano and tephra Xitle volcano is located on the lower slopes of Ajusco stratovolcano (3950 m) near the southern edge of Mexico City (Fig. 1). This area is part of the Chichinautzin Volcanic Field. Lavas erupted from Xitle produced the Pedregal de San Angel, a rocky terrain on which the southern suburbs of Mexico City are constructed (Canon-Tapia et aL 1995). Xitle may be described as a small cinder cone with a diameter of c. 500 m and a height of 140m above the surrounding ground. The Xitle lava around the Cuicuilco pyramid and the National University of Mexico (UNAM) comprises a complex series of pahoehoe flows
209
with flow units ranging in thickness from 0.2 to 12.0m. The lavas display a young morphology with well-preserved flow structures and little or no vegetation cover. Vesicles are very common, particularly towards the top of the flow units. Within the flow, structures such as explosion tubes, lava channels, pressure crests and tumuli are visible (Martin del Pozzo et al. 19976). Emplacement of the lava was controlled by topography: the Xitle cinder and scoria cone grew on the slope of the extinct Ajusco volcano and lava flowed down the slope to the N-NE until it reached the floor of the Basin of Mexico (Fig. 1), where it spread out over the settlements in the Cuicuilco and Copilco areas. During excavations in May-July 1997 for the construction of the Inbursa Tower located 365 m south of the Cuicuilco pyramid, pillow lavas were exposed on top of lake sediments, indicating that Xitle flows had reached the shore of a former lake (Fig. 3). The composition of the lava flows is basaltic and the lavas can be described as olivine basalts (Schmitter 1953; Badilla-Cruz 1977). Major element geochemical analysis of a lava sample near the Olympic Stadium (S-9), (Gonzalez-Huesca
Fig. 4. Plot of K2O v. SiO2 (Peccerillo & Taylor 1976), indicating that the composition of the Xitle lava (A) and tephras (•) are of basaltic composition with a calcalkaline trend. The lava sample S-9 was taken close to the Olympic Stadium on the University campus, (a) Tephra close to the crater; (b) tephra at Cuicuilco pyramid; (c) tephra at the Centre de Estudios para Extranjeros (CEPE-UNAM).
S. GONZALEZ ET AL.
210
Table 1. Major element analyses of Xitle lava (sample S-9) and tephras (samples A B and C) Locality
Si02
A1203
Fe203
MnO
MgO
CaO
Na20
K20
TiO2
P205
Total
S-9 A B C
51.03 52.89 52.70 49.24
16.17 14.32 16.13 13.19
8.55 8.49 8.63 11.06
0.15 0.13 0.13 0.16
7.41 10.42 6.87 12.90
7.67 7.04 8.30 6.10
3.70 3.35 3.76 3.08
1.13 1.08 1.20 1.04
2.02 0.96 1.25 1.40
0.51 0.26 0.33 0.53
98.34 98.64 98.89 98.74
Fig. 5. Map showing the distribution of the Xitle lavas, as well as the location of the three studied sections. Locality A, 1.5 km from the crater; locality B, Cuicuilco pyramid; locality C, CEPE-UNAM.
THE CUICUILO PYRAMID
211
Fig. 6. Histograms showing the grain-size distribution of tephra samples from the three studied localities.
1992) indicates that the Xitle lavas are calcalkaline in character (Fig. 4 and Table 1). Birth of the volcano started with the emplacement of a widespread sandy to silty, grey ash-fall deposit, distributed in a NE direction from the vent. This ash blanketed the original land surface as well as the pyramid. Chemical analysis of the tephra at different distances of the crater (Localities A B and C) indicates a basaltic composition and a calcalkaline trend (Fig. 4 and Table 1). Grain-size analyses were performed on the same tephra samples. These analyses document a decrease in grain size from sand to silt with distance from the source. At locality B a marked depletion in the coarse fraction can be interpreted as reworking of the tephra (Fig. 6). The volcanic activity of Xitle led to the abandonment of Cuicuilco and other nearby human settlements. However, the volcanic products (tephra and lava flow) protected the archaeological evidence from later destruction and looting, and provide a unique opportunity for obtaining a more complete picture of human life in the Basin of Mexico more than 1600 years ago. Cuicuilco pyramid archaeological sequence The chronology of cultural evolution in Prehispanic Mexico has been divided into four main periods. According to Coe (1994), the archaeological time scale can be divided as follows: Period Preclassic (200 BC-AD 250)
Important sites
Cuicuilco and Chalcatzingo Teotihuacan Classic (AD 250-750) Epiclassic (AD 750-1 150) Tula, Cholula and Xochicalco Postclassic (AD 1150-1521) Tenochtitlan-Tlatelolco, Texcoco and Tacuba
Cuicuilco is one of the oldest and most important ceremonial centres of the Preclassic period, situated on the shores of the Xochimilco freshwater lake (Fig. 1). It represents the initial phase of urbanism and the city-state complex in the Basin of Mexico. The pyramidal structures of Cuicuilco, like the main conical pyramid (named Temple 1 of Cuicuilco A), were built consecutively with stratified infills, which were covered with blocks or boulders of basaltic or andesitic lavas from lava flows of older adjacent volcanoes. Between the archaeological materials recovered from the excavations below the Xitle lava, the following elements were found: floors, infills, ramps, walls, drainage channels and a megalithic sculpture with remains of pigments. Other important cultural elements found include burials with different ritual practices, in groups of both sexes and ages, usually with offerings of pottery and ornamental objects made of jade, obsidian, basalt, flint, onyx and bone. The objects are evidence of contact and exchange with different regions and cultures of Mesoamerica. The earliest occupation in the Cuicuilco area dates back to 2100 BC, when people lived in small villages with incipient agriculture. Between 800 and 600 BC the main structure of the pyramid was constructed, in the form of a truncated cone (Heizer & Bennyof 1972); it is at this time that Cuicuilco became a city. The time of the abandonment of Cuicuilco has been a point of discussion, with two main hypotheses being proposed: (1) Cuicuilco was abandoned before the eruption of Xitle, as a consequence of the rise of the city of Teotihuacan in the NE part of the Basin of Mexico; (2) Cuicuilco was abandoned because of the volcanic activity of Xitle. The first hypothesis (Lopez-Camacho 1991; Cordoba et al. 1994) is supported by evidence found directly at the mounds and buildings of Cuicuilco B in the Olympic Villa area. Apparently, these mounds had been abandoned and undergone decay and erosion for some time
212
S. GONZALEZ ET AL.
before the eruption of Xitle. Lopez-Camacho and Cordoba et al. interpreted two 'bodies of debris' exposed during the first excavations of 1922-1924, at the southern base of the main pyramid in Cuicuilco, as evidence for the abandonment and decay of the structure before Xitle's eruption. We believe, however, that the second hypothesis is the correct one, and support the idea of the decline of Cuicuilco as a direct result of the eruption. Stratigraphic information coming from the new trenches excavated around Cuicuilco pyramid and from other outcrops strongly supports this version, and is presented in the next section. Stratigraphy and archaeology of the Xitle eruption To better understand the relationships between the volcanic stratigraphy and the archaeology of the area around Cuicuilco we studied in detail
several natural outcrops at different distances from the Xitle crater (Fig. 5). At the same time, and taking advantage of the excavations made by INAH, we studied new sections around the main conical pyramid (Fig. 7). We were particularly interested in the contacts between archaeological structures, and covering ash fall and lava flows. Location A is a geological outcrop located 1.5km SW of Xitle's crater (Fig. 8). At the base of the section are 120cm of a dark grey, sandy, well-bedded air-fall ash (Xitle tephra). The ash is bedded in layers that are 2-5 cm thick. A welldefined erosional unconformity indicates temporary cessation of the ash fall during the early stages of the eruption. A sample of the ash was taken 30 cm below the contact with the lava flow for chemical and granulometric analysis. These analyses revealed a basaltic composition and a dominance of sand-sized fraction (Figs 4 and 6). Charcoal is abundant within the ash layers and was radiocarbon dated at 1665 ± 65 years BP (see also Table 2). The ash is topped by a thin
Fig. 7. Plan of the circular pyramid of Cuicuilco, after Haury (1925; in Schavelzon 1993). The position of the five trenches excavated by INAH during the 1996 season is also shown.
Fig. 8. Stratigraphy at locality A 1.5km south of Xitle volcano, showing the Xitle tephra and lava flow. Charcoal samples within the ash-fall sequence were dated at 1665 ±65 years BP in this study.
S. GONZALEZ ET AL.
214
Table 2. New radiocarbon dates for Xitle's volcanic activity Locality
14
C age (years BP)
Calibrated results (2o)
^13C
A
A9587
1665 ±65
250-548 cal AD
-23.6
Bl
A8985
1675 ±40
252-296 cal AD 3 18-450 cal AD 482-500 cal AD 5 14-530 ca AD
-23.3
B2
A9586
1995 ±60
380calBC-350calAD 360-380 cal AD
-25.0
C
A9585
2295 ±115
760-620 cal BC 600-60 cal BC
-25.0
Locality
scoriaceous lava flow (35cm) from Xitle. This is followed by 220 cm of a vesicular lava flow. Location B is at the Cuicuilco pyramid, situated 7km SE of Xitle's crater. A total of five trenches were studied by us. The main pyramid and associated structures were first covered by a layer of basaltic ash fall (Xitle tephra) with a thickness ranging between 20 and 30cm (Fig. 4). Grain-size analysis of an ash sample taken in trench B3 (Fig. 6) indicates a depletion in the coarse fraction, if compared with ash from localities A and C. We interpret this as a result of reworking by wind and rain. The trenches Bl, B2 and B3 are those that best display the stratigraphic relationships between the archaeological buildings, the ash fall and the lava flow (Figs 9 and 10). Trench B4 was dug around the decorated andesitic column discovered in 1996 and trench B5 on top of the pyramid. Trench Bl (Fig. 9) is particularly important because it shows the constructional fill at the base of the section (120cm thick), which has been interpreted in the past as evidence of the state of decay of the pyramid before the eruption of Xitle (Lopez-Camacho 1991; Cordoba et al. 1994). The fill is overlain by an occupational soil horizon, which is 32 cm thick, and contains fragments of pottery and charcoal. The top 12cm of the occupational horizon display a light brown colour produced by a baking effect from the hot lava flow above. The Xitle tephra covers the occupational layer, with a thickness of 22cm. A charred maize plant was found at the contact, and yielded a 14C age of 1675 ± 40 years BP. The Xitle lava is 3 m thick in this area. The following observations were made: (a) The contact of the lava with what we believe is the last body of the pyramid is at a very consistent mean height of 3.50m
(over the main corridor). We believe that a low platform was in this position. During the excavations by Cummings in 19221924, the lava flow was broken using dynamite. This must have partially destroyed the low platform. (b) Trench Bl shows the interior filling of the platform, which can also be observed along 35% of the circumference of the excavated pyramid. The fill consists of basaltic and rhyolitic blocks of a homogeneous size set in a clay-rich matrix. In some places it is possible to observe that the platform was built in two stages. (c) The platform contains burials with offerings (pottery and various stone objects). For this reason it is impossible that it represents a collapsed sector of the pyramid. We believe that the interior of the platform and the ash layer on top have been misinterpreted in the past as material coming from the destruction and erosion of the pyramid. This interpretation has led to the conclusion that the main pyramid was abandoned before the eruption. Our observations do not support this hypothesis. Trenches B2 and B3 were excavated in the SE ramp of the pyramid (Fig. 10). They show basically the same stratigraphy: constructional fills rich in pottery and charcoal fragments at the base of the sections followed by the Xitle tephra. The tephra is 8.5cm thick in trench B2 and 4cm thick in trench B3. The sequence is topped by the Xitle lava flow, which varies in thickness between 3.20 and 3.70m. In trench B3 a very strong baking effect of the fill on the ramp can be observed. The baking reaches a depth of 30 cm and can be recognized by its brick-red colour. The contact between the ramp, tephra and lava flow is very clear and planar, following the shape of the ramp. The
Fig. 9. Locality Bl at Cuicuilco. Person is pointing at charcoal dated in this study at 1675 ±40 years BP. a, Scoriaceous basal breccia of lava flow; b, reworked air-fall ash; c, baked occupational surface; d, constructional fill of pyramid.
Fig. 10. Stratigraphic sections at localities B2 and B3 at Cuicuilco. Charcoal dated at 1995 ± 60 years BP was recovered from the constructional fill below the air-fall ash.
THE CUICUILO PYRAMID
217
study we obtained the following 14C dates (Table 2 and Fig. 12): 1665 ±65 years BP (location A burnt branches of wood within the tephra sequence); 1675 ±40 years BP (location Bl, on maize plant charred by the Xitle lava); 1995 ±60 years BP (location B2, charcoal fragment in constructional fill in contact with the tephra layer; 2295 ±115 years BP (location C, charcoal within dark soil below lava flow). In general, the 14C dates seem to cluster around 2000 years BP However, the dates directly associated with the Cuicuilco archaeological zone tend to be younger, together with the date obtained in this study near the crater (location A). In previous studies young ages were obtained on a root in a midden burned by the lava at Cuicuilco A (pyramid) (1536±65 years BP; Fergusson & Libby 1963), and for a midden in Cuicuilco B (1430 ±200 years BP; Crane & Griffin 1958). The young ages obtained during this study are both on material that was definitely burnt by the volcanic activity, which would suggest that they represent the true age of the eruption. However, an explanation for the clustering of ages around 2000 years BP still eludes us. The younger 14C dates support the archaeological sequence for Cuicuilco proposed by Age of the eruption Heizer & Bennyof (1958, 1972), who suggested The Xitle lava flow is certainly the most fre- that the eruption of Xitle occurred around quently dated volcanic feature of the Chichi- AD 400. nautzin volcanic field. None the less, there are As an independent way to assess the age of still uncertainties with regard to the age of the the eruption, we tried to date the sediments eruption. First radiocarbon determinations were baked by the lava at localities B3 (SE ramp at carried out during the early development of the Cuicuilco) and C (CEPE-UNAM) using optiradiocarbon method, by Arnold & Libby (1951) cally stimulated luminescence (OSL) (Duller and Libby (1955). They reported an age of 1996). At each site a block of baked sediment 2422 ±25 years BP on charcoal associated with 20 cm x 20 cm was carved from the section and the eruption of Xitle volcano. wrapped in foil to avoid additional exposure to More than 28 radiocarbon dates have been sunlight. In the laboratory the external surface reported in the past decades from charcoal asso- of the blocks of sediment was removed, because ciated with Xitle lavas (Urrutia-Fucugauchi it would have been exposed to light during 19960, &), with dates ranging from 1536 ±65 sampling. The material from within the block years BP to almost 4000 years BP being reported. was then disaggregated in sodium oxalate and The main problem seems to reside in the poor hydrogen peroxide, and settled in distilled water control on the depth within the soil with respect to obtain grains of 4-11 ^m. Initial luminescense measurements showed no to the lava flow at which samples have been collected. For example, at the Cuicuilco pyramid, significant OSL signal above the level of the evidence for human settlement spans several background on the Risoe automated TL-OSL centuries in the anthropogenic fills of the archae- reader that was used. The fundamental problem ological buildings. These often contain frag- was the absence of quartz or feldspar with a suitments of charcoal that are considerably older able grain size (>4 /mi) in the samples. Although some material was separated, we suspect that it than the volcanic activity of Xitle. The most recently reported 14C dates include may have been clay pellets rather than true siltthe following: 2030 ± 60 and 2090 ± 70 years BP sized material. This yielded signal levels that (Martin del Pozzo et al 19970); 1945 ±55 and were at about the noise level of the instrument. 2025 ±55 years BP (Cervantes-Laing & Moli- One possible problem was that the material was nero 1995; Delgado et al. 1998). During our so young that it was not possible to obtain a large lack of charred vegetation, reworked material, erosional features, and animal holes indicates that the ramp was still in use at the time of Xitle's eruption. Location C is situated 12km NE of the Xitle crater, in a garden within the grounds of the Centro de Estudios para Extranjeros (CEPE) on UNAM campus (Fig. 11). The section shows at the base at least 100cm of a dark brown soil. The soil is sandy-clayey, with abundant pottery and charcoal fragments. The upper part of the soil (10-20 cm) is brick-red in colour and indurated as a result of the baking effect of the lava flow on top. The Xitle tephra is of silt grade and is only 2.5 cm thick, whereas the lava flow is 7.80m thick. The outcrop is very extensive and measures at least 40m in length. Archaeological remains in the soils below the Xitle tephra and lava flow include floors, walls, pottery and bone fragments. This shows proof of the extension of the inhabited area at the time of the Xitle eruption, which was clearly much larger than the area officially delimited and protected as an archaeological site.
Fig. 11. Locality C near CEPE on UNAM campus. Charcoal within dark soil below the lava flow was dated at 2295 ± 115 years BP in this study.
Fig. 12. Graph showing archaeological time scale for central Mexico and recent 14C dates reported for the Xitle eruption. Points A Bl, B2 and C are the dates obtained during the present study. Points D and E are from Martin del Pozzo et al. (19970); points F and G from Cervantes-Laing & Molinero (1995), Delgado et al. (1998).
220
S. GONZALEZ ET AL.
signal from it. To test this, separate aliquots of the samples were irradiated in the laboratory using a 90Sr/90Y beta source. The doses were chosen such that they were equivalent to c. 500010000 years of natural radiation. However, even after this irradiation the samples still yielded a negligible OSL signal. These results suggest that the 4-11 /am material that was separated from the samples did not have a suitable mineralogy for the application of luminescence dating. Magnetic properties of anthropogenic soils and tephras During this project, magnetic properties of different materials were determined, to discriminate and correlate the Xitle tephra and the anthropogenic soils around the Cuicuilco pyramid. The magnetic properties determined included magnetic susceptibility of tephras and soils and Curie temperatures of the tephras. At some archaeological sites with long occupational sequences it is difficult to distinguish between an occupational surface (or floor) and a fill, and magnetic susceptibility measurements have been useful in differentiating them (Tite & Mullins 1971; Tite 1972). For this reason, we took a series of samples in each soil layer in trenches Bl, B2, B3, B4 and B5 around the main pyramid. The samples were air dried in the laboratory and were used to fill cylindrical plastic pots 2.5cm in diameter. The samples were measured using a Bartington Susceptibility Meter (MS-2). An example of the results obtained in the SE ramp (trench B2) is shown in Fig. 13 together with the stratigraphy. The results show that the magnetic susceptibility measurements represent a very useful tool in distinguishing occupational floors from fills. Generally, the floors have higher values with distinct peaks of magnetic susceptibility, suggesting high concentrations of magnetite, probably produced by fire. The fills, on the other hand, have lower susceptibilities. The maximum value was found at the contact between the original land surface (a deposit of fluvial origin containing rounded pebbles of andesite) and the constructional fill of the pyramid. This suggests that human activity probably occurred in this area for some time before the commencement of the construction of the pyramid. Magnetic susceptibility measurements were also carried out on tephra samples at localities A, B3 and C. The results show similar values for localities A (260xlO~ 6 SI) and C (233 x 10~6SI), whereas at locality B3 the value was
much higher (413 x 10 6 SI), indicating a different magnetic mineralogy. It is possible that the presence of additional magnetic minerals resulted from the baking of the tephra produced by the Xitle lava at this site. Curie temperature experiments were performed on the same tephra samples, which were heated to 700°C in a strong magnetic field and then allowed to cool. The results are presented in a graph of temperature v. magnetisation (Fig. 14). The curves from localities A and C are both dominated by paramagnetic minerals and show an increase in magnetization after heating. Locality B3 shows a thermomagnetic curve that is typical of magnetite, with a Curie temperature of c. 590°C. The values found for the magnetic susceptibility and Curie temperatures for localities A and C are very similar, whereas values near the pyramid (locality B3) are very different. The tephra collected at the pyramid is visibly reddened, indicating that baking occurred during the emplacement of the lava. The baking effect of the lava is locally very variable. Around locality C there are zones where the underlying sediments are considerably baked and others nearby seem unaffected. It would seem that this baking has produced magnetite in the tephra, possibly through oxidation. Another possible explanation is that during the reworking of the tephra at the Cuicuilco site magnetic minerals have been concentrated. These changes are noticeable only through the measurement of the magnetic properties, which shows the usefulness of this approach in understanding the processes of site formation in archaeological contexts. Conclusions New stratigraphic and archaeological evidence found during the excavations of 1996, when five trenches were excavated around the Cuicuilco pyramid, reinforces the view that the abandonment of this important archaeological zone resulted primarily from the eruption of the Xitle volcano, 7 km SW of the pyramid. The clean contact, with no evidence of charred vegetation or animal holes, between the archaeological buildings and the overlying tephra and lava is strong evidence in support of this hypothesis. Furthermore, the material found in front of the decorated andesite column at the southern base of the pyramid, which had been previously interpreted as rubble derived from the erosion of the pyramid and advocated as one of the most important pieces of evidence in support of the
THE CUICUILO PYRAMID
221
Fig. 13. Stratigraphy and magnetic susceptibility measurements in ramp SE of Cuicuilco (locality B2). The depth is with respect to the top of the pyramid.
abandonment and degradation of the pyramid long before the Xitle eruption (Lopez-Camacho 1991; Cordoba et a/. 1994), is in fact part of a low platform containing undisturbed burials and offerings. The shape of this platform has been cast in the lava flow on top, with a steep angle of c. 45° away from the pyramid. Despite the large number of determined radiocarbon dates (more than 35) the age of the Xitle eruption is still a matter of debate. The majority of the samples with a good stratigraphic control
cluster around an age of c. 2000 years BP. However, we have obtained two new dates with excellent stratigraphic control that suggest an age for the eruption of about 1670 years BP. The discrepancy of about 300 years cannot be explained at present with the available information, but we believe that the most reliable dates for the eruption are the younger ones. Efforts to obtain independent dates through luminescence dating of the baked soils were unsuccessful because the mineralogy of the
Fig. 14. Curie temperature experiments on tephra samples at localities A, B and C.
THE CUICUILO PYRAMID material proved to be inappropriate for luminescence dating. Knowledge of the exact age of the Xitle eruption is important to better understand the fall of Cuicuilco, the most important ceremonial centre in the Basin of Mexico at the end of the Late Preclassic-Early Classic period, and the rise of Teotihuacan in the NE of the Basin. This is especially so because around that time Popocatepetl stratovolcano, located 40km SE of Cuicuilco, experienced a major eruption (Siebe et al. 1996). Deposits of this eruption have been dated at c. 200 BC. Maybe a combination of these two important volcanic events in the south of the Basin of Mexico could explain why there was a major population shift to the north. The favourable geographical location of the Basin of Mexico, with temperate climate, fertile volcanic soils and abundance of water, has been responsible for early human settlement since Pleistocene-Early Holocene time. Volcanic activity in the central part of Mexico must have played an important role in determining the timing of large migrations in Prehispanic times as well as the rise and fall of cities such as Cuicuilco, Teotihuacan and Cholula (Siebe et al. 1996, 1997). Xitle tephra and lava flows covered the settlement of Cuicuilco, the first known urban centre in the Mexican Altiplano, sealing in this way a complete Prehispanic landscape and cultural remains. Archaeological investigations are difficult today, not only because they involve the breaking of the lava layer but also because the area is now heavily populated. Only 5% of the archaeological zone has been investigated so far, and it is to be hoped that much more information will be recovered in the future. We thank the team of archaeologists from INAH involved in the excavation of Cuicuilco during the 1996 season, particularly M. Perez-Campa, H. GomezRueda and S. Pefia. Work by C. Siebe was supported by UNAM-DGAPA (IN107196) and CONACyT (0264P-T9506). Radiocarbon dates were determined by A. Long and C. Eastoe at the University of Arizona in Tucson.
References ARNOLD, J. T. & LIBBY, W. F. 1951. Radiocarbon dates. Science, 113, 111-120. BADILLA-CRUZ, R. R. 1977. Estudio petrologico de la lava de la parte noreste del Pedregal de San Angel. Boletin de la Sociedad Geologica Mexicana, 38, 40-57. CANON-TAPIA, E., WALKER, G. P. L. & HERREROBERVERA, E. 1995. Magnetic fabric and flow direction in basaltic pahoehoe lava of Xitle
223
Volcano, Mexico. Journal of Volcanology and Geothermal Research, 65, 249-263. CERVANTES-LAING, P. & MOLINERO, R. J. 1995. Eventos volcdnicos al sur de la ciudad de Mexico. BSc thesis, Facultad de Ingenieria, UNAM, Mexico City. COE, M. 1994. Mexico, from the Olmecs to the Aztecs, 4th edn. Thames and Hudson, London. CORDOBA, C., MARTIN, A. L. & LOPEZ, J. 1994. Palaeolandforms and volcanic impact on the environment of Prehistoric Cuicuilco, Southern Basin of Mexico. Journal of Archaeological Science, 21, 585-596. CRANE, H. R. & GRIFFIN, J. B. 1958. University of Michigan radiocarbon dates III. Science, 128, 1117-1123. CUMMINGS, B. 1933. Cuicuilco and the Archaic culture of Mexico. University of Arizona Social Science Bulletin, 4(8). DELGADO, H., MOLINERO, R., CERVANTES, P. et al. 1998. Geology of the Xitle Volcano in Southern Mexico City - a 2000 year old monogenetic volcano in an urban area. Revista Mexicana de Crencias Geologicas, 15(2), 115-131. DULLER, G. A. T. 1996. Recent developments in luminescence dating of Quaternary sediments. Progress in Physical Geography, 20, 133—151. FERGUSON, G. J. & LIBBY, W. F. 1963. 1963 UCLA radiocarbon dates II. Radiocarbon, 5, 1-22. GONZALEZ-HUESCA, I. S. 1992. La variacion secular en Mexico central durante los ultimos 30000 anos por medio del estudio magnetico de lavas. PhD, National University of Mexico, Geophysics Institute, Mexico City. HEIZER, R. & BENNYHOFF, J. 1958. Archaeological investigations of Cuicuilco, Valley of Mexico 1956. Science, 127, 232-237. & 1972. Archaeological excavations at Cuicuilco, Mexico 1957. National Geographic Reports, 1955-1960, 93-104. LIBBY, W. F. 1955. Radiocarbon Dating, 2nd edn. Chicago University Press, Chicago, IL. LOPEZ-CAMACHO, J. 1991. Estratigrafia de la piramide de Cuicuilco en retrospectiva. Cuicuilco, 27, 35-46. MARTIN DEL Pozzo, A. L., CORDOBA, C. & LOPEZ, J. 19970. Volcanic impact on the southern Basin of Mexico during the Holocene. Quaternary International, 43/44, 181-190. , ESPINASA, R., LUGO, J. et al. 19976. Volcanic Impact in Central Mexico. Excursion Guide. IAVCEI General Assembly, Puerto Vallarta, Mexico, 15-19 January. PARSONS, J. R. 1989. Arqueologia regional en la cuenca de Mexico: una estrategia para la investigation futura. Anales de Antropologia, 26, 157-257. PECCERILLO, A. S. & TAYLOR, S. R. 1976. Geochemistry of Eocene calcalkaline volcanic rocks from the Kastamonu area, northern Turkey. Contributions to Mineralogy and Petrology, 58, 63-81. PEREZ-CAMPA, M., PASTRANA, A. & GOMEZ-RUEDA, H. 1995. Proyecto Arqueologico Cuicuilco. Archivo del Consejo de Arqueologia, INAH, Mexico City.
224
S. GONZALEZ ET AL.
PINA-CHAN, R. 1967. Official Guide of CopilcoCuicuilco. Institute Nacional de Antropologia e Historia, Mexico City. SCHAVELZON, D. 1993. La Piramide de Cuicuilco. Fondo de Cultura Economica, Mexico City. SIEBE, C, ABRAMS, M., MACIAS, J. L. & OBENHOLZNER, J. 1996. Repeated volcanic disasters in Prehispanic time at Popocatepetl, central Mexico: past key to the future? Geology, 24, 399-402. , MACIAS, J. L., ABRAMS, M., RODRIGUEZ, S. & CASTRO, R. 1997. Catastrophic Prehistoric Eruptions at Popocatepetl and Quaternary Explosive Volcanism in the Ser dan-Oriental Basin, East-Central Mexico. Excursion Guide. IAVCEI General Assembly, Puerto Vallarta, Mexico, Premeeting Excursion 4, 12-18 January. SCHMITTER, E. 1953. Investigation petrologica en las lavas del Pedregal de San Angel. Memorias del Congreso Cientifico Mexicano, 3, 218-237.
TITE, M. S. 1972. The influence of geology on the magnetic susceptibility of soils on archaeological sites. Archaeometry, 14, 229-236. & MULLINS, C. 1971. Enhancement of the magnetic susceptibility of soils on archaeological sites. Archaeometry, 13(2), 209-219. URRUTIA-FUCUGAUCHI, J. 1996a. Palaeomagnetic study of the Xitle-Pedregal de San Angel lava flow, southern Basin of Mexico. Physics of the Earth and Planetary Interiors, 97, 177-196. \996b. Comentarios sobre la edad del campo volcanico Pedregal de San Angel, Cuenca de Mexico - fechamientos por radiocarbono. GEOS, Union Geofisica Mexicana, June 96-98.
Volcanic disasters and cultural discontinuities in Holocene time, in West New Britain, Papua New Guinea ROBIN TORRENCE1, CHRISTINA PAVLIDES2, PETER JACKSON3 & JOHN WEBB3 1
Division of Anthropology, The Australian Museum, 6 College Street, Sydney, NSW 2010, Australia 2 School of Archaeological and Historical Studies, La Trobe University, Bundoora, Vic. 3083, Australia ^School of Earth Sciences, La Trobe University, Bundoora, Vic. 3083, Australia Abstract: An evaluation of the relationship between culture change and the history of volcanic activity from the Witori and Dakataua volcanoes in West New Britain province, Papua New Guinea, demonstrates the importance of studies focusing on long time spans to an understanding of cultural adaptation to volcanic disasters. Using a chronostratigraphy based on several techniques for matching tephras, the cultural responses to five volcanic events are compared and contrasted between the Willaumez Peninsula and Yombon, areas whose environment and proximity to the volcanoes vary significantly. Archaeological analyses of material show that human groups did not immediately adjust to the effects of the most severe volcanic events but abandoned both regions. In contrast, adaptation on a long-term basis may be indicated by the occurrence of a punctuated trend in lithic technology inferred to reflect a decrease in mobility and an increase in the intensification of subsistence practices. This pattern, combined with limited radiocarbon dating, suggests that the length of abandonment decreased after each eruption, probably because of changes in social organization and subsistence practices. The paper demonstrates the value of collaboration between archaeology and geology in the study of long-term human responses to natural hazards.
Most research into the role of natural catastrophes in prehistory focuses on single outstanding events and their relatively immediate consequences, known or inferred. Excellent case studies using this approach were presented at this conference; for example, the eruption of Santorini and its effects on Minoan civilization (Driessen & Macdonald this volume; Bicknell this volume), the effects of Popocatepetl on the rise of Teotihuacan (Plunket & Urunuela this volume); or of earthquakes causing the abandonment of sites (Waelkkens et al. this volume), In contrast, continuing research in West New Britain province, Papua New Guinea, is examining the effects of repeated volcanic eruptions on long-term patterns of cultural change over the past 6000 years. The aim of the multi-disciplinary work is to investigate whether human populations living in a hazardous environment, one where volcanic disasters are relatively frequent on a geological time scale, make long-term cultural adaptations to buffer themselves or whether they are simply buffeted
up and down on alternating waves of prosperity and disaster. Our initial conclusion is that the answer lies somewhere between these two extreme positions. Although all severe volcanic events have necessitated abandonment, reoccupation appears to have taken place more quickly after each of the events that we have studied. The roles of social organization and subsistence patterns are identified as critical variables in the process of reoccupation. The study illustrates the value to hazards research of archaeological studies that can provide a very long record involving multiple events and responses (see Grayson & Sheets 1979). West New Britain is an excellent place to study the interrelationship between disasters and human responses because during the mid- to late Holocene period large parts of the region experienced the effects of up to 13 eruptions from the Witori volcano and four from the Dakataua volcano: many of these have been characterized as very severe, high-magnitude events. Furthermore, recent detailed tephrachronological
From: McGuiRE, W. G., GRIFFITHS, D. R., HANCOCK, P. L. & STEWART, I. S. (eds) The Archaeology of Geological Catastrophes. Geological Society, London, Special Publications, 171, 225-244. 1-86239-062-2/00/ $15.00 © The Geological Society of London 2000.
226
R. TORRENCE ET AL.
(Torrence & Webb 1992; Machida et al 1996) and geochemical studies have provided the basis for constructing a chronostratigraphy of the region. In this paper we compare and contrast human responses to a succession of five major volcanic events, which both occurred in two regions. The Willaumez Peninsula and inland Yombon areas (Fig. 1) were chosen because they vary greatly in respect to local environment and proximity to the volcanoes concerned and detailed archaeological research is under way in both places (e.g. Torrence et al 1990; Pavlides 1993, 1996; Pavlides & Gosden 1994; Torrence & Boyd 1996; Torrence & Summerhayes 1996). It is fairly obvious that in the short term, major volcanic eruptions matter to local resi-
dents because they suffer loss of life, homes and means of living. In many cases people are forced to abandon a region for various lengths of time. There is no general agreement about the relationship between the long-lasting effects of natural hazards and the social organizations of the groups involved. Modern scholars of natural hazards predict that in developed countries with complex, state-level social organizations disasters can be mitigated through aid programmes, redistribution of resources, etc., whereas egalitarian societies with smaller social networks have a wide range of responses but are more likely to abandon an area (e.g. Chester 1993, table 8.4). A contrasting picture has been presented by Sheets et al. (1991), who have compared the prehistory of the Arenal valley in Costa
Fig. 1. Location of study areas and sites mentioned in the text.
VOLCANIC DISASTERS IN WEST NEW BRITAIN Rica with that of San Salvador and Panama. In the former case, despite 10 explosive eruptions during 4000 years, the cultural stability reconstructed from archaeological research is said to be remarkable. In contrast, the effects of major eruptions of Ilopango volcano in San Salvador had far-reaching consequences for the development of Mayan civilization, and the Baru volcano severely disrupted the Bariles chiefdom society in Panama. According to Sheets et al. (1991, p. 446) simpler societies 'appear to be more resilient in the aftermath of explosive eruptions' than are more complex societies, which are dependent on a built environment and economies based on 'occupational specialization, redistribution, and long distance trade routes'. They did note, however, that differences in the severity of the three cases, as largely witnessed by the depth of air-fall tephra, may also have been an important variable (Sheets et al. 1991, p. 462). The ancient cultures of Papua New Guinea that we are studying are similar to the Costa Rican egalitarian, subsistence farmers. Not surprisingly, we find that when viewed over thousands of years, they too appear to have been resilient in the face of volcanic disasters, as they continued to be self-sufficient farmers with a fairly simple level of social organization until recent times. When the data are viewed in more detail, however, it is clear that there are significant differences between the prehistoric societies of West New Britain and Costa Rica in the way they have reacted to a long history of volcanism. Unlike the Arenal case, for which there is no relationship between phases defined by cultural differences and volcanic events (Sheets et al. 1991, p. 462), in West New Britain there are cultural changes after at least two of the events (Torrence et al. 1990; Specht et al. 1991; Pavlides 1993, 1996; Pavlides & Gosden 1994). To explain this somewhat puzzling variation in long-term adaptation to volcanism among roughly similar societies occurring in
227
these two different parts of the world, the paper will examine the relationship between volcanic events and social organization focusing on the long-term record of West New Britain. Comparisons of the severity of volcanic events with strategies for reoccupation in terms of subsistence, settlement and social organization, using studies of lithic technology, use-wear or residues and the spatial distribution of cultural materials provide the basis for understanding cultural responses to volcanism. The volcanic background During the mid- to late Holocene period two volcanoes, Dakataua on the tip of the Willaumez Peninsula and Witori to the east on the Hoskins Peninsula, were among the most active among the many known volcanic centres in the island of New Britain (Fig. 1). Their eruptive histories have been the subject of recent research summarized in detail by Machida et al. (1996). Analyses of the tephrastratigraphy, the eruptive products, heavy minerals and refractive indices of the glasses for 13 Witori and four Dakataua eruptions comprise the basis for a chronostratlgraphy that can be used over the very broad region in West New Britain where tephras have been found in archaeological contexts (Specht et al. 1991; Pavlides 1993; Gosden & Pavlides 1994; Machida et al. 1996). The high-magnitude prehistoric events are reasonably well dated by radiocarbon from samples taken either from within flows or from soils directly under tephras. The type of eruption and their dates (calibrated) are summarized in Table 1. The data are derived from Machida et al. (1996), except for the case of the W-K1 eruption, for which a new date of 5204 ±85 (NZA 1570) has been obtained from site FAP on Garua Island (Table 2; Fig. 1). As this date has been corroborated by another unpublished determination from Yombon, we are confident
Table 1. Summary of major Holocene volcanic events Eruption
Approximate date* (calendar years)
Volume (km3)
Type
Dakataua volcano Dk
1000
10
phreatomagmatic, Plinian, ignimbrite forming
Witori volcano W-K4 W-K3 W-K2 W-K1
1400 1700 3600 5900
6 6 30 10
phreatomagmatic, Plinian, ignimbrite forming Plinian phreatomagmatic, Plinian, ignimbrite forming Plinian, ignimbrite forming
* Based on Machida et al. (1996), and Table 2 using CALIB Version 3.0.3.
228
R. TORRENCE ET AL.
Table 2. Radiocarbon dates for recolonization of study areas Location
Material
Lab. No.
Phase 5: soil developed on Dakataua FAO 1001/999 carbonized nutshell Beta 72139
14
C date
Calibrated date
1100 ±60
1062 (977) 939
Phases 3-4: soil FSZ 17/83 Level 2 spit 1 FSZ 17/96 Level 1 spit 1 FSZ13/92 Level 1 spit 1 FSZ 17/98 Level 1 spit 3 FSZ I unit A DK trans. FSZ 14/88 Level 1 spit 1
directly under Dakataua, or origin unknown
FSZ 14/88 FSZ 17/96 Level 2 spit 2 FSZ 13/92 Level 1 spit 2
carbonized nutshell Beta 72143
1 160 ± 60
1161 (1062)977
carbonized nutshell NZA 3730
1553±63
1522 (1410) 1349
carbonized nutshell NZA 6098
1798 ±69
1814(1710) 1614
carbonized nutshell NZA 3732
1834 ±64
1830(1731) 1635
carbonized nutshell Beta 72142
1990 ±60
1991 (1931) 1870
carbonized nutshell NZA 2851
2287 ± 64
2348 (2328) 2158
carbonized nutshell NZA 2852
2358 ±64
2365 (2348) 2332
carbonized nutshell NZA 3731
2361 ±75
2440 (2348) 2329
carbonized nutshell NZA 6099
2781 ±68
2949 (2860) 2783
Phase 4: soil developed on or mixed wi thin W-K3-4 FIF 5 Unit 2 charcoal Beta 62321 1460 ±120 Phase 3: soil developed on W-K2 FAO 1000/1000 Level 3 spit 2 carbonized nutshell NZA 3738
Reference
1494 (1331) 1278
2439 ± 64
2712 (2469, 2385, 2376) 2352
carbonized nutshell NZA 3729
2452 ± 67
2716 (2469) 2354
carbonized nutshell Beta-72140
2540 ± 60
2747 (2720) 2489
charcoal charcoal
Beta 47048 Beta 1545
2570 ±90 2575 ±100
2759 (2740) 2489 2766 (2740) 2486
Pavlides (1993, p. 57) Spechtetal. (1981, p. 14)
charcoal
Beta 62320
2580 ±80
2758 (2742) 2551
Pavlides & Gosden (1994, p. 609)
charcoal
SUA-1491
2760 ±210
3146 (2851) 2728
carbonized nutshell NZA 3733
2883 ±64
3101 (2975)2882
carbonized nutshell Beta 72144
3060 ±60
3351 (3261) 3169
carbonized nutshell NZA 3734
3030 ±69
3342 (3214) 3084
Phase 2: soil beneath W-K2 FAAK 4 Level 6 spit 1 carbonized nutshell Beta- 102967 3330 ±40 FAO 1000/1000 Level 6 spit 1 carbonized nutshell NZA 2901 3532 ±66
3642 (3562) 3473
FAO 990/990 Level 3 spit 4 FAQ 95/210 Level 3 spit 3 FGT/7 Unit 5 FGT/I FIF/3 Unit 4
FHC Layer 4 hearth FYSII Level 5 spit 1 FYSII Level 5 spit 3 FYS II Level 5 spit 4
3885 (3828, 3781, 3776) 3697
VOLCANIC DISASTERS IN WEST NEW BRITAIN
229
Table 2. (continued) Location
Material
Lab. No.
14
Calibrated date
C date
Phase 2: soil beneath W-K2 (continued) FAQ 45/49 Level 6 spit 2 carbonized nutshell NZA 2850 FAQ 45/49 Level 7 spit 6 carbonized nutshell NZA 3015
3566 ±64
3923 (3838) 3727
3894 ±69
4415 (4347,4325, 4303)4158
FYW/3 Unit 4
charcoal
Beta 62315
4040 ±140
FGT/I
charcoal
Beta 1544
4270 ±130
4819 (4517, 4468, 4453)4314 4978 (4840) 4578
NZA 1570
5204 ±85
6164 (5936) 5903
NZA 1905
5970 ± 76
6886 (6840, 6830, 6790) 6730
Phase 1: soil beneath W-K1 FAPD charcoal FGT/7 Unit 9 charcoal
that 5900CalBP is more accurate than the preliminary data presented by Machida et al (1996, p. 71). In addition, new dates from below and directly on top of the Dakataua tephra from archaeological contexts on Garua Island confirm the poorly provenanced date presented by Machida et al. (1996, pp. 71-27) (Table 2). The large magnitude of these events, which together produced about 80km3 of material (Machida et al. 1996, p. 78), must have had a profound effect on human settlement in the region. As shown in Table 1 and presented more fully by Machida et al. (1996), the frequency of Witori eruptions has increased steadily through time. 'We are unaware of similar eruption rates from such relatively small areas elsewhere' (Machida et al. 1996, p. 78). On the other hand, the scale of the eruptions in terms of the size and spatial extent of pyroclastic flows and Plinian deposits has decreased since W-K2. W-K1 and W-K2, at least, were caldera-forming events and with W-K4 probably erupted through a lake, although this is no longer present. Isopach maps have been constructed by Machida et al. (1996; pp. 73-75) for a portion of the north coast of West New Britain. Their maps for the Plinian tephras suggest easterly prevailing winds, which are common in low-latitude regions (Machida et al. 1996, pp. 73-75). As determined by a number of techniques discussed further below, the W-K2 tephra is the most widespread, reaching Lolmo cave in the Arawe Islands off the south coast (Gosden et al. 1994, p. 106; Fig. 1). Although W-K1 to W-K4 tephras reached the inland region of Yombon (Pavlides 1993, p. 57), the effects of the Dakataua tephra were restricted
Reference
Pavlides & Gosden (1994, p. 608) Specht et al. (1981, p. 14)
Pavlides (1993, p. 57)
to the north coast (Machida et al. 1996, p. 72). As the geological survey reported by Machida et al. (1996) was restricted to a relatively small region along the north coast, it is not possible to construct isopach maps for the entire region discussed in this paper. Outside that region our data are restricted to information on buried tephras obtained from small archaeological excavations, rather than large-scale surveys of exposures.
The study areas Knowledge about the spatial distribution of tephras derived from the geological fieldwork and laboratory studies reported by Machida et al. (1996) has been expanded by archaeological research in the Willaumez Peninsula, Yombon in the interior rainforests, and the Arawe Islands and Kandrian region in the south (Fig. 1). At a number of sites various Witori and one Dakataua tephra have been identified through tephrastratigraphy, refractive indices of the glasses (Pavlides 1993; Gosden et al. 1994; Machida et al. 1996) and scanning electron microscopy (SEM) microprobe analyses of individual glass shards. This paper focuses on a number of sites in the Willaumez Peninsula region and surrounding the Yombon airstrip (Fig. 1), where Witori tephras W-K1 to W-K4 have been well preserved. Adding the Dakataua (DK) event, which is visible only within the Willaumez Peninsula, provides an opportunity to compare and contrast human responses to volcanic disaster over a period spanning about 10000 years.
230
R. TORRENCE ET AL.
Archaeological research in the Yombon region began with reconnaissance and test excavations by Specht (Specht et al. 1981, 1983). This has been followed by a more thorough investigation at a wide range of sites directed by Pavlides (1993, 1996; Pavlides & Gosden 1994) (Fig. 1). Specht was also the first to undertake archaeological research in the Willaumez Peninsula and to recognize the significance of the presence of tephra layers in sites (Specht et al 1988, 1991). Building on initial work, Torrence (Torrence & Webb 1992; Torrence 1993; Boyd & Torrence 1996; Torrence & Boyd 1996) has since conducted a major field project on Garua Island, located just to the east of the peninsula (Fig. 1; see also Fig. 8, below), where a large number of additional archaeological contexts have been studied. Although field identification of tephras supplemented with refractive indices has proved useful for sourcing tephras interbedded in archaeological sites, particularly in the Willaumez Peninsula where relatively thick deposits are well preserved, additional geochemical data have been used to identify buried tephra layers from a wide range of archaeological contexts. Using a microprobe attached to a scanning electron microscope, Jackson carried out analyses of individual glass grains derived from field samples and separated through washing of soils in water. Only fresh unweathered grains were analysed wherever possible. Identifications are based on matching the geochemistry of grains from known tephras present at one main type section (locality 37 of Machida et al. (1996) and later revisited by Webb and Jackson) (Fig. 2) and several other key localities (samples provided by Machida) with that of tephra samples collected from archaeological contexts. Iron and calcium proved to be the most useful elements for differentiating tephras, and are relatively immobile during weathering. As can be seen in the plot of iron and calcium oxides, the two most widespread tephras, W-K1 and W-K2, are easily discriminated from each other and from W-K3 and W-K4, although there is some overlap between the latter two (Fig. 3). The Dk tephra is distinct from the Witori sequence because of the very much higher levels of both iron and calcium. Archaeological samples from sites on Garua Island (Fig. 4) and Yombon (Fig. 5) have been successfully matched with the type sections. The tephra stratigraphy has proved invaluable for archaeological research. First, the relative chronology provided by the presence of identifiable interbedded tephras is often all archaeologists have to go on, because acidic soils and
Fig. 2. Main section used in geochemical study for matching archaeological layers to volcanic events (compare locality 37 of Machida et al. (1996, p. 67)). high rainfall mean that organic preservation is exceptionally poor, and the main source of evidence, undiagnostic stone tools, is difficult to date. Second, the chronostratigraphy allows relatively detailed comparisons between very widely spaced sites. So, for example, the stratigraphies at Bitokara Mission (FRL) and Garua Island (FAO) on the Willaumez Peninsula region, where W-K1, W-K2 mixed W-K3-4, and Dk tephras have been identified using tephrastratigraphic, refractive indices and geochemical criteria can be correlated directly with sites around Yombon deep in the rainforest interior where W-K1, W-K2, and a mixed W-K3-4 have also been identified (Fig. 6). At FRL the W-K1 tephra has been redeposited,
VOLCANIC DISASTERS IN WEST NEW BRITAIN
231
Fig. 3. Results of SEM analyses of glass shards from type localities. and it is not preserved at the hilltop setting of FAO, probably because it was washed away. W-K1 has been noted only in redeposited contexts in drainages on Garua Island. At Yombon an additional tephra, dated somewhere between 14000BP and W-K1 is present, but its source is unknown (Pavlides & Gosden 1994, p. 607). A series of unprovenanced, pre-W-Kl tephras have also been noted on Garua Island, but as yet all archaeological material is stratified well above them. To compare and contrast the prehistories of Yombon and the Willaumez Peninsula a basic outline of phases has been constructed using tephras identified by physical or chemical characteristics as boundaries (see Table 3, below). In addition, a large number of radiocarbon dates has been obtained from most of the sites discussed below. A number of these are presented in Table 2 (see Specht et al 1991; Pavlides 1993; Pavlides & Gosden 1994). The Yombon archaeological sequence extends back into the Pleistocene period (Pavlides & Gosden 1994), but the
oldest Willaumez Peninsula material appears to be limited to the Holocene period and so this paper focuses on approximately the past 10000 years. Phases 4 and 5 are not distinguishable at Yombon because the Dk tephra is absent, but this is not a serious loss of resolution because, at only 600 years, phase 4 is much shorter than the others. Furthermore, in the Willaumez Peninsula the practice has been to collapse phase 4 into phase 3 because the W-K3-4 tephra has a very patchy occurrence. The differences in chronological divisions between the two regions cause some difficulty in correlating data, but in general do not affect the reconstruction of long-term patterns of prehistoric human behaviour, as will be demonstrated in the analyses presented below. Finally, the tephra chronology allows a comparison of the effects of the same volcanic eruption on human societies in the two study regions. As the distance to the Witori volcano varies and the two regions are characterized by different environments, different responses to volcanic events can be predicted. The variation in distance
232
R. TORRENCE ET AL.
Fig. 4. Results of SEM analyses of tephra layers from archaeological site FRL in the Willaumez Peninsula region.
is reflected by smaller average depths of tephra at Yombon than on the Willaumez Peninsula (Table 4). The Yombon region, located near the centre of the island on the southern foothills of the Whiteman Range, is characterized by mountainous, extremely rugged, karst topography with altitudes around 500 m above sea level and high rainfall (c. 6.5m annually). The region is currently covered with tropical rainforest and occupied by people who live in dispersed hamlets with a low-intensity system of shifting cultivation supplemented by hunting and gathering (Pavlides 1993, p. 55; Goodale 1995). In contrast, the archaeological sites investigated in the Willaumez Peninsula region (including Garua Island) are within 2 km of the coast and less than 100m above sea level. There are marked wet and dry seasons, with the bulk of the roughly 4.2 m of rainfall per year falling from December to April (Brookfield & Hart 1966, table 1). Shifting cultivation and dispersed settlement also appear to have characterized societies here before European contact, but it seems likely
that hunting and gathering played a smaller role than in the interior. Stone tool technology, subsistence and mobility The most common artefacts recovered from archaeological sites in West New Britain are stone tools. Analysis of changes in the technology of stone tools and attributes related to their uses can be used to infer changes in patterns of subsistence and patterns of mobility. In the Willaumez Peninsula region almost 100% of the stone tools are made from obsidian, which crops out naturally in the area; a number of obsidian quarries have been located and studied (Specht et al 1988; Torrence et al 1990, 1992, 1996). Obsidian from this region was transported large distances across the Pacific region beginning about 20 000 years ago (Summerhayes & Allen 1993) but the largest distribution network occurred during phase 3 (Summerhayes
VOLCANIC DISASTERS IN WEST NEW BRITAIN
233
Fig. 5. Results of SEM analyses of tephra layers from archaeological site FGT/7 in the Yombon region.
et al. 1993, 1998). Stone sources and quarries are also found in the Yombon region, where the raw material was nodular chert. This material was widely used and distributed as far as the Kandrian region of the south coast throughout the Holocene period. Obsidian brought in to Yombon from the Willaumez Peninsula and the Mopir source near the Witori volcano occurs in relatively small portions beginning from phase 1 but increases in frequency through time (Pavlides 1993, p. 58; 1996). Despite differences in the products and associated technology, caused largely by the nature of the two raw materials, the overall pattern of change in lithic assemblages throughout the past c. 10000 years has been remarkably similar in the two regions. On the basis of characteristics of the tools, waste by-products of manufacture, and changes in the location of quarrying and manufacturing sites, a shift with time can be observed in the nature and scheduling of manufacture (Torrence 1992; Pavlides 1996). In phases 1 and 2 at Garua Island and phase 2 at
Yombon well-worked tools with stemmed bases (Fig. 7a and b), presumed to be for hafting, were carefully flaked at special purpose quarry sites and then carried around, used and maintained over long periods of time. From phase 3 there is a noticeable change, to a technology in which unretouched flakes were produced, used and discarded in the same location using nodules of raw material that were obtained from a wide range of locations (not just quarries) and then stockpiled at sites located some distance from the sources. This trend continued and intensified until European contact, when metal tools largely replaced stone. The change from a curated or highly scheduled and planned stone technology to a casual, expedient use of raw material (Binford 1979) begins as early as phase 2. For example, on Garua Island in the Willaumez Peninsula region a special purpose obsidian quarry at site FAP is abandoned after the W-K1 event and obsidian manufacture is shifted to a number of small working areas, in the order of several square
Fig. 6. Chronostratigraphy for sites in the Willaumez Peninsula and Yombon regions is based on the comparison of sections such as these.
VOLCANIC DISASTERS IN WEST NEW BRITAIN Table 3. Chronological phases for West New Britain Phase
Tephra
Date
5 4 3 2 1
Dk W-K3-4 to Dk W-K2 to W-K3-4 W-K1 to W-K2 pre-W-Kl
1000-present 1700-1100 3600-1700 5900-3600 c. 10000-5900
Table 4. Average thickness of air-fall tephra present in archaeological sites (cm)
W-K1 W-K2 W-K3-4? Dakataua
Willaumez Peninsula
Yombon
7 0.50 0.20 0.75
0.35 0.35 0.15 not present
metres. Examples of these have been recovered at sites FAQ and FAO (Fig. 8). A detailed study of changes in technology at the well-stratified site FRL (Figs 6 and 8; Specht et al. 1988) has also detected a trend beginning in phase 2 (Torrence 1992). Although at FRL stemmed tools were produced in both phases 1 and 2 in phase 2 there is a decline in the number of cores with bifacial platforms, which are associated with the manufacture of stemmed tools. At the same time, the proportion of retouched artefacts and carefully worked multi-platform cores also decreases, whereas the percentage of single platform cores, which are used to produce unretouched flakes, increases (Torrence 1992, pp. 119-120). Phase 3 represents a much more major change. At this time in the FRL sequence stemmed tools and their associated core type disappear and the proportion of retouched artefacts in the assemblages drops dramatically. A few small retouched forms, some resembling stemmed tools, have been found on Garua Island in phase 3 but these also disappear in phases 4-5. From phase 3 at FRL single platform cores for the production of small unretouched flakes become the dominant core form. Unlike the previous pattern, in which either a series of production stages were required to carefully manufacture stemmed artefacts at the quarry or cores were rotated so that a large number of flakes were removed before the core was abandoned, from phase 3 there is almost no planning in the stone technology. Flakes are hit from unprepared cores, used on the spot, and then discarded at the same location (Torrence 1992).
235
Additional support for the model of a casual, expedient approach to tool manufacture comes from a core at FRL which can be almost completely refitted because all the flakes struck from it were found within 1 m of each other (Fullagar 1990). In phase 3 when the trend to expediency was greatly accelerated, systematic quarrying ceases at the mainland obsidian source at FDQ. At this time the use of small, often water-rolled beach cobbles and scavenging of previously flaked material becomes dominant; the practice continues up to the present (Torrence 1992, p. 115). The same trend from a planned, staged stone technology to an expedient system of production and use occurred at Yombon, although the raw material and the technology for producing stemmed tools differed from those in the Willaumez Peninsula. Beginning in phase 1, extraction and manufacture of chert tools was highly focused at the FGT/7 quarry, although tools were used and resharpened at other locations in the area. Furthermore, the staged reduction sequence is clearly evident in the spatial distribution of artefacts across sites in the region; for example, there are few cores deposited away from the quarry. However, the ratio of used and discarded tools to other flaking debris increases. In phase 2 the FGT/7 quarry is abandoned and production shifts to a number of smaller quarries or workshops (eg. FYV/1 and 2), which like those on Garua Island, contain far less waste from production than the earlier quarry. At the same time there is a more even distribution of material across the landscape, suggesting that staged manufacture and reduction has declined, although stemmed tools are the dominant tool type used at this time (Pavlides 1996). As with the Willaumez Peninsula, there is a major change at Yombon in phase 3 after the W-K2 tephra. Quarrying becomes less focused and scavenging of previously flaked chert and obsidian begins in phase 3 and increases through time. Stemmed tools disappear and unretouched flakes completely dominate the assemblages for the remainder of prehistory. As indicated by the dramatic decline in the frequency of flakes related to preparation of platforms, the care taken in producing flakes of particular shapes and sizes decreases after phase 2. The trend toward expediency continues up to phases 4-5 (Pavlides 1996). Torrence (1992; compare Pavlides 1996) has inferred that the trend in stone technology incorporating changes in raw material procurement, manufacture and discard can be explained as the result of intensification in land-use. She has hypothesized that in phases 1 and 2 highly
236
R. TORRENCE ET AL.
Fig. 7. Stemmed tools from the Willaumez Peninsula region (e.g. A from site FRL) are found in phases 1 and 2, whereas those found at Yombon (e.g. B site FIF/2) are found only within phase 2.
VOLCANIC DISASTERS IN WEST NEW BRITAIN
237
Fig. 8. Location of test excavations on Garua Island used in the distributional study of artefacts.
mobile groups were exploiting the tropical forest in a relatively casual manner, with plant collecting likely to have been the primary source of food. Consequently, tools were prepared in advance and carried around until the group was able to visit the quarries again. The level of forest management increased slightly between phases 1 and 2 and was dramatically intensified in phase 3 when a shift from primarily low-level management of resources to some gardening seems the best explanation for the change in stone technology. Following the change in subsistence practices, from phase 3 groups became more sedentary and raw material was brought or traded into settlements, stored, and used as needed. In phases 4 and 5 a greater proportion of stone was scrounged locally rather than obtained from source areas. Support for Torrence's hypothesis for a trend toward intensified land use is provided by usewear and residue studies on stone tools recov-
ered from a number of sites in the Willaumez Peninsula. Fullagar (1992) found that during phases 1 and 2 only a few tool-using activities were represented at sites. This pattern contrasted with phases 3-5, for which nearly all activities took place at the sites. Like Torrence (1992), he has interpreted this shift to mean a decline in mobility through time. In his view, earlier sites were occupied for short periods of time during which only a few activities took place, whereas in the later phases a wide range of behaviour took place at long-term residential locations (Fullagar 1992, p. 140). Distributional archaeology and settlement patterns The models derived from studies of stone tool technology concerning changes in subsistence and mobility patterns can be evaluated by
238
R. TORRENCE ET AL.
monitoring the spatial distribution of artefacts across the landscape. In addition, studying the spatial patterning of material provides important information about settlement patterns, and by inference, social organization. The basic assumption behind a distributional approach is that the spatial pattern of artefact discard is a product of how people move around, subsist and interact with themselves and their environment. If there are changes in the intensification of subsistence, as hypothesized above, then the landscape associated with each cultural phase should have a different spread of material. This hypothesis is very difficult to test in most parts of the world because the same landscape has been occupied over very long phases of time and it is not always easy to separate artefacts belonging to different chronological phases, particularly if they are waste by-products from stone tool manufacture and maintenance. As a result of the emplacement of the various Witori and Dakataua tephras in West New Britain, a single, unique landscape can be associated with each prehistoric phase and so distributional archaeology is a very appropriate method for studying patterns of land-use and settlement. In the Yombon region 15 small test excavations (each of 1 m2) have been carried out in four locations within an area extending over a 5km transect. The fieldwork concentrated on locating places where the density of stone artefacts was likely to be high as well as making some attempt to sample a range of local topographic features. The pattern that emerged from this is clear. In phase 1 evidence for lithic manufacture and use is present over approximately 60% of the study area and quarrying is highly focused. The phase 2 pattern extends to incorporate almost the entire study area. At this time there is an expansion in the locations where quarrying and tool manufacture took place and a more even spread of the different stages of tool reduction across the study area. After phase 3 there is a major change in the distribution of stone artefacts characterized by a contraction in the locations where they have been found. Out of 15 excavation trenches, artefacts were recovered from 12; however, only three of these sites yielded more than a handful of finds. This pattern represents a very low density scatter of artefacts amongst which there are a few clusters. This pattern continues into phase 4, at which time there is evidence for the use of the currently occupied villages (Pavlides 1996). The Yombon results are paralleled by excavations undertaken at sites with high concentrations of phase 3 artefacts on Garua Island (Torrence & Webb 1992; Torrence 1993). As in
the interior region, a contraction in the distribution of material was noted with phase 3 sites confined to small, dense concentrations of material, whereas phase 1 and 2 artefacts were spread evenly across all the trenches excavated. Following these initial findings, Torrence & Boyd (1996) have undertaken an explicit programme of research designed to study distributional patterns of artefacts across phase 2, 3-4, and 5 landscapes on Garua Island. In 1996 a series of 30, 1 m2, square test pits were excavated. These were placed along transects running north to south across the island and from the high plateau to the coast (Fig. 8). An attempt was made to sample a wide range of general physiographic zones as well as local topographic features. As the W-K1 tephra on Garua Island has only been found redeposited in drainages, phase 1 was not studied as part of the programme of test pits, although valuable technological information has been retrieved from deep soundings at sites FAP and FAO (Fig. 8). Furthermore, phases 3 and 4 have not been distinguished because of the patchy distribution of the mixed W-K3-4 layer. To compare artefact distributions on Garua Island between the various phases, frequencies of obsidian artefacts per O.lm 3 have been grouped into classes representing low, medium and high abundances. As demonstrated in Table 5 there is a radical shift in the distribution of material between phases 2 and 3 (\2 = 42.411, P< 0.0001). During phase 2 when mobility is thought to have been high, large quantities of material are distributed evenly over the landscape: 64% of the pits have large quantities of material. Phases 3-4 represent a dramatic shift to a highly focused distribution with high abundances noted in only one-quarter of the pits. In phase 5 the distribution of abundance classes maintains the same pattern (x2 = 2.539, P< 0.2809). The Yombon and Garua distributional data support the previous model for a highly mobile Table 5. Garua Island distributional archaeology Obsidian counts
Number of test pits Phase 2 No. %
High (>60) Medium (7-60) Low (0-6)
18 9 1
Total
28
64 32 4
Phases 3-4 No. % 5 6 9 20
25 30 45
Phase 5 No. % 6 11 10 27
22 41 37
VOLCANIC DISASTERS IN WEST NEW BRITAIN form of subsistence and settlement before phase 3. The widespread distribution of material demonstrates that, with the exception of raw material procurement and manufacture, during phases 1 and 2 no place in the landscape was used or occupied for a special reason or longer than any other one. After the W-K2 event, however, particular places were used repeatedly as the foci for tool manufacture and use. This pattern neatly fits Torrence's (1992, 1994) predictions for a decline in mobility. Furthermore, the decrease in the scale of land management can be assumed to signify an increase in manipulation or harnessing of plant resources. Although the concentrated distribution of artefacts in phases 3-5 can be partially explained in terms of changes in subsistence and landuse, there are further patterns in the data that suggest additional factors are involved. First, the paucity of artefacts outside of the sites is unusual even with longer occupations. It seems that, combined with a shift in mobility, there was also a change in the way landscapes were conceived. In other words, the almost total restriction of artefact discard to settlements must have had an important social component. Second, when combined with data from previous years' excavations, the 1996 distributional data from Garua also show that in phases 3-4 the highly focused clusters of material occur almost exclusively on easily defensible, isolated hill and ridge tops that are very close to the coast. As nowhere is very far from the sea, resource use seems an unlikely cause. It seems more likely that the location of residential activities in very specific locations must signify changes in social constructions of landscape and probably indicates the existence of social conflict. In phase 5 the artefact clusters move away from the coast to the high plateau region or on the mountain tops. Defence may again have been a significant factor, but the nature of social conflict must have altered, as the sites are in different settings. Yet again, within a rather similar subsistence pattern to the preceding one, the change in phase 5 indicates a new social conception of how the landscape was to be viewed and used. It is also interesting to note that a highly decorated style of pottery called Lapita makes a sudden appearance in the Willaumez Peninsula in phase 3. The same designs that are found in West New Britain are used on similar pottery at sites spread over an extremely large region of Melanesia and parts of Western Polynesia. It seems likely that the people who recolonized West New Britain after the W-K2 event brought with them a new social system in which shared symbols on pots played a part in maintaining
239
links between groups. Kirch (1988) has suggested that these social networks assisted in reducing risk for small, widely separated communities. Pottery production disappears during phase 4 and does not reoccur in West New Britain, adding weight to the hypothesis that, like the distribution of settlements, the social organization in phase 5 is different from that in the preceding phases. In summary, then, there is a contrast between the long-term trend exhibited by changes in stone technology and subsistence patterns and the nondirectional changes in social organization and settlement location. Studies of stone tool technology and function show a punctuated trend through time beginning in phase 2 but accentuated in phase 3. Beginning in phase 2 a planned and staged strategy for making and using stone tools changes to one in which raw material selection is casual and simple, unworked flakes are produced and used on the spot with little skill or effort. Coupled with Fullagar's (1992) study of tool function and analyses of spatial patterning of finds at Yombon and on Garua Island, we argue that a decrease in mobility and, by implication, in the amount of space exploited for food takes place. In other words, an increase in the intensification in the management of resources and a possible shift from a primary emphasis on hunting and gathering to gardening took place. Although the shift from phase 1 to 2 appears to have been gradual, there was a major acceleration in the process beginning in phase 3. Alongside the long-term trend in subsistence and mobility, non-directional changes in social organization are indicated by the shift from an almost continuous spread of material in phases 1 and 2 to the selection of very few kinds of setting for sites occupied from phase 3. In addition, the specific physical settings chosen for the tightly clustered discard of cultural material in phase 3 and phases 4-5 on Garua Island are different. In explaining the long-term patterns of cultural change in relation to differences in the nature and severity of the volcanic events W-K1-4 and Dk, the contrast between a trend in some forms of behaviour and non-directional change in others must be taken into account.
Rates of reoccupation compared with severity of event On one level, the data suggest that the massive volcanic eruptions that took place between each of the phases had very little effect on long-term patterns of human behaviour. Rather than a
240
R. TORRENCE ET AL.
pattern of growth and crash, we observe a general trend toward intensification in subsistence. Besides this trend, non-directional changes in social organization in phases 3-4 and 5 have been observed. At this stage, the data available indicate that the various societies represented in the different phases were basically egalitarian and roughly similar in character to those known for the recent past (Specht 1980; Goodale 1995). One might therefore conclude that, like the Costa Rican peasants studied by Sheets et al. (1991), the prehistoric societies of West New Britain were very resilient to periodic volcanic activity. We interpret the data differently. Unlike in Costa Rica, at Yombon and the Willaumez Peninsula, there is a record of abandonment after major eruptions. We also argue that the changes in subsistence can best be explained in terms of differing strategies for reoccupation, which we hypothesize are linked to varying forms of social organization whose origins cannot be explained within the framework of this research. In other words, with the exception of the post-Dakataua population, the recolonizers are most likely migrants from another area. Why these groups migrated into West New Britain at this time could be due to a wide range of factors involving cultural change outside our region. At this point we merely note that the nature of social organization of the new settlers was critical to the nature and tempo of recolonization after the severe volcanic events. Environmental changes caused by the emplacement of thick layers of tephra were probably also important factors. In the Arenal area the impacts of the eruptions were 'negligible at distances of more than 20 to 30km' from the volcano (Sheets et al. 1991, p. 461), and possible refuge areas close by have been identified (Sheets et al 1991, p. 462). In contrast, the impact of the W-K1 and W-K2 eruptions, in particular, were extremely widespread, as evidenced by the considerable depth of tephra in the Yombon region (Table 4), which is 90km away from Witori (Fig. 1). A comparison of the nature of reoccupation provides an important means for assessing the effects of volcanism on human behaviour and the role of subsistence and social organization in mitigating these. To begin with, the severity of the eruptions can evaluated using the volume of output (Table 1) and the depth of the tephras preserved at archaeological sites in the regions (Table 4). From most to least severe the events can be ranked as follows: W-K2; Dk (Willaumez Peninsula only); W-K1; and W-K3 and W-4 as equals. In terms of spatial scale of area affected, however, the Dk event was the least serious
because its tephra is restricted to the Willaumez Peninsula (Machida et al. 1996, p. 72). Given the presence of W-K2 tephra at Lolmo cave on the south coast 150km from Witori (Gosden et al. 1994, p. 106) (Fig. 1), large areas of West New Britain were obviously devastated by this eruption. It is questionable whether people living within a considerable radius of Witori could have withstood the effects of deep falls of tephra and the consequent destruction of the vegetation. The timing of reoccupation can be estimated by studying the radiocarbon dates for archaeological layers in the Willaumez Peninsula and Yombon regions (Table 2). However, when all the data are taken together, the ranking of the eruptions provides only a rough indication of the speed with which regions were recolonized. It seems likely that in explaining the longterm prehistory of the two regions, the nature of the human societies themselves was at least as important as the severity of the volcanic event. After the W-K1 event the forest would have regenerated completely before human populations were able to recover and recolonize: the earliest date for human occupation at Yombon is 1000 years and for the Willaumez Peninsula 1600 years after W-K1. After the W-K2 eruption the time for reoccupation is shortened, but the process is reversed, such that people return to the Willaumez Peninsula before they reach the interior: the earliest date for Yombon is roughly 800 years later compared with only about 250 years for the Willaumez Peninsula. The speed of reoccupation in the latter case, which is closer to the Witori volcano, is somewhat puzzling because the W-K2 event was hypothesized by Machida et al. (1996, p. 77) to have been 'the largest in the Holocene' and to have produced three times as much material as did W-K1 (Table 1). Not surprisingly, the much smaller W-K3 and W-K4 events appear to have had very little impact, although the absence of firm dates for these events and the paucity of dates for relevant horizons at archaeological sites make a thorough evaluation difficult. At Yombon there is a gap in the radiocarbon record from several hundred of years before W-K3 until just after W-K4. As with many of the sites on Garua Island (e.g. FAO, FAQ), at Yombon a number of sites were abandoned long before W-K3. Site FSZ on Garua Island, however, has a range of dates running from before W-K3 until the Dk event. It is also interesting that there is no preservation of the W-K3-4 layer at this site. Presumably, people living at FSZ were not much bothered by the tephra falls, did not abandon the site, and continued to sweep the ground
VOLCANIC DISASTERS IN WEST NEW BRITAIN clean as they had done previously (W-K2 tephra has also been swept away). Finally, the Dakataua event, which buried the entire Willaumez Peninsula region with a relatively thick layer of tephra (e.g. Fig. 8), was a high-magnitude event (Table 2). Nevertheless, dates from Garua Island for before and after the event are indistinguishable in terms of the radiocarbon chronology. The small spatial scale of the Dk tephra must be important for understanding the rapid speed of colonization. Nevertheless, unlike the response to the equally thick tephra levels of the W-K1 and W-K2 eruptions, when reoccupation was delayed until at least the forest had regenerated, people recolonizing the Willaumez Peninsula would have done so long before reforestation had proceeded very far. Social and environmental factors We propose a combination of environmental and social changes to explain the lack of fit between the severity of eruptions and the length of abandonment. Both of these factors are subject to random fluctuations so that it cannot be assumed that there would have been continuity with the past after the volcanic disasters we have studied. After each eruption (with the possibly exceptions of W-K3 and W-K4) the landscape of the two study areas would have been radically altered. Not only would the emplacement of relatively thick layers of tephra have significantly changed the topography by filling in swamps and valleys, covering over stone sources, creating coastal plains, etc., but also when combined with high rainfall and the destruction of forest cover, erosion could have greatly accentuated these effects. In contrast, the new soils formed on tephra would have been generally well drained and contained many nutrients (Machida et al. 1996, pp. 76-77). Furthermore, the regeneration of vegetation would have been influenced by random factors such as the potential source of seeds and which seeds were blown in. For all these reasons, it highly likely that the new forest would not have mimicked the previous one. Likewise, random factors would have determined which social group reoccupied the region after each event, especially for W-K1 and W-K2, whose effects would have been extremely widespread. The nature and likelihood of colonizers would have depended on processes taking place outside the study regions and probably beyond the island of New Britain itself. As both social and environmental factors would have been random, it is surprising that there is so much stability in the prehistory of the study regions.
241
During phases 1 and 2 a gradual trend toward reduction in mobility and intensification of land management is indicated by changes in stone tool technology and tool functions. Two possibilities can be suggested: the first hypothesis relies on changes in the environment and the second on alterations in social behaviour. Either the emplacement of the nutrient-rich W-K1 tephra or the regeneration of a different type of forest after the event could have affected the behaviour of a population reoccupying the area. If the new forest had higher levels of productivity, colonizers in phase 2 would have been able to derive a larger amount of food from a smaller area, thereby necessitating the slightly lower pattern of mobility that is exhibited in the data. A second explanation is that different forms of subsistence involving more intensive forest management were introduced by the new population, who had previously adopted this behaviour outside the region. At this stage, the information needed to test these hypotheses, i.e. the nature of the vegetation before and after the W-K1 event, is not yet available, although studies along these lines are in progress (Boyd et al. 1998). The changes in technology and subsistence after W-K2 (phase 3) continue the previously established trend toward intensification, but the character and speed of change is much more radical than before; for example, pottery is introduced, stone tool types disappear, and the whole pattern of artefact discard is transformed. Furthermore, despite having witnessed the most severe event in terms of both magnitude and spatial extent of tephra, the study areas are recolonized in less than half the time it took to replace the population after W-K1. It is difficult to explain this entirely in terms of differences in the nature of the regrowth forest, although this might have been a significant factor and needs to be tested. It seems more likely that a major difference in human behaviour is required to explain the changes after the W-K2 event. The earliest colonists in the Willaumez Peninsula study area after W-K2 occupied low-lying coastal regions and offshore islands (eg. site FYS in Table 2). Complete reoccupation including the inland region of Yombon and the hilltops of Garua Island did not take place for another 400 years. It is possible that because the first people returning to the area subsisted largely on marine resources rather than forest products, forest regeneration was not crucial to reoccupation of the region. In this view, people moved inland later when the forest regenerated. The hypothetical pattern would fit the predictive model of Swadling (1996), who noted that sites with the
242
R. TORRENCE ET AL.
distinctive Lapita pottery are located close to rich reefs. Unfortunately, we do not have good data on the effects of the various eruptions on Willaumez Peninsula region reefs. There are several difficulties with this strictly environmentally determined model. Gosden & Pavlides (1994) have argued that Lapita sites in the Arawe Islands, off the south coast of West New Britain, were not permanently occupied and that plant foods were important to the subsistence pattern: many seeds and nuts have been recovered from waterlogged deposits at the Arawe sites. On the other hand, when people reoccupied Garua Island and Yombon, they did not exploit the forest in the same way as before W-K2 because the stone tool technology and spatial patterning of material are markedly changed. The key factor must have been the social and cultural practices of the group that reoccupied the regions. First, a more intense system of land-use was introduced, which led to decreased mobility, repeated occupation of places, and a change to an expedient stone technology. Second, a different system of social interaction was introduced involving intra-regional exchange of obsidian and pottery (Torrence & Summerhayes 1997) and long-distance links with groups using similar decorative styles. The location of sites in defensible settings suggests that the maintenance of social ties was not without difficulties. Nevertheless, the introduction of a different social system allowed small groups to subsist in an environment that was radically altered and to reoccupy in less time than after the W-K1 event. A subsistence system that depended more highly on cultivation combined with the use of social ties through exchange probably helped dampen the effects of subsequent catastrophes. In contrast to W-K2, W-K3 and W-K4 had very little impact on human occupation of Garua Island and at Yombon, although in the latter case the timing of abandonment, if any, is unclear. Not only were the depths of tephra small (Table 1), as in the case of the Arenal Valley where the volcanic effects were seemingly negligible (Sheets et al. 1991), but also it seems likely that social strategies introduced after W-K2 in phase 3 created a large enough safety net such that the loss of resources could be coped with, perhaps through exchange networks or by seeking temporary refuge with people belonging to the same extended social network. Replacement of gardens with tubers brought in from elsewhere, rather than waiting for wild resources to regenerate, would also have permitted people to recover fairly quickly after a
tephra fall. It therefore seems likely that by the end of phase 3 subsistence had become much more dependent on cultivated resources than in previous phases. The role of gardening and of social relations might also explain reoccupation after the very severe Dk event. The depth of Dk tephra is much greater than for the W-K2 event, and would have had disastrous effects on vegetation and reef resources. On the other hand, the Dk tephra has not been recognized outside the Willaumez Peninsula region, so the spatial scale of the eruption may not have been very great. Groups reoccupied Garua Island immediately after the Dakataua eruption (Table 2) and settled on top of the thick layer of tephra. As the length of abandonment, if any, was very short, it is likely that the same social group returned to Garua Island. Nevertheless, some change was necessitated by the catastrophe because settlements moved from a primarily coastal to a mainly inland location. We hypothesize that the nature of social exchange and conflict had altered, but are unable to offer a convincing explanation for the change at this stage. In conclusion, the data presented here demonstrate that there is no simple and direct correlation between the severity of a volcanic event and human response over the long term. In contrast, the prehistory of human reoccupation in West New Britain illustrates the complex interrelationships between environmental and social factors. As both of these contain random elements, the outcome of individual events cannot be accurately predicted. One can, however, detect the importance of economic and social relations in buffering the effects of disasters. Economies based on cultivated crops are able to recolonize regions more quickly because they are not dependent on forest regeneration and can benefit directly from the advantages of tephra-rich soils. Societies with ties to other groups are able to buffer the effects of disasters either through exchange or by taking refuge. One of the most important implications of our findings is that the responses of relatively simple, egalitarian societies can vary enormously. In the case of the W-K1 and W-K2 events no adaptation took place: both regions were abandoned for relatively long periods of time. In contrast, later groups had slight variations in social systems that allowed them to cope with relatively extreme disasters, such as the W-K3, W-K4 and Dk volcanic eruptions. Clearly, it is difficult to make broad generalizations about the nature of cultural responses to disasters without a full understanding of the particular social and economic systems involved.
VOLCANIC DISASTERS IN WEST NEW BRITAIN Implications for long-term studies On a world scale where the collapse of civilizations receives most attention, volcanic eruptions among the egalitarian groups of West New Britain may appear insignificant because, on one level, like the Costa Rican peasants studied by Sheets et al (1991), a small-scale, tribal society carried on in much the same over roughly 10 000 years. We have presented an opposing view based on archaeological studies of stone tool technology, use and spatial patterning of artefacts. This paper has demonstrated that volcanic eruptions have had a very marked effect on prehistoric life in West New Britain. The punctuated history of severe volcanic events has markedly influenced the way human societies have exploited, reoccupied and adapted to the hazardous environment of West New Britain. Three of the five volcanic events that affected the two study areas of Yombon and the Willaumez Peninsula led to total abandonment of the regions. Surprisingly, the timing and nature of how populations returned cannot be directly linked to the severity of these events. Instead, changes in the nature of society best explain changes in subsistence and settlement patterns. Furthermore, variations in social and economic safety nets were crucial to how well populations adjusted to major volcanic events. To test the hypotheses presented here will require additional fieldwork and analyses, much of which has already begun. Studies of ecofacts, such as phytoliths and starch grains (Boyd et al. 1998; Therin et al. 1999), use-wear and residue analysis of stone tools (e.g. Pavlides 1996), and geomorphological studies of landscape change (Boyd & Torrence 1996) are all well under way and will assist in evaluating the roles of environmental and social factors in the nature of long-term changes in response to volcanic activity in West New Britain. Finally, we wish to stress that in addition to the more popular studies of one-off disasters and cultural collapses, a great deal can be learned about human adaptation to natural hazards when geologists and archaeologists combine their skills to study patterns and trends over long time scales. R.T. holds an Australian Research Council Senior Research Fellowship at the Australian Museum. C.P.'s research has been assisted by a La Trobe University post-graduate scholarship. Fieldwork on Garua Island and at Yombon has been supported by The Australia and Pacific Foundation, Earthwatch, Australian Research Council Major Grants to R.T. and Gosden and Specht and Australian Research Council Small Grant to Gosden and C.P. We thank the West New Britain Provincial Government for permission to carry
243
out fieldwork and for their assistance and encouragement; especially J. Namuno, J. Normu, C. Mare, C. Rukuva and J. Clengme. The National Research Institute and the National Museum have assisted with permits and research visas. Our hosts Garua Plantation, Kimbe Bay Shipping Company, Walindi Plantation, New Tribes Mission Sengseng, and Auwa, and Eliva hamlets at Yombon and Sisisil and Dulago villages provided many forms of invaluable support. This work would not be possible without the hard-working and loyal teams of excavators, among whom R. Mondol, able digger and negotiator in both study areas, deserves special mention and thanks. Comments from D. Gilbertson, B. Boyd and P. White, and responses to a seminar presented at the Department of Anthropology, Northwestern University, were especially helpful in the preparation of the paper. F. Roberts, R. Fullagar and M. Wei prepared the figures.
References BICKNELL, P. 1999. Late Minoan IB marine ware, the marine environment of the Aegean, and the Bronze Age eruption of the Thera volcano. This volume. BINFORD, L. 1979. Organization and formation processes: looking at curated technologies. Journal of Anthropological Research, 35, 255-273. BOYD, B. & TORRENCE, R. 1996. Periodic erosion and human land-use on Garua Island, PNG: a progress report. Tempus, 6, 265-274. , LENTFER, C. & TORRENCE, R. 1998. The archaeological palynology of wet tropics volcanic ash deposits: methodological issues and preliminary results from prehistoric West New Britain, PNG. Journal of Palynology, 22, 213-228. BROOKFIELD, H. & HART, D. 1966. Rainfall in the Tropical Southwest Pacific. Australian National University, Canberra. CHESTER, D. 1993. Volcanoes and Society. Arnold, London. DRIESSEN, P. & MACDONALD, C. F. 2000. The Eruption of the Santorini Volcano and its effect on Minoan Crete. This volume. FULLAGAR, R. 1990. A reconstructed obsidian core from the Talasea excavations. Australian Archaeology, 30, 79-80. 1992. Lithically Lapita. Functional analysis of flaked stone assemblages from West New Britain Province, Papua New Guinea. In: GALIPAUD, J.-C. (ed.) Poterie Lapita et Peuplement. ORSTOM, Noumea, 135-143. GOODALE, J. 1995. To Sing with Pigs is Human. University of Washington Press, Seattle. GOSDEN, C. & C. PAVLIDES, 1994. Are islands insular? Landscape vs. seascape in the case of the Arawe Islands, PNG. Archaeology in Oceania, 29, 162-171. , WEBB, J., MARSHALL, B. & SUMMERHAYES, G. 1994. Lolmo Cave: a mid- to late Holocene site, the Arawe Islands, West New Britain province, Papua New Guinea. Asian Perspectives, 33, 97-119.
244
R. TORRENCE ET AL.
GRAYSON , D. & SHEETS, P. 1979. Volcanic disasters and the archaeological record. In: SHEETS, P. & GRAYSON, D. (eds) Volcanic Activity and Human Ecology. Academic Press, New York, 623-633. KIRCH, P. 1988. Long-distance exchange and island colonisation: the Lapita case. Norwegian Archaeological Review, 21, 103-117. MACHIDA, H., BLONG, R., MORIWAKI, H. et al. 1996. Holocene explosive eruptions of Witori and Dakataua volcanoes in West New Britain, Papua New Guinea and their possible impact on human environment. Quaternary International, 35-36, 65-78. PAVLIDES, C. 1993. New archaeological research at Yombon, West New Britain, Papua New Guinea. Archaeology in Oceania, 28, 55-59. 1996. Transformations in stone: characterising the structure and organisation of Holocene stone tool assemblages in the rainforests of West New Britain, Papua New Guinea. Unpublished paper presented at the 3rd Lapita Conference, Port Villa, Vanuatu, 31 July-6 August 1996. & GOSDEN, C. 1994. 35000-year-old sites in the rainforests of West New Britain, Papua New Guinea. Antiquity, 68, 604-610. PLUNKET, P. & URUNUELA, G. The archaeology of a Plinian eruption of Popocatepetl volcano. This volume. SHEETS, P., HOOPES, J., MELSON, W. et al. 1991. Prehistory and volcanism in the Arenal area, Costa Rica. Journal of Field Archaeology, 18, 445-465. SPECHT, J. 1980. Preliminary report on archaeological research in West New Britain Province 1979-80. Oral History, 8, 1-10. , FULLAGAR, R. & TORRENCE, R. 1991. What was the importance of Lapita pottery at Talasea? Bulletin of the Indo-Pacific Prehistory Association, 11, 281-294. , , & BAKER, N. 1988. Prehistoric obsidian exchange in Melanesia: a perspective from the Talasea sources. Australian Archaeology, 27, 3-16. , LILLEY, I. & NORMU, J. 1981. Radiocarbon dates from West New Britain, Papua New Guinea. Australian Archaeology, 12, 13-15. -1983. More on radiocarbon dates from West New Britain, Papua New Guinea. Australian Archaeology, 16, 92-95. SUMMERHAYES, G. & ALLEN, J. 1993. The transport of Mopir obsidian to Late Pleistocene New Ireland. Archaeology in Oceania, 28, 144-8. , BIRD, J., FULLAGAR, R., GOSDEN, C., SPECHT, J. & TORRENCE, R. 1998. Application of PIXE-PIGME to archaeological analysis of changing patterns of obsidian use in West New Britain, Papua New Guinea. In: SHACKLEY, S. (ed.) Advances in Obsidian Glass Studies. Plenum, New York, 129-158. , , KATSAROS, A. et al. 1993 West New Britain obsidian: production and consumption patterns.
In: FANKHAUSER, B. & BIRD, R. (eds) Archaeometry: Current Australasian Research. Department of Prehistory, Research School of Pacific Studies, Australian National University, Canberra, 57-68. SWADLING, P. 1996. The distribution of Lapita sites and coral reef resources in the Solomon Islands. In: IRWIN, J., DAVIDSON, J., PAWLEY, A. & BROWN, D. (eds) Oceanic Cultural History: Essays in Honour of Roger Green. New Zealand Journal of Archaeology Special Publication, 237-239. THERIN, M., FULLAGAR, R. & TORRENCE, R. 1999. Starch in sediments: a new approach to the study of subsistence and land use in Papua New Guinea. In: GOSDEN, C. & MATHER, J. (eds) The Prehistory of Food. Routledge, London, 438-462. TORRENCE, R. 1992. What is Lapita about obsidian: a view from the Talasea sources. In: GALIPAUD, J.-C. (ed.) Poterie Lapita et Peuplement. ORSTOM, Noumea, 111-126. 1993. Archaeological research on Garua Island, West New Britain Province, Papua New Guinea, June-July 1993. Report submitted to official organizations within Papua New Guinea. 1994. Strategies for moving on in lithic studies. In: CARR, P. (ed.) The Organization of Technology. University of Michigan Press, Ann Arbor, MI, 123-131. & BOYD, B. 1996. Archaeological fieldwork on Garua Island, West New Britain, Papua New Guinea, June—August 1996. Report submitted to official organizations within Papua New Guinea. & SUMMERHAYES, G. 1997. Sociality and the short distance trader: intra-regional obsidian exchange in the Willaumez region, Papua New Guinea. Archaeology in Oceania, 32, 74-84. & WEBB, J. 1992. Report on archaeological research on Garua Island, West New Britain Province, Papua New Guinea, July-August 1992. Report submitted to official organizations within Papua New Guinea. , SPECHT, J. & FULLAGAR, R. 1990. Pompeiis in the Pacific. Australian Natural History, 23, 3-16. , , & BIRD, R. 1992. From Pleistocene to present: obsidian sources in West New Britain, Papua New Guinea. Records of the Australian Museum, Supplement, 15, 83-98. & SUMMERHAYES, G. 1996. Which obsidian is worth it? A view from the West New Britain sources. In: IRWIN, J., DAVIDSON, J., PAWLEY, A. & BROWN, D. (eds) Oceanic Cultural History: Essays in Honour of Roger Green. New Zealand Journal of Archaeology Special Publication, 211-224. WAELKENS, M., SINTURBIN, M., MUCHEZ, P. & PAULISSEN, E. 2000. Archaeological, geomorphological and geological evidences for a major earthquake at Sagalassos (SW Turkey) around the middle of the seventh century AD. This volume.
Tephrochronology of the Brooks River Archaeological District, Katmai National Park and Preserve, Alaska: what can and cannot be done with tephra deposits JAMES R. RIEHLE1, DON. E. DUMOND2, CHARLES E. MEYER3 & JEANNE M. SCHAAF4 1
US Geological Survey, 4200 University Drive, Anchorage, AK 99508, USA (e-mail:
[email protected]) 2 Department of Anthropology, 1218 University of Oregon, Eugene, OR 97403, USA 3 US Geological Survey, 345 Middlefield Road, Menlo Park, CA 94025, USA 4 National Park Service, 2525 Gambell Street, Anchorage, AK 99503 USA Abstract: The Brooks River Archaeological District (BRAD) in Katmai National Park and Preserve is a classical site for the study of early humans in Alaska. Because of proximity to the active Aleutian volcanic arc, there are numerous tephra deposits in the BRAD, which are potentially useful for correlating among sites of archaeological investigations. Microprobe analyses of glass separates show, however, that most of these tephra deposits are heterogeneous mixtures of multiple glass populations. Some glasses are highly similar to pyroclasts of Aniakchak Crater (160km to the south), others are similar to pyroclasts in the nearby Valley of Ten Thousand Smokes, and some are similar to no other tephra samples from the Alaska Peninsula. Moreover, tephra deposits in any one archaeological study site are not always similar to those from nearby sites, indicating inconsistent preservation of these mainly thin, fine-grained deposits. At least 15, late Holocene tephra deposits are inferred at the BRAD. Their heterogeneity is the result of either eruptions of mixed or heterogeneous magmas, like the 1912 Katmai eruption, or secondary mixing of closely succeeding tephra deposits. Because most cannot be reliably distinguished from one another on the basis of megascopic properties, their utility for correlations is limited. At least one deposit can be reliably identified because of its thickness (10cm) and colour stratification. Early humans seem not to have been significantly affected by these tephra falls, which is not surprising in view of the resilience exhibited by both plants and animals following the 1912 Katmai eruption.
The Brooks River Archaeological District deposits found there. ('Tephra'and'ash'are com(BRAD) comprises 56 acres (0.23km2) in the monly used as synonyms to mean particles of wilderness of Katmai National Park and Pre- volcanic origin that settle from an airborne serve on the northern Alaska Peninsula (Fig. 1). cloud. For consistency, we refer here to such Having more than 900 prehistoric surface pits airfall material as tephra and we restrict the term and depressions (Fig. 2), the BRAD qualifies as 'ash' to sand-size volcanic particles (0.0625a National Historic Landmark and is an impor- 2.0mm diameter; 'lapilli' are >2.0mm) regardtant site for the study of early humans in Alaska, less of mode of transport. 'Pyroclast' is a general Nine cultural phases are recognized over the term referring to fragments that formed by ex4500 years of site occupation, chiefly for salmon plosive volcanic activity and that occur in both and caribou harvest, and two older phases occur flow and fall deposits. Tephra deposits are also elsewhere in the region (Dumond 1981). pyroclastic deposits, even though some tephra The BRAD is located near the Aleutian deposits consist exclusively of old rock fragvolcanic arc; nine active volcanoes occur within ments that were incidentally caught in a steam 60km (Figs 1 and 3). Prehistoric human occu- explosion.) Thus, the question was asked durpants of the BRAD frequently experienced ing the earliest archaeological investigations tephra falls, based on the numerous tephra (see Dumond 1979), did prehistoric humans at
From: McGuiRE, W. G., GRIFFITHS, D. R., HANCOCK, .P. L. & STEWART, I. S. (eds) The Archaeology of Geological Catastrophes. Geological Society, London, Special Publications, 171, 245-266. 1-86239-062-2/00/ $15.00 © The Geological Society of London 2000.
246
J. R. RIEHLE ET AL.
Fig. 1. The location of the Brooks River Archaeological District in the Aleutian volcanic arc, northern Alaska Peninsula. Named volcanoes are those to which reference is made in the text.
BRAD suffer adversely from tephra falls? Additionally, would the BRAD tephra deposits provide a convenient field method for correlating among local sites and perhaps even provide a basis for constraining the ages of cultural deposits bounded by tephra deposits? Nowak (1968) concluded that many of the tephra deposits along the Brooks River could not be uniquely identified on the basis of megascopic field characteristics, mineral content and refractive index of the glass. Dumond (1979) reported that trace-element contents of the bulk samples also did not serve to distinguish the various deposits. Consequently, the usefulness of the BRAD tephra deposits as field markers is limited. Some deposits are, however, distinguishable; ash C (see Table 1 and Fig. 5 (below)), for example, can be consistently identified on the basis of its thickness, colour stratification and high stratigraphic position. However, most of the other deposits, especially where isolated with-
out a complete stratigraphic context, are not always unambiguously distinguishable on the basis of bulk and (or) megascopic properties. After several summers of sampling and dating, Dumond (1964) and Nowak (1968) concluded that there are at least nine prehistoric, Holocene tephra deposits in the BRAD. The radiocarbon data that serve to constrain the ages of these nine original deposits have been summarized by Dumond (1979, 1981) (Table 1). To minimize confusion, when referring to these original deposits we hereafter use the terminology 'Dumond B', etc. Our subsequent samples that are reported here are also labelled 'A', 'B', etc., but are prefaced by the site number as a Roman numeral. Present study In 1981, two of us (Riehle and Meyer) began sampling Holocene tephra deposits on the
THE BROOKS RIVER ARCHAEOLOGICAL DISTRICT
247
Fig. 2. More than 900 surface features of archaeological significance, ranging from 4500 years old to early historical, have been identified along the Brooks River in Katmai National Park and Preserve, Alaska. Locations and elevations of emergent lake and river terraces are from Dumond (1981, fig. 2.3). Triangles show localities of tephra samples reported herein.
Alaska Peninsula with the objectives of identifying which of the Aleutian arc volcanoes had been most active during Holocene time, and which widespread deposits might serve as stratigraphic marker horizons. As part of the Alaska Peninsula tephra study, we sought to correlate our samples with the original Dumond tephra deposits. Here, we report analytical results for BRAD samples, attempt to correlate deposits among BRAD archaeological sites, discuss constraints to their use for correlations, and evaluate possible sources of the tephra deposits.
Methods Correlation of tephra samples, that is, identification of all pairs that could be the same deposit from a set of
samples, requires a method for comparing their degree of similarity that is reliable, precise and convenient. The mineral content of a tephra deposit is easily measured, but because of differences in density and mean grain size, the proportions among minerals change as the ash cloud travels downwind (SarnaWojcicki et al. 1981). In addition, the main types of tephra (e.g. andesitic) typically have similar mineral contents. Thus, mineral composition is neither reliable nor typically diagnostic (precise). In contrast to mineralogy, the glass composition does not change from coarse-grained deposits on the flanks of a source volcano to fine-grained distant deposits (Carmichael & MacDonald 1961). The majorelement composition of a small amount of glass can be determined for use in correlation by electron microprobe, a technique that was experimental in the 1960s (Smith & Westgate 1969) but by 1980 had become conventional (e.g. Westgate & Gorton 1981). The technique measures the composition of as little as 10/mi3 of glass. Typically 10-20 such points, either
Fig. 3. Index showing locations of the Katmai-area volcanoes near the Valley of Ten Thousand Smokes. Site 16 is one of the few sites near the Valley where a number of Holocene tephra deposits have been found. Wind frequencies from Wilcox (1959).
THE BROOKS RIVER ARCHAEOLOGICAL DISTRICT
249
Table 1. Original Brooks River tephra deposits and their bounding radiocarbon ages; from Dumond (1979) Deposit label
Bounding 14C ages (in RC years BP)*
Thickness (cm)
Colour (damp)
A B
1912 (Katmai tephra)
20 1-2
white blackish grey
10
banded, shades of olive drab and grey
C D E F G H I J
230 ±80; 335 ±85; 450 ±60; 480 ±90 300 ±75; 670 ±105; 880 ±65; 975 ±120
1175 ±125; 1200 ±70 1225 ±130 1790 ±130; 1895 ±140 2110±350;2140±105 3052 ±250; 3088 ±200 3610 ±85; 3900 ±130 3840 ±130; 3900 ±120 3970 ±440 4240 ±250; 4430 ±110 7360 ±250
1-2
black, grey
1-2
black, grey
3
reddish black
3
yellow
1-2
black, orange, yellow
1-2 3
grey brown
* Only the dates that most closely constrain the age of each deposit are listed here. RC, radiocarbon.
spread over a large piece of glass or of different ash grains, are analysed and averaged. Thus, the microprobe provides the average composition of multiple glass shards, but also returns information about the spread of analytical values about the average so that mixtures of glass populations can be identified. In heterogeneous deposits, this is critical information that is not returned by bulk analytical techniques, regardless of whether they are for major or trace elements. Seven to nine oxides are obtained by microanalysis, which provides a quantitative basis for comparing samples that is both robust and discriminatory: among our 400 samples from the Alaska Peninsula, only 1% of the 79 800 possible pairs are sufficiently similar to be potential correlatives. Of these potential correlatives, about 10% are known from other evidence to be clearly inadmissible; for example, lapilli samples that are highly similar to one another, but that are from the flanks of two widely separated volcanoes. Equally important, only a small amount of sample is required for microanalysis; in theory, as little as a dozen small shards. These advantages of the technique permit reliable and precise comparison of thin, fine-grained tephra deposits with coarse deposits. The degree of similarity between two samples can be expressed as the similarity coefficient (s.c.), which is the average ratio of each pair of major-element concentrations in the two analyses (Borchardt et al. 1972). The lesser of the two concentrations is the numerator, so
the s.c. does not exceed one. Expressing the degree of similarity of two samples as a single value facilitates comparison of multiple sample pairs. Because analytical uncertainty increases as concentration decreases, we exclude from the calculation of the s.c. any oxide that is less than an arbitrary 0.40wt%. This is so that one pair of oxides having a large analytical uncertainty, such as MnO or TiO2, cannot unduly influence the average value. Each average analysis is normalized to 100% before calculating the s.c., so that glasses having different degrees of hydration or differing volatile contents or microvesicles can still be compared on the basis of their major-element contents (see footnote to Table 2). An s.c. value of 1.00 is rarely encountered even for samples of the same tephra deposit, because of both inherent variability and analytical uncertainty. Replicate analyses of sample splits show that values of 0.96-0.98 are typical of pairs of samples of the same deposit. In rare cases of unusually heterogeneous deposits, values as low as 0.93 have been reported from the same deposit (Riehle et al. 1992). As previously stated, high values of s.c. are occasionally encountered for samples that cannot possibly correlate as the same tephra deposit. In such cases, trace-element contents might provide a basis for distinguishing the deposits, but as for the major elements, only if glass separates are analysed. Glass may compose as little as 10% of an arc tephra deposit,
Table 2. Major-element composition of glass in Holocene tephra deposits from Site 16 in the Valley of Ten Thousand Smokes, Katmai National Park and Preserve, Alaska
Al
A2
Na20 MgO A1203 Si02 K2O CaO TiO2 MnO FeOT
4.04(6.5) 0.91(35) 14.5(10) 69.1(4.6) 2.50(0.8) 3.08(24) 0.71 n.d. 3.31(23)
3.94(9.2) 3.80 3.72(6.2) 4.15(4.8) 0.60(37) 1.45 1.12(24) 0.72(3.9) 12.4(9.3) 13.9 12.9(0.5) 13.6(1.5) 70.9(0.8) 67.5 65.5(3.6) 72.7(0.4) 3.28(0.4) 1.82 2.84(8.1) 2.34(3.6) 1.54(18) 4.30 3.40(15) 2.,71(6.4) 0.69 0.65 1.07 0.62 n.d. n.d. n .d. n.d. 2.97(41) 5.14 5.52(7.2) 2.,82(5.5)
Total c/o/h
98.2(3) 44/56/0
96.3(2)
A3
Bl
B2
98.6(1) 96.1(2) 41/57/2
99.,7(2) -
F3
Gl
Na20 MgO A1203 Si02 K20 CaO TiO2 MnO FeOT
4.34 1.55 14.8 64.2 1.49 4.83 1.03 n.d. 6.28
3.75(0.1) 3.45(3.4) 3.55(13) 2.11 2.98(6.6) 0.32(31) 0.72(2.0) 0.07 16.0(1.0) 11.1 12.1(6.1) 13.3(0.1) 57.4(1.5) 73.0(3.5) 70.7(1.1) 73.6 1.48(10) 2.40(2.4) 4.93 3.32(12) 7.01(9.1) 1.74(11) 2.85(1.6) 0.45 1.35 0.63 0.49 0.45 n.d. n.d. n.d. n.d. 8.29(5.0) 2.32(4.6) 3.05(1.3) 1.45
Total c/o/h
98.5 -
98.2(3) (60/40/0)
G2
96.7(2) -
G3
97.2(2) -
G4
B4
C
Dl
D2
4.60(0.6) 0.38(7.4) 14.0(1.5) 71.7(1.6) 3.25(3.0) 1.53(5.6) 0.41 n.d. 2.25(6.0)
4.50 2.34 15.6 61.3 1.69 5.37 1.32 n.d. 6.44
3.82(1.1) 0.44(98) 13.1(1.6) 71.1(3.4) 2.92(27) 2.33(37) 0.65 n.d. 2.55(16)
4.10(9.4) 0.29(25) 13.1(6.3) 74.1(0.5) 2.53(3.2) 2.04(22) 0.52 n.d. 1.32(86)
3.13(6.8) 3.97 4.16 0.23(9.2) 2.91 1.03 12.1(2.9) 16.4 15.4 73.5(1.2) 59.8 67.4 4.12(2.4) 1.70 2.63 0.82(25) 6.72 3.84 0.54 1.11 0.49 n.d. n.d. n.d. 1.91(25) 7.37 3.69
98.1(2)
98.5(1) 96.9(2) 40/60/0
97.9(3) (40/60/1)
96.4(2) -
G5
H
4.78 2.13 16.8 63.1 2.04 5.35 0.97 n.d. 5.37
3.56(3.5) 3.66(5.8) 3.58(1.0) 0.33(21) 0.47(30) 1.34(16) 12.3(1.7) 12.7(5.1) 12.5(13) 75.7(1.3) 75.7(1.8) 72.8(2.8) 3.13(10) 1.79(6.7) 1.65(4.3) 1.61(14) 2.38(18) 3.97(0.2) 0.35 0.37 0.40 n.d. n.d. n.d. 1.87(8.2) 1.86(22) 2.81(33)
94.2(1) 100.3(1) 98.8(8) 35/70/0 -
Jl
98.9(7) 53/45/2
J2
99.0(2) -
D3
D4
B3
El
E2
Fl
F2
3.66(9.1) 0.51(33) 13.3(3.6) 73.1(1.9) 2.91(8.9) 2.22(7.1) 0.52 n.d. 2.49(8.2)
3.03 0.21 11.6 74.3 3.08 1.,27 0,.29 n.d. 1.46
3.63(7.5) 0.26(55) 12.0(3.9) 75.6(1.8) 3.02(4.0) 1.38(39) 0.28 n.d. 1.58(28)
3.25(13) 0.12(41) 11.9(8.6) 75.8(2.7) 4.30(12) 0.34(23) 0.34 n.d. 1.62(17)
100.0(1)98.6(1) 98.7(6) 44/54/2 -
Kl
.K2
K3
97.7(2) -
95..2(1) 97.7(3) -
.LI
L2
3.50 4.32(13) 2.91 (32) 4.21(17) 4.02(14) 0.64 0.63(153) 1.09(30) 0.09(16) 0.27(40) 11.1 13.6(9.2) 14.4(9.6) 10.9(3.2) 12.5(6.7) 74.9(1.8) 74.0(1.7) 71.7(2.8) 66.0(2.4) 74.2 2.59(7.9) 3.84(0.7) 2.12(4 .4) 2.20(3.8) 2.70 1.40 2.81(12) 3.52(15) 1.13(72) 1.73(33) 0.48 0.54 0.85 0.59 0.66 n.d. n.d. n.d. n.d. n.d. 4.82(27) 2.46 1.82(25) 1.75(28) 2.36(11) 99.0(5) (40/60/0)
95.2(2) -
97.4(4) -
97.2(4) 43/56/1
96.5 -
L3
4.14 0.17 13.2 73.3 1.87 2.05 0.13 n.d. 0.91 95.7 -
Samples are arranged in stratigraphic order from youngest to oldest. For averages of two or more points, oxide concentrations measured by electron microprobe are followed by Icr analytical uncertainty (expressed as per cent of the reported value). Analytical conditions: 15kV, 0.01/xA sample current, and defocused beam and 15s count times to minimize loss of alkalis. Standards: glasses except for Mn2O3 and Ti-amphibole; internal standard rhyolitic glass RLS132. Difference from 100% is due to a combination of analytical imprecision, presence of elements not analysed, such as water and CO2, and potential occurrence of microvesicles in the analysed glass volume. Mafic-phenocryst proportions reported as clinopyroxene/orthopyroxene/hornblende (c/o/h) are based on point counts of 100-150 grains; values in parentheses are estimates, and a blank indicates no data. Dashes indicate an additional glass component that has the same mafic-phenocryst content as the first component. Plagioclase is ubiquitous and dominant and its abundance was not measured, n.d., not determined.
THE BROOKS RIVER ARCHAEOLOGICAL DISTRICT in which case so much bulk sample is required that comparison with thinly bedded deposits is precluded. Alternatively, the compositions of magnetite and ilmenite grains have provided a basis for distinguishing some deposits that have highly similar major-element contents (Downes 1985; Riehle et al. 1998). Only a few grains are needed for microanalysis. Lastly, despite the shortcomings of mineral content as a basis for correlating tephras, we do measure the proportions among the mafic minerals (clinopyroxene, orthopyroxene and amphibole) in many of our samples. This is because these minerals have roughly similar densities (3.35±0.10gcm~ 3 ). Thus, except where there is a large difference in the average grain size of these mineral types, their proportions to one another are more likely to remain constant during transport in the tephra cloud than are the proportions among plagioclase, glass and mafic minerals. Because of the imprecision of the method, however, we require a large difference in mafic-phenocryst proportions before precluding a correlation based solely on mafic minerals, especially where comparing samples of widely differing mean grain size. Mafic minerals are of little value for comparing deposits that are mixtures of multiple tephra falls.
Tephra sources in the Katmai region Likely sources of the BRAD tephra deposits include the volcanoes in the Katmai region, so we first discuss the available data on proximal tephra deposits of these volcanoes. Chemical analyses are available for the 1912 Novarupta eruption (Fierstein & Hildreth 1992); the Lethe tephra, a late Pleistocene regional marker deposit of uncertain source in the Katmai area (Pinney & Beget 1991); and our studies of Holocene tephra deposits. (Pinney's (1993) thesis includes glass compositions of latest Quaternary pyroclastic deposits in the Windy Creek tributary of the VTTS. However, only two of these, the latest Pleistocene Lethe tephra and Windy Creek ashflow deposit, have inferred local sources. Two other Holocene deposits are correlated with sources outside the VTTS region.) The 1912 eruption was remarkable for its size and for the distance (10km) between Mount Katmai, which collapsed following withdrawal of magma from the subsurface chamber, and Novarupta dome, the site of the actual eruption (Curtis 1968; Hildreth 1983; see Fig. 3). The 1912 eruption was also remarkable for the compositional variability of the magma. Three types of magma were erupted in a 60 h period: light-coloured rhyolite (77% SiO2 in bulk), dacite (70-72% SiO2), and dark andesite (60-63% SiO2). We sampled 11 tephra deposits in an upper Holocene section of loess and peat on the northeastern side of the Valley of Ten Thousand Smokes (VTTS) (Site 16, Fig. 3). Glass in all
251
but two of the deposits is highly heterogeneous (Table 2), in some cases containing as many as four different glass compositions (populations) in one deposit. The basic possible causes of such heterogeneity are: contamination (secondary mixing) after deposition; eruption of multiple magmas as in 1912; or inclusion in an eruption of material from a pre-existing lava dome, in which heterogeneity is the result of variable degrees of devitrification after dome emplacement. Secondary mixing of different tephra deposits should be identifiable by the appearance of older deposits as a component in younger deposits. However, of the 465 possible pairs among the 31 components in Site 16 deposits, only one pair has a similarity coefficient that is greater than 0.95. Thus, the heterogeneity that characterizes these Katmai deposits is not due simply to mixing of older deposits with younger deposits. Devitrification of the Novarupta rhyolitic dome after its emplacement in 1912 produced quartz, sodium-rich plagioclase and a Ca-Ferich clinopyroxene; potassium feldspar occurs only rarely in the devitrification mineral assemblage (Wiesneth & Eichelberger 1996) even though the rhyolitic glass contains 3% K2O. Thus, MgO and K2O have preferentially increased in the Novarupta glass as a result of devitrification whereas Na2O has decreased. Nearly all high-silica, low-MgO glasses from Site 16 show an increase of Na2O/K2O as MgO increases (Fig. 4a), indicating that these rhyolitic glasses are unlikely to be the result of devitrification of a single parent glass. The exception is two shards in 16-L. Deposit 16-L is also the only deposit at Site 16 to show an increase of (Na2O + K2O)/Al2O3 with MgO (Fig. 4b). This is relevant because the ratio of (Na2O 4- K2O)/ A12O3 in plagioclase is about the same as that in the rhyolitic glass, whereas the ratio in potassium-rich feldspar is twice that in the glass. Thus, variation of (Na2O + K2O)/A12O3 with MgO in these two 16-L glass shards is consistent with removal of Na-rich plagioclase but not K-rich feldspar, whereas variation in the other Site 16 samples must be due to other causes. To identify sources of the Site 16 deposits, proximal samples from each of the Katmai volcanoes are needed for comparison. We have, however, sampled proximal deposits from only two VTTS vents: Mount Griggs and Novarupta dome. Samples of proximal tephra deposits of the other volcanoes are not available, partly because of the harsh climate (Katmai winds scour even lapilli-sized deposits) and partly because of the lack of large, prehistoric tephraforming eruptions at these volcanoes. But we
252
J. R. RIEHLE ET AL.
occurred during the 1912 eruption, and (or) of closely succeeding eruptions of different magmas. Because correlatives of the Site 16 deposits have not been identified elsewhere on the Alaska Peninsula, it is unlikely that these are fartravelled deposits of distant volcanoes. And if not, these deposits probably document Holocene tephra-forming eruptions of the Katmai volcanoes that were small in volume.
Original Brooks River tephra samples
Fig. 4. Compositions of individual glass shards in Site 16 deposits (deposit C is not plotted because it has a single, homogeneous glass). The spread of glass compositions may be the result of variable devitrification of a dome, (a) If the hypothesized devitrification is like that in the rhyolitic dome of Novarupta, then both potassium and magnesium should increase in the residual glass relative to sodium, which is removed in sodium-rich plagioclase. Only one deposit, 16L, shows such a relationship, and only for two low-MgO shards (encircled), (b) These two points are also unique in the sample set for having an increase in the ratio of alkalis to aluminium with increasing MgO, which is permissive evidence for the removal of plagioclase but not potassium-rich feldspar.
can compare Site 16 deposits with the Lethe tephra, Novarupta dome, and the three kinds of 1912 tephra. Clearly, the Holocene deposits at Site 16 do not actually correlate with these Pleistocene or historical deposits; instead, the goal is to look for similarities that suggest the same source vent. Results (Table 3) indicate that the Site 16 deposits are dissimilar to nearly all other samples from adjacent sites to the north and south of the Katmai region, although some glass components are similar to samples of Griggs Volcano or to the Lethe or 1912 deposits. The combination of these two findings implies that most of the thin, fine-grained tephra deposits at Site 16 originated at one of the Katmai volcanoes. Additional work on tephra deposits around the VTTS is needed to identify specific sources of the deposits. For now, we tentatively conclude that the unusual degree of heterogeneity of these deposits is most likely the result of eruption of mixed or multiple magmas like that which
With time and by depletion of samples for other analyses, no samples of the original Dumond deposits remained for our microanalysis. So in 1997 we resampled BRAD deposits mainly at three sites: two on the 22 m terrace east of Brooks Lake and a third along Brooks River. The best exposure was at Site I the greatest number of well-preserved tephras were at Site II, and Dumond C was best exposed at Site III (Figs 2 and 5). We are confident that our sample III-C correlates with (that is, is the same as) Dumond C because of its thickness and distinctive colour, but we are less certain that any of our other samples necessarily correlate with specific Dumond deposits. Tephra was provisionally identified in the field by the megascopic recognition of a distinct layer of uniform grain size or, for fine-grained deposits, by a distinctively coloured layer. Except for III-C, all samples of pre-1912 BRAD tephra deposits are mixtures of volcanic grains, organic fragments, and subangular to subrounded mineral and rock fragments. Juvenile volcanic grains (those having glass) include dense vitrophyre, scoria and vesicular pumice. The median size of the volcanic grains in most pre-1912 BRAD deposits is fine ash (0.0625-0.25 mm diameter); a few are medium ash. Sample III-C is also fine ash but is typically 10cm thick, and it is likely that the heterogeneity of the other deposits, none of which is more than 2 cm thick, is the result of secondary mixing with adjacent deposits. Even for our contaminated samples, however, glass could be effectively separated from organic and mineral fragments by crushing and sieving to fine ash size, then immersing the grains in a high-density liquid (see SteenMclntyre 1977). Glasses in the resampled BRAD deposits are as heterogeneous as the deposits at Site 16 in the VTTS (Table 4). Deposits II-F and II-J, for example, have seven glasses. Some of the glass components are clearly contamination from underlying or overlying tephra deposits: II-B appears to be a mixture of overlying 1912 tephra
THE BROOKS RIVER ARCHAEOLOGICAL DISTRICT
253
Table 3. Comparison of glass compositions of Site 16 tephra deposits with other tephra deposits from the Alaska Peninsula and with Novarupta dome and proximal tephra deposits from Griggs Volcano (Riehle unpub. data), the Lethe tephra (Pinney & Beget 1991), and 1912 Katmai tephra deposits (Fier stein & Hildreth, 1992) 016A: 3 components are all low-silica rhyolites that differ widely in MgO and K2O contents but have similar Na2O contents. Al is similar to a component in Brooks River deposits and so probably has a local source. A2 is marginally similar to a homogeneous lapillus from the summit of Griggs Volcano. A3 is not similar to any other Alaska Peninsula samples, but consists of only a single analytical point. All 3 components are provisionally assigned a source at Griggs Volcano. 016B: 4 glass components ranging from 62 to 73% SiO2. Bl and B2 are dissimilar to all other Alaska Peninsula samples. B3 and B4 are each similar to 3 samples of lower to middle Holocene deposits of Aniakchak caldera. However, the Aniakchak samples are themselves similar to a large number of other samples (indicating a widespread occurrence), and B3 and B4 are not even marginally similar to those other samples; moreover, 016B is unlikely to be as old as mid-Holocene time. Thus, 016B is interpreted to be a heterogeneous deposit of an uncertain, Katmai-region source. 016C: a single, heterogeneous glass component. Marginally similar to only a proximal deposit of Griggs Volcano, and to the dominant component of 016E. Probably an upper Holocene deposit of Griggs Volcano. 016D: 4 glass components, 2 of which are rhyolitic and the other 2 (less silicic) are each only a single analytical point. Dl is marginally similar to only one other, Katmai-area deposit. D2 is marginally similar to only 16K2. D3 is similar to only Dumond Bl. D4 is marginally similar (coincidentally so?) to several samples of upper Holocene Aniakchak deposits from 150km to the south. We conclude most likely an uncertain, Katmai-region source. 016E: 2 glass components, both rhyolitic. The dominant component is marginally similar to each of Dumond El, 16C and a middle Holocene deposit 80km north of Katmai that is similar to no other sample. The dominant component is also marginally similar in glass composition, but not mafic-phenocryst ratio, to a component of the middle Holocene, Aniakchak caldera-forming deposits. The minor component is marginally similar to 2 other Katmai-region deposits that are dissimilar to any other deposits, to a Brooks River component (VI-B2), and to 16F1. Clearly, the deposit originated at one of the Katmai volcanoes; because of the marginal similarity to 16C, Griggs Volcano is a likely source. 016F: 3 glass components, 2 of which are rhyolitic and the third, a single analytical point, is andesitic. The dominant component is highly similar to components in 2 other, middle Holocene deposits on the northern Alaska Peninsula and to glass of the dacitic component of the 1912 Katmai tephra. It is also marginally similar to 16H and 16E2, and to VI-B2 from the Brooks River. The second component differs from the first component in having unusually low MgO and CaO and high K2O and, like the third component, is dissimilar to all other samples. An uncertain, Katmai-region source. 016G: 5 glass components that differ significantly in SiO2, MgO, and K2O contents. The dominant component, andesitic in composition, comprises 3 analytical points; 2 components are 2 points; and the fifth component is only one point. The dominant component is similar to a number of other andesitic glasses from the Alaska Peninsula, but which include a range of ages and inferred sources and so are of no use in establishing a source. The dominant component is also similar to Dumond Dl. The second component is marginally similar to 3 deposits to the south that are tentatively assigned to Chiginagak Volcano, as well as to a proximal deposit of Griggs Volcano. The third component is marginally similar to Brooks River deposits VI-C1 and VI-B1. The fourth component, only a single analytical point, is marginally similar to Brooks River deposit VII-E and to another Katmai-region deposit. The fifth component is unique because of its extremely low MgO, CaO and Na2O contents together with high K2O. Little can be concluded about the source(s) of this heterogeneous deposit with certainty, except that some components seem to be unique to the Katmai region. 016H: relative to other Site 16 deposits, a homogeneous deposit. The glass is similar to samples from the Brooks River (IV-D, VI-B2) and from 30km to the south of Site 16. Although the deposit is also marginally similar to three samples assigned to Chiginagak Volcano, it probably has a Katmai-area source. 016J: 2 glass components, both rhyolitic. The main component is highly similar to the main component of a middle Holocene deposit 40km north of the VTTS. The second component is unlike any other sample. Apparently a unique, middle Holocene tephra deposit that originated from a Katmai-area volcano. 016K: 3 glass components, all rhyolitic. The first component is marginally similar to another middle Holocene, Katmai-area deposit. The other 2 components are not similar to any other sample. Like 16J, probably a unique, Katmai-area tephra deposit. 016L: 3 glass components, 2 of which are each only one analytical point. The main component is dacitic and is marginally similar to several other samples, including some assigned to Veniaminof Volcano and to a component of the middle Holocene, caldera-forming eruption of Aniakchak Crater. The main component is also marginally similar to Brooks River sample IV-H, which cannot be as old as mid-Holocene time. Probably the main component had an unknown source in the Katmai region and is only coincidentally similar to deposits of Aniakchak and Veniaminof volcanoes. 'Similar' means a similarity coefficient (s.c.) of >0.95; 'marginally similar' is 0.93-0.94.
THE BROOKS RIVER ARCHAEOLOGICAL DISTRICT and the underlying deposit II-C. Probably most of the contamination occurred by biological activity or frost churning, as the amount of inter-tephra sediment between most of the tephra deposits at our sample sites is minimal (Fig. 5). To try to confirm resampling of Dumond B at a fourth locality we sampled a deposit between the base of 1912 tephra and the obvious correlative of Dumond C (as well as the correlative of C 'Ash C'). The sample 'Ash B(?)' is highly similar to III-C1 (correlative of Dumond C), whereas the upper part of 'Ash C' is similar to III-C3 and the base of 'Ash C' is highly similar to III-C2 and III-C3. Thus, as at Site II, 'Ash B(?)' at the fourth site is either the reworked top of the underlying Ash C or is a closely succeeding deposit of the same magma as formed Ash C. Because the presence of a separate tephra deposit between Dumond C and the 1912 Katmai tephra has been unequivocally shown at some archaeological sites (e.g. Dumond 1981, fig. 6.23; Harritt 1988, table 1), we do not dispute the existence of Dumond B. However, we concur with Nowak (1968, p. 33) that Dumond B is 'rarely distinct enough to be easily recognized'. Most of the identifiable components in the eight analysed BRAD deposits are highly similar to pyroclasts from Aniakchak Crater, 160km southwest of the BRAD (Fig. 1). Aniakchak has had the largest number of major tephra-forming eruptions during Holocene time of any volcano we have studied on the Alaska Peninsula: at a minimum, six during late Holocene time, nine during a period of intense activity that culminated with caldera formation at 3460 a BP (Miller & Smith 1987), four to six just before the onset of the caldera-forming activity, and several more during early Holocene time. Prevailing winds to 30 000 ft above the Alaska Peninsula are northeasterly (Fig. 3), so an abundance of Aniakchak deposits in the BRAD is not surprising. BRAD deposits II-D, II-I and II-J have major components that cannot be assigned to specific sources, and other deposits have minor components of unknown sources (Table 5). These unknown components may have originated at the Katmai-group volcanoes: if their sources
255
were more distant volcanoes, at least some coarse-grained correlatives should have been identified nearer the sources. In any case, few of these deposits of unknown sources are similar to any Site 16 glasses, which suggests that Site 16 deposits represent yet additional eruptions of Katmai volcanoes. It is readily understandable why Nowak (1968) and Dumond (1979) concluded that, with the exception of Dumond C (and, of course, the 1912 tephra), the BRAD tephra deposits are not reliably distinguishable from one another. Each is a mixture of multiple components. Whatever distinguishing characteristics the individual components may have had initially, they have surely lost by mixing.
Correlations of tephra deposits among BRAD sites A requirement of a National Historic Landmark is that, to preclude loss or damage of resources, all ground disturbances be preceded by competent archaeological assessment. Thus, as Katmai Park managers strive to maintain adequate visitor facilities in the face of growing Park visitation, the frequency of archaeological assessment of small areas has increased. Such limited compliance assessments do not have the scope of a larger investigation that might provide a critical stratigraphic context or that might yield datable organic materials. The ability to identify a specific tephra deposit, especially if its age were known, would be highly useful in these compliance studies. To test the potential applicability of tephra correlations to archaeological research at the BRAD, we sampled and analysed 10 'unknown' tephra deposits from four study sites (Fig. 6). One is the drainfield site east of Brooks Lake on a 19m beach-ridge terrace (site IV, Fig. 2). The other three sites are located within 100m of one another, on emergent 13.5-14.5m beach ridges of Naknek Lake near Ranger Headquarters (Sites V, VI and VII, Fig. 2). Analytical results for these samples are not reported because the question is only whether these samples correlate with other BRAD
Fig. 5. Holocene deposits at three sites (see Fig. 2 for locations) in the Brooks River Archaeological District, (a) Measured section at Site II on a c. 7000 year old, 22m emergent terrace. Capital letters to the right refer to analysed samples discussed in this paper, (b) Holocene tephra deposits at Site I (where they are better exposed than at Site II). The ofT-white, sandy material beneath modern sod at the top is 20cm of 1912 Katmai tephra. Labels are tephra deposits 'C, D, F, G, I, J'. (c) Exposure of Ash C (beneath the thumb) along the Brooks River (Site III). The stratification of Ash C into a pale yellow-grey base, a greenish grey centre and a greyish brown top should be noted. The three strata were separately sampled as Cl, C2 and C3 (Table 4). One-half metre of poorly sorted material beneath Ash C is cultural refuse of a pre-Ash C occupation; Ash C is overlain by a thin soil, which in turn is overlain by post-Ash-C refuse of a younger occupation. The 1912 Katmai tephra extends from top of fingers to top of bank.
Table 4. Major-element compositions of glass in tephra deposits from Sites II (all except C) and III (C) in the Brooks River Archaeological District
Bl
B2
B3
B4
B5
B6
CM
Cl-2
Cl-3
Cl-4
Na2O MgO A12O3 Si02 K2O CaO Ti02 MnO FeOT
3.98(2.8) 0.26(12) 12.2(2.0) 75.7(1.0) 2.85(2.6) 1.26(7.7) 0.32(18) 0.05 1.61(6.4)
3.83(3.7) 0.14(24) 11.9(2.1) 76.7(1.0) 3.20(2.8) 0.89(20) 0.20(42) 0.04 1.17(11)
3.92 0.35 12.6 74.9 2.72 1.53 0.33 0.04 1.78
4.67(3.5) 1.09(5.4) 14.7(0.8) 66.6(0.3) 2.84(2.0) 3.12(0.8) 0.97(4.5) 0.19 4.69(4.8)
4.49 2.09 14.1 62.4 2.26 4.18 1.23 0.23 7.17
4.26 2.74 15.4 58.6 1.55 6.34 1.07 0.18 7.40
4.11(2.0) 2.74(4.9) 16.2(1.8) 58.7(1.1) 1.67(4.3) 6.18(3.6) 1.18(11) 0.19 7.56(4.7)
4.45(3.5) 2.29(16) 15.5(3.6) 61.1(0.8) 2.00(8.6) 4.92(11) 1.05(22) 0.19 7.22(4.4)
4.42(4.8) 1.69(6.0) 15.3(3.1) 63.3(1.4) 2.28(4.3) 4.08(3.9) 1.01(14) 0.15 5.91(11)
4.34(7.5) 0.89(2.7) 15.0(4.2) 67.4(1.7) 3.11(16) 2.52(24) 0.73(16) 0.16 3.67(10)
4.05(3.7) 2.77(8.4) 16.2(2.2) 58.1(2.0) 1.59(7.5) 6.37(5.9) 1.11(13) 0.19 7.70(6.1)
Total
98.2(16)
98.1(5)
98.2
98.9(3)
98.2
97.5
98.5(3)
98.7(5)
98.1(5)
97.8(4)
98.1(18)
C2-2
C3-1
C3-2
C3-3
C3-4
Dl
Na2O MgO A1203 Si02 K2O CaO Ti02 MnO FeOT
4.62 1.23 16.0 66.2 2.67 3.38 0.71 0.17 4.51
4.15(3.9) 2.74(8.2) 16.3(1.4) 58.8(1.8) 1.59(8.9) 6.24(5.4) 1 .12(6.3) 0.19 7.53(8.4)
4.83(2.7) 0.81(6.6) 15.2(1.8) 68.8(0.5) 2.69(3.0) 2.54(5.3) 0.70(13) 0.16 3.36(5.6)
3.88 2.92 15.7 55.7 1.47 6.6 1.20 0.20 8.19
4.74 1.85 15.7 63.0 1.82 4.55 0.89 0.21 5.67
4.61(10) 0.55(19) 14.4(4.8) 70.7(2.4) 2.70(9.5) 2.00(18) 0.53(19) 0.09 2.48(13)
Total
99.6
98.7(9)
99.1(7)
95.9
98.4
98.1(15)
D2 4.07 0.16 12.5 78.3 3.11 1.07 0.26 0.06 1.28 100.8
C2-1
D3
D4
El
E2
E3
E4
4.48 1.64 15.6 61.4 2.18 4.43 1.19 0.20 7.22
3.92 0.34 12.7 74.3 2.81 1.73 0.43 0.06 2.09
4.59(11) 0.57(20) 14.2(4.8) 70.9(1.2) 2.46(11) 2.23(17) 0.56(21) 0.10 2.62(12)
4.71(4.5) 1.10(5.6) 15.8(0.4) 67.2(0.5) 2.45(3.8) 3.21(0.5) 0.76(9.4) 0.16 3.81(1.4)
4.27(3.7) 2.85(5.0) 16.4(0.6) 58.1(1.5) 1.50(1.7) 6.27(3.7) 1.34(8.0) 0.21 7.02(6.1)
4.56 1.36 15.4 65.1 2.55 3.58 0.98 0.20 4.89
98.4
98.4
98.2(7)
99.2(4)
98.0(3)
98.7
E5
Fl
F2
F3
F4
F5
F6
¥1
Gl
G2
G3
HI
H2
Na2O MgO A1203 Si02 K20 CaO Ti02 MnO FeOT
3.84 0.38 12.8 74.6 2.85 1.80 0.45 0.05 2.00
4.17(3.6) 2.70(6.6) 15.8(1.8) 57.2(1.7) 1.58(4.2) 6.11(5.8) 1.35(5.3) 0.22 8.55(7.0)
4.31(5.4) 2.16(13) 16.3(6.6) 59.6(0.1) 1.79(14) 5.56(9.3) 1.19(8.7) 0.18 7.26(9.0)
4.27(7.0) 1.89(13) 15.1(4.0) 61.0(1.0) 2.18(2.8) 4.69(3.7) 1.31(7.1) 0.16 7.70(12)
4.35 1.22 14.5 63.6 2.69 3.61 1.15 0.15 6.65
4.16 1.48 13.5 69.4 2.15 3.26 0.53 0.10 3.62
4.02(2.6) 0.58(6.2) 13.5(0.7) 71.3(1.4) 2.41 (0.7) 2.37(4.9) 0.50(20) 0.06 2.68(6.2)
3.57 0.40 12.9 74.5 2.85 1.76 0.41 0.05 1.90
4.76(2.5) 0.39(9.7) 14.5(1.4) 70.8(0.7) 3.25(5.3) 1.54(5.1) 0.38(23) 0.12 2.42(3.6)
3.76(3.6) 0.30(16) 12.5(1.4) 75.1(0.6) 2.73(19) 1.52(15) 0.32(21) 0.03 1.65(8.1)
4.79 1.20 15.3 66.3 2.55 3.20 1.00 0.15 4.45
4.66(5.7) 1.44(22) 15.5(1.2) 63.9(1.6) 2.40(7.6) 3.84(3.7) 1.04(10) 0.15 4.98(12)
4.89 0.78 14.7 69.2 3.01 2.20 0.76 0.15 3.46
Total
98.8
97.7(4)
98.4(3)
98.3(4)
97.9
98.3
97.4(3)
98.3
98.2(7)
97.9(5)
98.9
97.9(7)99.2
H3
H4
11
12
13
Jl
J2
J3
J4
J5
J6
J7
Na2O MgO A1203 Si02 K2O CaO Ti02 MnO FeOT
3.95(9.0) 0.65(15) 13.3(5.4) 72.7(1.1) 2.52(9.1) 2.27(15) 0.54(12) 0.08 2.73 (9.7)
4.67 0.93 15.5 65.5 2.80 3.29 1.09 0.15 4.49
4.17(0.1) 0.22(5.2) 11.8(0.6) 75.8(0.9) 2.95(1.4) 1.07(1.6) 0.32(17) 0.04 1.64(4.3)
4.32(3.3) 0.67(12) 13.6(2.0) 71.0(1.3) 2.37(2.2) 2.44(9.5) 0.57(10) 0.08 2.93(5.5)
4.16(8.9) 0.48(11) 12.9(1.1) 73.2(0.9) 2.62(9.7) 2.06(16) 0.51(8.2) 0.06 2.34(7.2)
5.12 0.84 16.0 64.0 2.28 4.28 0.83 0.08 4.78
4.46(5.6) 0.66(8.8) 13.6(2.2) 70.6(1.0) 2.35(3.0) 2.48(7.2) 0.58(11) 0.09 2.73(7.3)
3.19(15) 0.06 12.6(7.8) 73.8(3.3) 4.91(16) 0.83 (60) 0.18 0.01 0.60(20)
4.36 0.54 13.3 72.1 2.52 2.13 0.58 0.05 2.47
4.01 0.12 12.0 77.0 3.24 0.82 0.30 0.03 1.44
3.42 0.15 11.4 74.2 3.76 0.89 0.22 0.03 1.15
4.06 0.35 12.5 73.1 2.90 1.85 0.65 0.03 2.62
Total
98.7(4)
98.4
98.0(3)
98.0(10)
98.3(4)
98.2
97.6(13)
96.1(3)
98.1
99.0
95.2
98.1
Details same as in explanation for Table 2 except that mafic-phenocryst ratios are not reported because all samples are mixtures of multiple components, and standard deviations are not reported for components having fewer than 3 analytical points.
258
J. R. RIEHLE ET AL.
Table 5. Comparison of samples from the Alaska Peninsula with samples of Brooks River tephra deposits at Sites II and III, based on similarity of glass compositions Deposit B (historical): Bl and B3 are similar to 1912 dacitic Katmai tephra and B2 is similar to 1912 rhyolitic tephra. B3 is also highly similar to Deposit G2. B4 (3 points) and B6 (2 points) are similar to components of upper Holocene deposits south of the Katmai region that are correlated with Aniakchak Crater; B6 is also highly similar to underlying Deposit Cl. We conclude that B at Site II is chiefly 1912 tephra; there is also a minor amount of tephra similar to that of Aniakchak Crater but whether this is a separate tephra fall similar to underlying Deposit C or is contamination by Deposit C is unknown. Deposit C (about 400 years old): The major component of each of 3 subsamples (Cl-1, C2-1 and C3-1) is highly similar to many samples of upper Holocene deposits of Aniakchak Crater. The other components of Cl, C2 and C3 are similar to some components of these, or other, Aniakchak deposits. We conclude that the 3 layers of deposit C represent 3 closely succeeding, late prehistoric deposits of Aniakchak Crater. Deposit D (900-1200 years old): Dl and D2 are similar to no other samples (except underlying El). D3 is marginally similar to a few deposits of several volcanoes on the southern Alaska Peninsula. D4 is similar to several, upper Holocene Katmai-area deposits and to Dumond E5 and F7. We conclude that deposit D is a Katmai-area deposit of unknown source. Deposit E (1200-1800 years old): El is similar only to overlying Dl. E2, E3 and E4 are similar to middle and late Holocene deposits of Aniakchak Crater. E5 and D4 are similar to a few local deposits. We conclude that the two deposits represent a slightly younger tephra of local origin (D) and a slightly older deposit of Aniakchak Crater (E). Deposit F (2100-3100 years old): Fl is highly similar to several samples of the mid-Holocene, caldera-forming eruptions of Aniakchak Crater, as well as other, lower and upper Holocene Aniakchak deposits. F2 is similar to a few deposits of Aniakchak Crater of a range of ages. F3 is highly similar to two middle Holocene deposits of Aniakchak Crater. F4 and F5, both one-shard analyses, are only marginally similar to a few other samples. F6 is similar to only one, lower Holocene deposit from elsewhere in the Katmai region and is marginally similar to the 1912 andesitic fallout. F7 is similar to two other, upper Holocene deposits in the Katmai region and to Dumond E5 and D4. We conclude the deposit is largely or entirely tephra of the 3460 year BP, caldera-forming eruptions of Aniakchak Crater or a closely succeeding Aniakchak eruption. There may also be a minor amount of locally derived tephra (F4-F7; but only F6 consists of more than two analytical points). Deposit G (3800-4000 years old): Gl is similar to a few middle Holocene deposits of Aniakchak Crater. G2 is similar only to 1912 dacite and to a few other Brooks River samples. G3 is marginally similar to a few mid- and upper Holocene deposits of Aniakchak Crater. We conclude the deposit is a mixture of Aniakchak and locally derived tephra. Deposit H (3800-4400 years old): HI is highly similar to some middle Holocene, pre-caldera-forming deposits of Aniakchak Crater and (or) of Black Peak (about 4000 yrep). H2 is similar to some middle and upper Holocene deposits of Aniakchak Crater. H3 is similar to some middle Holocene, Katmai-region deposits including F6. H4 is marginally similar to only 3 other samples. We conclude that deposit H is mainly middle Holocene, Aniakchak tephra and possibly a minor amount of a local tephra. Deposit I (roughly 5500 years old): II is marginally similar to 1912 dacitic tephra, to 16-F1, and to overlying B2 and Dl. 12 is similar to the Lethe tephra, to several middle Holocene deposits in the Katmai region, and to two other, upper Holocene deposits at Brooks River, as well as to overlying H3 and underlying J2.13 is marginally similar to only 16-E and to Dumond J4. We conclude that this is a middle Holocene deposit of local origin. Deposit J (roughly 6500 years old): Jl and J3 are similar to no other samples. J2 is highly similar to the Lethe tephra, to a few middle Holocene, Katmai-region deposits, and to Dumond 12. J4 is marginally similar to 1912 andesitic tephra, to a lower Holocene deposit 80km north of Katmai, and to overlying Dl, El, H3 and 13. J5 is marginally similar to only overlying B2 and J6 is marginally similar to 16-K1. We conclude that this is a lower Holocene deposit of local origin. 'Similar' means a similarity coefficient >0.95; 'marginally similar' is 0.93-0.94. Ages refer to the original Dumond deposits (1979) (see Table 1), which may or may not correlate with samples from Sites II and III.
samples (analyses available from J. Riehle). Of the 10 'unknown' samples, only three, VI-B, VI-C and V-H, have s.c. values >0.95 with one another (Table 6). Two of these samples are succeeding deposits at Site VI that clearly cannot be the same deposit. The poor degree of correlation among these closely adjacent sites
means that not all tephra falls are reliably preserved at every BRAD site, as a result of local, minor unconformities in combination with the thin, fine-grained nature of the deposits. Moreover, only V-C1, V-H, VI-B1 and VI-C1 are similar to a major component of a Site II deposit, Thus, these 10 'unknown' deposits include at
Table 6. Similarity coefficients for pairs of some upper Holocene tephra deposits from the Brooks River Archaeological District, Alaska IV-D
IV-D IV-E1 IV-E2 IV-F IV-H VII-C1 VII-C2 VII-E VI-B1 VI-B2 VI-C1 VI-C2 V-C1 V-C2 V-H
IV-E1
IV-E2
IV-F
IV-H
VII-C1
VII-C2
VII-E
VI-B1
VI-B2
VI-C1
VI-C2
V-C1
V-C2
V-H
0.96
0.97
0.96 0.97
Only values >0.95 are included because lower values typically preclude correlation as the same deposit. (To find the coefficients of a particular sample with every other sample, first read down the column for that sample until the dash is intersected, then continue along the row to the right. For locations of sample sites, see Fig. 2; for stratigraphic settings, see Fig. 6.) Samples that are similar to the original Dumond deposits (Table 4) are listed below because the table is not large enough to include all components of these deposits. Samples '16' are from Site 16 in the Valley of Ten Thousand Smokes (Table 2). IV-D, similar to II-E5 and II-F7 (neither is a major component); IV-E1, marginally similar only to II-C1-3; IV-E2, no similarity coefficient >0.94 with any other sample; IV-F, similar to a number of Aniakchak deposits, including highly similar to some upper Holocene deposits; IV-H, highly similar to some middle Holocene Aniakchak deposits; VII-C1, similar to some Aniakchak deposits and to 16A1; VII-C2, similar to two, upper Holocene Aniakchak deposits; VII-E, similar to two, lower Holocene deposits to the south of the BRAD; VI-B2, marginally similar only to 16H; VI-B1, highly similar to Lethe tephra, 11-12 and II-J2, and two lower Holocene deposits from the Katmai region (see V-H); VI-C1, same as VI-B1; also similar to 16G3; VI-C2, similar to several middle Holocene Aniakchak deposits and to II-H1; V-C1, similar to several lower Holocene Aniakchak deposits and to II-C1-3; V-C2, similar to two, upper Holocene Aniakchak deposits; V-H, similar to Lethe tephra, to some lower Holocene Katmai-region deposits, and to 11-12 and J2; correlation with 11-12 or J2 or with the lower Holocene deposits is clearly impossible, because of the much greater age of these deposits (>5500 years BP) compared with the age of emergence at site V (>2500 years BP).
260
J. R. RIEHLE ET AL.
Fig. 6. Upper Holocene deposits at four sites of archaeological investigations, Brooks River Archaeological District; locations are plotted in Fig. 2. The age at the base of sections 5, 6 and 7 is not known with certainty, but based on the site elevations within 4 m of the present surface of Naknek Lake, cannot be more than about 3000 years old (Dumond 1981, fig. 2.4). Letters to the right of each section are labelled samples discussed in the text.
least six deposits in addition to the original Dumond deposits.
Biological effects of the 1912 Katmai eruption The record of prehistoric occupation at Brooks River affords an opportunity to study the longterm response of humans to tephra falls. Dumond (1979) noted the lack of correlation of periods of abandonment of the BRAD with times of tephra eruptions, and concluded that the volumes of these tephra falls were insufficient to have had a significant impact on humans there.
The 1912 Katmai eruption provides a useful model to illustrate the relationship between thickness of tephra fall and the significance of its impact on humans, animals and vegetation. The 1912 Katmai eruption comprised three main eruptive pulses that occurred over a period of 60 h, beginning on 6 June (Hildreth 1983). Winds to high levels were mainly east directed during this time, and the main axis of the tephra deposit extends over northern Kodiak and Afognak Islands (Fig. 7; Fierstein & Hildreth 1992). The ashflow in the VTTS was emplaced during the first eruptive pulse; some of the apparent ashfall material within 10km of the VTTS is actually surge deposits that were swept
THE BROOKS RIVER ARCHAEOLOGICAL DISTRICT
261
Fig. 7. Thickness of tephra deposited during 1912 eruption of Novarupta (Katmai) volcanoes, and representative effects on humans, animals and plants related to tephra thickness.
from the moving ashflow, rather than being true ashfall (that is, carried aloft by the uprising eruption column). Both the ashflow and the ashfall material were erupted from the Novarupta vent at the head of the VTTS, not from Katmai Caldera, which did, however, collapse as a result of withdrawal of support because of eruption of magma from Novarupta vent (Curtis 1968; Hildreth 1983, see fig. 3). Humans directly affected by the eruption were located mainly in four areas: Katmai village, 30 km southeast of Novarupta (Fig. 8); a pair of settlements on the Ukak and Savonoski Rivers of which one was sited near the foot of the ashflow deposit, 20km northwest of the vent; Douglas village, 80km to the east; and Kodiak village, 180 km east (Fig. 7). Most of the Douglas and Katmai village inhabitants were at summer fishing camps on Kaflia Bay. The few Katmai villagers left behind fled in fear early on 6 June as the frequency of premonitory earthquakes increased. They were in bidarkas (kayaks) at Cape Kubugakli when the eruption occurred.
Remarkably, no known deaths were directly attributable to the eruption (Martin 1913; Griggs 1922) despite its size, the largest of the twentieth century (estimated Volcano Explosivity Index of six; Simkin et al 1981), and the proximity of people to the vent. There are several reasons for the lack of fatalities. First, none were close enough to be burned by contact with tephra or by the ashflow or its surge, unlike victims 20-30 km from Mount St Helens in 1980 (Rosenbaum & Waitt 1981). Individuals fleeing Savonoski and Katmai villages reported heat from falling tephra, but they were able to continue their journeys until they were beyond the tephra fall (Griggs 1922, p. 17). Second, although there was flooding on drainages heading in the vicinity of Mount Katmai or the VTTS (Griggs 1922; Hildreth 1983), flood volumes were insufficient to damage either Savonoski village or Katmai village. Third, individuals in the zone of heaviest tephra fall were rescued before serious dehydration or starvation. People at Kaflia Bay were rescued by the
262
J. R. RIEHLE ET AL.
Fig. 8. Katmai Village 3 years after the 1912 eruption. Because of the thick primary tephra fall (nearly 1 m at this site only 30 km from the source) and to subsequent instability of the Katmai River channel, the site was abandoned. (Photograph by G. C. Martin, in Griggs 1922.)
steamer Redondo 3 days after the end of the eruption (Revenue-Cutter Service 1913, p. 125), and those from Savonoski village paddled their bidarkas for 2 days to Naknek, beyond significant tephra fall 60km to the west (Fig. 7). The effects of the tephra on animals and vegetation, of major importance to a subsistence culture, depend critically on thickness of tephra and the particular species. Virtually all animals either died or left areas of heaviest tephra fall on the Alaska Peninsula; Griggs (1922) observed bear and fox tracks on the beach of Katmai Bay in 1915, but the only signs of animal life in the upper reaches of the Katmai River valley were occasional birds. By 1919, however, nesting birds, mice and ground squirrels in the same area had increased notably (Griggs 1922, p. 164). Cattle at Kodiak, where 20cm of tephra fell, survived the eruption but then were removed from the island until pastures had fully revegetated 2 years later (Griggs 1922, p. 44). In general, mammals in the areas of heaviest tephra on Kodiak Island were not seriously affected except for malnutrition (Evermann 1914), although smaller mammals may have suffered more heavily than larger ones (Erskine 1962). Caribou on the Alaska Peninsula abraded their teeth on volcanic ash to the point of starvation after a relatively minor eruption of Aniakchak Crater in 1931; many new-born calves were lost as the
herd migrated from the area of ashfall (Trowbridge 1976). Probably the same occurred in areas of even light tephra on the Alaska Peninsula after the 1912 eruption. Caribou graze on low-standing mosses and vegetation whereas moose browse on shrubs and young trees, which are more easily cleared of tephra by wind and rain. Effects of tephra on salmon are complicated because different species (mainly coho, sockeye and pinks in this part of Alaska) have different life cycles and spawning ages, and require different types of spawning beds. Salmon were just beginning to enter streams on the Peninsula and in the Kodiak Islands at the time of the eruption. Those in areas of more than about 10cm tephra either suffocated in tephra-laden waters or returned to the sea, from which they periodically attempted to re-enter the streams (Evermann 1914). Streams in areas of lighter tephra generally cleared in time to permit late spawners to enter in 1912, whereas those in areas of heavy tephra on the Peninsula were still unsuitable for spawning because of eroding tephra and unstable beds as much as 5 years after the eruption (Griggs 1922, p. 161). Young salmon spend their first few months to 3 years in fresh water; those in deeper streams, or sockeyes in lakes at the heads of streams, had better odds of survival than those in shallow
THE BROOKS RIVER ARCHAEOLOGICAL DISTRICT
263
Fig. 9. In areas of less than about 25 cm of tephra fall, some species such as this Devil's Club survived by extending from old root systems to the surface of the tephra. The new roots developing just below the new ground surface should be noted. (Photograph by R. F. Griggs, in Griggs 1922.)
streams or in areas of heavier tephra. Survival depends partly on the food supply, which also affects nonmigratory species such as trout; some streams on western Afognak Island (Fig. 7) were devoid of fish food a year after the eruption (Evermann 1914). The impact of the eruption on the salmon resource was not fully realized, however, until several years after the eruption. Sockeye returns to the Kodiak Islands began to decline in 1915 (pink salmon spawn after 2 years, sockeyes and coho, after 3-6 years) and continued to decline until 1920 (Eicher & Rounsefell 1957). Thus, immediately after the eruption, salmon were still
available as a food resource although opportunities to take them from streams in areas of 20cm tephra or more were probably limited. Additionally, marine shellfish were killed in significant numbers in areas of heavy tephra, and even cod were reported to have left traditional grounds near Kodiak Island (Evermann 1914, p. 62). The details of vegetative recovery, recorded by botanist Robert Griggs (1922), are as varied as was the impact on salmon. In areas of heavy tephra, tree limbs were broken, grasses and shrubs were buried, and landslides and floods excavated or deeply buried trees growing in
264
J. R. RIEHLE ET AL.
Fig. 10. Willows were able to develop new (adventitious) roots at the surface of the tephra, even when buried to depths of 2m. Here, the roots were exposed by subsequent erosion of the tephra. (Photograph by D. B. Church, in Griggs 1922.)
floodplains and at the foot of steep slopes. On Kodiak Island, grasses and shrubs had recovered within 2-3 years; recovery was not as much by reseeding as by sending up shoots from existing root systems (Fig. 9). Some plants survived as much as 3 years of burial (Griggs 1922, p. 51). Some willows (Fig. 10) survived even deep burial by sending out adventitious roots just below the new ground surface. Experiments confirmed that the tephra had few available nutrients but would support plant growth if properly fertilized. A striking demonstration of this characteristic was provided by a slow-growing tree from a Kodiak bog, which in the first few years after the eruption more than doubled its diameter (Griggs 1922, p. 53). The benefit of the tephra to the tree was due not to fertilization, but instead to the temporary suppression of grasses at the site, which competed for available nutrients. Summary and conclusions (1) Because of its proximity to the active Aleutian volcanic arc, the Brooks River Archaeological District has experienced numerous
tephra falls throughout the past 4500 years of human occupation. At least 15 prehistoric, Holocene tephra deposits, the oldest about 6500 years, can be recognized. (2) One deposit (Dumond C about 400 years old) is sufficiently thick, distinctive and reliably preserved to be useful as a marker bed. The other deposits, however, are thin and fine grained, and either are not preserved at every locality or are less readily distinguishable in the field on the basis of their megascopic characteristics. (3) Microprobe analyses of glass separates show that most of the deposits are chemically heterogeneous, either the result of eruption of mixed or heterogeneous magma as occurred during the 1912 eruption, or because of mixing of succeeding tephra deposits by biological activity and frost action. (4) About half of the tephra deposits originated at Aniakchak Crater, 160km to the southwest and one of the most frequently active volcanoes on the Alaska Peninsula. Other deposits consist mainly of glass that is either similar only to known deposits of Katmai volcanoes, or is unique to the Katmai region. Because of the heterogeneous nature of these tephra deposits
THE BROOKS RIVER ARCHAEOLOGICAL DISTRICT
and the lack of widespread marker deposits, specific ages and sources for most of the prehistoric Katmai deposits cannot be identified at this time. (5) Floral and faunal recovery from the effects of the 1912 eruption of Katmai (Novarupta) volcano varied in detail, depending on the thickness of the tephra and on species-specific characteristics. Thus, effects of volcanic eruptions on ancient subsistence populations may be expected to have varied from case to case. In general, however, recovery was rapid in areas that had less than 10cm of ashfall, which explains the lack of correlation between ashfalls and periods of abandonment of the Brooks River site. References BORCHARDT, G. A., ARUSCAVAGE, P. J., & MILLARD, H. T., JR 1972. Correlation of the Bishop ash, a Pleistocene marker bed, using instrumental neutron activation analysis. Journal of Sedimentary Petrology, 42, 301-306. CARMICHAEL, I. S. E. & MACDONALD, A. 1961. The geochemistry of some natural acid glasses from the North Atlantic Tertiary volcanic province. Geochimica et Cosmochimica Acta, 25, 189-222. CURTIS, G. H. 1968, The stratigraphy of the ejecta from the 1912 eruption of Mount Katmai and Novarupta, Alaska. In: COATS, R. R., HAY, R. L. & ANDERSON, C. A. (eds) Studies in Volcanology. Geological Society of America, Memoir, 116, 153-210. DOWNES, H. 1985. Evidence for magma heterogeneity in the White River Ash (Yukon Territory). Canadian Journal of Earth Sciences, 22, 929-934. DUMOND, D. E. 1964. Archaeological Survey in Katmai National Monument, Alaska 1963. Report to the National Park Service, Western Region. Department of Anthropology, University of Oregon. 1979. People and pumice on the Alaska Peninsula. In: SHEETS, P. D. & GRAYSON, D. K. (eds) Volcanic Activity and Human Ecology. Academic Press, New York, 373-392. 1981. Archeology on the Alaska Peninsula: the Naknek Region, 1960-1975. University of Oregon, Anthropological Papers, 21. EICHER, G. J., JR & ROUNSEFELL, G. A. 1957. Effects of lake fertilization by volcanic activity on abundance of salmon. Limnology and Oceanography, 2, 70-78. ERSKINE, W. F. 1962. Katmai. Abelard-Schuman, London. EVERMANN, B. W. 1914. Alaska Fisheries and Fur Industries in 1913. Report of the US Commissioner of Fisheries. US Government Printing Office, Washington, DC. FIERSTEIN, J. & HILDRETH, W. 1992. The plinian eruptions of 1912 at Novarupta, Katmai National Park, Alaska. Bulletin of Volcanology, 54,646-684.
265
GRIGGS, R. F. 1922. The Valley of Ten Thousand Smokes. National Geographic Society, Washington, DC. HARRITT, R. K. 1988. The Late Prehistory of Brooks River, Alaska. University of Oregon, Anthropological Papers, 38. HILDRETH, W. 1983. The compositionally zoned eruption of 1912 in the Valley of Ten Thousand Smokes, Katmai National Park, Alaska. Journal of Volcanology and Geothermal Research, 18, 1-56. MARTIN, G. C. 1913. The recent eruption of Katmai volcano in Alaska. National Geographic Magazine, 24, 131-181. MILLER, T. P. & SMITH, R. L. 1987. Late Quaternary caldera forming eruptions in the eastern Aleutian arc, Alaska. Geology, 15, 434-438. NOWAK, M. 1968. Archaeological dating by means of volcanic ash strata. PhD dissertation, University of Oregon. PINNEY, D. S. 1993. Late Quaternary facial and volcanic stratigraphy near Windy Creek, Katmai National Park, Alaska. MSc thesis, University of Alaska, Fairbanks. & BEGET, J. E. 1991. Late Pleistocene volcanic deposits near the Valley of Ten Thousand Smokes, Katmai National Park, Alaska. In: REGER, R. D. (ed.) Short Notes on Alaskan Geology 1991. Alaska Division of Geological and Geophysical Surveys, Professional Report, 111, 45-54. REVENUE-CUTTER SERVICE 1913. Annual Report for the Fiscal Year Ended June 30, 1912. US Government Printing Office, Washington, DC. RIEHLE, J. R., MANN, D. H., PETEET, D. M., ENGSTROM, D. R., BREW, D. A. & MEYER, C. E. 1992. The Mount Edgecumbe tephra deposits, a marker horizon in southeastern Alaska near the Pleistocene-Holocene boundary. Quaternary Research, 37, 183-202. , WAITT, R. B., JR, MEYER, C. E. & CALK, L. C. 1998. The age of formation of Kaguyak Caldera, eastern Aleutian arc, Alaska, estimated by tephrochronology. In: GRAY, J. & RIEHLE, J. R. (eds) Geologic Studies in Alaska by the US Geological Survey, 1996. US Geological Survey, Professional Paper, 1595. ROSENBAUM, J. G. & WAITT, R. B., JR 1981. Summary of eyewitness accounts of the May 18 eruption. In: LIPMAN, P. W. & MULLINEAUX, D. R. (eds) The 1980 Eruptions of Mount St Helens, Washington. US Geological Survey, Professional Paper, 1250, 53-68. SARNA-WOJCICKI, A. M., WAITT, R. B., JR, WOODWARD, M. J., SHIPLEY, S. & RIVERA, J. 1981. Premagmatic ash erupted from March 27 through May 14, 1980 - extent, mass, volume, and composition. In: LIPMAN, P. W. & MULLINEAUX, D. R. (eds) The 1980 Eruptions of Mount St Helens, Washington. US Geological Survey Professional Paper, 1250, 569-576. SlMKIN, T., SlEBERT, L., McCLELLAND, L., BRIDGE, D.,
NEWHALL, C. & LATTER, J. H. 1981. Volcanoes of the World. Smithsonian Institution, Washington, DC.
266
J. R. RIEHLE ET AL.
SMITH, D. G. W. & WESTGATE, J. A. 1969. Electron probe technique for characterizing pyroclastic deposits. Earth and Planetary Science Letters, 5, 313-319. STEEN-MclNTYRE, V. 1977. A Manual for Tephrochronology. Colorado School of Mines, Golden, CO. TROWBRIDGE, T. 1976. Aniakchak Crater. In: RENNICK, & PERRY (eds) Alaska's Volcanoes: Northern Link in the Ring of Fire. Alaska Geographic, 4, 70-73. WESTGATE, J. A. & GORTON, M. P. 1981. Correlation techniques in tephra studies, In: SELF, S. &
SPARKS, R. S. J. (eds) Tephra Studies. D. Riedel, Dordrecht, 73-94. WIESNETH, D. W. & EICHELBERGER, J. C. 1996. Vapor phase crystallization in rhyolite lava from Novarupta dome, Katmai National Park, Alaska. EOS Transactions, American Geophysical Union, 77, 770. WILCOX, R. E. 1959. Some effects of recent volcanic ash falls with especial reference to Alaska. US Geological Survey Bulletin, 1028-N, 409-476.
Endemic stress, farming communities and the influence of Icelandic volcanic eruptions in the Scottish Highlands R. A. DODGSHON1, D. D. GILBERTSON2 & J. P. GRATTAN1 1
Institute of Geography and Earth Sciences, University of Wales Aberystwyth, Aberystwyth SY23 3DB, UK (e-mail:
[email protected]) 2 The Nene Centre for Research, University College Northampton, Northampton NN2 7AH, UK Abstract: This paper explores present understanding of the possible impacts that volcanic eruptions in Iceland might have had upon the environments and traditional farming systems of the Highlands and Islands of Scotland, before 'the Clearances' of the late 18th and 19th centuries AD. It reconstructs both the nature of the impacts and the character of the risks that might have been faced by subsistence communities within the historical period from such Icelandic volcanic eruptions, and as such serves to redirect a research emphasis that has previously been principally focused upon the European Bronze Age. The study also emphasizes that it is inadequate to envisage the impacts from volcanic aerosols as threats to the community to be understood solely along a continuum of environmental hazards. For example, in the historical period before the Clearances, the wider social, political and economic contexts of subsistence economies affected can be shown to have raised or lowered the thresholds at which environmental risks became real or were turned into subsistence crises. In times past, as now, the capacity of people to cope with such environmental vicissitudes would have varied according to a complex of pre-disposing factors, their recent experiences, attitudes and perceptions, political and social relationships, health and wellbeing (especially their susceptibility to respiratory problems), economy, education and memory, and general inventiveness and resilience. Unlike much earlier research, which focused upon the European Late Bronze Age and emphasized global climatic change and its regional-scale consequences, this account of more recent times emphasizes the small scale, the importance of local pre-disposition and contingency, and hence the likely patchiness and indeterminacy of consequences on the ground of distant volcanic eruptions. The paper concludes that in the historical past, for a variety of environmental, agricultural, social and political reasons, some communities in the Highlands and Islands would have already been typically at risk of a subsistence crisis one year in every four or five. Hence a particular group of people could have been at notable further risk if a significant quantity of volcanically derived noxious and toxic materials had fallen upon them. As a result, for both habitats and human populations in historical times, the consequences of an Icelandic volcanic eruption are likely to have varied from place to place and from time to time. This analysis also suggests that it is difficult to envisage that any postulated region-wide abandonment of settlement in the British Isles might be attributable, directly or indirectly, to the distal impacts of volcanic eruptions in Iceland.
This paper explores the ability of traditional farming systems in the Highlands and Islands of Scotland, before'the Clearances'of the late 18th and early 19th centuries, to absorb and buffer the community from the stresses that might have been introduced as a result of the eruption of volcanoes in Iceland. The magnitude of the potential problem posed by such external events is difficult to estimate. Over the last decade, hypotheses describing the possible impacts have focused upon induced climatic deterioration and
the impacts of phytotoxic compounds both adsorbed on tephra and in solution in acid precipitation. The roles of prevailing synoptic meteorological situations, as well as the inherent susceptibility of the affected habitats, have also been explored. It would be wrong to see these hazards to traditional subsistence communities as something to be described and measured solely along a continuum of environmental hazards, perhaps ranging from the local and relatively minor to the regional and severe.
From: McGuiRE, W. G., GRIFFITHS, D. R., HANCOCK, P. L. & STEWART, I. S. (eds) The Archaeology of Geological Catastrophes. Geological Society, London, Special Publications, 171, 267-280. 1-86239-062-2/00/ $15.00 © The Geological Society of London 2000.
268
R. A. DODGSHON, D. D. GILBERTSON & J. P. GRATTAN
Traditional societies trying to maintain a subsistence economy within the difficult, often acidified and unproductive habitats of the north and west of Britain, would always have faced considerable risk. Detailed examination of the social, political and economic contexts of the traditional farm and township economies presented in this paper demonstrates that their social contexts could serve to raise or to lower the threshold at which risks became real and were turned into subsistence crises. Although this paper focuses upon the historical period from the advent of the Scottish clan system in the 12th and 13th centuries AD down to the time of the Clearances at about 1820, the intellectual starting point for this particular approach lies elsewhere, in the coming together over the last decade of several previously distinct and sometimes contentious areas of scholarship: volcanic eruptions, induced global climatic change, and postulated abandonment of upland settlement in the Late Bronze Age in parts of upland Britain; the development of tephrochronology within the UK; reports of the impacts of actual eruptions on historical communities in Iceland; investigations of acid precipitation and more general air pollution upon acidified ecosystems; holistic investigations of documentary sources concerning societies and their environments. Volcanic activity, induced climatic deterioration and the abandonment of upland settlement A variety of environmental and human mechanisms have been advanced to explain the numerous remains of medieval and post-medieval settlement that exist in upland and moorland Britain. Non-catastrophic explanations include land-holding and manorial policies (Austin 1985; for Dartmoor), climatic change (Beresford 1981; Dartmoor), crop failure as a result of persistent bad weather (Parry 1975, 1978, 1981, 1985; Parry & Carter 1985; for the Lammermuir Hills), as well as famine, disease, the Black Death, the Dissolution of the Monasteries, civil strife and invasion (Bell & Walker 1992). It is perhaps surprising to learn of hypotheses that suggest that volcanic eruptions in Iceland may have brought about injurious events to the habitats and populations of Scotland or Ireland, far distant downwind, across the breadth of the eastern Atlantic. Such possibilities are even more unexpected given that the known Holocene eruptions of Icelandic volcanoes were not particularly large, when viewed on a global scale (see Simkin & Siebert 1994).
Nevertheless, several hypotheses exploring these ideas have been advanced over the past decade, and the possible impacts of volcanic eruptions in Iceland (especially Hekla 3 and 4) upon the prehistoric, rather than historical, subsistence communities and environments of the north and west of Britain have been major subjects of archaeological, palaeoecological, palaeoclimatological and stratigraphic research (Burgess 1984, 1985, 1989). These ideas have benefited from the discovery in Scotland over the last decade of tephra from several Icelandic volcanic eruptions. This has provided impetus for seeking further volcanic catastrophic mechanisms by which to seek to explain postulated change in the archaeological and historical records (e.g. Hammer et al. 1980; Dugmore 1989; Blackford et al 1992; Dugmore & Newton 1992, 1996; Dugmore et al. 1992, 1995, 1996; Pilcher & Hall 1992, 1996; Hall et al 1993, 1994; Edwards et al 1994; Grattan & Gilbertson 1994; Gilbertson 1995; Pilcher et al 1995; Grattan et al 1996; Lowe & Turney 1997). At present, no less than 20 Holocene tephras from Icelandic eruptions, of both historical and prehistoric date, are known to have been deposited in northwest Europe and have the potential to form significant stratigraphic markers (Dugmore et al 1996). It is clear that some may have been associated with impacts on the ground that are detectable in the palaeoecological records (Blackford et al 1992). For example, environmental disruption following Icelandic eruptions has been proposed to explain decadal-length reductions in tree-growth increment sequence preserved in the Irish bog-oak record (see Pilcher et al 1984; Baillie 1988, 1989; Baillie & Munro 1988). Some, but not all, of this work was prompted by a proposal by Burgess (1989) that an apparent widespread abandonment of Late Bronze Age settlement on moorland and upland sites in Scotland and elsewhere at the end of the second millennium BC (see also Parker Pearson (1993)) was associated with catastrophic climatic change generated by an eruption, termed Hekla 3, at c. 1159BC, of the Icelandic volcano Hekla. It is difficult to assert with real confidence that any such widespread and synchronous abandonment of settlement did in fact take place in upland Scotland in particular, or upland Britain in general. For example, with the aid of a series of theoretical models, Grattan et al. (1999a) suggested in a paper presented to the Geological Society of London in 1994, that a widespread, prehistoric abandonment of uplands and moorlands in Britain was not a likely consequence of the eruption of either the Hekla 3 or 4 event. Starting from the archaeological information,
ICELANDIC VOLCANIC ERUPTIONS IN THE SCOTTISH HIGHLANDS Young & Simmonds (1996) showed that the evidence of abandonment in northern Britain was not compelling. In their important review of marginality and the nature of prehistoric settlement in the north of England they questioned the extent to which there actually was widespread and sudden abandonment of uplands and whether or not it was 'instantaneous'. They have emphasized that on innumerable occasions people have coped with very difficult situations because they chose not to leave their homes, or that in other situations, communities under pressure have often been supported by the wider and more distant communities in the society of which they form part. Young & Simmonds also indicated that many, perhaps misleading, assumptions lie in the use of the term 'marginal' in the descriptions and analysis of past peoples, economies and habitats of the north and west of the British Isles. More recently, Buckland et al. (1997) called into question the association of Hekla 3 with the putative Bronze Age settlement abandonment. They pointed out the problems in linking this eruption with particular ice-core acidity peaks, or narrow rings in the Irish bog-oak chronology. Starting from a broader perspective, Renfrew (1990) cautioned: 'it is necessary to recognize and discount the common tendency among archaeologists and historians to assume a causal link between distant and widely-spaced events of which they may have knowledge. An eruption here, a destruction there, a plague somewhere else - all are too easily linked in hasty surmise.' Neither has it proved easy to detect reliable, independent and corroborative palaeo-environmental evidence of the various postulated environmental processes and impacts that have been conjectured as agents of settlement change. Indeed, sometimes, the conclusions published concerning the ecological impact of just one ash layer on broadly similar peat-land habitats can be surprisingly diverse. For example, the report of a clear deleterious biological impact by the fall of volcanic ash of Hekla 4 (c. 2395-2279 BC; Pilcher et al. 1995) on pine forests in northeast Scotland given by Blackford et al. (1992) was followed by a statement of there being no such relationship in the north of Ireland (Hall et al. 1994), and even later studies in the north of Scotland further complicate the issue (Charman et al. 1995). Obviously, these differences may reflect genuine differences in impact, rather than point to inconsistencies and deficiencies in the evidence. These results also emphasize the need for a clearer understanding of the mechanisms through which the distal fall-out of volcanic eruptions may produce recognizable impacts
269
upon the landscapes of the British Isles. One mechanism through which volcanic fall-out from Icelandic eruptions may have affected habitats in the north and west of Britain, which has been explored in detail recently, focuses upon local, small-scale variations in habitat susceptibility and the environmental focusing of pollutants by the synoptic meteorological circulation (Grattan & Charman 1994; Grattan & Gilbertson 1994; Grattan & Pyatt 1994; Grattan et al. 1996, 1998). Particular reference has been made to document-based studies of the eruption and distant impacts of aerosols and fall-out from the Laki fissure eruption of AD 1783. This mechanism is explored in detail below.
Impacts upon people and their environment: issues of scale and susceptibility Global climatic changes? Initially, global climatic change produced by the eruption Hekla 3 was suspected to be the mechanism that promoted the volcanically generated equivalent of a nuclear winter for the Late Bronze Age communities in Scotland (Burgess 1989). For highland societies presumed to be living at the margins of sustainable subsistence farming in these remote and difficult environments, this sudden, but extended climatic deterioration, was suspected to have triggered complete collapse. In fact, there are no theoretical or empirical grounds for assuming that the Hekla 3 eruption, nor indeed any eruption that has occurred during the Holocene period was capable of causing such a phenomenon (see Mass & Portman 1989). For example, the massive eruption of Mt Pinatubo, in 1991 (McCormick et al. 1995; Robock & Mao 1995), resulted in a mean global surface air temperature reduction of 0.5°C, which is well within one standard deviation of the temperature variation that occurs normally on a year-to-year basis. It is possible that the prevailing synoptic atmospheric circulation over northwest Europe could have resulted in either an increase or decrease of this effect over northwest Britain (Kelly et al. 1984; Lamb 1992). However, given the variability of weather of northwest Britain, it is unlikely that a volcanically induced atmospheric circulation fluctuation, from any Icelandic volcanic eruption of Holocene time, would have been prominent. It is unlikely that, in themselves, climatic fluctuations generated by Icelandic volcanic eruptions would have been
270
R. A. DODGSHON, D. D. GILBERTSON & J. P. GRATTAN
sufficient to have brought about episodes of catastrophic social distress or environmental catastrophe in the subsistence-based communities of Britain in the historical past. Nor is it likely that such climatic fluctuations would have been of sufficient magnitude to account for the palaeo-environmental phenomena that have been tentatively associated with various Icelandic volcanic eruptions (Baillie 1988; Blackford et al 1992; Charman et al 1995).
Accounts of the impact of volcanic aerosols and tephra in Iceland and Europe Global climate change is not the only mechanism that has been invoked to explain settlement abandonment and environmental change in the wake of Icelandic volcanic eruptions. Across Europe, the environmental impacts of volatile compounds released during the Laki fissure eruption in AD 1783 were particularly damaging and provide clues to the manner in which past eruptions may have affected habitats and people. Impacts upon the European subsistence communities living immediately adjacent to these Icelandic eruptions are especially revealing. For example, detailed reading of the accounts by Thorarinsson (1967, 1979), Andresson (1984), Blong (1984) or Ogilvie (1986) provides convincing descriptions of both the direct and indirect downwind effects of noxious and toxic volcanically produced volatile materials and gases upon the subsistence farmers and fisherfolk of that island in the recent historical period. Many of the medical and veterinary symptoms and problems reported in Iceland after the 1783 eruption of the Laki fissure are recognizable as the results of the deposition, inhalation or ingestion of sulphuric, hydrochloric and hydrofluoric acids, affecting cows, sheep, crops, fodder and grazing, as well as people, their well-being, economies and societies (Thorarinsson 1971, 1979, 1981; Devine et al. 1984; Petursson et al. 1984; Palais & Sigurdsson 1989; Thordarson & Self 1993). Of critical importance for understanding the possible impacts of Icelandic volcanic eruptions on the British Isles, these accounts also reveal how the problems caused by one eruption in one area can spread through many aspects of the life of the affected communities and then be communicated to distant coastal communities by the emigration of distressed people. People fleeing the catastrophe caused the lands of the interior of Iceland to become desolate, whereas the coastal communities were impoverished as their
general resource base became overburdened by the influx of these people. The contamination of pastures obviously continued for a number of years, and it is evident that some crops were more susceptible than others to this natural pollution. Even so, these grim events should not automatically be assumed to have affected, or been communicated, throughout all the country of Iceland, or to have been the same from one eruption to the next. Thorarinsson (1971) showed that the vast majority of farms that were permanently abandoned after the eruption of Hekla in AD 1104 were found within 25km of the volcano, although one site 60 km distant was never resettled. He was also able to use their depth of burial in tephra as a more general measure of the severity of impact of the eruption, and related this aspect to the probability that the site would be reoccupied. For example, for the six Icelandic eruptions he studied since the AD 1104 eruption, the general pattern was that a burial in 10cm of tephra caused abandonment for up to 1 year, 15cm for a period of 1-5 years, and 20cm for periods of some decades. In northern and western areas of the British Isles the tephra isopachs are so fine that they cannot even be measured in millimetres. Caution must therefore be exercised before one accepts the assumption that any tephra fall in Britain was the sole cause for extensive and persistent settlement abandonment. Even the most severe of the Icelandic experiences would suggest that human responses to such a tephra-fall situation in the British Isles is likely to have been far more complex than a simple abandonment of property. Apart from the impacts of measurable tephra fall, volcanic gases may also have a marked effect upon the environment. Of key importance to the argument examined here is the fact that in 1783, volcanic gases emitted by the eruption in Iceland were transported to Europe, where they caused considerable respiratory distress to susceptible people and damage to susceptible crops, trees and fish (Grattan & Brayshay 1995; Grattan et al. 1998; and Fig. 1). Very detailed descriptions of severe acid damage to vegetation, insects, people and property have been left by a number of scientists, of which the recently rediscovered records of two Dutchmen, Brugmans (1787) and Van Swinden (1786), deserve particular emphasis. In mainland Europe the volcanic gases were manifested as a 'dry fog', 'acid fog' or 'sulphurous fog'. For example, Brugmans (1787) noted: 'On many days after the 24th June, in both the town of Groningen and countryside there was a strong, persistent fog... the fog was very dense and accompanied by a
ICELANDIC VOLCANIC ERUPTIONS IN THE SCOTTISH HIGHLANDS
271
Fig. 1. The location of currently known reports of environmental damage to plants and crops across Europe attributable to noxious or toxic materials originating from the Laki fissure eruptions in Iceland of 1783, and volcanoes in southern Italy (Camuffo & Enzi 1995) and perhaps east Germany (Grattan et al. this volume).
272
R. A. DODGSHON, D. D. GILBERTSON & J. P. GRATTAN
very strong smell of sulphur... many people in the open air experienced an uncomfortable pressure, headaches and experienced a difficulty breathing exactly like that encountered when the air is full of burning sulphur, asthmatics suffered to an even greater degree: On the morning of the 25th the land offered an aspect of severe desolation, the green colour of the plants had disappeared and everywhere the leaves were dry,... some were covered in spots, others changed gradually while some leaves dried up completely Another noticeable change was that in a moment the colour could change from
green to brown, black, grey or white. Afterwards a great quantity of leaves fell.' This account is typical of many that were written at the time and it is now clear from them that long-distance transport and deposition of materials from the Laki fissure eruption in 1783 to the European mainland took place and had significant deleterious effects upon the health of both susceptible crops and people across much of Europe (Grattan & Charman 1994; Grattan & Gilbertson 1994; Grattan & Brayshay 1995; Grattan et al. 1998). Further detail concerning the pollution event that occurred in AD 1793
Fig. 2. The concentration of volcanic gases over Europe in AD 1783. (Synoptic chart adapted from Kington (1988).)
ICELANDIC VOLCANIC ERUPTIONS IN THE SCOTTISH HIGHLANDS has been given by Camuffo & Enzi (1995) and Stothers (1996). It is evident from the example of the AD 1783 Laki fissure eruption that toxic and poisonous aerosols from Iceland were capable of causing modest levels of damage to animals, crops and people far downwind across the Atlantic in northwestern Europe when concentrated by stable air conditions (Fig. 2). Present biogeographical knowledge suggests that, in general, such impacts are likely to have been most evident in the more acidified ecosystems of the British Isles (Gorham 1987; Skiba et al 1989; Bull 1991; Smiths al. 1993).
The social context of risk and subsistence: the example of the Highlands and Western Islands It is therefore against a background of minor, probably undetectable, climatic fluctuations, minor tephra falls and probably patchy impacts of acid rain or fog, that we must assess the risk posed to the pre-'Clearances' subsistence economies of the Highlands and Western Islands by past Icelandic volcanic eruptions. Nevertheless, tephra falls and acid fogs and rain are documented for the historical period and it is therefore against a background of Scottish society in these regions during the historical period that the impact of volcanic eruptions is now considered. Trying to maintain a subsistence economy within a difficult, often marginal environment involved considerable risk for all traditional societies, were they in Iceland or in the north and west of Britain. It is wrong to see these risks as something to be measured solely along a continuum of environmental hazards, ranging from the local or minor to regional or severe. The wider social context of a subsistence economy could serve to raise or lower the threshold at which risks became real or were turned into subsistence crises. Attempts to understand more fully the character and scale of the difficulty faced by past communities with subsistence economies when presented with influxes of noxious or toxic volcanic materials in the study region can be substantially aided by a detailed consideration of the social, political and economical characteristics of an example. It is particularly fortunate in this context that the farm or township economy of the communities that lived in the western Highlands and Islands of Scotland over the late medieval and early modern periods are relatively well known through
273
recent research (Dodgshon 1993, 1994). By any standards, these traditional Highland communities lived in an endemically risk-laden world, but their subsistence crises had as much to do with their socio-political situation as with their natural environment and its perturbations. Three particular factors contributed to this heightened vulnerability.
The political context of subsistence crises The traditional pre-Clearance township in the Western Highlands and Islands was organized for self-sufficiency, with only a marginal involvement in marketing and other onward distribution of products. Inevitably, this sort of relatively closed, subsistence economy, or what some would call a 'natural economy', was more at risk of recurrent subsistence crises than more openmarket integrated economies that could exchange or trade away their problems with other regions. The crises that confronted such an economy were rarely self-correcting as a result of grain or meal flowing from areas of surplus to areas of deficit via regional markets or other exchange networks. However, this traditional Highland and Hebridean society was also a kin-based society organized around the power of clan chiefs. As with chiefly societies elsewhere, what can be called the chiefly economy provided local communities with a form of social storage and an insurance against risk (climatic, ecological, volcanic, or immediately human, social or political). Communities paid landowners large quantities of food (oatmeal, bear (= barley), cattle, sheep, cheese, whisky, fish, etc.) either as a form of rent or via obligations of hospitality (known as coid-oidche). In times of surplus, chiefs consumed such renders through the maintenance of household or fighting men, by extravagant displays of feasting, and, in the case of items of value such as cattle, by using them to sustain alliances and fosterage deals. In times of crisis, meanwhile, chiefs could use what had accumulated in the girnal houses, or food stores, to offset local deficits. This two-way flow of food was imbued with an instrumental value, securing the power of a chief over his kinsmen. As one source put it, the Steward of the Southern Isles was a great man simply because of the food which he had gathered in (Martin 1703, p. 98). The geological metaphor of MacLeod's Tables, opposite Dunvegan Castle, served a similar purpose for the MacLeod of MacLeod, proclaiming the scale of his hospitality. One sideeffect was that inter-clan feuding became
274
R. A. DODGSHON, D. D. GILBERTSON & J. P. GRATTAN
focused around food. In character, feuding involved not only constant warring between clans by the fighting men of a clan, men who were supported by the food renders gathered in by a chief, but also, the routine destruction of a rival's food base: a case of fighting with food over food (Dodgshon 1995, pp. 99-109). The ideological overlay to the basic use value of food, however, had begun to change by the start of the 17th century. Legislation such as the Statutes of lona of 1609 controlled the behaviour of chiefs, banning them from maintaining large households and from excessive feasting, both prime uses of food gathered in as rent (Register of the Privy Council X, 1891, pp. 773781). In response, landlords began to market the renders gathered in as rent, or to convert them into cash payments, forcing tenants to market them instead. However, for the majority of farmers, especially those in the far west of the region, marketing involved livestock whereas their basic subsistence continued to depend upon arable. Indeed, all the signs are that dependence on arable increased, as local populations increased and as rents previously based on a range of farm produce now became loaded onto livestock, leaving more arable output for consumption. In these changing circumstances, the risks facing highland communities must have increased. On the one hand, traditional riskaversion strategies based upon chiefly networks of support during subsistence crises were weakened as food rents were converted to cash and more grain was consumed on the farm. More and more, chiefly support took the form of rent rebates rather than food handouts. Only exceptionally is it possible to find landowners buying in meal to offset crises of subsistence. On the other hand, market supply networks worked their way into the region only slowly. In fact, marketing brought new problems, as the more remote islands and even the more distant parts of the mainland began to suffer the downside of a market-based economy, the diseconomies of location. Smaller islands that had maintained large populations and equally large arable sectors, such as Pabbay, Taransay and Ensay off the southern and southwestern corner of Harris, were bound to suffer once production strategies were affected by prevailing market prices.
The marginality of production A second reason why the western Highlands and IsJands were risk-laden before 1820 lies in the low average levels of farm output. Where available data enable us to analyse returns on
seed, it is clear that most parts had only modest returns. Across the region, oats (the basic subsistence crop) averaged returns of 2-3 times on seed, whereas bear yielded 3-4 times, although locally there were islands like Tiree, where returns were lower. Systematic data collected for every township on the island during a 4 year period in the mid-1760s suggests yields of 1:2.2 for oats and 1:3.5 for bear (Dodgshon 1993). Although the island was reported as suffering from soil exhaustion by the mid-18th century, the figures were probably typical of many less fertile Hebridean islands by this point. To a degree, the widespread use of the spade or cashcrom for cultivation provided a small yield bonus in return for a heavy investment of labour compared with the plough. However, it still left communities with only modest surpluses after grain (approximately one-third of the total output) had been paid as rent and seed (again one-third) had been removed. In these circumstances, communities would have experienced crises more frequently, as even modest swings in climate or adverse weather patterns dug deep into what were narrow margins of subsistence. It is this lack of a comfortable annual surplus, coupled with the problems of grain storage in a physical climate that was increasingly hostile, that lay behind the subsistence problems of the region. As Dr Johnson put it during a visit to Mull, 'the consequences of a bad season here is not scarcity, but emptiness, and they whose plenty was barely a supply of natural or present need, when that slender stock fails, must perish with hunger' (Johnson 1971, p. 137). It was often said about the clan system in the western Highlands and Islands that it cultivated men more than land. This was another way of saying that landowners had a vested interest in populating their estates with men who could farm and fight, and therefore tried to absorb population growth back onto the land. This had the effect not just of expanding a clan's food base for its chief, but also of spreading subsistence across a wider range of environments. The outcome was that many environments that would not normally have been cultivated became settled, but at a risk. The extensive cultivation of river floodplain deposits, sandy soils or machair, steep slopes, high and exposed ground can all be documented, but so too can the environmental damage and subsistence crises that followed storms, flooding and the unexpected occurrence of unseasonable cold or wet weather. The scale of this damage can be reconstructed through contemporary sources. In an environment in which cultivable land was scarce, the cultivation of gravelly river floodplain deposits
ICELANDIC VOLCANIC ERUPTIONS IN THE SCOTTISH HIGHLANDS was always likely to be an attractive option. In terms of possible volcanic impacts on habitats, their soils were likely to be relatively nutrient enriched and buffered. The problem was that pushing cultivation close to river banks increased the risk of flooding and erosion during storms. Many communities must have taken a calculated view of the risks, accepting that (as in Glen Shira) where they grew oats one year, they fished for salmon another. Some estates tried to control the risks through restrictions on the cropping of land close to rivers and by compelling tenants to share in the task of building retaining walls. However, such controls could not have been effective because mid-18th century reports suggest that large amounts of arable had already been eroded in areas such as Coigach. Comparable problems of environmental stress were caused by over-ploughing and over-grazing of machair shell-sands in the Hebrides and the northern and western shorelines of the mainland. This environment was perhaps the most effectively buffered habitat in the region against the impact of volcanically derived acid volatiles. The scale of the erosion problem that affected these habitats is defined by a mid-18th century survey of Tiree. The survey reports over 1000 acres of land lost by sand-blows, whose impact reportedly owed as much to the pressure created by ploughing, grazing and the harvesting of plants for thatch, dyeing, etc., as to the power of storms (Scottish Record Office, Edinburgh, RHP 8826/2, General description of the Island of Tiree, by James Turnbull). Although not understating the ability of storms to destroy machair without any trigger from human activity, the latter undoubtedly accentuated the risks. Certainly, contemporary attempts to control the situation saw it as being as much a human as an environmental problem. The damage caused by the storms of 1697 was all the greater where it acted on fragile areas that had been under intensive cultivation. In the case of the former, its impact can be monitored through rentals for MacLeod of MacLeod's estate. The estate included among its possessions a cluster of islands to the south and east of Harris, whose land use included substantial amounts of machair pasture. Rentals drawn up at the close of the 17th century make it clear that islands such as Pabbay lost substantial amounts of arable (one estimate puts the figure at 300 acres) in the storms of 1697, the island being reassessed downwards from a capacity of 16 pennylands to only 10 (Dunvegan Castle, MacLeod of MacLeod Papers, 2/487/15, Rental of Harris 1698; 2/487/18, Rental of Harris 1703). The same storm devastated arable on the
275
western side of Bernaray, as well as Illeray on Baleshare, North Uist, with extensive parts of the sandy arable that lay to the north side of Illeray being swept away and other parts being overwhelmed by freshly blown sand (McKay 1980, p. 13). Later hurricane-force storms, such as the one in February 1749, were reported by Dr James Walker (1764; see McKay 1980) as having caused extensive damage to the arable on Barra, especially on the Eoligarry isthmus (Gilbertson et al 1996), now adjacent to the island's airstrip, with settlements having to be moved. The problem caused by its 'extensive Fields of blowing sand' that 'turn and wheel, and move over the Country, in very hurtfull way' was seen as endemic on the island thanks largely to its 'forbidding' climate and sandy soils, and created an annual deficit in grain for the island (Walker 1764; see McKay 1980, pp. 186-187). For townships on South Uist, there was a widespread temptation to harvest shell-sand from the seaward side of the township and to add it to the peatier soils inland. Thesum effect was that large areas of machair along the west side of the island were regularly destabilized.
Frequency of subsistence crises The frequency of subsistence crises, primarily understood in terms of climatic fluctuations in the Highlands and Islands, can only be appreciated when the wider human context is taken into account. This frequency can be established in a number of ways. Historical accounts provide generalized reference to the years when harvests were bad and starvation widespread across the country as a whole. From such sources, for instance, it is known that between 1550 and 1600, there occurred 17 bad harvests (i.e. 1550-1552, 1560, 1562-1563, 1568, 15711572, 1585-1587 and 1594-1596), and more occurred between 1630 and 1660 (i.e. 1622-1623, 1634-1635, 1640, 1649-1651 and 1655-1656) (Lythe 1960, pp. 16-23; Parry 1978, p. 162). Although it has been argued that poor harvests became less frequent after 1660 thanks to an improvement in yields, this was less true of the Highlands. In the latter, bad harvests were a feature of the late 1670s and of the so-called King William's lean or ill years, 1695-1702. Crop failures followed by subsistence crises are also reported for 1709, 1740-1741, 1745-1747, 1756, 1760-1761, 1778, 1782-1783, 1796 and 1799-1800. The coincidence of periods of famine with Icelandic volcanic eruptions is rare
276
R. A. DODGSHON, D. D. GILBERTSON & J. P. GRATTAN
Table 1. The incidence of famine in the Highlands of Scotland and volcanic eruptions 1550-1800 Period of famine
Total volcanic eruptions
Total volcanic eruptions VEI> 3
Total Icelandic volcanic eruptions
1550-1552 1560 1562-1563 1568 1571-1572 1585-1587 1594-1596 1622-1623 1634-1635 1640 1649-1651 1655-1656 1695-1707 1709 1740-1741 1745-1747 1756 1760-1761 1778 1782-1783 1796 1799-1800
20 6 5 1 4 15 10 5 7 9 21 6 30 6 6 8 1 12 6 15 9 20
0 0
1
0 0 0 0 0 0 0 0 0 0 0 1 5 0 0 1 0 0 0
1
0 0
1
0 0 2 1 1 0 2 1 1 1 0 1 0 0
1
2 0
1
(Table 1) and in all cases the eruptions follow rather than cause the initial distress. The opportunity for a more systematic understanding of how often Highland subsistence economies may have been stressed by fluxes in output is provided by available prices for basic foodstuffs. Data on oatmeal prices in Perthshire between 1630 and 1820 provide an excellent long-run series. They show that prices experienced sharp upward surges approximately once a decade throughout the 17th and for much of the 18th century, with at least 14 price peaks being identifiable between 1630 and 1780. By the late 18th century, trends are complicated by the general but strong inflection of prices that took place during the Napoleonic Wars, 17931815, but even during the Wars, the general price is broken by sudden and powerful surges in price (based on http://www.ex.ac.uk/~ajgibson/Scotdata/prices/fi). Of course, not all these sudden price jumps can be attributed to poor harvests. Other factors were involved, but marked yearto-year shifts in farm output are likely to have been at the root of such movements. Estate documents provide an obvious source of data for such subsistence crises. What they add to our perspective is the importance of the scale of the problem. On the one hand, region-
wide crises are well depicted through rentals and estate correspondence. The succession of poor harvests known as King William's lean or ill years, 1695-1702, have been shown to have killed many people in northeast Scotland, but there is also good reason to believe that many perished further west in the Highlands and Islands. Certainly Martin (1703) reported that they killed many of the poorer landholders and cottars in the Hebrides. Detailed rental data show that those rents actually collected fell dramatically, with arrears adding up to two-thirds of what was supposed to have been paid. Close examination of the available information concerning why this calamity occurred indicates that whole townships, such as Kelsay and Kilneave, lay waste for 1 or 2 years, while the crisis worked itself out. Others, like Upper Stincha, lay partially waste (Cawdor Castle, Campbell of Cawdor Papers Bundle 21, Rent of Hay 1703-1707). To the east, on the mainland of Argyll, subsistence problems were experienced during these crisis years by the inland townships of Glenorchy, with the Breadalbane estate having to arrange for meals to be distributed amongst the townships (Scottish Record Office, Campbell of Barcaldine Muniments, GDI70/629). Equally widespread in its effects was the subsistence crisis of the mid17408, 1744-1747. In Mull, Cregeen found that one-quarter of all the Duke of Argyll's farms on Mull were partially or wholly left to waste in 1747 because of poor harvests, as were many of the farms on the mainland around Loch Awe (Cregeen 1969, p. 127). To the northwest, correspondence amongst the Macleod of Macleod MSS also makes it clear that this same subsistence crisis led to famine on Skye. One letter written in 1745 talks of 'most' of the inhabitants of Trumpen More and Beg (Vatternish) being at risk of perishing 'for want of Bread' if relief was not provided within a few days (Dunvegan Castle, MacLeod of MacLeod Papers, 4/151, Letter 12 May 1745). The exceptionally poor season of 1722 also created famine problems for the Macleod of Macleod estate, with 'most part of the inhabitants' in Duirinish Waternish having 'nothing to eat or sow in yr Ground' (Dunvegan Castle, MacLeod of MacLeod Papers, 4/304/1, Letter 28 April 1772). In another letter, it was described as a 'universal' crisis, tenants being 'without Sowing without Bread to support Nature, without money or credit'. It goes on to say that 'hundreds will starve' (Dunvegan Castle, MacLeod of MacLeod Papers, 4/304/2). The widespread collapse of farm output during the poor harvest of 1782, and the loss of stock during the poor winter of 1782, shortly before the eruption of the Laki fissure
ICELANDIC VOLCANIC ERUPTIONS IN THE SCOTTISH HIGHLANDS the following summer, are particularly well documented. Indeed, their devastation in the Highlands prompted a Government Commission and led to the formation of the Highlands and Islands Agricultural Society with the express intention of improving the region out of such subsistence crises. Reports to the Commission suggested that the failure of the 1782 crop was brought about by a combination of a very poor summer for crop growth, followed inevitably by a very late harvest and then, soon after the harvest had begun, heavy snow fell. To compound matters, the following spring and summer were stormy and wet (Handley 1963, p. 72). At present, the significance of the impact of acid volatiles from Laki upon the highlands is largely unknown, beyond a comment of uncertain significance in Geikie (1893) that in 1783 so much volcanic ash fell in parts of Caithness that it was remembered as 'the year of the ashie'. Especially revealing are the reports from northern Scotland which suggest that many farmers and their families abandoned their farms and 'were forced to beg or perish' (Scottish Record Office, Edinburgh, E746.86, Letter from the Minister of Fodderty, 1783). Though the Commission provides evidence of grain and pease being shipped into the region, what stands out from the Commission's evidence and other sources is the extent to which one season's failure could precipitate such a severe subsistence crisis, and the inadequacy of local markets at this time to provide market-led solutions. In addition to providing information on largescale crises, these estate documents indicate that this region was afflicted by numerous local crises. This was in the nature of its natural environment, with storms or sudden downpours devastating crops in one glen, but not in another, or devastating crops on exposed farms but not on more sheltered farms. A survey of rental and estate accounts provides ample evidence for this sort of local disaster. In 1733, for example, eight tenants in the township of Barbreck Lochou petitioned the Earl of Breadalbane, 'showing their losses by a storm and asking for allowance, according to the custom of the country' (Scottish Record Office, Edinburgh, GDI 12/10/1/3/56). Likewise, in 1793, tenants on Seil and Luing, near Oban, reportedly lost half their crops through storms. As a result of the combination of major regional crises and intense local crises, the region is readily seen as risk laden. One reliable estimate for the 1760s, put the risk of a poor harvest as averaging one year in four (i.e. Burl's Letters (1754); see Jamieson 1876, Vol. 1, p. 158). Given the tight margin on
277
yields, such a frequency of risk must have easily turned to crisis when poor harvests followed each other. Conclusions Scottish Highland society during the historical period clearly operated in a continuum of risk, and it is within this continuum that the potential impact of volcanic ejecta must be considered. As a result of the particular conditions of the eruption and the synoptic meteorological situation, toxic and noxious volcanic acid gases and volatiles from Iceland would have been introduced into northern Britain in a patchy, discontinuous manner to habitats, people and communities that also varied greatly in their susceptibility to these inputs. It is evident that during the historical period, subsistence farming in the north and west of Scotland was an enterprise fraught with risks. These risks were environmental, social, political and economic, and were buffered by a complex support system and social obligations, which operated across wide areas. These features can be seen to have occurred, with varying degrees of success, throughout the historical period. In times of subsistence crisis, whole townships were abandoned, albeit temporarily. It is important to note that many people died of famine in the Highlands and Islands of Scotland without the injurious external influences of volcanic materials from Iceland. It can be seen above that famine was frequent and often severe, and that it only occasionally coincided, as in 1783, with the eruption of Icelandic volcanoes (Table 1). It is, however, telling to observe that the Laki fissure eruption of 1783 coincided with a famine that had already been taking place in Scotland for a year and that the contemporary documents concerning Scotland read by the authors give no mention of the role of volcanic gases or ashfall in either exacerbating the famine or its consequences; this despite the widely reported preoccupation across Europe with respiratory problems, plant and animal damage and spectacular atmospheric effects, now associated with this protracted volcanice ruption (Grattan & Charman 1994; Grattan & Brayshay 1995; Grattan et al. 1998). However, although this analysis suggests that explanations of widespread settlement and palaeoecological change in the British Isles that depend only upon Icelandic volcanic eruptions appear inadequate, it is clear that within the continuum of risk described above, Icelandic volcanic eruptions may on occasion have played
278
R. A. DODGSHON, D. D. GILBERTSON & J. P. GRATTAN
a role in intensifying subsistence crises in this region in the historical, and perhaps in the prehistoric, period.
References ANDRESSON, S. H. 1984. Athstoth Einokunarverslunarinnar. In: GUNNLAUGSSON, G. A., GUDBERGSSON, G. M., THORARINSSON, S., RAFFNSON, S. & EINARSSON, T. (eds) Skdftareldar 1783-1784. Mai Og Menning, Reykjavik [in Icelandic with English summaries]. AUSTIN, D. 1985. Dartmoor and the upland village of the South-West of England. In: HOOKE, D. (ed.) Medieval Villages: a Review of Current Work. Oxford University Committee for Archaeology Monographs, 5, 71-79. BAILLIE, M. G. L. 1988. Irish oaks record volcanic dust veils drama. Archaeology in Ireland, 2, 71-74. 1989. Hekla 3: how big was it? Endeavour, 13, 78-81. & MUNRO, M. 1988. Irish tree-rings, Santorini and volcanic dust veils. Nature, 332, 344-346. BELL, M. & WALKER, M. J. C. 1992. Late Quaternary Environmental Change: Physical and Human Perspectives. Longman, Harlow. BERESFORD, G. 1981. Climatic change and its effects upon the settlement and desertion of medieval villages in Britain. In: DELANO SMITH, C. & PARRY, M. L. (eds) Consequences of Climatic Change. University of Nottingham, Department of Geography, Nottingham, 30-39. BLACKFORD, J. J., EDWARDS, K. J., DUGMORE, A. J., COOK, G. T. & BUCKLAND, P. C. 1992. Icelandic volcanic ash and the mid-Holocene pine pollen decline in northern Scotland. Holocene, 2, 260-265. BLONG, R. J. 1984. Volcanic Hazards. Academic Press, London. BRUGMANS, S. J., 1787. Natuurkundige verhandeling over een zwavelagtigen nevel den 24 Juni 1783 innde provincie van stad en lande en naburige landen \vaargenomen. (A physical treatise on a sulphuric smog as observed on the 24th of July 1783 in the province of Groningen and neighbouring countries.) Leyden. BUCKLAND, P. C., DUGMORE, A. J. & EDWARDS, K. J. 1997. Bronze Age myths? Volcanic activity and human response in the Mediterranean and North Atlantic regions. Antiquity, 71, 581-593. BULL, K. 1991. Critical load maps for the U.K. NERC News, July, 31-32. BURGESS, C. B. 1984. The prehistoric settlement of Northumberland. In: BURGESS, C. B. & MIKET, R. (eds) Between and Beyond the Walls: Essays on the Prehistory of North Britain in Honour of George Jobey. Edinburgh University Press, Edinburgh, 126-175. 1985. Prehistoric climate and upland settlement. In: SPRATT, D. & BURGESS, C. (eds) Upland Settlement in Britain. British Archaeological Reports, 143, 195-230.
1989. Volcanoes, catastrophe and the global crisis of the late second millennium, BC. Current Archaeology, 111, 325-329. CAMUFFO, D., & ENZI, S. 1995. Impacts of clouds of volcanic aerosols in Italy during the last 7 centuries. Natural Hazards, 11, 135-161. CHARM AN, D. J., GRATTAN, J. P., WEST, S. & KELLY, A. 1995. Environmental response to tephra deposition in the Strath of Kildonan, northern Scotland. Journal of Archaeological Science, 22, 799-809. CREGEEN, E. R. 1969. The Tacksmen and their successes. A study of tenurial re-organisation on Mull, Morven and Tiree in the early eighteenth century. Scottish Studies, 13, 93-144. DEVINE, J. D., SIGURDSSON, H., DAVIS, A. N. & SELF, S. 1984. Estimates of sulfur and chlorine yield to the atmosphere from volcanic eruptions and potential climatic effects. Journal of Geophysical Research, 89, 6309-6325. DODGSHON, R. A. 1993. Strategies of farming in the western Highlands and Islands of Scotland prior to crofting and the clearances. Economic History Review, XLVI, 679-701. 1994. Budgeting for survival: nutrient flow and traditional Highland farming. In: FOSTER, S. & SMOUT, H. C. (eds) The History of Soils and Field Systems. Scottish Cultural Press, Aberdeen, 83-93. 1995. Modelling chiefdoms in the Scottish Highlands and Islands prior to the '45. In: ARNOLD, B. & GIBSON, B. (eds) Celtic Chiefdom: Celtic State. New Directions in Archaeology Series. Cambridge University Press, Cambridge, 99-109. DUGMORE, A. J. 1989. Icelandic volcanic ash in Scotland. Scottish Geographical Magazine, 105, 168-172. & NEWTON, A. J. 1992. Thin tephra layers in peat revealed by X-radiography. Journal of Archaeological Science, 19, 163-170. & 1996. Ideas and evidence from tephra. In: GILBERTSON, D. D., KENT, M. & GRATTAN, J. P. (eds) The Outer Hebrides: the Last 14,000 Years. Sheffield Academic Press, Sheffield, 45-50. , LARSEN, G. & NEWTON, A. J. 1995. Seven tephra isochrones in Scotland. Holocene, 5, 257-266. , NEWTON, A. J., EDWARDS, K. J., LARSEN, G., BLACKFORD, J. J. & COOK, G. T. 1996. Longdistance marker horizons from small-scale eruptions: British tephra deposits from the AD 1510 eruption of Hekla, Iceland. Journal of Quaternary Science, 11, 511-516. , & SUGDEN, D. E. 1992. Geochemical stability of fine-grained tephra in Iceland and Scotland. Journal of Quaternary Science, 1, 173-183. EDWARDS, K. J., BUCKLAND, P. C., BLACKFORD, J. J., DUGMORE, A. J. & SADLER, J. P. 1994. The impact of tephra: proximal and distal studies of Icelandic eruptions. Muncher Geographische Abhandlungen, B12, 108-126. GEIKIE, A. 1893. Text-Book of Geology. Macmillan, London. GILBERTSON, D. D. 1995. Studies of lithostratigraphy and lithofacies: a selective review of research
ICELANDIC VOLCANIC ERUPTIONS IN THE SCOTTISH HIGHLANDS developments in the last decade and their applications to geoarchaeology. In: BARHAM, A. J. & MACPHAIL, R. I. (eds) Archaeological Sediments and Soils: Analysis, Interpretation and Management. Institute of Archaeology, University College London, London, 99-144. , GRATTAN, J. P. & SCHWENNINGER, J.-L. 1996. A stratigraphic survey of the Holocene coastal dune and machair sequences. In: GILBERTSON, D. D., KENT, M. & GRATTAN, J. P. (eds) The Outer Hebrides: the Last 14,000 Years. Sheffield Academic Press, Sheffield, 72-101. GORHAM, E. 1987. Group summary report: wetlands. In: HUTCHINSON, T. C. & MEEMA, K. M. (eds) Effects of Atmospheric Pollutants on Forests, Wetlands and Agricultural Systems. Springer, Berlin, 631-636. GRATTAN, J. P. & BRAYSHAY, M. B. 1995. An amazing and portentous summer: environmental and social responses in Britain to the 1783 eruption of an Icelandic volcano. Geographical Journal, 161,125-134. & CHARMAN, D. J. 1994. Non-climatic factors and the environmental impact of volcanic volatiles: implications of the Laki Fissure eruption of AD 1783. Holocene, 4, 101-106. & GILBERTSON, D. D. 1994. Acid-loading from Icelandic tephra falling on acidified ecosystems as a key to understanding archaeological and environmental stress in northern and western Britain. Journal of Archaeological Science, 21, 851-859. & PYATT, F. B. 1994. Acid-damage caused by the Laki Fissure eruption - an historical review. Science of the Total Environment, 151, 241-247. , CHARMAN, D. & GILBERTSON, D. D. 1996. The environmental impact of Icelandic volcanic eruptions: a Hebridean perspective. In: GILBERTSON, D. D., KENT, M. & GRATTAN, J. P. (eds) The Outer Hebrides: the Last 14,000 Years. Sheffield Academic Press, Sheffield, 51-58. , GILBERTSON, D. D. & CHARMAN, D. J. \999a. Modelling the impact of Icelandic volcanic eruptions upon the prehistoric societies of northern and western Britain. In: FIRTH, C. & McGuiRE, W. (eds) Volcanoes in the Quaternary. Geological Society, London, Special Publications, 161, 147-160. , & DILL, A. 1999/7. 'A fire spitting volcano in our dear Germany': documentaty evidence for a low-intensity volcanic eruption of the Gleichberg in 1783.' This volume. HALL, V. A., PILCHER, J. R. & MACCORMAC, F. G. 1993. Tephra-dated lowland landscape history of the north of Ireland, AD 750-1150. New Phytologist, 125, 193-202. , & 1994. Icelandic volcanic ash and the mid-Holocene Scots pine (Pinus sylvestris} decline in the north of Ireland: no correlation. Holocene, 4, 79-83. HAMMER, C. U., CLAUSEN, H. B. & DANSGAARD, W. 1980. Greenland ice sheet evidence of post-glacial volcanism and its climatic impact. Nature, 288, 230-235. HANDLEY, J. E. 1963. The Agricultural Revolution in Scotland. Burns, Glasgow.
279
JAMIESON, R. (ed.) 1876. Burl's Letters from the North of Scotland, first published in 1754. 2 vols. Edinburgh. JOHNSON, S. 1971 edition. A Journey to the Western Islands of Scotland. Yale University Press, New Haven, CT. KELLY, P. M., WIGLEY, T. M. L. & JONES, P. D. 1984. European pressure maps for 1815-16, the time of the eruption of Tambora. Climate Monitor, 13, 76-91. KINGTON, J. A. 1988. The weather for the 1780s over Europe. Cambridge University Press, Cambridge. LAMB, H. H. 1992. First essay at reconstructing the general atmospheric circulation in 1816 and the early nineteenth century. In: HARINGTON, C. R. (ed.) The Year Without a Summer. Canadian Museum of Nature, Ottawa, Ont., 355-357. LOWE, J. J. & TURNEY, C. S. M. 1997. Vedde Ash layer discovered in a small lake basin on the Scottish mainland. Journal of the Geological Society, London, 154, 605-612. LYTHE, S. G. E. 1960. The economy of Scotland in its European setting 1550-1625. Oliver and Boyd, Edinburgh. MARTIN, M. 1703. A Description of the Western Isles of Scotland, reprinted 1934. Mackay, Stirling. MASS, C. F. & PORTMAN, D. A. 1989. Major volcanic eruptions and climate: a critical evaluation. Journal of Climate, 2, 566-593. MCCORMICK, P. M., THOMASON, L. W. & TREPTE, C. R. 1995. Atmospheric effects of the Mt. Pinatubo eruption. Nature, 373, 399-404. McKAY, M. M. (ed) 1980. The Rev. Dr John Walker's Report on the Hebrides of 1764 and 1771. John Donald, Edinburgh. OGILVIE, A. E. J. 1986. The climate of Iceland 17011784. Jokull, 36, 57-73. PALAIS, J. M. & SIGURDSSON, H. 1989. Petrologic evidence of volatile emissions from major historic and pre-historic volcanic eruptions. In: BERGER, A., DICKINSON, R. E. & KIDSON, J. W. (eds) Understanding Climate Change. Geophysical Monograph, American Geophysical Union, 52, 31-53. PARKER PEARSON, M. 1993. Bronze Age Britain. Batsford-English Heritage, London. PARRY, M. L. 1975. Secular climatic change and marginal land. Transactions of the Institute of British Geographers, 64, 1-13. 1978. Climatic Change, Agriculture and Settlement. Dawson, Folkestone. 1981. Evaluating the impact of climatic change. In: DELANO SMITH, C. & PARRY, M. L. (eds) Consequences of Climatic Change. University of Nottingham, Department of Geography, Nottingham, 3-16. 1985. Upland settlement and climatic change: the medieval evidence. In: SPRATT, D. & BURGESS, C. B. (eds) Upland Settlement in Northern Britain. British Archaeological Reports, 143, 195-229. & CARTER, T. R. 1985. The effect of climatic variations on agricultural risk. Climatic Change, 1, 95-100.
280
R. A. DODGSHON, D. D. GILBERTSON & J. P. GRATTAN
PETURSSON, G., PALSSON, G. A. & GEORGSSON, G. 1984. Um Eiturahrif. In: GUNNLAUGSSON, G. A., GUDBERGSSON, G. M., THORARINSSON, S., RAFFN-
SON, S. & EINARSSON, T. (eds). Skdftareldar 17831784. Mai Og Menning, Reykjavik [in Icelandic with English summaries]. PILCHER, J. R. & HALL, V. A. 1992. Towards a tephrochronology for the Holocene of the north of Ireland. Holocene, 2, 255-259. & 1996. Tephrochronological studies in northern England. Holocene, 6, 100-105. , BAILLIE, M. G. L., SCHMIDT, B. & BECKER, B. 1984. A 7272-year tree-ring chronology for western Europe. Nature, 312, 150-152. , HALL, V. A. & MACCORMAC, F. G. 1995. Dates of Holocene eruptions from tephra layers in Irish peats. Holocene, 5, 103-110. REGISTER OF THE PRIVY COUNCIL OF SCOTLAND, X, 1613-1616. 1891. H. M. Register House, Edinburgh. RENFREW, C. 1990. Climate and Holocene culture change: some practical problems. Philosophical Transactions of the Royal Society of London, Series A, 330, 657-663. ROBOCK, A. & MAO, J. 1995. The volcanic signal in surface temperature observations. Journal of Climate, 8, 1086-1103. SIMKIN, T., & SIEBERT, L. 1994. Volcanoes of the World, 2nd edn. Geoscience Press, Tucson, AZ. SKIBA, U., CRESSER, M. S., DERWENT, R. G. & FUTTY, D. W. 1989. Peat acidification in Scotland. Nature, 337, 68-69.
SMITH, C. M. S., CRESSER, M. S. & MITCHELL, R. D. J. 1993. Sensitivity to acid deposition of dystrophic peat in Great Britain. Ambio, 22, 22-26. STOTHERS, R. B. 1996. The Great Dry Fog of 1783. Climatic Change, 32, 79-89. THORARINSSON, S. 1967. The Eruption of Hekla 19471948. The Eruptions of Hekla in Historical Times. H. F. Leiftur, Reykjavik. 1971. Damage caused by tephra fall in some big Icelandic eruptions and its relation to the thickness of the tephra layers. In: KALOYEROPOYLOY, A. (ed.) Acta of the First International Scientific Congress on the Volcano Thera, September 1969, 213-236. 1979. On the damage caused by volcanic eruptions with special reference to tephra and gases. In: SHEETS, P. D. & GRAYSON, D. K. (eds) Volcanic Activity and Human Ecology. Academic Press, New York, 125-159. 1981. Greetings from Iceland. Ash-fall and volcanic aerosols in Scandinavia. Geografiska Annaler, 63A, 109-118. THORDARSON & SELF, S. 1993. The Laki [Skaftar Fires] and Grimsvotn eruptions in 1783-85. Bulletin Volcanologique, 55, 233-263. VAN SWINDEN, M. 1786. Observations sur quelques particularites meteorologiques de 1'annee 1783. Memoir es de I'Academic Roy ale des Sciences, Turin, 1784-1785, 113-140. YOUNG, R. & SIMMONDS, T. 1996. Marginality and the nature of later prehistoric upland settlement in the north of England. Landscape History, 17, 5-16.
Comparison and cross-checking of historical, archaeological and geological evidence for the location and type of historical and sub-historical eruptions of multiple-vent oceanic island volcanoes S. J. DAY 1 , J. C. CARRACEDO2, H. GUILLOU3, F. J. PAIS PAIS4, E. RODRIGUEZ BADIOLA5, J. F. B. D. FONSECA6 & S. I. N. HELENO6 1
Benfield Greig Hazard Research Centre, Department of Geological Sciences, University College London, Gower Street, London WC1E 6BT, UK
2
Estacion Volcanologica de Canarias, CSIC, Aptdo. Correos 195, La Laguna 38206, Tenerife, Canary Islands, Spain 3
Centre des Faibles Radioactivites, Laboratoire Mixte 91198 Gif-Sur-Yvette,
4
University of La Laguna, La Laguna, Tenerife, Canary Islands, Spain 5
6
CEA-CNRS,
France
Museo Nacional de Ciencias Naturales, CSIC, Madrid, Spain
Grupo de Fisica da Terra e do Ambiente Instituto Superior Tecnico, Lisbon, Portugal Abstract: Oceanic island volcanoes, like many others, have many small vents scattered over their flanks in addition to, or in place of, large summit vents. These small vents are commonly monogenetic and many eruptions of this type of volcano involve activity at more than one such vent: lines of volcanic cones are often produced by eruptions fed by dykes, and if more than one dyke is emplaced during an eruption these vents can be along different alignments, and many kilometres apart. Identifying which vents were produced in which eruptions is an important problem in reconstructing the development of multiplevent volcanoes. Reconstruction of historical and sub-historical eruptions of two oceanic island volcanoes, the Cumbre Vieja volcano on La Palma in the Canary Islands and Cha das Caldeiras volcano on Fogo, Cape Verde Islands, has indicated that historical eyewitness accounts and archaeological evidence can be extremely valuable adjuncts to detailed geological studies. In contrast, secondary accounts including folk memories and earlier accounts in the scientific literature are commonly inconsistent both with the eyewitness accounts and with the results of detailed geological mapping, archaeological evidence or other historical documents. A common source of error is confusion of the vents of historical eruptions with older but larger or more prominent volcanic vents that lie along the same line of sight as viewed from adjacent settlements or from convenient viewpoints. Examples of this are provided by mis-locations of the vents of the AD 1677 eruption on La Palma and of some of the vents of the AD 1951 eruption on Fogo. Another problem arises when the location or style of eruption on a volcano has changed in early historical time, as has occurred on Fogo. The differences between early historical accounts of eruptions on this volcano and the more detailed accounts of more recent eruptions has led to the discrediting of the former by some researchers, whereas geological studies have supported the early historical accounts.
Most oceanic island volcanoes, and many other dominantly basaltic volcanoes, have a large number of small volcanic vents on their flanks. These vents are typically monogenetic (formed in a single eruption) and feed lava flows that make up the bulk of the volume of the volcano and commonly present the dominant (and certainly most frequent) volcanic hazard. A discrete summit vent
complex is present on some volcanoes of this type: examples include the Teide-Pico Viejo summit complex on Tenerife, Canary Islands (Ancochea etal 1990), and the Pico do Fogo summit cone on Fogo, Cape Verde Islands (see discussion below), In other cases, however, the summit crater complex is a relatively minor feature superimposed on a deeper rift system and largely resulting from
From: McGuiRE, W. G., GRIFFITHS, D. R., HANCOCK, P. L. & STEWART, I. S. (eds) The Archaeology of Geological Catastrophes. Geological Society, London, Special Publications, 171, 281-306. 1-86239-062-2/00/ $15.00 © The Geological Society of London 2000.
282
S. J. DAY ET AL.
subsidence as a result of dyke injection along the rifts (such as the caldera of Kilauea, in Hawaii; Ryan et al. 1983), and in many cases, such as the Cumbre Vieja volcano of La Palma and the recent volcanic edifice of El Hierro, both in the Canary Islands, there is no summit complex at all (Carracedo 1994, 1996^,6; Carracedo et al. 1997). The summit regions of these volcanoes are instead formed by the intersection of the volcanic rift zones in which the vents are concentrated. These rift zones overlie the swarms of dykes along which the magma erupted at the monogenetic vents ascends to the surface. The positions of the dyke swarms and volcanic rift zones on volcanoes of this type is a function
of their overall structure. It is well established that the orientations of dykes are closely related to the principal stress directions in the regions through which they propagate. The most notable papers on this subject include those by Anderson (1935), Nakamura (1977), Pollard (1987), Chevalier & Verwoerd (1988), McGuire & Pullen (1989) and Tibaldi (1995). Carracedo (1994, 19960, b) has argued that in many cases oceanic island volcanoes have a triple-rift ('Mercedes Star') structure governed by radial deformation above an inflating subvolcanic magma reservoir. However, Carracedo et al. (\991a,b, in prep.) and Day et al. (1996, in prep.) have shown in a detailed study of the Cumbre Vieja volcano,
Fig. 1. Location map of southern La Palma, showing the distribution of historical and young prehistoric vents, with names and ages indicated.
ERUPTIONS ON MULTIPLE-VENT VOLCANOES La Palma, that this triple-rift structure may be modified on time-scales of thousands of years or less in a process of rift reorganization caused by weakening of the edifice and consequent nearsurface dominance of topographic-gravitational stresses (McGuire & Pullen 1989). Such a process may be an important indicator of volcano instability and a long-term precursor to giant lateral collapse of the edifice concerned. The feeder dyke swarms of the Cumbre Vieja volcano, like those of many other highly active volcanoes, are not exposed. In such cases it is necessary to infer their positions and orientations from the distributions of surface volcanic vents and their elongations and other morphological features, using methods described by Nakamura (1977) and Tibaldi (1995), respectively. An important part of this procedure is to identify vents formed in the same eruptions, so as to correctly infer the sub-surface dyke geometry. Where the vents are close together or overlapping it is possible to show that they are co-eruptive by detailed mapping of the vents and their eruptive products. However, where they are more widely spaced it is generally only possible to show that they are related from historical accounts. In a number of cases, such as those of the 1949 eruption of the Cumbre Vieja (Bonelli Rubio 1950), the 1785 and 1951 eruptions of Fogo (Feijo 1786; Ribeiro 1960; see also below) and the three eruptions of 17041706 in Tenerife (Fritsch & Reiss 1868), vents identified as being active in the same eruption are kilometres or even tens of kilometres apart. In some cases, notably the 1949 Cumbre Vieja eruption and the early 18th century activity on Tenerife, even the vent elongation directions are not the same because the vents are fed by dykes emplaced into distinct stress domains. Analysis and reinterpretation of historical accounts of volcanic eruptions in the light of geological studies is a standard procedure in volcanological and volcanic hazard studies. The historical accounts contribute in particular to an understanding of precursory seismicity, toxic gas emissions and other aspects of the eruptions that leave little or no geological evidence at the surface. They also provide important data on the absolute duration and timing of events within periods of eruptive activity. Recent outstanding examples include the reinterpretation of the AD 79 eruption of Vesuvius (Sigurdsson et al. 1982, 1985). However, these studies typically deal with central-vent volcanoes, in which case there is little doubt about the location of the volcanic vents. In the case of multiple-vent volcanoes, in contrast, vent locations are commonly uncertain
283
and different accounts and the geological evidence may provide conflicting vent locations. In such cases it is necessary to be able to test the reliability of historical accounts and related evidence such as local traditions with regard to such basic features of historical eruptions as the location of the vents involved. It should be noted that if a vent is wrongly identified as belonging to a particular historical eruption all subsequent geological work on that vent, and the volcanic rocks erupted from it, will inevitably give an erroneous impression of that eruption. Incorrect location of vents may also obscure the reconfiguration of volcanic rift zones. In this paper we show how historical accounts of various types, including eyewitness accounts, early maps and local traditions in the settlements around the eruption sites, may be tested using geological and archaeological evidence.
The 1677 eruption of the Cumbre Vieja volcano, La Palma Six out of the 12 historical eruptions in the Canary Islands, in the five centuries since occupation of the islands by the Spanish, have occurred on the Cumbre Vieja volcano of La Palma. This volcano forms the southern third of this island, rises almost 2 km above sea level and has a subaerial volume of some 200km 3 ; it also probably extends well below sea level. All radiometrically dated rocks from the Cumbre Vieja are less than 130 ka old (Carracedo et al. 1996, 1991 a, b; Guillou et al. submitted) and it is at present the most active volcano in the archipelago. Detailed mapping (Carracedo et al. 19976, submitted) has shown that since about 7000 years BP, activity of the Cumbre Vieja has been concentrated along the main N-S ridge of the volcano and from E-W trending en echelon fissure swarms on its western flank. Of the historical eruptions, those of 1646, 1677, 1949 and 1971 involved eruptions from the main ridge, as did a prehistoric eruption c. AD 1480 (Hernandez Pacheco & Vails 1982); whereas west flank fissures were active in 1585, 1712 and 1949 (see Fig. 1 for locations of these eruptions). A number of these eruptions have been poorly located, but in this paper we concentrate upon the 1677 eruption. The 1677 eruption occurred close to the town of Fuencaliente at the southern end of the island: the name of the town alludes to a hot spring on the coast nearby, the Fuente Santa, which before its destruction in the eruption formed the basis of a famous medicinal spa. Local tradition (repeated in a number of previous works
284
S. J. DAY ET AL.
(Hernandez Pacheco & Vails 1982; Romero 1991) as well as in tourist literature) has it that the large volcanic cone of San Antonio formed in this eruption. This cone, one of the two or three largest at present visible in the Cumbre Vieja, is mainly composed of scoria and spatter but has a phreatomagmatic surge deposit of several metres thickness preserved on its rim and western flank (see Fig. 4, below). The existence of this deposit provided the initial
geological clue that the local tradition might be in error. The occurrence of a surge-forming explosive phase in the 1677 eruption would have inflicted considerable damage upon Fuencaliente, less than 1 km away. No such disaster has been recorded. Following the recognition of this anomaly, detailed geological mapping, radiometric dating and a search for eyewitness accounts of the eruption were carried out. The results of this work have been described in
Fig. 2. Map of the pre-1971 geology of southernmost La Palma, showing the various units of the 1677 eruption.
ERUPTIONS ON MULTIPLE-VENT VOLCANOES detail by Carracedo et al (1996); here they are summarized and compared with the results of archaeological investigations carried out independently by the fourth author.
Geological mapping and radiometric dating Geological mapping was carried out partly on the ground and partly by photogeological interpretation of pre-1971 aerial photographs, as a
285
large part of the 1677 lava field was covered by the products of the 1971 Teneguia eruption (Afonso et al. 1974). The results of this work are shown in Fig. 1. The mapping shows that the two youngest pre-1971 vents in this area are a small, NW-SE elongated scoria cone produced by strombolian eruptive activity on the NE side of the San Antonio cone at about 600m above sea level, and a group of spatter vent fissures on the SW side of the San Antonio cone at about 500m
Fig. 3. Views of the 1677 eruption vents and of prehistoric rocks around the San Antonio cone, (a) General view from Fuencaliente of the San Antonio cone and of the upper (northeastern) vent of the 1677 eruption. 'Photo' marks the location of (d). (b) View of the cluster of spatter vents of the 1677 eruption, located on the southwest side of the San Antonio cone. The proximity of these vents to an old sea-cliff should be noted; the coastal exposures of the 1677 lavas lie at the foot of this cliff (see Fig. 2). (c) View from the southwest rim of the interior of the San Antonio crater, with Fuencaliente in the background. The collapse of the southern flank of the 1677 strombolian cone and of the older phreatomagmatic deposits have filled the crater to its present depth, (d) Road cut in the flat ground between the San Antonio cone and Fuencaliente (see (a)), showing strombolian pyroclastic deposits (lapilli and scoria) of the 1677 eruption resting directly on Fuencaliente vents lavas, without intervening pyroclastic deposits from the San Antonio cone, whose rim is less than 200 m from this point.
286
S. J. DAY ET AL.
Fig. 3. (continued}
above sea level (Fig. 3a and b). The latter fed a total of eight discrete lava flows, which together make up the young pre-1971 lava field and lava platform at the coast (Fig. 2). Deposits of the upper, northeastern vent are well exposed on the NE rim of the San Antonio crater, suggesting that they have in part collapsed into it, and clearly overlie weathered phreatomagmatic deposits and other rocks of the old cone (Fig. 3c). There is no evidence to support the interpretation (Hernandez Pacheco & Vails 1982) of young vents on the coast about 2km north west of Fuencaliente as also having been formed in the 1677 eruption. These vents appear to be hornitos or rootless vents in a coastal lava platform produced by lavas erupted from a
group of sub-historical vents upslope from Las Indias: one of these lavas has been dated at 3 ± 2 ka using the K-Ar technique (Carracedo et al \991a, b; Guillou et al submitted). The San Antonio cone can also be shown to be older than a number of other eruptive units in the area. The cone is a substantial topographic feature, rising some 200m above the surrounding terrain on its south and west sides and some 1200m across. A number of younger lavas from vents located within the present-day town of Fuencaliente (Fuencaliente vents lavas in Fig. 2) flowed downhill to the San Antonio cone and were deflected around it; they also overlie a lava erupted from the foot of the San Antonio cone on its western side. They are therefore inferred
ERUPTIONS ON MULTIPLE-VENT VOLCANOES to be younger than it, and this is confirmed by the lack of phreatomagmatic surge deposits lying on top of the Fuencaliente vents lavas (Fig. 3d). These deposits are at present confined to the rim and western flank of the cone (Fig. 4) but their thickness and well-developed low-angle surge cross-bedding indicate the occurrence of a violently explosive phase towards the end of the eruption that formed the San Antonio cone,
287
which would have spread thinner, more distal equivalents of these deposits over a wide area. Neither the San Antonio cone nor the Fuencaliente vents lavas have been dated as yet, but a still younger group of lavas, the Montana del Fuego lavas, have been dated at 4 ± 3 k a (2cr error) by the K-Ar method and by two independent 14C datings, 3255 ±140 a and 3350 ± 50 a (2cr errors) (Carracedo et al. 1991 a, b;
Fig. 4. Phreatomagmatic deposits of the San Antonio cone. (A) Bedded phreatomagmatic deposits perched on the western flank of the San Antonio cone; (B) surge cross-bedding within these deposits.
288
S. J. DAY ET AL.
Guillou et al. submitted). It is therefore likely that the San Antonio cone is substantially more than 3000 years old, consistent with the welldeveloped weathering beneath the younger strombolian deposit (Fig. 3c). The geological evidence therefore indicates that the San Antonio cone is a relatively old feature and the most plausible candidates for the vents of the 1677 eruption are the much smaller vents on the NE and SW sides of the San Antonio cone. In contrast to the San Antonio eruption, which had a violently explosive phase as noted above, the relatively mild effusive and strombolian activity indicated by the geological features of these vents is consistent with the eyewitness accounts of the eruption, as will be shown below. Archaeological evidence Further evidence for the pre-1677 origin of the San Antonio cone comes from archaeological excavations around the cone. The pre-Hispanic
culture of the Canary Islands, associated with a people of North African extraction, the Guanches, disappeared very soon after the occupation of La Palma by the Spanish in the last decade of the 15th century AD. Occurrences of Guanche pottery and other artefacts therefore provide a stratigraphic indicator of a pre-1500 age. Figure 5 shows the distribution of Guanche finds around the San Antonio cone, including pottery; petroglyphs (distinctive carvings on rock faces); post-hole circles and other remains of the foundations of clusters of Guanche dwellings; and artificial caves excavated in soft but cohesive pyroclastic units such as the phreatomagmatic ashes and used as refuges by the Guanches. The distribution of these artefacts and dwelling-sites clearly indicates the pre-Hispanic age of the San Antonio cone and the Fuencaliente vents lavas. The only units that demonstrably overlie Guanche remains are the lava flows from the SW vents of the 1677 eruption.
Fig. 5. Map of the distribution of Guanche archaeological sites and finds around the San Antonio cone.
ERUPTIONS ON MULTIPLE-VENT VOLCANOES
Eye-witness accounts of the eruption and other historical documents The earliest document relevant to the location of the 1677 eruption dates from some 80 years earlier. A map produced in 1586 and accompanying an account of the 1585 eruption at Jedey, some 10km north of Fuencaliente (Torriani 1592), shows a hill named San Antonio just south of Fuencaliente and in approximately the same location as the volcanic cone given that name today. Torriani was a military engineer tasked with construction of fortifications in the island; his skills would have included surveying. Other named topographic features on his map are correctly located, giving some confidence in the accuracy of his positioning of the San Antonio cone. This documentary evidence for the existence of the San Antonio cone before 1585 is consistent with the geological and archaeological evidence that it pre-dates the occupation of La Palma by the Spanish. The contemporary accounts of the 1677 eruption were written by the civil and ecclesiastical authorities of the area and must therefore be interpreted with care. None the less four accounts in particular have survived which can be used in conjunction with geological evidence to reconstruct the course of the eruption: •
•
•
•
Relaciones (Accounts) written by the priest Juan Pinto de Guisla, describing the 1646 (Volcan Martin) and 1677 eruptions of the Cumbre Vieja. A copy of the original, transcribed in 1806, has been given by Lorenzo Rodriguez (1987); translations into German and French were published by Von Buch (1825) and Von Buch (1836), respectively. Gazeta del Ayuntamiento de La Palma (Gazette of the La Palma community council) of 2 December 1677, describing the development of the eruption from 13 to 26 November 1677 and written by an anonymous council official (Anonymous 1677). A copy of the original again has been published by Lorenzo Rodriguez (1987); a translation into French has been published by Webb & Berthelot(1839). Cartas (letters) sent by the chief bailiff of the Spanish Inquisition in La Palma, Antonio Pinto de Guisla, to Bishop B. Garcia Ximenez, on 29 November, 6 and 24 December 1677, and 17 and 27 January 1678. Carta sent by the vicar Melchior Brier to the same Bishop, dated 30 November 1677. The originals of these last two have been lost but copies are preserved in the archives of the Bishop (Anonymous 1678).
289
The single most important piece of information in these accounts describes how the people of Fuencaliente went to the top of Montana del Corral (the name used for Montana San Antonio in that account) to view vents that had opened at its base and that, fortunately after they had returned, a large vent opened in the mountain, which 'had it caught them unawares would have engulfed them in a voracious fire'. This clearly indicates that two vents were involved in the eruption, a lower one at the base of the San Antonio cone and an upper one near its summit. These positions match those of the two young vents on either side of the San Antonio crater and indicate that both were active in the 1677 eruption. The various accounts describe the early phases of the eruption, from its start near sunset on 17 November 1677 until the destruction of the Fuente Santa by lava flows on 26 November, in some detail. The eruption was preceded by intense seismic activity from 13 to 15 November, and emission of 'hot air with a smell of sulphur' from fissures that developed on either side of Montana del Corral/Montana San Antonio from 15 November onwards. These fissures were described as opening in a few minutes to a width 'difficult to jump'. Their opening was accompanied by strong earthquakes, which also caused the collapse of the church tower in Fuencaliente. The earthquakes, and the unusual opening of the fissures some days before to the eruption, may reflect bulging of the steep slopes south of Fuencaliente as a dyke was emplaced beneath them. The 1677 vents lie on an overall trend bearing about 030°, which may indicate the orientation of this dyke at depth, whereas the NW-SE trend of the individual vents (Fig. 2) is parallel to the local topographic contours and indicates the influence of topographicgravitational stresses. Subsequent events are shown in map and schematic form in Figs 6 and 7. The Fuente Santa was destroyed on 26 November, apparently by the southern branch of flow LIII-2 (Fig. 2), and with the loss of this important economic resource interest in the activity declined. It will be noted that two out of the four sets of documents listed above deal solely with the first 2 weeks of the eruption, which continued for a further 2 months afterwards. It appears that once the most serious damage had been done, interest in the eruption declined as no further resources of note were threatened. This is the converse of the situation in the later and much larger 1730-1736 eruption of Lanzarote (Carracedo & Rodriguez Badiola 1991; Carracedo et al 1992), where the poorly
290
S. J. DAY ET AL.
Fig. 6. Maps showing the different stages in the development of the San Antonio area in general and of the 1677 eruption in particular; it should be noted that the distribution of lavas in areas covered by the 1971 Teneguia eruption is based on interpretation of pre-1971 aerial photographs.
ERUPTIONS ON MULTIPLE-VENT VOLCANOES
291
Fig. 7. Time-line depicting the events of the 1677 eruption, based on the historical eyewitness accounts cited in the text.
documented post-1731 phases of the eruption occurred after complete evacuation of the area of the eruption and an at least partial breakdown of the local authority. Throughout the eruption the upper vent appears to have emitted lapilli scoria and spatter in strombolian style activity, with three pronounced explosions toward the end of the first
stage of activity on 23 November. These explosions were preceded by seismic activity, which may record drainback of magma from the upper vent, as activity at the upper vent ceased for 2 days from 23 to 25 November. Gases (most probably carbon dioxide) emitted from the upper vent seem to have ponded in depressions in the hummocky ground between San Antonio
292
S. J. DAY ET AL.
and Fuencaliente itself: the deaths by asphyxiation of cattle, rabbits, birds and a shepherd were recorded in December 1677. The lower vents erupted a series of lava flows from successive vents slightly offset from one another (Figs 2 and 6), with accumulations of spatter around the vents themselves. In one of the earliest recorded experiments in volcanology, markers were emplaced at known intervals in advance of a lava flow front, and the velocity of its advance was estimated as about 20 m in half an hour, or about 0.01 m s"1; this value is similar to more recent estimates of flow front velocity for basaltic lava flows (Kilburn et al. 1995). There may have been hiatuses or periods of reduced effusion rate at the lower vents, but details are scarce. The last phase of the eruption, producing the L IV-1 and -2 flows shown in Fig. 2, ended abruptly on 21 January 1678.
Why was the San Antonio cone erroneously identified as the vent of the 1677 eruption? Although the detailed mapping and other studies described above were carried out only very recently, the eyewitness accounts of the 1677 eruption and other relevant documents were not lost, and indeed were widely published and translated in the early 19th century, by Von Buch (1825, 1836) in particular. The problem, therefore, is why the main San Antonio cone has come to be identified as the vent of the 1677 eruption in local tradition and both the scientific and popular literature. A hint is provided by the view of the San Antonio cone from the north side of Fuencaliente, shown in Fig. 3a. In this view the northeastern vent of the 1677 eruption lies just in front of the older cone, whereas the lower, southwestern vents are on the opposite side (see also Fig. 2). Thus the 1677 vents and the older San Antonio cone all lie along the same line of sight from Fuencaliente, and furthermore the older cone is the most prominent topographic feature in that direction. It is therefore easy to become confused about what precisely is being pointed out when the position of the eruption is indicated from viewpoints in the vicinity of Fuencaliente. Although this 'line of sight' error could have been corrected on the basis of a detailed examination of the contemporary eyewitness accounts alone, or on the basis of the geological and archaeological evidence alone, we emphasize that the use of both archival and field work together provides a much stronger case for the interpretation of the geology of the San Antonio
cone and its surroundings that we have outlined here, and has removed many ambiguities from the interpretation. The reconfiguration of the Cha das Caldeiras volcano, Cape Verde Islands, within historical time and consequences for interpretation of early historical accounts The Cha das Caldeiras volcano (also known as the Pico do Fogo, but this name is misleading for reasons discussed by Day et al. (submitted)) is on the island of Fogo in the Cape Verde archipelago. It is by far the most active volcano in that archipelago and the only one from which historical eruptions have been recorded. Like the Cumbre Vieja, it is an essentially alkali basaltic volcano, and during the five centuries since first colonization of the Cape Verdes by the Portuguese as many as 26 eruptions have been recorded. Although some of these records are doubtful, it is likely that some minor eruptions may have gone unrecorded in the earlier part of the historical period. This is because organized colonization of the eastern, volcanically active part of the island did not take place until the end of the 18th century, and the remote region within which the summit of the active volcano lies was not occupied until several decades later still. For much of the earlier period, settlement was confined to the vicinity of Sao Felipe in the southwest of the island. The 1995 eruption on Fogo has led to a number of studies at present at various stages of completion (IICT 1997; Heleno da Silva & Fonseca 1999). An interim account of the eruption has been provided by Silveira et al. (1995). Older published works include those by Machado (19650, b), Machado & Torre de Assuncao (1965) and Torre de Assuncao et al. (1967), but perhaps the most detailed work on the island is that by Ribeiro (1960). This provides a detailed description of the geography of the island, as well as an invaluable summary of the historical accounts of volcanic and seismic activity and a detailed description of the 1951 eruption. Many of the eyewitness accounts of early (pre1785) historical eruptions found by Ribeiro were written by sea-captains and other mariners. The first scientific account of an eruption was written by a chemist and natural scientist, Feijo, who wrote a valuable account of the 1785 eruption (Feijo 1786), which is discussed further below. Although he catalogued the earlier accounts, Ribeiro (1960) questioned their accuracy because the descriptions were inconsistent with
ERUPTIONS ON MULTIPLE-VENT VOLCANOES the locations and styles of activity of the more recent and more precisely described eruptions. It is therefore of interest to consider whether the geology of the volcano supports Ribeiro's scepticism or whether it indicates that a change in the style and distribution of activity has genuinely taken place. The more recent activity has involved multiple-vent eruptions and the formation of numerous monogenetic volcanic vents and provides a number of instances where vents have been mis-located and lost entirely for reasons that are also of interest in the context of this paper. The geology of Fogo island and the Cha das Caldeiras volcano The Cape Verde archipelago as a whole is associated with a mantle plume: the best account of its tectonic setting has been provided by White (1989). The present-day location of the plume is inferred from the geoid anomaly and other geophysical data to be in the southwest of the archipelago, close to the position of Fogo island and the adjacent island of Brava, which contains numerous morphologically young volcanic vents and has experienced numerous episodes of seismic unrest in historical time (Heleno da Silva et al 1997) although no eruptions have occurred there. Fogo island is located at the eastern end of a submarine platform connecting it and Brava to the west. It is roughly circular in shape and some 25km across. Limited exposures of basement rocks on both islands indicate that this platform is composed of a pyroxenite-carbonatite intrusive complex and associated volcanic rocks. Built up on this platform and its steep submarine flanks and forming most of the island is a steepsided volcanic edifice, the Monte Amarelo volcano. This edifice is composed of alkaline basic and intermediate lavas (nephelinites, basanites and tephrites), with rare phonolitic lavas and domes. These rocks were erupted from an inferred small summit vent complex and from well-developed volcanic rift zones. These zones correspond to dyke swarms trending c. 030°, 150°, 240° and 300° (the last two possibly forming a single broad east-west-trending rift zone) which are exposed in the near-vertical cliffs, up to 1 km high, that bound an east-facing collapse structure some 8km wide and extending 10 km inland from the coast of the island. This collapse structure removed the summit of the Monte Amarelo volcano, which was originally some 3 km high above sea level: the highest point on the rim of the collapse structure, Ponto Alto
293
do Norte, is at present some 2650m above sea level. Since this collapse the collapse scar has been largely filled in by the growth of a younger volcanic edifice, the Cha das Caldeiras volcano (Fig. 8). The most prominent feature of the Cha das Caldeiras volcano is a steep-sided summit cone, the Pico do Fogo itself, which rises to 2829m above sea level and some 1200m above the general level of the volcano. This cone is located in the centre of the Monte Amarelo collapse scar, with its summit only just over 5 km from the coast: the average slope on this side of the island between the summit and the sea is no less than 28°. The summit cone consists of thick units of volcanic spatter, spatter-fed lavas, coarse scoria and thin-bedded lapilli and yellow phreatomagmatic ash. It has not erupted since 1785 and has since that time undergone considerable erosion by slope failures and rockfalls, particularly in the vicinity of a N-trending zone of active fumaroles on its northern side and in the headwalls of gully systems on its eastern side. The sequences exposed in these slope failure scars are discussed further below in the context of interpretation of the early historical accounts. The growth of this summit cone has isolated the western part of the collapse scar from the sea, and ponding of lavas between the cone and the cliffs of the collapse scar has produced a plain, the Cha das Caldeiras, which is between 1.6 and 1.8km above sea level. 'Cha das Caldeiras' translates as Plain of Craters, referring to the many monogenetic scoria and spatter cones that have formed on it. The locations and vent elongation directions of these vents are shown in Fig. 9. Many of these vents form the up-rift end of inferred volcanic rift zones that extend outside the collapse structure as indicated by the locations (Fig. 8) and elongation trends of young, post-collapse monogenetic scoria and spatter cones and associated lavas on the outer slopes of the old Monte Amarelo volcano. The most active zones trend along approximate bearings of 030°, 150° and 240-270°, broadly corresponding to the rift zones of the Monte Amarelo volcano. However, with the exception of the 1995 eruption to the SW of the summit cone, which took place along the 240°-trending rift system, the most recent eruptions (identified from post1785 eyewitness accounts and the results of reconnaissance geological mapping, which are generally consistent with one another as discussed in the next section) are confined to within the Monte Amarelo collapse structure and are aligned along N-S-trending fissures, which commonly are arranged in en echelon systems. The
294
S. J. DAY ET AL.
Fig. 8. Sketch map showing the principal geological features of Fogo, Cape Verde Islands.
sense of offset within these en echelon sets indicates that the dykes feeding these systems lie along the old 030°- and 150°-trending swarms at depth but rotate as they propagate upwards into conformity with an east-west extensional stress system. This pattern is most easily explained in terms of eastward (seaward) displacement, during eruptions, of a block bounded by the fissure system on the west and strike-slip faults along the
boundaries of the older collapse structure. The implication that the eastern flank of Fogo may have recently entered a phase of instability makes it especially important to determine whether the change in eruptive style implied by the discrepancy between the early historical accounts of eruptions and the more recent activity is real or, as suggested by Ribeiro (1960), a symptom of the unreliability of these prescientific accounts.
ERUPTIONS ON MULTIPLE-VENT VOLCANOES
295
Table 1. Summary of early historical eruptions ofFogo, based on citations of original sources by Ribeiro (1960) and Machado (1965b); further details are given by those workers Date
Location of vents
Eruptive phenomena
Productions of eruption
1500?
Summit and both flanks
Explosive activity, with sustained paroxysms
Dust, coarse ash and scoria* covering entire island
1564
Summit cone?
Explosive
All of island covered in ash
1580-1585?
Summit cone
Explosive and effusive; 'glowing and giving off flames by day and night'; continuous immense 'flames'; 'rivers of fire' emitted by volcano
Lava flows; probable spatter around summit (note reference to incandescence).
1596
Summit cone?
Explosive
Major ash fall in northeast of island and at sea
1604
Summit and flanks of summit cone
'Flames and sulphurous vapours'
1662 or 1663? (account dated 1664)
Summit and two other vents
Explosive and effusive
1680
Summit?
Major earthquakes; large eruption with great explosion of 'lavas'
1683? (account of 1680 eruption)
Summit
'Great flames'
1689
Summit (and flanks?)
Explosive; 'flames', smoke, sulphurous clouds
Scoria* deposits covering the sea (?)
Explosive; 'smoke' during day, clouds of incandescent sparks at night
Scoria* deposits
'Ash' and rocks ejected from volcano (displacing people from the island)
Large incandescent blocks (like iron slag) rolling down slopes; lava flows entered the sea
Large eruption
1675
1693
Ash fall in west? (Migration of inhabitants to Brava)
1695
Summit?
Explosive; 'fire' visible at night, 'smoke' during day
1697
Summit or upper flanks
Explosive; 'flames' emitted from heights of volcano
1699
Summit
Explosive; thick clouds of 'smoke' visible 50 km away during day; 'flames' at night
1712 (start of 1713 eruption?)
Summit
'Smoke'
1713
Summit
Explosive; 'immense' clouds of 'smoke', visible 90km away on clear days; 'flames' visible at night
Between 1721 and 1725
Summit
Explosive and effusive; large blocks ejected to great height and torrents of 'sulphur' flowing down flanks of cone
Incandescent blocks (spatter), lava flows, cinders (scoria)
1761
?
1769 or 1774
South flank of summit cone
Effusive?
Lavas?
* Described as 'pedra-pomes' (pumice) in original accounts, but as no pumice is found on Fogo this is most likely to refer to highly vesicular basaltic scoria and has been interpreted accordingly.
296
S. J. DAY ET AL.
Early historical accounts of eruptions of the Cha das Caldeiras volcano The eruptions from about 1500 to 1769 and the sources of accounts of them are summarized in Table 1. Table 1 is based upon the versions of these accounts given by Ribeiro (1960) and Machado (19656). Most of these accounts, as noted above, were written by sea-captains and other mariners (including the pirates Anthony Sherley and William Dampier). Ribeiro (1960) failed to find any accounts written by local authorities. The main points that these various accounts reveal can be summarized as follows: •
Eruptions were prolonged and frequent for much of the period in question, although there appear to have been periods of repose or relative quiet from 1500 to 1564 and from 1604 to 1662 or 1663. Most accounts indicate that the eruptions were regarded by the inhabitants of Santiago as normal or frequent occurrences, suggesting that activity was semi-continuous for much of the period. However, strong earthquakes in 1680, associated with a large eruption, led to emigration of some of the inhabitants of Fogo to Brava. Furthermore, the oft-quoted use of the Pico do Fogo as a 'lighthouse' by mariners is based only on the account of the 1662-1663 eruption ('at night it is a mariner's lighthouse by reason of the flames which are thrown without cease from a very high peak') by Andre de Faro (1664; republished in 1945 and quoted by Machado (19656)). It is not clear from this whether de Faro intended to state that the volcano was actually in regular use as a navigation aid or if he sought to draw an analogy with the fires on high headlands that were used as navigator's lights at the time. • Many accounts referred to explosive eruptive activity. Falls of 'cinders' and 'stones' both on the island and at sea around it are frequently mentioned. Incandescent rocks and large blocks are mentioned as being ejected over large areas in the accounts of eruptions in 1662 or 1663 and 1721-1725(7), and the eruption in the period 1580-1585 may Have been similar. As noted above, Ribeiro (1960) questioned the veracity of these accounts, in particular that of the eruption in the period 1721-1725 (by an Englishman resident in the Cape Verdes in that period, Roberts), in part because the style of activity is very different from that of more recent eruptions. In contrast, accounts of eruptions in the period 1675-1713 refer instead to
generation of high eruption columns, described as 'thick clouds' or 'clouds of smoke', visible several tens of kilometres away, indicating a change in the style of activity in that period, the nature of which is discussed further below. During eruptions in this period the glow from incandescent material, generally referred to as 'fire' or 'flames', was visible only at night. The eruption of 1596 is also notable as having produced a widespread fall of ash. • Some accounts also referred to lava flows descending to the coast down the steep eastern slope of the island. In some cases (eruption of 1604) these were considered to originate from multiple vents. In most other cases the sites of vents were not clearly defined, although the eruption of 1769 has been located at Monte Laipo or Monte Lorna on the SE side of the Pico do Fogo, on the basis of a comment in Feijo's account of the 1785 eruption to the effect that the previous eruption was on the south side of Pico do Fogo. There are a number of very young vents in this area that could be the 1769 vent(s) (Fig. 9). • The activity was generally considered to originate from Pico do Fogo and certainly from within the old collapse structure (hence the poor view of the eruptions from Sao Felipe, as noted above). Ribeiro considered it likely that, in fact, most eruptions occurred on the flanks of Pico do Fogo and in the Cha das Caldeiras. He was particularly sceptical of the account of Roberts, which described flows of 'sulphur' descending the cone of Pico like 'torrents of water', and of Roberts' claim that the Pico had not existed before the eruption in the 1721-1725 period. Despite Ribeiro's scepticism regarding the reliability of these early accounts, their nonscientific nature and the genuine lack of information regarding the topography of the interior of the island, it is none the less possible to relate many features of the accounts to the geology of the Cha das Caldeiras volcano and of the Pico do Fogo itself. Perhaps most interesting is the emphasis in many accounts upon more or less explosive activity at the summit of the Pico, with the ejection of large amounts of incandescent rock in many eruptions. Allowing for the cultural background of the observers, these descriptions match the inferred mode of eruption of the welded spatter and (probably spatter-fed) lava units that make up much of the sequences exposed in the gullies on the flanks of the Pico do Fogo summit cone.
ERUPTIONS ON MULTIPLE-VENT VOLCANOES
297
Fig. 9. Map of vent locations and elongation directions around the Pico do Fogo and Cha das Caldeiras, with dates of identified historical vents.
298
S. J. DAY ET AL.
The sequence in these gullies shows a consistent stratigraphic sequence, which is exposed around the northern and eastern flanks of the volcano. Part of this sequence is shown in Fig. 10. The uppermost part of the sequence is composed of two or more coarse spatter and spatter-fed lava flow units, the uppermost of which is some 20-30 m thick (Fig. 11) and can be traced around the entire summit of Pico do Fogo. These welded units are separated by blocky scoria beds. Below them is about 30m of dark lapilli and yellowish to grey finely laminated ash in a plane-bedded sequence, which in turn overlies further spatter and spatter-fed lava. This sequence can be matched to variations in the descriptions of the historic eruptions that are summarized in Table 1. Accounts of the eruptions of 1662 or 1663 and of 1721-1725 place special emphasis upon the eruption of abundant 'Pedras incandescentes' ('incandescent rocks', interpreted as meaning spatter), 'pedra-pomes' (literally, pumice, but more likely to refer to vesicular scoria, as no true pumice occurs in any of the recent rocks of Fogo) and lava flows from the summit of the volcano. Roberts' account of the eruption that occurred between 1721 and 1725
indicates that this final major summit eruption was especially voluminous: this matches the uppermost spatter and spatter-fed lava unit in the sequence exposed around the summit of the cone. If these uppermost lava and welded spatter units are attributed to a major eruption in 17211725, then the thinner units below may be related to eruptions around 1700, and the finely bedded ash and lapilli unit exposed beneath these on the north and northeastern flanks of the Pico can be attributed to eruptive activity in the period 1675-1680, when the accounts refer more to clouds of smoke, earthquakes and great explosions. The eruption of this largely phreatomagmatic unit would have produced black rather than incandescent eruption columns and widespread ash and lapilli falls. This may have contributed to emigration from the island during the eruption of 1680. The accounts of the 1662-1663 eruption, when the ejection of 'large glowing rocks' is recorded, are consistent with the presence of further spatter and spatter-fed lava units below the ash and lapilli unit (Fig. 10). Alternatively, the finely bedded ash and lapilli unit could correspond to the eruptions of 1500(?) or 1596, notable for the wide distribution of ashfall deposits, but in this case a greater number of
Fig. 10. View of western wall of main gully on north side of Pico do Fogo (see Fig. 9 for location) showing sequence consisting of upper spatter and lava unit, middle finely bedded lapilli and phreatomagmatic ash unit (c. 30m thick), and lower spatter and lava unit. (See text for correlations of these units with historical eruptions.)
ERUPTIONS ON MULTIPLE-VENT VOLCANOES
299
Fig. 11. Main (>20m thick) spatter and clastogenic lava unit on north flank of Pico do Fogo, close to summit (see Fig. 9 for location), interpreted as having formed in an eruption between 1721 and 1725. This same unit forms the irregular crags all around the summit of Pico do Fogo (Fig. 15).
discrete units of spatter and/or scoria would be expected to occur above it in the sequence beneath the 1721-1725 spatter unit. Whatever the exact dates of formation of these eruptive units the existence of discrete units in the sequence making up the Pico implies a number of similarly discrete eruptions separated by periods of repose. This is consistent with Ribeiro (1960), who interpreted the eyewitness accounts as indicating discrete eruptions, and inconsistent with other interpretations (such as that by Machado, 19650, b) that attach more significance to the description of the Pico as a 'mariner's lighthouse' and consider the early historical activity to have been continuous. In this context it is also to be noted that the welded spatter and spatter-fed lava units that make up much of the Pico would have to be erupted at a very high rate to undergo welding and rheomorphism; such an eruption rate could not be maintained continuously or otherwise the Pico would be far larger than it actually is. The many points of agreement between the early historical accounts and geological features of the sequence forming the Pico do Fogo summit cone, and in particular the match between the variations between accounts of different eruptions and the sequences of different lithological
units, make it clear both that the Pico do Fogo summit cone was the principal site of activity in the early historical period and that the early historical accounts are in fact accurate in many respects. The scepticism of Ribeiro (1960) regarding the veracity of these accounts does not seem to be justified by the geological observations described here.
Recent flank eruptions of the Cha das Caldeiras volcano: more examples of the line-of-sight error The last eruption on Fogo to involve activity at the summit crater of the Pico do Fogo was that of 1785. This eruption, which is also the first eruption of the volcano to be described in detail (Feijo (1786); this account has been transcribed in full by Ribeiro (I960)), was primarily a fissure eruption involving activity at as many as nine vents along the SW-NE-trending volcanic rift zone. Three of these vents were located to the north of the northern boundary of the older collapse structure; the 1785 eruption was also the most recent eruption to involve vents outside this structure. The summit activity appears to
300
S. J. DAY ET AL.
Fig. 12. Sketch map of Fogo island showing the Cha das Caldeiras and the vents of the 1785 eruption, from Feijo (1786). Parts of original key translated from the original.
ERUPTIONS ON MULTIPLE-VENT VOLCANOES have been relatively minor: Feijo describes the generation of a small dark (non-incandescent) eruption column, which deposited fine ash at the start of the eruption. No deposits from this activity are now recognizable at the summit of Pico do Fogo, most probably because any surfaces flat enough for it to have accumulated and been preserved have since been covered by fresh lapilli from more recent eruptions, notably those of 1951 and 1995. Feijo's account included a sketch map and a sketch view of the volcano from the east, reproduced here as Figs 12 and 13. These are sufficiently detailed that the vents of the 1785 eruption can be identified with some confidence (Fig. 9) and the general 030° trend of both vent elongation directions and the overall alignment of vents recognized. A feature of particular interest in the account written by Feijo is the occurrence of vents close to the village of Mosteiros on the northeast coast of the island. A number of flows around Mosteiros still lack significant vegetation, even though they are located on the wet side of the island. Despite the youthful appearance of these flows and the clear evidence for their formation in 1785 from Feijo's account, they have disappeared from most recent literature: they do not, for example, appear in the maps of historical lava flows contained in papers by Machado (I965a,b), Machado & Torre de Assuncao (1965) and Silveira et al (1995). The subsequent eruptions, of 1799, 1816, 1847, 1852 and 1857, all involved relatively small groups of vents, or single vents, mainly to the north of Pico do Fogo (the exception being the 1857 eruption, which occurred southeast of the Pico (Ribeiro (1960; see Fig. 9). Owing to the proximity of the different vents and a thick blanket of lapilli from the 1951 and 1995 eruptions, the identities of the different eruption sites are not entirely clear: particular problems
Fig. 13. Sketch of Fogo in the first stages of the 1785 eruption, viewed from the east; from Feijo (1786).
301
are associated with vents that may date from 1816 or 1847, and the locations of these differ from those presented by Torres et al. (1997). The evidence for the vent locations in Fig. 9 will be discussed in more detail elsewhere (Day et al. in review). It is, however, clear that most of these vents are aligned along a north-south dyke swarm distinct from that feeding the older NEor NNE-trending vents, such as those of the 1785 eruption, which are cross-cut by these younger vents. The interpretations of the sites of the 1951 eruption have varied rather more despite the recent date of this eruption and extensive observations of the later phases of the eruption in particular. Two vents, Monte Preto de Cima on the northwest side of the Pico do Fogo and Monte Orlando to the south (Fig. 9), have been consistently recognized as being active in 1951, and the very fresh appearance of these vents and their products and the fact that the only deposits that overlie them are from the 1995 eruption are entirely consistent with this. However, two other vents appear to have been active in the early phases of the eruption, and recognition of the correct location and even existence of these has been uneven. The first documents relating to the 1951 eruption are an account, and accompanying oblique aerial photographs, by Francisco Mendes, the meteorologist based on the island of Sal at the time. The photographs were taken on the second day of the eruption, 13 June 1951: Mendes' sketches based on these appear here as Fig. 14. Taken in a sequence as the aircraft flew from north to south towards and past the east coast of the island, descending as it did so, the photographs indicate the presence of two very active vents on the south side of the volcano aligned in a broadly southeasterly direction and a less-active vent on the west side of the volcano. This vent appears to be somewhat to the south and west of the position of Monte Preto de Cima, which in the view of Fig. 14c would appear on the Cha das Caldeiras approximately in line with the high point on the rim of the old collapse structure, Ponto Alto do Norte, rather than being hidden from view by the Pico do Fogo. The existence of vents in the western part of the Cha das Caldeiras during the 1951 eruption is confirmed by geological observations (Torres et al. 1997; Day in prep.). The flows from the undoubted 1951 vents at Monte Preto de Cima (Fig. 9) merge with flows that originate from a north-trending line of very small vents to the southwest, on the lower western slopes of Pico do Fogo. It appears that these vents were those
302
S. J. DAY ET AL.
Fig. 14. Sketches by Francisco Mendes, based on his aerial photographs, showing the vents of the 1951 eruption on the afternoon of 13 June 1951 (second day of the eruption), with names of topographic features added. (See text for discussion.) active on the second day of the eruption and that the dyke feeding them subsequently propagated further to the north and opened the Monte Preto de Cima vent. As Monte Preto de Cima is markedly lower than the vents to the southwest, it is likely that activity at the higher vents would have then ended. This must have occurred early in the eruption, before systematic observations of the activity began on 26 June: Monte Preto de Cima was certainly very largely in existence by 28 June. The vents to the southwest are small and have, since 1951, been partly covered by rockfalls from the Pico do Fogo and more recently by lapilli from the 1995 eruption. It is therefore perhaps not entirely surprising that they were not recognized by, amongst others, Ribeiro (1960), although Machado & Torre de Assuncao (1965) and Torre de Assuncao et al. (1967) indicated them as possibly having formed in 1857; this is inconsistent with their merging relationship with the 1951 Monte Preto de Cima flows. The first to correctly identify them as 1951 flows were Silveira et al. (1995). A more remarkable error has arisen with respect to the second 1951 vent, besides Monte Orlando, on the southern side of the Pico do Fogo. The eyewitness accounts by Mendes and others, including the island administrator L. S. Rendall, which were summarized by Ribeiro
(1960) all clearly indicate that this second vent was to the southeast of Monte Orlando on the floor of the Cha das Caldeiras, as in Ribeiro's own account and photographs. Monte Rendall, as the vent was named, ceased its activity well before the end of activity at Monte Orlando and was subsequently partly destroyed by collapse into the extensive Monte Orlando lava flows. It now forms a low hill isolated within the 1951 lava flows (Fig. 15), chiefly distinguished by the presence of a dry fissure system, which also extends well to the south of the cone to the site of a possible third vent, which may have been active in the very early stages of the eruption but is now almost completely buried by the Monte Orlando lavas (Fig. 9). The location and appearance of Monte Rendall has been nevertheless well documented by Ribeiro (1960) and it is therefore remarkable that as early as the mid-1960s, in the papers by Machado (1965#), Machado & Torre de Assuncao (1965) and Torre de Assuncao et al. (1967), the location of Monte Rendall was becoming confused. This situation has persisted to the present, the most recent published case being the paper by Silveira et al. (1995). That study in particular clearly identified 'Monte Rendall' as a vent due east of Monte Orlando and also situated on the lower slopes of Pico do Fogo (Fig. 16; see also Fig. 15). This vent is that named Monte Lantisco on older maps and by Ribeiro (1960). It is markedly older than Monte Orlando: lapilli from Monte Orlando overlie a well-developed weathering surface on the rim and western flank of the vent. In addition, lavas from a vent higher on the flank of Pico do Fogo (behind and slightly to the left of Monte Lantisco in Fig. 15) enter the breached northern crater of Monte Lantisco and curve around its western flank, where they disappear beneath 1951 scoria and lava flows. These lavas are also overlain by boulder screes produced by rockfalls from the slopes of Pico do Fogo. These field relationships are shown in Fig. 16. The flows concerned may well be those that formed in the eruption of 1769 or possibly 1774 (Ribeiro 1960; Torres et al. 1997). The reason for the confusion appears to be the same as that which has caused confusion on La Palma and is apparent from Fig. 15. The only road into the Cha das Caldeiras runs along the foot of the southern boundary cliff. Monte Rendall and Monte Lantisco lie along the same line of sight when viewed from this road, from which the photographs in Fig. 15 were taken. It appears that, as in the case of the 1677 vent and the San Antonio cone on La Palma, a young vent has become confused with an older but
ERUPTIONS ON MULTIPLE-VENT VOLCANOES
303
Fig. 15. View of the southern vents of the 1951 eruption from the road along the southern side of the Cha das Caldeiras. Monte Orlando at left, Monte Rendall in middle ground with Monte Lantisco behind; Pico do Fogo in background with remnant slabs of last major spatter eruption (1721-1725?; see text) around summit.
more prominent volcanic vent along the same line of sight.
The importance of integrated field (geological and archaeological) and historical studies in reconstructing volcanic eruptions The examples described in this paper provide some important pointers as to how to best carry out studies of historical eruptions of multiplevent volcanoes with a view to assessing volcanic hazards and volcanic structure. Special points to bear in mind are the following: •
Fig. 16. Sketch map of the Monte Orlando-Monte Rendall-Monte Lantisco area on the south flank of Pico do Fogo, showing the field relationships that demonstrate that Monte Lantisco is an early historical or prehistoric vent.
Even basic points such as the location of vents may become confused, especially where they are in proximity to older and more prominent vents, and the 'line-of-sight effect', which appears to have operated in both La Palma and Fogo, may cause confusion. All aspects of the historical accounts should be checked even where these accounts are of recent date (it will be noted that confusion over the identity of Monte Rendall on Fogo occurred within two decades of the eruption in which it formed).
304
•
•
•
•
•
•
S. J. DAY ET AL. In multiple-vent eruptions, smaller vents or those that are active only at the beginning of an eruption may be overlooked in contemporary accounts. Conversely, once the most serious damage has been done (as in the case of the destruction of the Fuente Santa by the 1677 eruption) or the area of the eruption has been completely evacuated or civil authority broken down (as in the case of the 1730-1736 eruption of Lanzarote), interest in the eruption may decline and the later stages of activity be less well documented. Outside observers such as sea-captains, provided they happen to pass by during an eruption, may provide as good descriptions of volcanic activity as local observers. Indeed, if the latter are not directly affected by eruptive activity, they may not record events at all. In contrast, the practice of keeping a ship's log habituates mariners to recording any observations of unusual natural phenomena. A corollary is that the historical archives of seafaring nations may provide as many records of volcanic activity as the archives of the countries in which the activity takes place. Where the style or location (or both, as in the case of the ending of summit activity and reorganization of volcanic rift zones that has taken place on Fogo) of activity changes, this is likely to be reflected in the eyewitness accounts, even when these are written by nonscientists or date from a pre-scientific period. These accounts should not be dismissed because they are inconsistent with more recent and better described activity, but checked against the geological evidence. Archaeological evidence can provide important stratigraphic constraints on the age and timing of activity, especially where the limits of archaeological periods are well defined (through, for example, conquests, socio-economic revolutions or cultural developments of known date that are reflected in preserved artefacts). Cross-checking against geological and archaeological evidence indicates that whereas primary eyewitness accounts, even if written by non- or pre-scientific observers, are accurate within their limits more often than not, local traditions and other secondary accounts (even if written by geologists and other scientists, unless based on detailed fieldwork rather than interpretation of archival evidence or these same traditions) are often in error, as a result of such things as confusion of different vents along the same
line of sight. These errors may develop remarkably quickly. Provided these points are borne in mind, archival and related investigations of historical eruptions of multiple-vent volcanoes can, as noted in the introduction, provide important information on aspects of eruptions that cannot be investigated by means of later fieldwork, most notably the absolute timing and duration of events within the eruptions. There is therefore much potential for collaboration between geologists, archaeologists and historians in the investigation of these relatively small-volume but potentially very significant eruptions. Fieldwork on La Palma was funded by the Spanish DGICYT Research Project PB92-0119, by European Union Environment Programme Project EV5V-CT920170, and the Viceconsejeria de Medio Ambiente of the Canarian Government. Fieldwork on Fogo by S.J.D. was funded by a grant from the Calouste Gulbenkian Foundation to J. Fonseca. We gratefully acknowledge the help of P. N. Perez, mayor of the town of Fuencaliente, who provided old manuscripts, drawings and copies of eyewitness accounts related to the 1677 eruption.
References AFONSO, A., APARICIO, A., HERNANDEZ PACHECO, A. & RODRIGUEZ BADIOLA, E. 1974. Morphology evolution of Teneguia volcano area. Estudios Geologicos, Teneguia special volume 19-26. ANCOCHEA, E., FUSTER, J. M., IBARROLA, E. et al. 1990. Volcanic evolution of the island of Tenerife (Canary Islands) in the light of new K-Ar data. Journal of Volcanology and Geothermal Research, 44, 231-249. ANDERSON, E. M. 1935. The dynamics of the formation of cone sheets, ring dykes and cauldron subsidence. Proceedings of the Royal Society of Edinburgh, 56, 128-157. ANONYMOUS 1677. Gazeta del Ayuntamiento de La Palma, 2 de diciembre de 1677 (Gazette of the City Council of La Palma, 2nd December 1677). (Copy of the original, found in the archive of the Marquis of Guisla Guiselin (Lorenzo Rodriguez 1987).) 1678. Edictos, Cartas, Instrucciones que se han despachado a los benificiados e parrocos en todos este obispado desde principios del ano 66 (1666) en adelante. Archives Parroq. La Concepcion, La Laguna, Arch. B. Garcia Ximenez, libro Q. (Edicts, letters and instructions that were sent to the beneficiaries and parish priests in this bishopric from the first day of 1666 onwards. Parochial Archives of the Archbishop B. Garcia Ximenez, Church of the Holy Conception, La Laguna, volume Q.) BONELLI RUBIO, J. M. 1950. Contribucion al estudio de la erupcion del Nambroque o San Juan (isla de La
ERUPTIONS ON MULTIPLE-VENT VOLCANOES Palma) (Contribution to the study of the eruption of Nambroque or San Juan (island of La Palma). Institute Geografico y Catastral, Madrid. CARRACEDO, J. C. 1994. The Canary Islands: an example of structural control on the growth of large oceanic-island volcanoes. Journal of Volcanology and Geothermal Research, 60, 225-241. 19960. A simple model for the genesis of large gravitational landslide hazards in the Canary Islands. In: McGuiRE, W. J., JONES, A. P. & NEUBERG, J. (eds) Volcano Instability on the Earth and other Planets. Geological Society, London, Special Publications, 110, 125-135. 19966. Morphological and structural evolution of the western Canary Islands: hotspot-induced three-armed rifts or regional tectonic trends? (Reply to comment by MARTI et at.) Journal of Volcanology and Geothermal Research, 72, 151-162. & RODRIGUEZ BADIOLA, E. 1991. Lamarote: La erupcion volcanica de 1730. Cabildo Insular de Lanzarote. , DAY, S. J. & GUILLOU, H. 1997a. Rapid growth and instability of an oceanic island volcano: the Cumbre Vieja rift, La Palma, Canary Islands. Volcanic Activity and the Environment, Symposium of the IAVCEl General Assembly, Puerto Vallarta, Mexico, 19-24 January 1997. , , & GRAVESTOCK, P. J. 19976. Geological map of the Cumbre Vieja Volcano (La Palma, Canary Islands). Viceconsejeria de Medio Ambiente del Gobierno de Canarias, Santa Cruz, Tenerife. , , & RODRIGUEZ BADIOLA, E. 1996. The 1677 eruption of La Palma, Canary Islands. Estudios Geologicos, 52, 345-357. , RODRIGUEZ BADIOLA, E. & SOLER, V. 1992. The 1730-1736 eruption of Lanzarote, Canary Islands: a long, high-magnitude basaltic fissure eruption. Journal of Volcanology and Geothermal Research, 53, 239-250. CHEVALLIER, L. & VERWOERD, W. J. 1988. A numerical model for the mechanical behaviour of intraplate volcanoes. Journal of Geophysical Research, 93, 4182-4199. DAY, S. J., CARRACEDO, J. C. & GUILLOU, H. 1996. Reconfiguration of the Cumbre Vieja volcano, La Palma: advance warning of a giant landslide? High Magnitude Events: Landslides, Floods and Earthquakes, Session of Applied Geoscience Meeting of Geological Society, London, Warwick, 15-18 April 1996. Geological Society, London, pp.xx-xx. DE FARO, A. 1664. Perigrinacao de Andre de Faro a Terra dos Gentios (The Travels of Andre de Faro in Heathen Lands). Republished 1945, Bertrand, Lisbon. FEIJO, J. S., 1786. Memoria sobre a nova irrupcao volcanica do Pico da ilha do Fogo que deve servir de Suplemento a Letra Filosofica No. 5 sobre o mesmo objecto. Archivo Historico Ultramarino, Lisboa, Cabo Verde: Cartografia - mapas nao catalogados (Memoir on the new volcanic eruption of the Pico of the Island of Fogo, supplementary to the 5th
305
Philosophical Letter on the same subject. Overseas Historical Archive, Lisbon: Cape Verde Cartography Section - unserialized maps) (complete transcript, with original diagrams, has been given by Ribeiro (I960)). FRITSCH, K. V. & REISS, W. 1868. Geologische Beschreibung der Insel Tenerife, ein Beitrag zur Kenntnisse Vulkanischer Gebirge. Wuster, Winterthur. HELENO DA SILVA, S. I. N. & FONSECA, J. F. B. D. 1999. A seismological investigation of the Fogo volcano, Cape Verde Islands: preliminary results. Volcanology and Seismology, in press. , FOULGER, G. R., BARROS, I. J., QUERIDO, A., WALKER, A. B. & FONSECA, J. F. B. D. 1997. Seismic activity in Fogo and Brava islands, Cape Verde. Proceedings of the International Congress on the Volcanic Eruption of 1995 in Fogo Island, Cape Verde. IICT, Lisbon, 79-91. HERNANDEZ PACHECO, A. & VALLS, M. C. 1982. The historic eruptions of La Palma island (Canaries Archipelago). Revista Universidad dos Acores, 3, 83-94. IICT 1997. A erupcao vulcanica de 1995 na Ilha do Fogo, Cabo Verde (On the volcanic eruption of 1995 on the island of Fogo, Cape Verde). Institute do Investigacoes Cientificas Tropicales, Lisbon. KlLBURN, C. R. J., PlNKERTON, H. & WlLSON, L.
1995. Forecasting the behaviour of lava flows. In: McGuiRE, W. J., KILBURN, C. R. J. & MURRAY, J. (eds) Monitoring Active Volcanoes. UCL Press, London, 347-368. LORENZO RODRIGUEZ, J. B. 1987. Noticias para la historia de La Palma, tomo I (Notes on the History of La Palma, Vol. I). Inst. Estudios Canaries, Cabildo Insular, La Palma. MACHADO, F. 1965a. Mechanism of Fogo volcano, Cape Verde Islands. Garcia de Orta (Lisboa), 13, 51-56. \965b. Vulcanismo das Islas de Cabo Verde e das Outras Ilhas Atlantidas. (Volcanism of the Cape Verde Islands and other Atlantic Islands.) Junto de Investigacoes do Ultramar, Lisboa, Estudos, Ensaios e Documentos, 117. & TORRE DE ASSUNCAO, C. F. 1965. Carta geologica de Cabo Verde (na escala de 1/100000); noticia explicativa da folha da ilha do Fogo estudos petrograficos. Garcia de Orta (Lisboa), 13, 597-604. McGuiRE, W. J. & PULLEN, A. D. 1989. Location and orientation of eruptive fissures and feeder-dykes at Mount Etna; influence of gravitational and regional tectonic stress regimes. Journal of Volcanology and Geothermal Research, 38, 325—344. NAKAMURA, K. 1977. Volcanoes as possible indicators of stress orientation: principle and proposal. Journal of Volcanology and Geothermal Research, 2, 1-16. POLLARD, D. D. 1987. Elementary fracture mechanics applied to the structural interpretation of dykes. In: HALLS, H. C. & FAHRIG, W. F. (eds) Mafic Dyke Swarms. Geological Association of Canada Special Paper, 34, 5-24.
306
S. J. DAY ET AL.
RIBEIRO, O. 1960. A Ilha do Fogo e as suas erupcoes, 2nd edn (The island of Fogo and its Eruptions). Memorias, serie geografica I. Junta de Investigacoes do Ultramar, Ministerio do Ultramar, Lisbon. ROMERO, C. 1991. Las manifestaciones volcanicas historicas del Archipielago Canario (Historic volcanic activity in the Canarian Archipelago), 2 vols. Consejeria Political Territorial, Santa Cruz de Tenerife. RYAN, M. P., BLEVINS, J. K., OKAMURA, A. T. & KOYANAGI, R. Y. 1983. Magma reservoir subsidence mechanics: theoretical summary and application to Kilauea Volcano, Hawaii. Journal of Geophysical Research, 88, 4147-4181. SIGURDSSON, H., CAREY, S., CORNELL, W. & PESCADORE, T. 1985. The eruption of Vesuvius in AD 79. National Geographic Research, 1, 332-387. , CASHDOLLAR, S. & SPARKS, R. S. J. 1982. The eruption of Vesuvius in AD 79: reconstruction from historical and volcanological evidence. American Journal of Archaeology, 86, 39-51. SILVEIRA, A. B., SERRALHEIRO, A., MARTINS, I. et al. 1995. A erupcao da Cha das Caldeiras (Ilha do Fogo) de 2 de Abril de 1995. Proteccao Civil, 7, 3-14. TIBALDI, A. 1995. Morphology of pyroclastic cones and tectonics. Journal of Geophysical Research, 100, 24521-24535.
TORRE DE ASSUNCAO, C. F., MACHADO, F. & CONCEICAO SILVA, L. 1967. Petrologia e vulcanismo da ilha do Fogo (Cabo Verde). Garcia de Orta (Lisboa), 15, 99-110. TORRES, P. C., MADEIRA, J., SILVA, L. C., BRUM DA SILVEIRA, A., SERRALHEIRO, A. & MOTA GOMES, A. 1997. Carta geologica das erupcoes historicas da Ilha do Fogo: revisao e actualizacao. A erupcao vulcanica de 1995 na Ilha do Fogo, Cabo Verde. IICT, Lisbon, 119-132. TORRIANI, L., 1592. Descripcion e historia del reino de las Islas Canarias (Description and History of the Realm of the Canary Islands). Translation of the original by A. Cioranescu 1978. Editorial Goya. VON BUCH, L. 1825. Physicalische beschreibung der Kanarischen Inseln. Berlin. 1836. Description physique des Isles Canaries. Levrault, Paris. WEBB, B. & BERTHELOT, S. 1839. Histoire Naturelle des lies Canaries, 2 vols. Paris. WHITE, R. S. 1989. Asthenospheric control on magmatism in the ocean basins. In: SAUNDERS, A. D. & NORRY, M. J. (eds) Magmatism in the Ocean Basins. Geological Society, London, Special Publications, 42, 17-27.
'A fire spitting volcano in our dear Germany': documentary evidence for a low-intensity volcanic eruption of the Gleichberg in 1783? J. P. GRATTAN1, D. D. GILBERTSON2 & A. DILL3 1
The University of Wales, Aberystwyth, Institute of Geography and Earth Sciences, Aberystwyth SY23 3DB, UK (e-mail:
[email protected]) 2 Nene Centre for Research, University College Northampton, Northampton NN2 7AH, UK 3 Geldenaaksevest 44, B-3000, Leuven, Belgium Abstract: This paper presents documentary evidence suggesting that the most recent volcanic activity in Germany may have occurred just over 200 years ago, rather than the 11 000 years held currently (Ulmener Maar, West Eifel). Several descriptions recounted here suggest that a mountain in Germany, the Gleichberg, may have erupted in the early summer of 1783. The reports of the volcanic event are laden with detail that would tempt the reader to accept them as genuine descriptions of an eruption, had the event been located in an historically active volcanic region. However, there are several reasons to suggest that the report of the Gleichberg eruption was a complex hoax, written to exploit the fear and panic generated by the dry fog present over much of Europe at the same time, which had its origins in the Laki fissure eruption. Geologically, the Gleichberg forms part of the Grabfeld, and is of Tertiary volcanic origin. There are no compelling geological reasons to suggest that this area has been tectonically active in recent times. The convincing detail of the report is used to illustrate the pitfalls waiting for geologists, historians and archaeologists who are using historical documents and folklore to explore the impact of volcanic eruptions upon ancient peoples and environments; time may lend weight to documents and folklore, which in reality may deserve none.
To evaluate the impact that volcanic activity may have had upon past human societies that are being investigated through their archaeological remains, it is necessary to possess accurate and reliable eruption chronologies, as well as detailed knowledge of the volcanic centres that have been active during human history or prehistory. This is not as straightforward as it may seem. Although volcanic eruption chronologies have been dramatically improved in recent years, our knowledge of volcanic eruptions from archaeological or geological evidence beyond the 19th century is often minimal (Simkin & Siebert 1994). Supporting evidence may also be sought from historical documents, but the interpretations of these are also open to challenge. The problems faced by archaeologists and historians working with such sources can be illustrated by a study of the hypothetical 'Gleichberg eruption' in 1783, an event in Germany that was widely reported across Europe by non-scientists. The year 1783 is already well known in volcanological history for the eruption of the Laki fissure in Iceland (Thordarson & Self 1993). In Iceland, the impact of the eruption upon both society and environment was severe (Thorarins-
son 1981), but the Laki fissure eruption also had dramatic impacts downwind, across the eastern Atlantic, upon the distant peoples and environments of Europe. Research using newspapers, private journals and scientific reports has revealed that from mid-June and throughout much of July, a dry acid fog composed of gases emitted by the eruption was distributed across much of Europe (Camuffo & Enzi 1995; Grattan & Brayshay 1995; Stothers 1996), with some regions experiencing higher concentrations of noxious aerosols as a result of the prevailing synoptic meteorological conditions (Grattan & Gilbertson 1994). In some cases, the impacts of the dry acid fog upon societies and environments in Europe were dramatic and have been the subject of extensive research by the authors, who have read hundreds of contemporary newspaper descriptions of the phenomena, as well as many more personal reflections recorded in diaries made at the time (Grattan & Charman 1994; Grattan & Gilbertson 1994; Grattan & Pyatt 1994; Grattan & Brayshay 1995; Grattan et al. 1996, 1999). Amongst all these documentary references to dry sulphurous fogs, crop damage, leaf loss, asthma epidemics and the
From: McGuiRE, W. G., GRIFFITHS, D. R., HANCOCK, P. L. & STEWART, I. S. (eds) The Archaeology of Geological Catastrophes. Geological Society, London, Special Publications, 171, 307-315. 1-86239-062-2/00/ $15.00 © The Geological Society of London 2000.
308
J. P. GRATTAN, D. D. GILBERTSON & A. DILL
Fig. 1. Location of the Gleichberg and proximal volcanic provinces.
Fig. 2. Geological map of the Gleichberg and surrounding district. Adapted from Von Berthold (1989).
Fig. 3. Geological cross-section of the Gleichberg, Adapted from Von Bernd (1989).
EVIDENCE FOR A VOLCANIC ERUPTION OF THE GLEICHBERG extermination of insects, there are references to eruptive activity on a volcanic mountain in Germany, the Gleichberg (Figs 1-3). These references are puzzling, as there is no other suggestion that the German volcanic centres have been active since the Ulmenar Maar event in the Eifel volcanic field (Hajdas et al. 1995; Zolitschka et al. 1995). This paper explores the nature of the report, considers its veracity and assesses the significance of such accounts for geologists and historians who are exploring the past by 'excavating words' (Weisburd 1985). Descriptions of the 'Gleichberg eruption' In early July 1783, several German newspapers reported the following letter from Hildburghausen: 'A report from Hildburghausen, of 24th June: Here is news from a recent strange natural phenomenon in our area. The Gleichberg, about 2 hours from here, is surely well known to you as a barometer substitute as its periods of smoking always announce subsequent rainfall. Since about Easter smoking has been stronger than ever before and is increasing every day. As a result thick clouds prevail in the whole area between Romhild and Hildburghausen, corresponding to a trip of 8 hours. All the woods in this area are white rather than green, and the whole sky appears to be dominated by pulverised or sublimed limestone. The clouds consist of sulphur killing everything, whereby sun and moon rise and set blood red. Since about 8 days ago terrible frightening thumps have been occurring within the Gleichberg, resembling explosions from cannons, until recently the Gleichberg opened up below thick sulphur clouds, and continuous terrible rumbling and roaring can be heard all around. In all the churches of the area special services are held, and the terrified inhabitants of the surrounding villages have fled as they fear the whole Gleichberg might finally collapse, or cause further misfortune. I shall report on future occurrences, as now we also have a fire spitting volcano in our dear Germany' (translated from Frankfurter Staats Ristretto, 12 July 1783, pp. 108, 456). Interest in the 'event' was so widespread that it was reported by correspondents and appeared in several English newspapers. 'Hildburghausen, July 4th: Mount Gleichberg, situated in our neighbourhood, has since Easter continually thrown out thick sulphurous vapours, and during the last eight days a violent noise has been frequently heard within the mountain ... another opening has since appeared, from which also issues a thick sul-
311
phurous smoke' (The Whitehall Evening Post, 9-12 August 1783). 'Hildburghausen, July 4th: Mount Gleichberg, in our neighbourhood, affords at present a singular and terrible phenomenon; the vapours which continually surround it are increased much, and form a thick mist which extends 8 leagues. This mist, which has destroyed the verdure of our woods, and has been substituted with a whitish tint, is, without doubt, by the scent, formed of sulphurous exhalations ... An aperture is formed, from which arises a very thick sulphureous (sic) smoke, which, with the subterraineous (sic) noise ... gives room to apprehend a new volcano' (The Morning Herald and Daily Advertiser, 12 August 1783). If unchallenged, these descriptions would leave little room to doubt that an eruption, albeit of low intensity, had taken place on the Gleichberg, with the consequences described above. However, another German newspaper, the Meiningen Wochentliche Nachrichten, disputed the report of the eruption. A letter published one week later on 19 July repeated the account published in the Frankfurter Staats Ristretto, but concluded: 'Our eyes do not see anything and our ears do not hear anything of the flight of the frightened inhabitants, the threatened collapse of the Gleichberg ... (nor the) thick sulphur smoke, that has covered the woods with a white blanket.'
The regional context: eruptive phenomena and social unease in Germany and Europe, June-August 1783 The passages translated above from the Frankfurter Staats Ristretto and the Meiningen Wochentliche Nachrichten are the most detailed accounts yet discovered in this study. It is likely that the passages printed in the English newspapers are essentially copies of the piece from the Frankfurter Staats Ristretto, as 18th-century English newspapers relied heavily on the Continental European Press for much of their foreign news (Wiles 1965). In the 6 weeks that followed the report in Frankfurter Staats Ristretto, British and other European newspapers contained hundreds of descriptions of reddened skies, dry fogs, crop damage and tremendous thunderstorms. Present research has indicated that these phenomena were largely the result of aerosols and gases emitted by the Laki fissure eruption in Iceland and transported to Europe by regional air circulation (see Grattan & Gilbertson (1994) and
312
J. P. GRATTAN, D. D. GILBERTSON & A. DILL
Grattan & Bray shay (1995), for a discussion of the newspaper coverage for this period). It may therefore be possible that the report of the 'Gleichberg eruption' simply attempted to capture and exploit the mood of the moment, which in many cases was one of panic. Diarists, correspondents and newspapers of this time tended to refer to the 'superstitious dread' with which the 'common people' viewed the dry fog and the strange appearance of the sun and moon. That the dry fog from Iceland also reached many areas of Germany is suggested by the following passages extracted from the Meiningen Wo'chentliche Nachrichten. 'Meiningen, July llth: Over almost all areas of Germany, the one fog, or the so-called 'majestic' smoke stretches out. The same has filled our horizon for three weeks. A thunderstorm, which we had here . . . and the following strong rain ... were not able to suppress it' (Meiningen Wochentliche Nachrichten, 12 July). 'Letter from Mannheim, 1 July: Since 17 June a fog has persisted day and night and can certainly be considered an extraordinary phenomenon. The eldest people cannot remember having ever experienced anything similar. This fog comes from the north-east, and is as common in the mountains as it is on the plains. It is also very dry, which is proved by the hygrometer. In the morning and after 7 o'clock at night, the sun has a red colour like glowing iron, and during the day it is very pale, with a stifling heat' (Meiningen Wochentliche Nachrichten, 12 July). These phenomena appear to have triggered such a degree of superstitious panic in the Meiningen area, as well as elsewhere, that a lecture on the nature of fogs was also published to allay the fears of the public: 'Because the sun now has a glowing red colour during sunrise and sunset, ... we will, in response to current superstitions present the following information to calm those of our readers who are afraid' (Meiningen Wochentliche Nachrichten, 12 July). The report of the Gleichberg eruption must therefore be seen against a broader background of Europe-wide phenomena, which sometimes triggered panic and superstition in Britain, France, and Germany, and in the Meiningen region in particular. It is in this context that we must examine whether the original account is likely to be fact or fiction. Geology The GroBer Gleichberg, 50.23°N 10.37°E, is a mountain, 679 m high, of volcanic origin, which
is sited on one of a group of volcanic fissures, known collectively as the 'Grabfeld', which lie to the south of Hildburghausen (Fig. 1) and which are eastern outliers of the German Volcanic System (Hoppe 1974). The fissures have been identified as being sited on Tertiary-early Quaternary, intra-plate volcanic activity (Weber 1955; Kastner 1974; Grumbt & Liitzner 1983; Ziegler 1990). The summits of both the GroBer and Kleiner Gleichberg are formed of basalt, which has to some extent been quarried, and examination of the published geological sources indicates that there is no doubt that the Gleichberge peaks are the result of volcanic activity (Figs 2 and 3). Until recently, the presence on the mountain of a Soviet military base hindered close scientific scrutiny. The area was also in the 'Sperrzone' or protection zone erected along the borders of the former East Germany and to which public access was severely restricted. As a whole, this group of hills has therefore attracted little attention from geologists, especially in comparison with the attention given to the Vogelsberg and Eifel volcanic structures located to the west (Fig. 1). The most recent volcanic activity in Germany in the recent Quaternary is the eruptions associated with the Laacher See tephra, c. 12 000 years ago, and Ulmener Maar, c. 11 000 years ago (Lippolt 1983; Van den Bogaard & Schmincke 1985; Hajdas et al. 1995; Lotter et al 1995; Zolitschka et al 1995), in the Eifel region. These events are widely held to mark the end of active volcanism in the German Volcanic System. Nevertheless, these German volcanoes cannot be considered to be completely inactive. Many hot springs testify to geothermal activity, and the frequent emanations of CO2 and helium isotropy in source waters in the Eifel region are evidence that the younger volcanic regions of Germany cannot be considered entirely extinct (Oxburgh & O'Nions 1987). It is extremely unlikely that this Quaternary activity could in any way be associated with the Tertiary volcanic fissures of the Grabfeld, which are over 2 Ma old, and certainly cannot be used to suggest that the Grabfeld fissures could have erupted in the recent historical past.
Fact or fiction? The detail contained in the original report is superficially convincing, and were it to have originated from an area of acknowledged recent volcanic activity it would probably have been accepted as genuine with only cursory scrutiny. The detail contained in the newspaper article is
EVIDENCE FOR A VOLCANIC ERUPTION OF THE GLEICHBERG remarkable because it suggests that the correspondent had witnessed an actual volcanic eruption and noted a range of associated phenomena, such as emissions of sulphur, tephra and water vapour. Such incidental, but important detail, seems out of place in a deliberate hoax. For instance, the report describes clearly events that can be reasonably interpreted as precursory volcanic activity, which could have eventually culminated in a low-intensity eruptive event of the type reported at the Gleichberg in June 1783. The unknown correspondent also suggested that the mountain was already known for periodic episodes of smoking and, in particular, that it was used locally as a 'barometer substitute'. This property might suggest that water vapour was being released from the ground during periods when the air temperature was cold. It is consistent with a ground surface heated from beneath by geothermal energy. Such an assumption is consistent with observations made in many areas of the world where geothermal heating is an accepted fact of life. Similarly, apparently convincing descriptions of tephra emission are given, with the tree foliage in the area turning from green to white in colour, and accounts of the air being filled with a pulverized limestone. The colour change also suggests the impact of acid deposition from sulphur clouds emitted in an eruption (Wilcox 1959; Caput et al 1978; Lang et al. 1980; Wisniewski 1982). However, it is important to note that similar phenomena were also noted in Britain and the Netherlands on 21-23 June by several reliable witnesses (Cullum 1784; Swinden 1786; Brugmans 1787; White 1789) and appear to be associated with gas emissions from the Laki fissure eruption in Iceland (Grattan & Gilbertson 1994; Stothers, 1996; Grattan et al \999a, b), and there may therefore be a copy-cat element in the Gleichberg account. It might also be argued that the damage to the vegetation observed around the Gleichberg was also due to the deposition of atmospheric acids emitted from the Laki fissure eruption. However, to date, all the known accounts of severe acid damage to vegetation, as opposed to descriptions of the dry fog attributable to the Laki fissure eruption, have come from relatively low-lying coastal districts around the North Sea (Grattan et al. 1999, 1998). It is less certain that the red colour of the sky at sunrise and sunset reported from Hildburghausen were entirely the consequences of the putative Gleichberg eruption, as the skies across Europe were spectacular at this time (Grattan & Brayshay 1995), probably also because of the material erupted from the Laki fissure in Iceland.
313
That the report was given at least shortterm credence is evident in the nature of the original publication. The Frankfurter Staats Ristretto was a serious, not a frivolous, newspaper. The position of the article in the newspaper does not appear to be pandering to taste or popular hysteria. For example, the description of the eruption was placed distinct from, but set amongst, detailed coverage of important domestic and international political and military affairs. Neither have we detected evidence that the actual text is a copy-cat account of events elsewhere, perhaps in southern Italy or Iceland. The date of the Gleichberg 'eruption' may also be important. Published on 12 July, the description of the eruption claims to have been written in Hildburghausen on 24 June 1783. The report may have been written and was certainly published shortly after a dry acid fog appeared in Europe; in Germany on 17 June and in Britain, France and the Netherlands, on 24 June. The intention of the correspondent could have been to exploit the fear and panic that was becoming widespread in Europe from early July as the gases emitted in the Laki fissure eruption obscured the sun, damaged crops and generated breathing difficulties across Europe (Thorarinsson 1981; Grattan & Charman 1994; Grattan & Gilbertson 1994; Grattan & Pyatt 1994; Grattan & Brayshay 1995; Grattan et al. 1996, 1999, 1998; Stothers 1996). It is also sensible to enquire, if the report was a hoax, 'why set it at the Gleichberg?' It is not at all clear that this mountain had been accepted as volcanic in origin at this stage of the 18th century. It should also be remembered that in the 18th and early 19th century, a powerful school of thought contested the volcanic origin of the German basalts, holding instead that they were the result of submarine sedimentation (Geikie 1893; Wagenbreth 1967; Von Bernd 1989). It may be that the Gleichberg was remote enough from Frankfurt, where the report of the 'eruption' was initially published, to prevent detailed editorial scrutiny. It is perhaps wise to sound a note of caution. There is no geological evidence to suggest that the Grabfeld fissures have been active since formed in the Tertiary, and, perhaps more compelling, the Meiningen Wochentliche Nachrichten, which was published much closer to the site of the proposed volcanic activity, announced itself to be unaware of any such event or of its social and environmental consequences. If it was a real event, it would surely have attracted proper scientific attention, and, tellingly, local historians have no record of the 'eruption'. Perhaps the event described was not a volcanic
314
J. P. GRATTAN, D. D. GILBERTSON & A. DILL
eruption but a forest fire, the impact of which became entangled in the phenomena associated with the gases emitted by the Laki fissure eruption. One such fire in the Mosel valley in the 1990s, was accompanied by descriptions of glowing lava blocks (M. Frechen pers. comm.), which will no doubt confuse researchers in the 2190s. Recent investigations in local museums and discussions with local historians have failed to unearth further confirmation of the Gleichberg event, an outcome suggesting that the original reports described above were either an elaborate hoax or the result of a genuine misunderstanding. Conclusion Although far from conclusive, newspaper articles in Germany suggest that the Gleichberg, a mountain of volcanic origin, near Hildburghausen, may have erupted in 1783. However, a detailed analysis of the text and an understanding of the wider context in which the report was made suggests that the report of the eruption was written to exploit the fear and interest generated by the presence in the atmosphere of a dry fog composed of volcanic gases that originated in Iceland, and is therefore a complex and clever hoax. Field research is needed to test this assertion but available geological data and current geological understanding make it unlikely that the 'eruption' will be confirmed, and the true nature and cause of the 'Gleichberg eruption' will remain a geological mystery. The case of the 'Gleichberg eruption' serves to illustrate some of the difficulties faced by archaeologists, historians and geologists attempting to determine the role and influence of volcanic activity upon ancient peoples, environments and cultures. One cannot rely entirely upon tradition and folklore, nor upon relatively recent detailed documentary material. It is clear that to test the veracity of historical or legendary accounts of volcanic activity one must possess or obtain a detailed understanding of the context in which the passage was written or the legend laid down. The passage of time lends authenticity to written accounts, which may not truly merit serious consideration. The authors are indebted to M. Frechen for detailed and helpful comments on the text, and to the staff of the German Historical Institute, London, in particular C. Freeman, for their assistance in this research. A. Gunst and B. Kulessa provided invaluable assistance in the translation of numerous German texts.
References BRUGMANS, S. J. 1787. Natuurkundige verhandeling over een zwavelagtigen nevel den 24 Juni 1783 inn de provincie Groningen van stad en lande en naburige landen waargenomen. Ley den. CAMUFFO, D. & ENZI, S. 1995. Impacts of clouds of volcanic aerosols in Italy during the last 7 centuries. Natural Hazards, 11, 135-161. CAPUT, C., BELOT, Y., AUCLAIR, D. & DECOURT, N. 1978. Absorption of sulphur dioxide by pine needles leading to acute injury. Environmental Pollution, 16, 3-15. CULLUM, J. 1784. Of a remarkable frost on the 23rd of June, 1783. Philosophical Transactions of the Royal Society. Abridged Volume, 15, 604. GEIKIE, A. 1893. Text-Book of Geology. Macmillan, London. GRATTAN, J. P. & BRAYSHAY, M. B. 1995. An amazing and portentous summer: environmental and social responses in Britain to the 1783 eruption of an Iceland volcano. Geographical Journal, 161(2), 125-134. & CHARMAN, D. J. 1994. Non-climatic factors and the environmental impact of volcanic volatiles: implications of the Laki fissure eruption of AD 1783. Holocene, 4(1), 101-106. & GILBERTSON, D. D. 1994. Acid-loading from Icelandic tephra falling on acidified ecosystems as a key to understanding archaeological and environmental stress in northern and western Britain. Journal of Archaeological Science, 21(6), 851-859. & PYATT, F. B. 1994. Acid damage in Europe caused by the Laki fissure eruption - an historical review. Science of the Total Environment, 151, 241-247. , CHARMAN, D. & GILBERTSON, D. D. 1996. The environmental impact of Icelandic volcanic eruptions: a Hebridean perspective. In: GILBERTSON, D. D., KENT, M. & GRATTAN, J. P. (eds) The Environment of the Outer Hebrides of Scotland: the Last 14000 Years. Sheffield Academic Press, Sheffield, 51-58. , BRAYSHAY, M. & SADLER, J. P. 1998. Modelling the impact of past volcanic gas emissions. Quaternnaire, 9(1), 25-35. , GILBERTSON, D. D. & CHARMAN, D. J. I999a. Modelling the impact of Icelandic volcanic eruptions upon the prehistoric societies of northern and western Britain. In: FIRTH, C. & McGuiRE, W. (eds) Volcanoes in the Quaternary. Geological Society, London, Special Publication, 161, 109-124. GRUMBT, E. & LUTZNER, H. 1983. Saxonian tectonics and basalt volcanism between the Thuringian forest and the fore Rhone. Zeitschrift fur Geologise he Wissenschaften, 943-954. HAJDAS, I., ZOLITSCHKA, B., IVYOCHS, S. D. et al. 1995. Radiocarbon dating of annually laminated sediments from Lake Holzmaar, Germany. Quaternary Science Reviews, 14, 137-143. HOPPE, W. 1974. Geologie von Thuringen. Hermann Haack, Gotha-Leipzig.
EVIDENCE FOR A VOLCANIC ERUPTION OF THE GLEICHBERG KASTNER, H. 1974. Jungtertiarer Vulkanismus. In: HOPPE, W. (ed.) Geologic von Thiiringen. Hermann Haack, Gotha-Leipzig, 782-789. LANG, D. S., HERZFELD, D. & KRUPA, S. V. 1980. Responses of plants to submicron acid aerosols. In: TORIBARA, T. Y., MILLER, M. W. & MORROW, P. E. (eds) Polluted Rain. Plenum, New York. LIPPOLT, H. J. 1983. Distribution of volcanic activity in space and time. In: FUCHS, K. (ed.) Plateau Uplift: the Rhenish Shield - A Case History. Springer, Berlin, 112-120. LOTTER, A. F., BlRKS, H. J. B. & ZOLITSCHKA, B.
1995. Late Glacial pollen and diatom changes in response to 2 different environmental perturbations - volcanic eruption and Younger Dryas cooling. Journal of Palaeolimnology, 14, 23-47 OXBURGH, E. R. & O'NiONS, R. K. 1987. Helium loss, tectonics and the terrestrial heat loss budget. Science, 237, 1583-1587. SIMKIN, T. & SIEBERT, L. 1994. Volcanoes of the World, 2nd edn. Geoscience Press, Tucson, AZ. STOTHERS, R. B. 1996. The Great Dry Fog of 1783. Climatic Change, 32, 79-89. THORARINSSON, S. 1981. Greetings from Iceland: ash falls and volcanic aerosols in Scandinavia. Geografiska Annaler, 63A, 109-118. THORDARSON, TH. & SELF, S. 1993. The Laki [Skaftar Fires] and Grimsvotn eruptions in 1783-85. Bulletin Volcanologique, 55, 233-263. VAN DEN BOGAARD, P. & SCHMINCKE, H.-U. 1985. Lacher See tephra: a widespread isochronous late Quaternary tephra layer in central and northern Europe. Geological Society of America Bulletin, 96, 1554-1571. VAN SWINDEN, M. 1786. Observations sur quelques particularites meteorologiques de Fannee 1783.
315
Memoires de I'Academie Royale des Sciences, Turin, 1784-1785, 113-140. VON BERND, W. B. 1989. Zum geologischen Aufbau der Gleichberglandschaft. Bikourgion, 7-13. 1989/7. Forschungsschichtliche Beziehungen zwischen Geologic und Archaologie am kleinen Gleichberg. Bikourgion, 39-47. VON BERTHOLD, W. 1989. Allgemeiner geologischer Uberlick iiber das Gleichberggebiet. Bikourgion: Zur Geologic des Gleichberggebietes, 14-22. WAGENBRETH, O. 1967. Die Entwicklung des geologischen Weltbildes in den letzten 200 Jahren. Forschungen und Fortschritte, 41, 365-371. WEBER, H. 1955. Einfiihrung in die Geologic Thiiringens. Deutscher Verlag der Wissenschaften, Berlin. WEISBURD, S. 1985. Excavating words - a geological tool. Science News, 127, 91-94. WHITE, G. 1789. The Natural History of Selbourne, reprinted 1977. Penguin, London. WILCOX, R. E. 1959. Some effects of the recent volcanic ash falls with special reference to Alaska. US Geological Survey, Bulletin, 1028-N, 409-476. WILES, R. M. 1965. Freshest Advices: Provincial Newspapers in England. Columbus, OH. WISNIEWSKI, J. 1982. The potential acidity associated with dews, frosts and fogs. Water, Air and Soil Pollution, 17(4), 361-377. ZIEGLER, A. 1990. Geological Atlas of Western and Central Europe. Maatschappij, The Hague. ZOLITSCHKA, B., NEGENDANK, J. F. W. & LOTTERMOSER, B. G. 1995. Sedimentological proof and dating of the Early Holocene volcanic eruption of Ulmener Maar, Germany. Geologische Rundschau, 84, 213-219.
This page intentionally left blank
Volcanic soils: their nature and significance for archaeology PETER JAMES1, DAVID CHESTER1 & ANGUS DUNCAN2 1
Department of Geography, University of Liverpool, P.O. Box 147, Liverpool L69 3BX, UK 2 Centre for Volcanic Studies, University of Luton, Luton LU1 3JU, UK Abstract: Whereas previous reviews of volcanic soils are biased in favour of those in tephra, the present paper examines the nature of weathering and pedogenesis in both tephra and lava. The classification of volcanic soils is discussed and examples are described of the response of pedogenesis to variations in climate, drainage, topography, vegetation and type and age of parent material. Archaeological implications considered include the distinctive properties of soils in tephra and the problems these may pose for laboratory analysis, evidence from buried soils, the ages of soils and their rates of development, and the fertility and erosion of volcanic soils.
An important consequence of volcanic activity is the formation of soil in the rocks and sediments that result from volcanic eruptions. It has been estimated that 27% of post-Archaean sedimentary rocks are tephra (including ash, pumice, volcanic bombs, lapilli and scoria (Fisher & Schmincke 1984)) currently covering about 0.84% of the earth's surface (Leamy 1984). Some 80% of this area is potential crop land (Mizota & van Reeuwijk 1989). Lava would appear to be more extensive than tephra, occurring in regions of recent volcanic activity and as Precambrian to Quaternary rocks including continental flood basalts, such as those of the Columbia River Plateau and the Deccan (see Fig. 1). Soil productivity has been one factor accounting for the density of human settlement in many of these areas of volcanic rocks and associated sediments. Despite the great area of land covered by lava, reviews of the literature on volcanic soils are biased strongly in favour of those developed in tephra. The probable reasons for this emphasis are the distinctive character of the products of tephra weathering, the technical challenge they have presented the laboratory analyst, the fact that many soils developed in lava and other rocks have received additions of tephra, and that tephra may form dateable stratigraphic markers, in some cases of considerable extent. Reviews of tephra soils include those by Ugolini & Zasoski (1979), Tan (1984), Wada (1985), Lowe (1986), Mizota & van Reeuwijk (1989) and Shoji et al (1993). Gibbs (1980) and Molloy (1993) discussed many examples of tephra and lava soils in New Zealand. The work by Shoji et al. (1993) is the most detailed review of tephra soils, but is
concerned chiefly with humid temperate regions. Mohr & van Baren (1954) and Mohr et al (1972) described the profile morphology and mineral composition of tephra and lava soils from many locations in the tropics. The collection of papers edited by Fernandez-Caldas & Yaalon (1985) dealt with both tephra and lava soils, but there is no substantial review concentrating on the latter. In contrast to the literature on tephra soils, that on lava soils includes few reports dealing with cool temperate environments. The archaeologist's first, and often most important, impressions of a soil are gained in the field. The general appearance of a soil, its profile morphology, reflects the operation of chemical, biological and physical processes upon parent material. A soil may reveal evidence of past environmental change, whether climatic, geological or human induced. It may reflect its relative age, or it may contain material that will yield a date relating to its age. If the soil contains features resulting from former human activity it is normally possible to distinguish these from natural pedogenic features. In this review we concentrate on the properties and profiles of volcanic soils. A note on classification is essential, as the terms applied to these soils are both many and confusing. We discuss soils in tephra and in lava separately and at some length because, as far as we are aware, this is the first review attempting to cover the full range of volcanic soils. We also consider several pedological aspects that have significance for archaeological research. Some of these are related to fertility, such as the nature and rates of weathering and soil development, and the susceptibility of soils to erosion. The relief of volcanic landforms has
From\ McGuiRE, W. G., GRIFFITHS, D. R., HANCOCK, P. L. & STEWART, I. S. (eds) The Archaeology of Geological Catastrophes. Geological Society, London, Special Publications, 171, 317-338. 1-86239-062-2/00/ $15.00 © The Geological Society of London 2000.
Fig. 1. Global distributions of principal lava flows and tephra deposits (based on Hyndman (1972), Derry (1980) and Chester (1993)).
VOLCANIC SOILS AND ARCHAEOLOGY an important influence on soils and their use. Some soil properties have less obvious implications: the potential of volcanic soils to hold chemical and magnetic signatures of human activity is of interest in geoarchaeological research; there are problems of interpreting the results of certain conventional laboratory procedures when they are applied to tephra soils.
Soil nomenclature For the reader with limited experience of soil science there are many good texts. Most of the pedological terms we use have been explained, for example, by Brady & Weil (1994). Because of the confusing variety of soil nomenclature in the literature, we use the classification, Soil Taxonomy (Soil Survey Staff 1997), where feasible. The scheme, summarized by Brady & Weil (1994), is the most comprehensive that attempts to have global application. To those unfamiliar with this system and its predecessors, the nomenclature will appear strange, yet it serves as a shorthand notation of the essential characteristics of a soil. Where insufficient data make it impossible to classify a soil correctly within Soil Taxonomy we follow the nomenclature used in the works cited, and in many instances traditional names for soils are given, as these are likely to be familiar to many workers in fields other than pedology. We also use the generic terms 'volcanic soils', 'tephra soils' and 'lava soils', which are informal but brief and clear in meaning. In Soil Taxonomy the status of the most distinctive volcanic soils (these being very largely in tephra) has been revised as knowledge of their mineralogy and chemistry has improved with advances in analytical techniques. Before 1990,
319
volcanic soils meeting certain physical and chemical criteria were included in the Andept suborder of the order Inceptisols (soils of limited development) (Soil Survey Staff 1975), but in 1990 they were allocated the highest category, the soil order of Andisols (Soil Survey Staff 1990). These soils, which have the 'andic' properties summarized in Table 1, are largely, but not exclusively, developed in tephra. There are andic or vitrandic (Soil Survey Staff 1997, pp. 23, 135) subgroups in all orders of Soil Taxonomy, with the exception of Vertisols (dark, swelling clays). The early developments in the classification of tephra soils by US workers between 1930 and 1970 have been summarized by Simonson & Rieger (1967). In the classification scheme of FAO-UNESCO (1989), designed for the 1: 500 000 Soil Map of the World, the major soil group, Andosols (a word partly derived from Japanese), is approximately equivalent to the Andisols of Soil Taxonomy. Some national soil classification schemes, such as those for Japan (Shoji et al 1993, p. 95) and New Zealand (Hewitt 1989), include classes for volcanic soils. The importance of tephra in controlling certain soil properties irrespective of climate is reflected in the fact that in FAO-UNESCO and Soil Taxonomy classifications, Andosols and Andisols, respectively, are the only major classes of soils in which the geological origin of the parent material is recognized. This is convenient for workers interested in volcanic soils, but many soils developed in volcanic materials, particularly lava but including tephra and other types of volcaniclastic sediment, do not meet the criteria for inclusion in these classes and so are grouped with other soils typical of the region in which they occur.
Table 1. Chief criteria defining andic soil properties in Soil Taxonomy; modified from Keys to Soil Taxonomy (Soil Survey Staff 1997, pp. 23-24) Most horizons that have andic soil properties consist of mineral soil materials; some consist of organic soil materials but must have <25% (by weight) organic C and meet one or both of the following requirements: (1) in the fine-earth fraction, all of the following: (a) Al plus 0.5 Fe percentages (by ammonium oxalate) totalling 2.0%; and (b) a bulk density, measured at 33kPa water retention, of <0.90gcm~ 3 ; and (c) a phosphate retention of <85%; or (2) in the fine-earth fraction, a phosphate retention of >25%, >30% particles of 0.02-2.Omm, and one of the following: (a) Al plus 0.5 Fe percentages (by ammonium oxalate) totalling >0.40 and, in the 0.02-2.Omm fraction, >30% volcanic glass; or (b) Al plus 0.5 Fe percentages (by ammonium oxalate) totalling >2.0% and, in the 0.02-2.00 mm fraction, >5% volcanic glass; or (c) Al plus 0.5 Fe percentages (by ammonium oxalate) totalling between 0.40-2.0 and, in the 0.02-2.Omm fraction, enough volcanic glass so that the glass percentage, when plotted against the value obtained by adding Al plus 0.5 Fe percentages in the fine-earth fraction, falls within an area defined graphically in Keys to Soil Taxonomy
P. JAMES, D. CHESTER & A. DUNCAN
320
Soils in tephra Volcanic rocks form predominantly from the solidification of silicate melts ranging in composition from <45% silica (by weight) to >75% (Table 2). The chemical and mineralogical composition of magma erupted from any one centre may vary considerably over time. The data in Table 2 illustrate the variations in chemical composition of lavas and tephra, particularly in their content of silica, iron and the bases, Mg, Ca, Na and K. Tephra and lava may be geochemically very similar, as the data show, and both are composed largely of silicate minerals and glass, but the rapid cooling of tephra results in its being predominantly glassy. Because of the high glass content and fragmental nature of tephra, it weathers rapidly in moist environments. This results in the formation of distinctive secondary materials and in the rapid formation of soil. Lava and tephra may be mixed: ash and lapilli typically accumulate in depressions within the rough surfaces of lava, and ash may be deposited upon developed lava soils. A significant feature common to well-developed soils in
tephra and lava is a high content of clay (data for tephra soils are given in Table 3a).
Weathering of tephra and formation of secondary minerals Tephra occurs to significant depths on the present land surface in areas of active or recently active volcanism (Fig. 1). The major characteristics of tephra soils are determined largely by the nature of their finest and most reactive constituents, the secondary materials produced from the weathering of glass. Under favourable conditions of moisture and temperature, Al, Si and Fe are released by hydrolysis from volcanic glass more rapidly than they are able to crystallize. The secondary materials so produced tend to be non-crystalline, but include poorly ordered (also called 'paracrystalline' or 'short-rangeorder') as well as amorphous materials. They are primarily the hydrous aluminosilicates, allophane and imogolite; the iron hydroxide, ferrihydrite and opaline silica. The secondary
Table 2. Representative geochemical data for lava and tephra
wt% Wt%
IBas
2Bas
3Tra
4 And
5 Dae
6Rhy
7T-Rhy
8T-Bas
Si02 A1203 Fe203 FeO MnO MgO CaO Na20 K2O TiO2 P205+ H20
51.8 14.8 3.9 7.3 0.2 7.1 10.6 2.4 0.7 1.1 0.1 nd nd
45.4 14.7 4.1 9.2 0.2 7.8 10.5 3 1 3 0.4 nd nd
63.7 14.1 2 6 0.3 <0.05 1.3 6.3 5.2 0.9 0.1 nd nd
57.2 18.2 3.5 2.9 0.2 2.6 7.1 4 2.1 0.7 nd nd nd
64.6 16 4.1 0.9 0.1 3 5.1 3.2 1.8 0.5 0.1 0.9 0.2
74.2 13.3 0.9 0.9 0.1 0.3 1.6 4.2 3.2 0.3 0.1 0.8 0.2
74.9 14.1 1.9 nd nd 0.37 2.15 2.83 2.25 0.17 0.02 nd nd
53.4 20.5 9.34 nd nd 3.8 8.73 3.25 1.11 0.85 0.18 nd nd
99.3
99.9
99.6
100.5
99.99 ppm: Sr Rb Zr Y Ba Sc La
98.69
101.16
H2cr Total
100
335 66 118 3 810 2 19
594 25 177 14 273 23 10
Data for 1-6 (lava) from Hughes (1982) (relevant table numbers are quoted); data for 7 and 8 (tephra) from Rose et al. (1981, p. 196). Major elements expressed as oxides (wt%) and trace elements expressed as ppm. nd, not determined. IBas, average of 21 Lesotho basalts, Karoo System (table 10.7); 2Bas, average of 35 alkali basalts, Hawaii (table 9.6); 3Tra, trachyte from the Kenya dome (table 10.3); 4 And, average of five andesites, Mt Egmont, New Zealand (table 11.5); 5 Dae, dacite, Tauhara complex, New Zealand (table 11.6); 6Rhy, average composition of rhyolites from Taupo Volcanic Zone, New Zealand (table 11.6); 7T-Rhy, tephra: Plinian W-tephra unit, Atitlan, Guatemala, rhyolite component; 8 T-Bas, tephra: W-tephra unit, Atitlan, Guatemala, basalt component.
Table 3a. Selected data on rates of soil development in tephra New Zealand Molloy (1993, p. 69) presents the following data for a sequence of soils in rhyolitic air-fall tephra (the Ngauruhoe tephra is andesitic): Soil property
Extent of leaching Age (years) % clay in B horizon % allophane in B % halloysite in B Bulk density (dry: gar1) % P retention
Raw volcanic soils Ngauruhoe — Taupo strong <1800 3 <1 0 0.8 60
Pumice soils
weak 1800 4 2 0 0.8 38
Volcanic loams
Tihoi
Ohaupo
strong 1800 3 1 0 0.9 62
weak to moderate strong 4000-20000 4000-20000 30 27 15 26 12 0 0.6 0.5 93 95
Mairoa
Volcanic loamy clays Patumahoe
Naike
moderate >50000 85 5 40 0.9 93
strong > 50 000 90 1 87 1.0 73
Other data from New Zealand: Vanatu basaltic tephra has almost completely altered to a depth of 1 m in c. 5000 years (Neall 1977) Sandy soils with A and C horizons developed in 500-750 years; AC or ABC sandy loam soils in 3000-5000 years; ABC profiles of loam containing amorphous clay in 5-20 ka; and ABC clay soils in 20-100 ka (Gibbs 1968) In a cold environment, one half of the mass of andesitic glass has altered in 7000 years (Neall 1977) Pollen and lake sediment evidence show that following destruction by ignimbrite, air-fall tephra and fire during the 1850 years BP Taupo eruption, the native podocarphardwood forest regenerated completely within 200 years (Wilmhurst & McGlone 1996) Japan Northeast Japan: humic A, Bw and C horizons may develop in cool temperate climates in periods of between several centuries and 1000 years (Ugolini & Zasoski 1979; Shoji et al. 1993, p. 62); melanic and fulvic A horizons take several thousand years (Shoji et al 1993, p. 60) Warm humid Kyushu: Andisols have reached maturity in 5000 years (Wada & Aomine 1973) Washington Cascades Spodosols have formed in 4000 years (Wada & Aomine 1973) St Vincent A clayey soil, 1.2-2.4 m thick, developed in andesitic ash aged 4000 year (Hay 1960) Martinique Nearly 30% of the glass has been altered to fine material (0-20 ^m) to a depth of 20-80 cm in Mt Pelee tephra in c. 2,000 years (Quantin et al. 1991) Costa Rica Well-developed soils can form in tephra in <2000 years (Nieuwenhuyse et al. 1993) Ecuador Under a rainforest climate, productive soils may develop in as little as 10 year (Colmet-Daage 1967)
322
P. JAMES, D. CHESTER & A. DUNCAN
minerals occur largely in the clay size fraction (<2//m) and form coatings and pore-infillings in the matrix of both weathering tephra and soil profile. Their influence on soil chemical and physical properties results from their fine particle size and the large area and other characteristics of their surfaces, which render them more reactive than crystalline clays. Complexes of Fe and Al with humus are also important. The non-crystalline materials listed above may occur together (Jongmans et al. 1995), but are favoured by different soil conditions: allophane, imogolite and ferrihydrite occur in soils of pH >5, whereas Al-Fe-humus complexes and opaline silica occur at lower pH (Mizota & van Reeuwijk 1989; Takahashi et al 1993). Factors in the secondary mineral composition other than pH include: the composition of the parent tephra; stage of soil formation; depth within the soil; thickness of any overburden; climate; effects of leaching that determine, for example, the silica concentration of the soil solution; and the influence of humus that forms complexes with Al and Fe and with some clay minerals (Wada 1985). Non-allophanic Andisols have andic properties but contain significant amounts of crystalline layer silicates; they are strongly acid and can cause Al toxicity (Dahlgren et al. 1993, p. 113). In the less humid climates with a dry season small amounts of Al-humus complexes, allophane and imogolite form, and appreciable amounts of halloysite (a kaolinite subgroup 1:1 structure mineral) and authigenic 2:1 clay minerals are produced by seasonal resilication. Where secondary materials are predominantly non-crystalline, soil properties include strong aggregation and low dispersivity of clay particles; well-developed blocky or finer structure in the upper solum; high porosity; low bulk density (as low as O.lgcm" 3 in Hawaiian allophanic soils (Ugolini & Zasoski 1979)); high water retention; and friable, non-sticky, non-plastic and non-hard consistence. Important chemical properties are variable charge and high phosphate retention. Soils with variable charge are well buffered and have a large pH-dependent cation exchange capacity (CEC). Under humid regimes, exchangeable bases are easily leached, but where allophane contents are high, pH (measured in water) tends not to fall below 5.0 (Nanzyo et al 1993, p. 146). The physical properties of allophane soils do not show the dependence upon exchangeable cations important in soils with crystalline layer clays. In addition to non-crystalline materials, a number of secondary aluminosilicate layer clays and iron oxyhydroxides are common in tephra soils and are important in modifying their
physical and chemical properties. Halloysite is the dominant clay mineral where Si is not removed by leaching, for example in areas of low rainfall (Wada 1985), and particularly in the lower parts of soil profiles (Wada 1977). In older soils of warm, humid environments, halloysite is believed to be transformed from allophane (Violante & Wilson 1983; Wada 1985). It may in turn be dehydrated and transformed to kaolinite (Violante & Wilson 1983), which is common in tephra soils, (Dahlgren et al 1993, p. 121). Gibbsite (an Al hydroxide with crystal structure) probably forms from the desilication of amorphous materials. It is important in warm environments and may occur as white concretions (e.g. Tejedor Salguero et al 1979). The 2:1 and 2:1:1 layer silicate clays commonly found in Andisols include smectite, vermiculite and mica. In addition to authigenic formation within the soil, a potential source of such clays (and of finegrained quartz) is aeolian deposition, in some cases following long-distance transport (Mizota & Matsuhisa 1985). Layer silicate clays have greater fixed charge, adsorb low amounts of anions and are less well buffered than the amorphous minerals. They are more easily dispersed, both in nature and in the laboratory, and are therefore more readily eluviated. Tephra soils with significant amounts of secondary layer silicates may not qualify as Andisols in Soil Taxonomy. The likelihood of transformation of allophane and imogolite to halloysite, kaolinite, gibbsite and 2:1 type clay minerals increases with soil development (Mizota & van Reeuwijk 1989, p. 12). Ferrihydrite is more likely to crystallize to goethite and haematite as Fe becomes available from metal-humus complexes.
Profile morphology of soils in tephra Detailed descriptions of the profile morphology of Andisols from a wide variety of climatic environments have been presented by Mizota & van Reeuwijk (1989). One of the most striking features of the soils in humid regions, ranging from cool temperate west Alaska (Simonson & Rieger 1967) to temperate (Shoji et al 1993, p. 23) and subtropical and tropical regions (e.g. Mohr & van Baren 1954; Mohr et al 1972; Xavier de Faria 1974; Vega 1980; Chartres & Pain 1984; Madeira et al 1994) is a very dark, often deep A horizon (see Brady & Weil (1996, p. 10) for definition of A, B and C soil horizons), in which a high content of well-humified organic matter is held in stable complexes with Al and Fe (Mizota & van Reeuwijk 1989; Nieuwenhuyse et al. 1993). The darkest Andisol topsoils
VOLCANIC SOILS AND ARCHAEOLOGY (epipedons), classed as melanic in Soil Taxonomy (Soil Survey Staff 1997, 5), are amongst the blackest soils in the world. They may reach 1 m in depth and contain 15-30% humus (Ugolini & Zasoski 1979, p. 85). Melanic epipedons tend to be associated with grassland and are reported as having developed from brown fulvic epipedons which existed under former forest in New Zealand and Japan (Shoji et al., 1993, pp.49, 54). A horizons of tephra soils are lighter in less humid environments (e.g. reddish yellow topsoils of Ultisols in Hawaii; Ugolini & Zasoski (1979)), where recent deposition of tephra has been frequent, or where soils have suffered erosion, sometimes as a result of cultivation. The colour of B horizons of tephra soils is determined by the mineralogy of the tephra and the coatings and masses of secondary minerals, particularly of the iron oxyhydroxides. Goethite and ferrihydrite impart yellowish brown to reddish brown colours. Red colours have also been ascribed to ferrihydrite. Haematite produces the red colours of many soils in warm climates, the development of these colours being the product of rubefaction (Duchaufour 1977). The colour of fresh tephra, and therefore of C. horizons in most well-drained tephra soils, varies from white to black, depending upon the colour of the glass and the mafic mineral content. Basalt and basaltic andesite tephra are black to reddish brown; andesite, trachyte, dacite and rhyolite tephra, with non-coloured glass and low amount of mafic minerals, are light in colour. The colour, as well as the degree of weathering and other properties of tephra, may vary considerably in one locality, even in one outcrop (e.g. Cronin et al. 1996). Poorly drained tephra soils exhibit the dull colours of reduced iron. The texture of A and B horizons in well-developed Andisols is dominated by clay, although its effect on consistence is modified by the formation of stable aggregates. This aggregation of the amorphous materials and their intimate association with humus favour a strong structural development, particularly in the A horizon. The eluviation of dispersible layer silicate clays produces Bt (argillic) horizons in soils of a range of climatic environments. The development of these horizons may be age dependent in some environments, or it may be a relict feature, reflecting the influence of a former climate more favourable to eluviation. Clay-rich B horizons may have prismatic structure. The nature of soil development in tephra may divert from Andisol formation to another soil order by the influence of regional or local climate, soil drainage, vegetation change, human use, or simply by transformations, particularly of
323
non-crystalline materials, that occur as soils evolve. The properties and drainage of alluvial, colluvial and other sediments derived from tephra may favour non-Andisol soil development. All 11 orders of Soil Taxonomy occur in materials derived from both tephra and lava: for example, all are present on Hawaii (Macdonald et al. 1981). Base desaturation, acidification and podzolization of tephra soils are described as rainfall increases both with elevation and with latitude. The trend may be accelerated by acidophyllous vegetation, as in southwest Alaska, where Andisols (Cryandepts) under grassland are rapidly podzolized when colonized by spruce (Simonson & Rieger 1967). The reverse transition of Spodosol (podzol) to Andisol is also possible after replacement of acid vegetation (Shoji etal 1993, p. 64). Soil sequences in mountain regions extend from variations within Andisols into other soil orders as rainfall and leaching increase, as temperature decreases and as vegetation changes with elevation. An allophanic Andisol-nonallophanic Andisol-Spodosol sequence is related to the Thornthwaite precipitation-evaporation index in northeast Japan (Takahashi & Shoji 1996). Inceptisols and Spodosols occur in tephra in New Zealand (Parfitt et al. 1983; Molloy 1993). On Mt Vulture in southern Italy, Andisols in Pleistocene tuffs are replaced by Spodosols under fir forest above 1200m (Lulli & Bidini 1980). In drier conditions on the lower slopes of the same volcano, Mollisols (dark soils of grasslands) and Alfisols (with clay-enriched subsoils; of medium to high base saturation), respectively, occur where soils are more base rich and where layer clays have been eluviated to give distinct argillic horizons. A shift toward layered clay minerals is associated with transition to these and other orders in tropical mountains: for example, Inceptisols and Ultisols (acid, highly weathered soils with eluviated clay, formed on old land surfaces) in Costa Rica (Martini 1976) and Cameroon (Delvaux et al. 1989); Mollisols in the Rwandan highlands (Mizota & Chapelle 1988); Mollisols, Vertisols and Oxisols (known by many other names, including red latosols and tropical red earths) in Kenya (Wielemaker & Wakatsuki 1984) and soils described as Latosolic Andosols in Papua New Guinea (Haantjens et al. 1967). The importance of volcanic soils in the tropics is reflected in the many references to them by Mohr & van Baren (1954), Mohr et al. (1972) and Young (1976). Much detail was given for a number of soil orders, particularly Andisols, Oxisols, Ultisols and Vertisols, in both tephra and lava. Hypothetical models presented by
324
P. JAMES, D. CHESTER & A. DUNCAN
Mohr & van Baren (1954) for tephra soil genesis in a range of tropical environmental conditions may be misleading, and were not considered by Mohr et al. (1972), but the soil profiles represented in the models are based on much field experience and are a good indication of the great variety of tephra soils in tropical regions.
Soils in lava As a soil parent material, lava differs significantly from tephra because typically it contains subordinate glass and is a coherent rock; it may be porous, but to a lesser extent than most
tephra; where it occurs in fragments as in aa lavas, the blocks tend to be larger than those that occur in tephra. In many aa flows, interstices occur between blocks, forming a honeycomb structure. Pedogenesis is strongly influenced by the nature and properties of the lava (Fernandez-Caldas & Yaalon 1985) but there is no distinctive and unifying control as in the case of the non-crystalline materials in Andisols. There *s' as a consecmence> no class *n Soil Taxonomy associated especially with lava soils. Nevertheless, little altered lava flows are distinctive soil-forming environments and are similar throughout their global distribution. A high clay content is common to all well-developed
Table 3b. Selected data on rates of soil development in lava North Island, New Zealand (Molloy 1993) Rangitoto Island: 200-c. 800 years BP basalt 'barely weathered', but supports > 200 plant species. Mainland: Quaternary basalts, centuries in age, carry Entisols ('raw volcanic soils'). Quaternary basalts, millennia in age, carry friable clay-rich (up to 80% clay) soil high in allophane, gibbsite, goethite and hematite; dominated by kaolin-type minerals where moderately to strongly leached, and red where hematite content is significant Older Quaternary basalts carry highly acid soils rich in Fe/Al oxides and kaolinite; gibbsite where high classed almost as bauxite Tertiary and Upper Cretaceous andesites and basalts carry compact clays (up to 90% clay, predominantly kaolin-type); hematite reddening in some horizons Methana, Greece 2250 years BP andesite aa flow has partially lichen-covered rock surfaces and pockets of brown Entisols. Well-developed Rhodoxeralfs occur in Pleistocene lavas and derived alluvial fans (James et al. 1997) Mt Etna Basalt flows: 1992, 1981 and 1971 flows: no soil; 1928 flow: 2cm organic-rich soil developed under mosses and lichens; 1809 flow: 35cm soil in crevices between aa blocks under diverse flora; 1566 flow: 40cm soil with crumb structure in top 5cm; 1536 flow: 80cm soil under diverse flora; >8000 years BP flow: >80cm soil in agricultural terrace with weak blocky structure, very dark grey, moist throughout; most of these lava soils contain lapilli or other tephra types Israel Alkali-olivine basalt: weathering crust only several mm thick under profuse lichen growth; weathering zone is several cm thick in vesicular glassy basalt (Singer 1978) Gran Canaria Youngest basalts of several thousand years age have rugged surfaces with little weathering (Limbrey 1967) Queensland Soil development 'effectively zero' on 13000-year-old surfaces in south; average deepening rate of red soils in basalt is c. 0.3m Ma"1 (Isbell et al. 1976; Pillans 1997) Oxisols occur in 22-55 Ma basalts; in north, no mineral soil has formed on Holocene basalts (Isbell et al. 1976) Pinacate Field, Sonora Desert, Mexico Lava flows, 140 to 1.1 Ma little altered (Slate et al. 1991) Afar Depression, Djibouti Basalt flows, 1-4 Ma little altered (Rouby et al. 1996) Cameroon 17-year-old pahoehoe lava in Cameroon with luxuriant vegetation (Mohr & van Baren 1954, p. 343) Costa Rica Blocky andesite ISkaBP: soil has 0.8m deep, dark A horizon (Jongmans et al. 1995) Cook Island, Polynesia Oxisols occur in 16.6-18.9 Ma lavas (Kirch 1996)
VOLCANIC SOILS AND ARCHAEOLOGY soils in lava, and high CEC and base saturation characterize many soils in basalt. Where tephra fall has occurred, non-crystalline materials may have significant effects on lava soil properties (e.g. Sandor 1987; Lorenzoni et al. 1995). Even without addition of tephra, glass-rich lava may weather to produce soils with andic properties (e.g. Jongmans et al. 1995).
Weathering and pedogenesis in lava Rates of weathering and pedogenesis in lava of different climatic environments are compared in Table 3b. Common observations are the slow rate of alteration of lava as compared with tephra, particularly in dry environments, and the importance for plant growth, and therefore for soil development, of moisture in vesicles, crevices and depressions in the lava (e.g. Rees 1979). The authors observed (in 1998) that moist pockets of mineral and organic material, wind blown and washed into crevices on areas of pahoehoe morphology on the 1992 basalt flow of Mt Etna (Sicily), support the growth of several species of vascular plants. Valuable additional nutrients, particularly nitrogen, are added by rabbits that graze land adjacent to the flow and defecate on the bare lava surface. The first plant colonizers of lava surfaces in all climatic environments, however, tend to be lichens. Mosses and lichens dominate early stages of vegetation succession on basalt flows of Mt Etna dating <200 years. In addition to producing acids, the fruticose lichens and particularly mosses facilitate soil development by forming a mat across interstices between lava blocks and by trapping wind- and water-deposited material. Beneath moss on the aa lava of the 1928 flow up to 2 cm of humic soil has accumulated in depressions upon and between blocks of lava of 60 cm maximum length. On the same flow, smaller blocks (modal length c. 20 cm) create a more favourable habitat where vascular plants are common with the bryophytes, and as much as 6 cm of soil has accumulated. Wasklewicz (1994) described the effect of lichens on the biochemical environment of olivine basalt surfaces in Hawaii: the acids they secrete cause the dissolution of olivine, whereas in the absence of lichens, olivine in weathering rinds disappears more slowly than adjacent clinopyroxene and plagioclase (an exception to Goldich's widely accepted mineral stability series (Goldich 1938)). In alkali-olivine basalts of Israel, weathering is a function of drainage, with thin reddish brown crusts several millimetres in depth forming on rock surfaces where drainage is rapid.
325
The crusts, formed under profuse lichen growth, are only several millimetres thick. They contain chiefly smectite, kaolinite and halloysite, with trace amounts of hematite and other minerals. A thicker weathering zone, several centimetres deep and of similar mineralogy to the thin crust, forms in vesicular glassy basalt. On watersaturated surfaces, the weathering zone is deeper and the only clay is smectite (Singer 1978). Sherman & Uehara (1956) demonstrated the significance of micro-variations in moisture status in the weathering of olivine in basalt boulders in Hawaiian soils. On upper surfaces of the boulders examined, kaolin-type clay developed where leaching had removed bases and soil was acid. On the undersides of boulders, montmorillonite-type clay formed in moister conditions where bases, particularly Mg, were not removed. Associations of red soils (especially Oxisols) with kaolinite as the chief clay mineral and black, montmorillonite-rich soils (especially Vertisols) reflecting the controls of drainage and leaching have been described from many tropical and subtropical environments. Lava surfaces beneath well-developed soils in many regions are reported as 'fresh' or with a weathering rind of only millimetres in thickness (e.g. Singer 1966; Limbrey 1967; Jongmans et al. 1995). In such soils the weathered material may be mixed with soil by fauna or may be washed from the rock surface. In some cases soil overlying unweathered lava may have formed in tephra, colluvium or other material deposited upon the lava. In the humid tropics, weathering over millions of years may have produced a deep alteration of the lava bedrock. Oxisols occur in lavas altered to depths of 20m or more, for example, in Cameroon (Geze, in Mohr & van Baren 1954). In tropical environments, where moisture is abundant in the weathering zone, basalt, andesite and dacite show concentric weathering of corestones in a matrix of red earth (Mohr & van Baren 1954). Relatively young blocky andesite (<18000 years old) in humid tropical Costa Rica, in contrast, has been hydrolysed to produce only fine coatings of allophane and gibbsite (Jongmans et al. 1995). The rock is porphyritic, with fine ground mass comprising glass, pyroxenes, plagioclase and opaque iron minerals. Weathering is most intense in the soil B horizon. Coatings on boulders in B and C horizons are continually removed as they become fragmented and mixed, particularly by fauna, into the soil matrix. The A and B horizons qualify as andic on the basis of P retention, bulk density and ammonium oxalate extractable Al and Fe. The soil is a Pachic Melanudand (Soil Survey Staff
P. JAMES, D. CHESTER & A. DUNCAN
326
Table 4. Chemical composition ofplinthite compared with its basalt parent rock, India (after data of Satanayarana & Thomas; presented by Mohr et al. 1972) SiO2
Rock Plinthite
50 23.5
A12O3
25.1 19.6
Fe2O3
8.6 41.2
Ti02
1.0 2.8
CaO
10.0 0.2
MgO
3.0 0.4
H2O
1.1 12.0
Ratios
a
b
15.4 1.5
4.6 0.7
Ratios: a is total SiO2/Fe2O3; b is Al2O3/Fe2O3.
1997, p. 161), having the dark topsoil similar to that of many tephra soils. A common feature of tropical soils is an anoxic subsoil and the formation of plinthite, a highly weathered material that may harden to ironstone on drying. Mohr et al. (1972, p. 197) presented data comparing the composition of plinthite and its basaltic parent rock (see Table 4). The loss of silica, Al and bases, and the relative gain in iron are marked. In arid and semi-arid environments, weathering and pedogenesis in lava proceed at a 'snail's pace' (Pillans 1997). Early Holocene basaltic lava flows in tropical semi-arid north Queensland have no mineral soil development whatsoever (Isbell et al. 1976). On stable c. 13000-year-old surfaces, soil development is effectively zero and small-scale ropy flow structures of the lava are widely preserved. The average rate of soil deepening in red soils on basalts ranging up to 5.9 Ma is computed to be 0.3m Ma"1 (Pillans 1997). Rates in unconsolidated aeolian and fluvial sediments in the same region are 1-3 orders of magnitude greater. As Pillans pointed out, the slow rate of soil formation on the basalts is easily exceeded by any accelerated erosion associated with land-use practices. Weathering rates of historic basalt lavas in Hawaii (Jackson & Keller 1970) and development of soils on 6-8 Ma basalts in Senegal (Nahon & Lappartient 1977) are similar to those determined by Pillans for north Queensland. Slow rates of weathering rind development on basalt clasts have been reported from Bohemia (Cernohouz & Sole 1966) and the western USA (Colman & Pierce 1981). In desert climates, basalt lavas of the Pinacate field, northwestern Sonora, Mexico (Slate et al. 1991), and those dating from 1 to 4 Ma years in the Afar Depression of Djibouti (Rouby et al. 1996) are little altered. Under a Mediterranean semi-arid climate, soil and geomorphological processes have made little impact on andesite flows dating from 2250 BP on Methana, Greece (James et al. 1997). Although well-developed lava soils may contain sufficient amorphous material to have a
significant effect on soil characteristics (e.g. Molloy 1993), the formation of layer silicate clay minerals is an important element of lava soil development. In Oxisols the silicate clay is chiefly low-activity kaolinite, but elsewhere 2:1 layer silicates impart stickiness and plasticity to the soil. The layer clays are more readily dispersed than the amorphous materials of Andisols and may be eluviated in tropical, subtropical and temperate environments. The eluviated clay accumulates in argillic horizons of Ultisols and Alfisols (e.g. Eswaran 1972; James et al. 1997). Because of a lower content of allophanic material, stable humus-metal complexes are less important in lava soils than in Andisols. The melanic epipedon, therefore, does not normally form, but dark, humus-rich mollic (well structured with high base saturation) or umbric (as mollic, but with low base saturation) epipedons occur (e.g. Lulli & Bidini 1980). Ferrihydrite weathered from rock crystallizes in well-drained environments to goethite or hematite, which colour A and particularly B horizons. Crystalline phyllosilicate clays are less water retentive than the non-crystalline materials of Andisols, but 2:1 layer clays swell and shrink on wetting and drying. Anion adsorption tends to be negligible and exchangeable cations play a more significant role in soil processes.
Examples of soils in lava Lava soils are represented in all orders of Soil Taxonomy, although the majority described in the literature occur in the middle and low latitudes. Variation is considerable within regions where sequences of soils occur along gradients of elevation, drainage, topography and lava age. A broad range of examples are discussed below. In regions of Quaternary volcanic activity, lava soils commonly contain sufficient tephra to affect their physical and chemical characteristics. The few descriptions of lava soils in cool humid environments include Inceptisols in the basalts of Antrim, Northern Ireland (Smith
VOLCANIC SOILS AND ARCHAEOLOGY 1957), and in lavas in Scotland (Smith 1962), which have been scoured by Late Pleistocene glacial ice. The Antrim soils are brown with welldeveloped crumb structure and high CEC, and contain amorphous clay material. In Scotland, the soils are brown and reddish brown loams of low base status and high CEC. Immature dark brown loams and peats (Entisols and Histosols) have been described from Iceland on basalts that have received tephra inputs (Helgasson 1962). Brown clay loams, developed in basalt beneath pine forest in the southern taiga of Siberia, are of near-neutral reaction and are high in Fe, Mg and Ca (Gradusov & Targul'yan 1962). The lava is fissured and weathered at depths of 120-1000 cm, where free calcium carbonate occurs. These base-rich lava soils resist the leaching and podzolizing processes so effective in other soils of this environment. Oxisols and Vertisols are the most widespread soils of Neogene and Pleistocene lavas in the tropics, and are also important in Pleistocene tephra. Many Oxisols occur on very old surfaces, for example on 16.6-18.9 Ma basalts in the Polynesian Cook Islands (Kirch 1996). In southern Queensland most basalts carrying Oxisols date from 22 to 55 Ma (Connah & Hubble, in Isbell et al. (1976)). They contain large amounts of non-crystalline Fe or Al oxyhydroxides and non-sticky, low-activity silicate clays, and intense leaching has removed much of the original silica. A number of soil orders occur along a basalt toposequence in humid tropical Nicaragua (Eswaran 1972). The basalt comprises calcic plagioclase with minor amounts of olivine, augite, magnetite and glass. An increase in soil development occurs from steep, eroded to gentle, stable slopes. Entisols occur in the steep sites. Weathering of rock commences in cracks, with glass disappearing first; other primary silicate minerals gradually diminish in size. Coatings of iron-stained kaolinite, some of which is eluviated from the A horizon, form on rock surfaces in the C horizon. Root decay, faunal mixing and soil settling probably account for homogenization of the A horizon. No B horizon has formed. Inceptisols, developed on more stable slopes where runoff is less significant, have a blocky cambic B horizon (weakly developed with some colour change). Some very weathered feldspar crystals survive. Ultisols represent a more advanced stage of pedogenesis on undulating landscapes where percolation of rainwater is significant. Clay eluviation results in well-developed argillic horizons with cutans of yellow to orange ironbearing clay. The secondary iron is largely amorphous, with minor amounts of hematite
327
and goethite. In Rhodudults, the soil matrix is coloured red by the amorphous iron. Oxisols have developed on stable flat sites. They are red and reddish yellow, or paler in less well-drained sites; soil structure is weakly developed; concretions and nodules of metal oxides are frequent. Halloysite occurs in the weakly developed soils of the sequence, kaolinite and gibbsite in the more evolved. A similar topographic pattern of Entisols and Oxisols occurs in the highlands of Papua New Guinea (Haantjens et al. 1967). In tropical north Queensland, Oxisols vary with regional rainfall on the gently sloping lava surfaces of the Pliocene and Pleistocene shield volcanoes (Isbell et al. 1976). Intense weathering has produced acid red soils of high clay content, which increase in depth from < 1 m to 10m with an increase in rainfall from < 700 mm to > 1 200 mm. The transition from soil to underlying unweathered basalt is gradual. Deeper soils (>2m) tend to have dark red to dark reddish brown, friable loam to clay loam A horizons with granular or crumb structure. These may harden on exposure to ironstone. B horizons are dark red, blocky clay. Kaolinite is the dominant clay mineral in most of the soils; in higher rainfall areas, gibbsite occurs in low to moderate amounts, increasing with depth in the profile. Ironstone nodules occur in all soils. The significance of bedrock type is evident in many accounts of local and regional variations in volcanic soils. In some regions, however, certain pedogenetic trends may converge despite variation in lava type. In Hawaii, for example, climatic conditions greatly influence soil formation over short distances, yet, as pointed out by Macdonald et al. (1981), subsoils so enriched in alumina as to be commercially exploitable bauxite have developed from almost the entire range of volcanic rocks on the islands. Vertisols, dark, base-rich soils with large amounts of swelling 2:1 layer clays and commonly a subsoil horizon of CaCO3 deposition, are found mostly in subhumid to semi-arid environments, on gentle slopes and in sites with slow drainage. They occur with Oxisols and soils of other orders in topographic or climatic sequences in the tropics (e.g. Mohr & van Baren 1954; Isbell et al. 1976), and in subtropical and Mediterranean climatic regions (e.g. Dan & Singer 1973; Fernandez-Caldas et al. 1981). Seasonal shrinking and swelling of Vertisols cause two kinds of disturbance that would affect human artefacts present within the soil or upon its surface. First, cracks open during the dry season, to widths of >10cm at the surface and to depths of 1 m or more (Brady & Weil 1994, p. 83). Topsoil and surface material slough into
328
P. JAMES, D. CHESTER & A. DUNCAN
these, penetrating the subsoil. On rewetting, the swelling soil is distorted and forced upward as a result of the increased subsoil volume, resulting in mechanical turbation of the soil and in the formation of microrelief mounds and depressions on the surface, termed gilgai in Australia (Hallsworth et al. 1955). Deeper-seated distortion of deeply weathered Neogene basalt in the Darling Downs of eastern Australia is attributed to rheid flow of clayey subsoil resulting from loading and the formation of piercement (diapiric) structures in the clay, called mukkara by Paton (1974). He gave no estimate of the time scale of such major distortion, but if it has occurred in late Quaternary times, as seems likely, any archaeological record may have been considerably disturbed. Cracking to widths of 1 cm or more probably occurs in most clay-rich soils weathered from lavas, whether or not Vertisols, in warm environments with markedly seasonal rainfall, such as in Mediterranean-type climates. The process may have a significant effect on the distribution of small artefacts such as sherds where they are kicked by animals across bare soil surfaces. An analysis of pedogenesis in lavas in a dry and hot mid- to low-latitude environment has been presented from the Sonoran Desert of Baja California by Graham & Franco-Vijcaino (1992). The rocks are Neogene and Pleistocene rhyolitic and basaltic andesites and basalts. Annual rainfall varies from 113 to 137mm and is seasonal. Soils are Aridisols (Vertic and Petrocalcic Paleargids). In both basalt and rhyolite soils, ochric epipedons (light coloured, low organic content) overlie natric (argillic, high in Na + ) and petrocalcic (carbonate-cemented) horizons. Clay content is highest (47%) in the basalt soils, which have prisms in the B horizon separated by cracks of 5-10 mm width, which extend up to 50 cm in vertical length, but which do not extend through the loamy A horizon. Clays are chiefly smectite with some kaolinite. The presence of allophane and imogolite could not be confirmed by the researchers. Secondary Fe, derived from olivine and hornblende, varies from 2.3 to 2.5%. Free CaCC>3 occurs throughout both basalt and rhyolite soils and dominates Bk horizons. As Ca is a minor element of the rhyolite, it is believed to have been wind transported from playas. Petrocalcic and argillic horizons may have been formed during wetter conditions of the late Pleistocene or early Holocene period. The volcanic soils and soil-forming environments of Mio-Pliocene, Pleistocene and Holocene basalts on subtropical Tenerife have been described in detail by Tejedor Salguero et al
(1979) and Fernandez-Caldas et al. (1981). There is a strong climatic control on soil distribution, though those workers observed that the influence is not a simple one, as it includes effects of earlier climatic change. The extent of soil development varies with age of lava and is also complicated by colluviation, tephra fall on lava soils, and erosion in areas of intensive cultivation. Rainfall varies significantly between the semi-arid southern and humid northern slopes, and with elevation on both. In the north, two separate sequences were described on the Neogene and Quaternary rocks, respectively. With increasing elevation (to >2500 m), and a shift in climate through semi-arid-humid subtropicalhumid temperate-dry temperate, Vertisols give way to Alfisols and Inceptisols (fersiallitic soils: Duchaufour 1977) and these to Ultisols (ferrallitic soils: Duchaufour 1977). Upward through this sequence, soils become deeper, and their colour redder. CEC, pH and the SiO2/A\2O3 ratio in the clay fraction decrease. Smectite, important in the Vertisols, decreases upslope, giving way to halloysite and gibbsite. The soils are clay rich throughout the sequence, the clay being eluviated in the Alfisols. Soil structure decreases in size from prismatic in the Vertisols, to blocky-prismatic in the Alfisols and to blocky in the Ultisols. The last are described as similar to Ultisols in basalts of the tropics. They are clayey, moderately to strongly base desaturated, of medium to fine blocky structure and have white gibbsite concretions (
3 occur at the base of the B horizon. The sequence of younger soils on the northfacing slope of Tenerife, developed in Quaternary basalts covered in recent basaltic and phonolitic tephra, is, with increasing elevation, Dystrochrepts (brown soils) containing halloysite, and dark, humus-rich desaturated Andisols (Dystrandepts). On the south-facing slope, lava and tephra are similar to those on the north slope, but here climate is arid and semi-arid subtropical near sea level, ranging with elevation through semi-arid to dry temperate. Alfisols and Inceptisols (fersiallitic and brown soils) occur above the semi-arid subtropical zone.
VOLCANIC SOILS AND ARCHAEOLOGY The effects of the dry subtropical climate below 400m are increasing alkalinity with CaCO3 and CaSO4 deposition in Camborthids (chestnut soils) and sodic soils. The carbonate deposits, including crusts, in the subsoils of the Alfisols, Vertisols, Camborthids and sodic soils, are attributed to soil-water movement in a former, moister climate. Dates of between 20 and 30 ka were derived for carbonate nodules. The high pH (8-9) and high exchangeable Na+ in the lowland soils reflect the present arid climate. In Mediterranean climatic regions Entisols, Inceptisols, Alfisols, Vertisols, and, at higher elevations, Spodosols, are developed in lavas. On the basalts of the Golan, Israel, the distribution of soils has been modelled in relation to sequences of topography, rainfall and lava age (Dan & Singer 1973; Singer 1978). Rainfall is seasonal, annual totals varying from 300 to 900mm. Deep, dark brown Vertisols predominate on gently undulating, early Pleistocene cover basalts. They have high CEC, low organic C and, in lower rainfall areas, have carbonate accumulations in subsoil horizons. The clay is chiefly smectite. Aquerts (hydromorphic grumosols) occur in depressions that are wet throughout the year. Steeper slopes carry Entisols. Where rainfall increases, smectite declines and kaolinite and secondary Fe and Al increase. On more recent Quaternary basalts and under a more humid Mediterranean climate, soils are Alfisols (Rhodoxeralfs and Haploxeralfs). These are moderately acid (pH5.5-6.5), with kaolinite and vermiculite clays that have been eluviated to form argillie horizons. The volcanic soils of the Methana peninsula, in the Peloponnese, have been described by James et al. (1997). The peninsula is composed largely of steep domes and flows of dacite and andesite dating from 2250 years BP to c. 1 Ma, with volcanic agglomerates partially filling many depressions in this relief. Tephra deposits are few. Annual rainfall is currently c. 400 mm. Fluvial erosion of the volcanic slopes, chiefly during late Pleistocene time, has produced alluvial fans and inter-dome basin fills. Soil-forming processes and profiles are similar to certain of those described above for Tenerife and Israel (Vertisols, which tend to be a feature of older landscapes, do not occur, although cracking clay soils are widespread). On Methana, the chief processes are the weathering of lava to clay, incorporation and humification of organic matter, rubefaction, and the illuviation of clay and calcium carbonate to form argillic and calcic horizons. Pedogenetic pathways are similar between dacite and andesite. In this relatively young landscape, age and slope of the land sur-
329
face and sediment type exert strong controls on the distribution of soils. Soils range in stage of development from late Holocene Entisols on the youngest lava flow to Pleistocene Rhodoxeralfs on gentler slopes of the oldest flows and derived sediments. Clay content, redness and amount of secondary calcium carbonate and crystalline iron oxides increase in the B horizon with soil age. In the cooler and moister climate of the higher slopes, soils tend to be browner than those of the lowlands and are slightly to moderately acid.
Soils buried by lava Much less is known about palaeosols buried by lava flows than those buried by tephra. Despite baking ('fritting') by hot lava, significant elements of the soil may be preserved. Singer & Ben-Dor (1987) have described one of a number of soils buried by lava flows in the Late Pliocene to Late Pleistocene basalts of the Golan Heights. Eight such buried soils occur in one cliff section. Chemical and physical properties of the buried soil examined are not unlike those of the modern soil, a dark brown Vertisol in basalt, although bulk density of the buried soil is higher, the colour redder and it is less dispersive because of compaction and iron oxide cementation. A dominance of maghemite and hematite is attributed to baking. Manganese has been transformed into coatings and pinhead-sized nodules. Root channels remain, although in this particular palaeosol there is no organic matter. The surface of the soil at its junction with the overlying lava comprises a platy layer, 0.5cm thick. Singer & Ben-Dor have also reviewed possible origins of red clay layers intercalated between lava flows that are not of pedogenic origin. Baked sediments below lava flows in Idaho have been dated using thermoluminescence (Forman et al. 1994). 14 C dating could be applied to organic matter within baked soils, but as experience by the present authors on Mount Etna has shown, organic matter may not have survived.
Volcanic soils: implications for archaeology Tephra Tephra is of special interest to the archaeologist because it is deposited largely from the air, and may be deposited repeatedly at one site. Despite its very destructive force, tephra may be one
330
P. JAMES, D. CHESTER & A. DUNCAN
of the best preservative agencies known to the archaeologist, 'blanketing the remains in their pristine state' (Renfrew 1979, p. 565). In addition to buildings and chattels of unfortunate victims, the soil, its content of fossil remains, and its record of climatic, hydrological and ecological environment may be preserved largely unaltered for periods of hundreds to thousands of years. The micromorphological characteristics of palaeosols preserved by tephra on Santorini permitted Pomel (1983) to infer climatic fluctuations during the past lOOka. Palaeosols have been used to correlate tephra in a sequence dating from 64 years BP to 150kaBP in North Island, New Zealand. The palaeosol features include high clay contents, particularly of halloysite, Mn nodules, soil structure, fossil organisms and red, yellow and dark colours (Campbell 1986). The degree of soil development has been examined by chemical analysis (Birrell & Pullar 1973). From the palaeosol evidence, Vucetich (1968) interpreted the effects of late Quaternary climatic change. Tephra layers may be dated using either documentary records or radiometric methods applied both to the tephra itself and to incorporated materials, and may be correlated between sites on the basis of chemical and optical properties. Relative dating of tephra is possible on the basis of thickness of hydration rinds on glass fragments (Steen-Mclntyre 1981a, b). A tephra layer at c. 9m depth in the sediments of Franchthi Cave on the Argolid peninsula of Greece, which separated horizons with cultural remains, was correlated chemically and optically by Vitaliano et al. (1981) with the Grey Campanian tuff, for which dates of between 25 and 40 ka had been derived. Tephrochronology has been applied to the dating of farm ruins, shifts in settlement patterns and evolution of farmhouse types in Iceland. The Landnam tephra layer dating from c. 1100 years BP occurs throughout south and southwest Iceland, and is an important time marker for the study of vegetation changes following the arrival of people and their livestock (Thorarinsson 1981). Volcanic and chert stone artefacts on the island of Flores, Indonesia, have been dated at between 0.80 ±0.07 and 0.88 ±0.07 Ma by zircon fission-track dating of the tuff deposit in which the tools were buried. The dates match those suggested earlier by palaeomagnetic determination and biostratigraphy, as noted by Morwood et al. (1998). Those workers claim that the evidence proves the existence of Homo erectus in that region during Early to Mid-Pleistocene time and that these hominids must have been capable of making repeated sea crossings.
When analysing tephra soils or analytical data derived from them, the worker must bear in mind the difficulties resulting from a high noncrystalline component. Soil texture is difficult to compare between soils. It is one of the most fundamental characteristics of a soil and is, after colour, the soil property most likely to be described by the archaeologist. Meaningful measurement of the proportion of the clay fraction and its comparison with that of other soil orders is virtually impossible because of the strong aggregation of the colloidal fraction and its resistance to dispersion, made irreversible by drying. One approach to particle-size characterization using laser diffraction has been described by Buurman et al. (1997). In Soil Taxonomy, substitutes are used for textural classes of soils with andic properties (Soil Survey Staff 1997, p. 498). Other irreversible changes on drying include increases in acidity, average pore size and water transmission, and a decrease in water retention (Maeda et al. 1977). CEC values obtained for Andisols by conventional methods are often difficult to interpret (Wada 1989), and the concepts of water availability are also difficult to apply (Ugolini & Zasoski 1979). The prospective techniques involving the use of near-surface chemical and other soil properties as evidence of past human activity have potential application to tephra soils. The high P retention of soils with andic properties (as in agricultural terraces of the Colca Valley, Peru, referred to below) is an advantage in this regard.
The analysis of soil chronosequences in tephra and lava Although it is impossible completely to isolate the factor of time from the other controls upon soil development (parent material, climate, topography, drainage, vegetation and human activity), considerable effort has been made by pedologists to quantify rates of soil formation and to model the progress of soil development to maturity (steady state) in different climatic environments. The relative ages and age correlation of soils in an area have potential significance for archaeology in several respects. First, where the construction of the volcanic landscape has overlapped with the time scale of human settlement, the dating and correlation of the landscape elements may be essential to an understanding of the distribution of archaeological sites. Second, where weathering and pedogenesis in humid regions are rapid a soil may change significantly over time scales of millennia
VOLCANIC SOILS AND ARCHAEOLOGY or even centuries. Unless a soil reached a steady state some time ago, its present character may not be that of the soil exploited by people in earlier times. Third, the rates of weathering and pedogenesis of a deep tephra will determine when soluble nutrients become available to support a new plant succession, and when a new soil is ready for people to return to cultivate crops (see Plunket & Urunuela, this volume). For millennia, farmers will have appreciated the importance of soil age in regions of episodic tephra deposition. The absolute age of lavas and tephra may be determined by radiocarbon dating where organic material is preserved in palaeosols underlying lava or was incorporated in tephra. Volcanic materials older than a few tens of thousands of years may be dated using K-Ar, 40Ar/39Ar and U-series (Scott 1989). Where significant soil erosion and deposition have occurred, however, the age of the present land surface may be considerably less than that of the underlying rock or sediment: it is likely to be the same as, or much closer to, the age of the topsoil. Contrasting rates of soil development in lava and in tephra are illustrated by the data in Tables 3a and 3b. For tephra, time scales as short as 10 years for a productive soil in a tropical rainforest climate and several centuries to 1000 years for a fully developed profile in cool temperate Japan are quoted. In contrast, it is the slow rate of development that is highlighted for soils in lava. The great ages for Oxisols in lavas reflect the time scale of development of these soils and also the occurrence of lavas of this age. There are many such data published on rates of change in tephra soils, but relatively few for lava soils. It is easier to quantify rates in the former, at least in humid environments, because development is relatively rapid and the ages of major tephra deposited in historical time are generally known. For detailed study of changes in soils over time, a technique has evolved from the analysis of chronosequences in many geological environments, although there appear to be very few examples dealing with volcanic materials. A number of approaches have developed, based upon measures of individual age-dependent soil properties, particularly as they are distributed down-profile, and upon the use of such properties in the calculation of indices of soil development. An index based upon field observations was developed by Harden (1982) for alluvial surfaces in the southwest USA. Properties measured down-profile related to colour, structure, texture, eluviated clay, secondary carbonate, stickiness, plasticity and the thickness of soil horizons.
331
Many studies involving non-volcanic materials have used laboratory analytical as well as field data, and different approaches are taken to deriving an index of soil development (e.g. Birkeland 1984a, b; Dorronsoro & Alonso 1994; Vidic & Lobnik 1997; and references given in these). The variables chosen for volcanic soils would depend upon the climatic environment and upon whether the soils were in lava or tephra. The variables used by Harden would be appropriate for many volcanic soils, with measures of organic carbon, pH and iron oxihydroxide phases and amorphous materials (e.g. ammonium oxalate extractable Al, Fe and Si) added. Thickness of weathering rinds on volcanic stones in many situations is proportional to the time since they were deposited with the sediment (Colman & Pierce 1981; Colman, 1986). Where soils are dated at stages through a sequence, a mathematical chronofunction may be derived, relating measures of either single properties or of a soil development index to soil age (Bockheim 1980). A common form of chronofunction for both whole soil development and for changes in single properties is the logistic curve (Birkeland 1984a), usually best modelled by a logarithmic or power function. A feature that emerges from many chronosequence studies is a stage when the rate of soil development decreases significantly, as indicated both by many individual properties and by whole profile indices. A steady state is not reached in all environments, however, despite evolution over long time scales (e.g. Busacca 1987). Soils evolve in response both to extrinsic forces, such as climatic change, and to intrinsic thresholds (Muhs 1984). There remain important questions about soil evolution that chronosequence analyses are helping to answer. Despite attendant problems, the approach is appropriate, if not essential, to understanding soils of many volcanic landscapes. It proves valuable for the correlation of landform age in geomorphological research (e.g. Harrison et al. 1990), and for archaeology it has the potential to provide both a model from which to determine relative age of soils across an area, and data on the rate of soil development that may be of value in post-dieting soil conditions in times past. The selection of sites for chronosequence analysis and the interpretation of the results must be approached with great care. In an ideal chronosequence of soils, sites are comparable in all respects except age. Minor variations in these aspects are of little significance, but in complex volcanic terrains it may be difficult to identify a sequence of sites with comparable soil-forming environments. Other problems presented by
332
P. JAMES, D. CHESTER & A. DUNCAN
these terrains are the alteration of soils by truncation and burial resulting from erosion, and the potential diversion of pedogenetic pathways by deposition of tephra. An alternative to examining single variables or to deriving a single index of development that subsumes many complex relationships is the use of principal components analysis. This reduces the amount of data, whilst at the same time defining essential characteristics of soils, and provides component scores for samples taken down each profile. These may be examined in a number of ways. Where two components dominate the explanation of the variance in the data, variations down each profile are summarized efficiently by plotting scores for each sample against these components. Differences in the degree of soil development are reflected in the pattern of sample scores for each profile across the graph. 'Anomalies' caused by truncation, burial or by the effect of some other local factor tend also to be shown in variations in sample scores down each profile. This approach was tested for 124 samples for 35 profiles in volcanic materials on Methana, using only four key variables. It proved successful in showing relative degrees of soil development on Quaternary lava flows, domes and alluvial fan surfaces (James et al. 1997); it showed the effects of inferred truncation and detected the local elevational gradient toward cooler and moister environments. Fertility and productivity of volcanic soils There is little point in attempting broad generalizations about the quality of volcanic soils, save to say that they vary from unproductive in their initial stages of development or following severe erosion, to being sufficiently fertile in many situations to support intensive agriculture and dense settlement. Thus it is true that 'Andisols are among the most productive soils in the world' (Shoji et al. 1993, p. 209), and that, after many centuries, Vesuvius remains 'an attractive fertile garden' (Fisher et al. 1997, p. 231), but also that, in El Salvador, 'the myth that ashfall improves soils certainly is not true' (Sheets 1985, p. 55). The fertility of volcanic soils for cultivation, grazing or forestry depends upon the many influences on soil processes discussed above. Thus, for example, the rapid changes in humid climate tephra soils could conceivably run full cycle from sterile ash to cultivable soil, and to sterile eroded subsoil within centuries. With the soil, the local economy may develop, flourish and decay, or at least be forced to adopt different strategies to survive (Sullivan & Downum 1991).
Advantages of Andisols for agriculture include high water retention, stable structure, good aeration and drainage, ease of cultivation and rapid release of nutrients. Provided a tephra deposit is sufficiently thick to support a mature soil and to withstand the disturbance that accompanies agricultural use, particularly on moderate slopes, the principal constraints are likely to be chemical, including low per cent base saturation, suboptimal levels of Mn, Cu and Fe, and deficiencies of B, Mo, Zn, Co, Se and Cu (Wada 1985; Kanva 1995). Because of the high buffering capacity of Andisols, it may not be practical to raise pH significantly by liming. P deficiency may be serious because of fixing by non-crystalline materials. Unlike soils dominated by silicate layer clays showing repulsion of nitrate, allophanic Andisols with low organic C can retain nitrate. This leads to high recovery of applied N fertilizer (Nanzyo et al. 1993, p. 146). Mineralization of both N and C from soil organic matter, however, may be slow, nitrification rates being depressed by P deficiency (Boudot et al. 1988). Al toxicity to plant roots is often observed in non-allophanic Andisols (Shoji et al. 1993, p. 154). Tephra soils other than Andisols are likely to have the limitations for agriculture of their soil orders. Thus acidity and nutrient deficiency will occur in Spodosols. Soils in regions of volcanic activity may receive supplements of nutrients from aerosol deposition for some months after an eruption (Prendez et al. 1994). The immediate effects of tephra deposition on agriculture depend largely on the thickness of the deposit, as illustrated by the zonation around Ilopango volcano in the Zapotitan Valley, El Salvador (Sheets 1985). The third-century AD eruption buried sites close to the volcano beneath 50m of tephra, extinguishing all life. At a distance of 75 km, burial by 1 m forced the abandonment of sites; further away, cultivation by digging sticks was not possible. Between 100km and several hundred kilometres, there was some beneficial conditioning of soils. The rate of ecological recovery from barren tephra lands depends largely upon climate. Less than 50 years after the 1883 Krakatau eruption, the island was covered in forest, which today is very dense (Fisher et al. 1997, p. 243). It is known that seeds of 11 species survived burial for 60 years (Whittaker et al. 1995). Shoji et al. (1993, p. 45) noted how the pioneer species on tephra differ from those on consolidated rocks, being higher plants that benefit from the high water availability in vesicular and fine-grained tephra, and from rapid release of nutrients by weathering. Examples are given of early recovery (within 1 year of eruption) of vegetation on tephra in Japan.
VOLCANIC SOILS AND ARCHAEOLOGY A major limiting nutrient is N so that N-fixing plant species are early colonizers. The fertility of volcanic soils has been increased and maintained for centuries under careful management in some farming systems. Agricultural terraces built up to 4000m on the steep slopes of the Colca Valley, Peru, between 1400 and 1700 years BP remain fertile, even where abandoned (Sandor 1987). The original hillslope soils, developed in Cretaceous and early Tertiary andesite and rhyolite tuff, were mature, containing cambic, calcic and argillic horizons, despite the steep slopes. The terrace soils are Mollisols, with topsoils of fine-structured and earthworm-worked loams, conditioned by manuring. Levels of residual P are high, with total P of 10 g kg"1. The most fertile soils in lava occur in the orders Mollisols, Alfisols, Inceptisols and Vertisols, the least fertile being generally those in Aridisols, Spodosols, Ultisols and Oxisols, which may have both chemical and physical limitations. Lava soils, particularly those in basalt with high CEC and base saturation, are commonly more fertile than neighbouring soils in other materials. The high clay content of many lava soils, particularly of Vertisols, renders them extremely difficult to till by non-mechanized methods. Limbrey (1990), however, has described how Vertisols are suited to wild wheat growth, and quoted Unger-Hamilton (1989), who proposed that Early Natufian people in southwest Asia were harvesting wild-type cereals from cultivated basaltic soils. Limbrey considered that if early farmers had devised methods of weed control, they may have used Vertisols without having to cultivate them. The barren surfaces of young lava flows do not everywhere exclude human activity. Shepherds have constructed small buildings from aa blocks on the 1971 basalt flow of Mt Etna. Also on Etna, the prickly pear (Opuntia ficus-indied) has been planted to hasten break-up of the rock to permit the planting of almonds, figs, pistachios and, later, vines (King 1973, p. 161) . On Methana, the agricultural landscapes on volcanic soils contrast markedly with the degraded slopes and almost entirely abandoned terraces of limestone outcrops on the peninsula. There is evidence that farmers carried volcanic soil to enrich terraces built in the terra rossa of the limestone slopes (James et al. 1997). In view of the details of the distribution of Early Helladic and Classical-Hellenistic artefacts on terraced slopes, it is possible that some of the terrace systems were in existence in Classical times. Whether or not the terraces one sees on Methana today were in existence nearly 2500
333
years ago, the soils on parts of Methana have sustained agriculture for 5000 years, since the Early Helladic period.
Erosion and instability of volcanic soils Erosion has probably affected most volcanic soils that have been cultivated, and many that have been grazed. Rapid removal of the results of long periods of soil development may follow climatic change or human disturbance of vegetation. Because of steep slopes and erodible materials, erosion rates in many volcanic areas are high. Phases of past instability are reflected in evidence of burial and truncation in hillslope soil profiles. Cerdrero & Dramis (1996) classed susceptibility to landslides, which include lahars and large gravitational movements, as moderate to very high on volcanoes, and low to moderate in gentler volcanic reliefs. In tephra with high hydraulic conductivity, high pore-water pressure is a chief factor leading to slope instability. Problems for many human activities are caused by high water content of Inceptisols developed in tephra in Hawaii. The amount of water may reach six times the weight of other soil material, commonly rendering the soil thixotropic (Macdonald et al. 1981). The low strength and expansion of Vertisols, which result in landslides and in damage to roads, buildings and other structures (as reported from Hawaii: Macdonald et al. (1981)), have archaeological implications additional to those considered above for these soils. Despite their fine texture, Vertisols may also suffer erosion by overland flow. Udic Pellusterts and Udic Chromusterts in the Darling Downs, Queensland, may be severely eroded under high-intensity summer rainfall (Fairburn & Wockner 1986). When left exposed after cultivation for wheat, the soil is susceptible to structural breakdown, surface sealing, sheetwash and rilling. Erosion rates on slopes of 10° may reach 621 ha"1 yr"1, but are greatly reduced when crop stubble is left in place to protect the soil from rainsplash. Erosion by wind deflation and overland flow tends to be more severe in tephra than in lava soils (Ugolini & Zasoski 1979; Molloy 1993; Inbar et al. 1994). In the aftermath of eruptions, denuded slopes and fresh tephra deposits are prone to severe erosion by water. Plumes of sediment washed through rivers from the bare slopes near the crater of the Soufriere Hills Volcano of Montserrat have been described as 'severe and dramatic' (Brosnan 1997). Such high rates decline with time as sediment supply is reduced and as slopes stabilize (Rees 1979; Inbar et al. 1994).
334
P. JAMES, D. CHESTER & A. DUNCAN
The speed and erosive power of overland flow are reduced by several means, especially by agricultural terracing, constructed in volcanic soils in Central and South America, islands of the Atlantic and Pacific, the circum-Mediterranean, Japan and south-east Asia. The effectiveness of terraces is attested by their survival for centuries in many of these areas. Where they are maintained successfully, and where terrace surfaces slope at only a few degrees, erosion by water may be limited to rainsplash and removal of <2mm material by unconcentrated surface flow. A common feature of the terraces of Methana, where cultivation of the lava soils has been abandoned, are stone lag deposits formed where fine soil has been selectively removed. The process has highlighted the existence of sites of archaeological interest, the stone covers containing many artefacts, particularly sherds, dating from the Early Helladic to the Modern period (James et al. 1994). The rough and rocky slopes of Methana would appear to have been at high risk of the climatically and anthropogenically forced erosion that has marked Late Quaternary environmental change in the Mediterranean region. Methana has some of the ingredients of severe erosion risk: mountainous relief, a markedly seasonal rainfall with high-energy rainstorms, and a long history of agricultural use. Yet, except on the steepest slopes, where they have formed during periods of past landscape instability but are now largely stabilized, gullies are absent from almost all terraced volcanic slopes on Methana. Rills are rare on the terrace systems. It is possible that, since local intensive agriculture began in the Early Helladic period, gullying and rilling, the damaging processes of incisive erosion, have not been significant on the agricultural soils of volcanic Methana. The lava rock and soils have probably afforded protection against erosion for many of the Early Helladic, Classical-Hellenistic and Late Roman sites, of which most survive on agricultural terraces, many on moderate slopes, and a few on the crests of very steep slopes. The lava has lent protection in several ways: it has furnished abundant building blocks for terrace-wall construction, it has weathered to a cohesive soil that is not very erodible, and it has provided rough, irregular surfaces to many slopes that break up and absorb overland water flow. Conclusions 'Volcanic terrains have the most complex surface environments on earth' (Orton 1996,
p. 485). They also range from continental plateaux to small oceanic islands, and include areas of active volcanoes, where fresh tephra and lava lie unaltered, as well as areas of ancient rocks, where volcanoes are long extinct. They occur in all major climates. Volcanic soils are as diverse as their environments. They are related by the origin of their parent materials, and many soils in Quaternary tephra tend to have features in common that result from their distinctive mineralogy. Yet volcanic soils vary greatly in their appearance and chemical and physical characteristics. Some are infertile; many are rich, having sustained agriculture under conservative farming for thousands of years. As well as being extremely diverse, volcanic soils include those, now known as Andisols, that remained inscrutable to the soil scientist until recent decades. Today, there is a sufficient body of theory to enable the archaeologist working in tephra to explain the nature of the soil profile revealed by excavation. Investigations of diversity and change in soils across volcanic landscapes, and of how these may relate to patterns in archaeological data, may begin by examining the soils in relation to the frameworks of soil parent material, relief, drainage, vegetation, and landform age. The less accessible data on relative ages of volcanic soils and on the impacts of past human action and environmental change are to be obtained from detailed analysis of soil profiles. The authors thank T. J. Van Andel and E. M. Bridges for their invaluable comments on the first draft of the paper, and S. Mather for producing the map.
References BIRKELAND, P. W. 19840. Soils and Geomorphology. Oxford University Press, New York. 1984&. Holocene soil chronofunctions, Southern Alps, New Zealand. Geoderma, 34, 115-134. BIRRELL, K. S. & PULLAR, W. A. 1973. Weathering of paleosols of Holocene and Late Pleistocene tephras in central North Island, New Zealand. New Zealand Journal of Geology and Geophysics, 16, 687-702. BOCKHEIM, J. G. 1980. Solution and use of chronofunctions in studying soil development. Geoderma, 24, 71-85. BOUDOT, J.-P., BRAHIM, A. B. H. & CHONE, T. 1988. Dependence of carbon and nitrogen mineralization rates upon amorphous metallic constituents and allophanes in highland soils. Geoderma, 42, 245-260. BRADY, N. C. & WEIL, R. R. 1996. The Nature and Properties of Soils, llth edn. Prentice-Hall, London.
VOLCANIC SOILS AND ARCHAEOLOGY BROSNAN, D. M. 1997. Sustainable Ecosystems Institute, Portland, OR: Internet web page, http: //www. sei. org/impacts. htm. BUSACCA, A. J. 1987. Pedogenesis of a chronosequence in the Sacramento Valley, California, USA: I. Application of soil development index. Geoderma, 41, 123-148. BUURMAN, P., DE BOER, K. & PAPE, TH. 1997. Laser diffraction grain-size characteristics of Andisols in perhumid Costa Rica: the aggregate size of allophane. Geoderma, 78, 71-91. CAMPBELL, I. B. 1986. Recognition of paleosols in Quaternary periglacial and volcanic environments in New Zealand. In: WRIGHT, P. V. (ed.) Paleosols: their Recognition and Interpretation. Blackwell Science, Oxford, 208-241. CERDRERO, A. & DRAMIS, F. 1996. The contribution of landslides to landscape evolution in Europe. Geomorphology, 15, 191-211. CERNOHOUZ, J. & SOLE, I. 1966. Use of sandstone wanes and weathered basalt crust in absolute chronology. Nature, 212, 806-807. CHARTRES, C. J. & PAIN, C. F. 1984. A climosequence of soils on late Quaternary volcanic ash in Highland Papua New Guinea. Geoderma, 32, 131-155. CHESTER, D. K. 1993. Volcanoes and Society. Edward Arnold, London. COLMAN, S. M. 1986. Levels of time information in weathering measurements, with examples from weathering rinds on volcanic clasts in the Western United States. In: COLMAN, S. M. & DETHIER, D. P. (eds) Rates of Chemical Weathering of Rocks and Minerals. Academic Press, Orlando, FL, 379-393. & PIERCE, K. L. 1981. Weathering Rinds on Andesitic and Basaltic Stones as a Quaternary Age Indicator, Western United States. US Geological Survey Professional Paper, 1210. COLMET-DAAGE, F. 1967. Characteristiques de quelques sols d'equateur derives de cendres volcaniques. Cahiers ORSTOM, Serie Pedologique, 5, 353-392. CRONIN, S. J., NEALL, V. E., STEWART, R. B. & PALMER, A. S. 1996. A multiple parameter approach to tephra correlation, Ruapehu volcano, New Zealand. Journal of Volcano logy and Geothermal Research, 72, 199-215. DAHLGREN, R., SHOJI, S. & NANZYO, M. 1993. Mineralogical characteristics of volcanic ash soils. In: SHOJI, S., NANZYO, M. & DAHLGREN, R. (eds) Volcanic Ash Soils: Genesis, Properties and Utilization. Elsevier, Amsterdam, 101-139. DAN, J. & SINGER, A. 1973. Soil evolution on basalt and basic pyroclastic materials in the Golan Heights (northern Israel). Geoderma, 9, 165-192. DELVAUX, B., HERBILLON, A. J. & VIELVOYE, L. 1989. Characterization of a weathering sequence of soils derived from volcanic ash in Cameroons - taxonomic, mineralogical and agronomic implications. Geoderma, 45, 375-388. DERRY, D. R. 1980. World Atlas of Geology and Mineral Deposits. Mining Journal Books, London.
335
DORRONSORO, C. & ALONSO, P. 1994. Chronosequence in Almar River fluvial-terrace soil. Soil Science Society of America Journal, 58, 910-925. DUCHAUFOUR, P. 1977. Precis de Pedologie. Masson, Paris. ESWARAN, H. 1972. Micromorphological indicators of pedogenesis in some tropical soils derived from basalts from Nicaragua. Geoderma, 1, 15-31. FAIRBURN, D. M. & WOCKNER, G. H. 1986. A study of soil erosion on Vertisols of the Eastern Darling Downs, Quueensland. Australian Journal of Soil Research, 24, 135-158. FAO-UNESCO 1989. FAO-UNESCO Soil Map of the World, Revised Legend. World Resources Report 60. FAO, Rome. FERNANDEZ-CALDAS, E. & YAALON, D. H. 1985. Volcanic soils: weathering and landscape relationships of soils on tephra and basalt. Catena Supplement, 7. , QUANTIN, P. & TEJEDOR SALGUERO, M. L. 1981. Sequences climatiques des sols derives de roches volcaniques aux ties Canaries. Geoderma, 26, 47-62. FISHER, R. V. & SCHMINCKE, H. U. 1984. Pyroclastic Rocks. Springer, Berlin. , HEIKEN, G. & HULEN, J. B. 1997. Volcanoes: Crucibles of Change. Princeton University Press, PRINCETON, NJ. FORMAN, S. L., PlERSON, J., SMITH, R. P., HACK-
ETT, W. R. & VALENTINE, G. 1994. Assessing the accuracy of thermoluminescence for dating baked sediments beneath late Quaternary lava flows, Snake River Plain, Idaho. Journal of Geophysical Research - Solid Earth, 99(B8), 15569-15576. GIBBS, H. S. 1968. Volcanic Ash Soils in New Zealand. New Zealand Department of Scientific and Industrial Research Information Series, 65. 1980. New Zealand Soils: an Introduction. Oxford University Press, Wellington. GOLDICH, S. S. 1938. A study in rock weathering. Journal of Geology, 46, 17-58. GRADUSOV, B. P. & TARGUL'YAN, V. O. 1962. Minerals of the clay separates in soils developed on the igneous rocks of Eastern Sayan. Soviet Soil Science, 1962(11), 1237-1245. GRAHAM, R. C. & FRANCO-VIJCAINO, E. 1992. Soils on igneous and metavolcanic rocks in the Sonoran Desert of Baja California, Mexico. Geoderma, 54, 1-21. HAANTJENS, H. A., REYNDERS, J. J., MOUTHAAN, W. L. P. J. & VAN BAREN, F. A. 1967. Major Soil Groups of New Guinea and their Distribution. Department of Agricultural Research, Koninklijk Institut voor de Tropen, Amsterdam, Communication, 55. HALLSWORTH, E. G., ROBERTSON, G. H. & GIBBONS, F. R. 1955. Studies in pedogenesis in New South Wales. VII. The 'gilgai' soils. Journal of Soil Science, 6, 1-21. HARDEN, J. W. 1982. A quantitative index of soil development from field descriptions: examples from a chronosequence in central California. Geoderma, 28, 1-28.
336
P. JAMES, D. CHESTER & A. DUNCAN
HARRISON, J. B. J., MCFADDEN, L. D. & WELDON, R. J. 1990. Spatial soil variability in the Cujon Pass chronosequence: implications for the use of soils as a geochronologic tool. Geomorphology, 3, 399-416. HAY, R. L. 1960. Rate of clay formation and mineral alteration in a 4000 year old volcanic ash soil on St Vincent, B.W.I. American Journal of Science, 258, 354-368. HELGASSON, B. 1962. Basaltic soils of south-west Iceland I. Journal of Soil Science, 14, 64-72. HEWITT, A. E. 1989. New Zealand Soil Classification (version 2.0). DSIR Division of Land and Soil Sciences, Technical Record, D2 DSIR. HUGHES, C. J. 1982. Igneous Petrology. Elsevier, Amsterdam. HYNDMAN, D. W. 1972. Petrology of Igneous and Metamorphic Rocks. McGraw-Hill, London. INBAR, M., HUBBS, J. L. & Ruiz, L. U. 1994. The geomorphological evolution of the Paricutin cone and lava flows, Mexico 1943-1990. Geomorphology, 9, 57-76. ISBELL, R. F., STEPHENSON, P. J., MURTHA, G. G. & GILLMAN, G. P. 1976. Red Basaltic Soils in North Queensland. Commonwealth Scientific and Industrial Research Organization, Australia, Division of Soil, Technical Paper, 28. JACKSON, T. A. & KELLER, W. D. 1970. A comparative study of the role of lichens and 'inorganic' processes in the chemical weathering of recent Hawaiian lava flows. American Journal of Science, 269, 446-466. JAMES, P. A., ATHERTON, M. P., HARVEY, A. M., FIRMIN, A. & MORROW, A. 1997. The physical environment of Methana: formation, exploitation and change. In: MEE, C. & FORBES, H. (eds) A Rough and Rocky Place: the Landscape and Settlement History of the Methana Peninsula, Greece. Liverpool University Press, Liverpool, 5-32. , MEE, C. B. & TAYLOR, G. 1994. Soil erosion and the archaeological landscape of Methana, Greece. Journal of Field Archaeology, 21, 395-416. JONGMANS, A. G., Verburg, P., NIEUWENHUYSE, A. & VAN OORT, F. 1995. Allophane, imogolite and gibbsite in coatings in a Costa Rican Andisol. Geoderma, 64, 327-342. KANVA, M. B. 1995. A review of properties, nutrient supply, cultivation and management of volcanic soils, with particular reference to Papua, New Guinea. Papua New Guinea Journal of Agriculture, Forestry and Fisheries, 38, 102-123. KING, R. 1973. Sicily. David and Charles, Newton Abbot KIRCH, P. V. 1996. Late Holocene human-induced modifications to a Central Polynesian island ecosystem. Proceedings of the National Academy of Sciences of the USA, 93, 5296-5300. LEAMY, M. L. 1984. Andisols of the world. In: Congresso Internacional de Suelos Volcanicos: Communicaciones. Universidad de La Laguna, Secretario de Publicaciones, Serie Informe 13, 368-387. LIMBREY, S. P. 1967. The basaltic soils of Gran Canaria: fossil, relict and recent soils and their
environmental significance. PhD thesis, University of London. 1990. Edaphic opportunism? A discussion of soil factors in relation to the beginnings of plant husbandry in south-west Asia. World Archaeology, 22, 45-52. LORENZONI, P., MlRABELLA, A., BlDINI, D. & LULLI,
L. 1995. Soil genesis on trachytic and leucititic lavas of Cimini Volcano complex (Latium, Italy). Geoderma, 68, 79-99. LOWE, D. J. 1986. Controls in the rates of weathering and clay mineral genesis in air fall tephra: a review and New Zealand case study. In: COLEMAN, S. M. & DETHIER, D. P. (eds) Rates of Chemical Weathering of Rocks and Minerals. Academic Press, Orlando, FL, 265-330. LULLI, L. & BIDINI, D. 1980. A climosequence of soils from tuffs on slopes of an extinct volcano in southern Italy. Geoderma, 24, 129-142. MACDONALD, G. A., ABBOTT, A. T. & PETERSON, F. L. 1981. Volcanoes in the Sea. Hawaii University Press, Honolulu. MADEIRA, M., FURTADO, A., JEANROY, E. & HERBILLON, A. J. 1994. Andisols of Madeira Island, (Portugal). Characteristics and classification. Geoderma, 62, 363-383. MAEDA, T., TAKENAKA, H. & WARKENTIN, B. P. 1977. Physical properties of allophane soils. Advances in Agronomy, 29, 229-264. MARTINI, J. A. 1976. The evolution of soil properties as it relates to the genesis of volcanic ash soils in Costa Rica. Soil Science Society of America Journal, 40, 895-900. MIZOTA, C. & CHAPELLE, J. 1988. Characteristics of some Andepts and andic soils in Rwanda, Central Africa. Geoderma, 41, 193-209. & MATSUHISA, Y. 1985. Eolian addition to soils and sediments of Japan. Soil Science and Plant Nutrition, 31, 369-382. & VAN REEUWIJK, L. P. 1989. Clay Mineralogy and Chemistry of Soils Formed in Volcanic Material in Diverse Climatic Regions. International Soil Reference and Information Centre, Wageningen, Soil Monograph, 2. MOHR, E. C. J. & VAN BAREN, F. A. 1954. Tropical Soils. Interscience, London. , & VAN SCHUYLENBORGH, J. 1972. Tropical Soils: a Comprehensive Study of their Genesis. Mouton-Ichtiar Baru-Von Hoeve, The Hague. MOLLOY, L. 1993. Soils in the New Zealand Landscape: the Living Mantle. New Zealand Society of Soil Science, Canterbury. MORWOOD, M. J., O'SULLIVAN, P. B., AZIZ, F. &
RAZA, A. 1998. Fission track ages of stone tools and fossils on the east Indonesian island of Flores. Nature, 392, 173-176. MUHS, D. R. 1984. Intrinsic thresholds in soil systems. Physical Geography, 5, 99-110. NAHON, D. & LAPPARTIENT, J.-R. 1977. Time factor and geochemistry in iron crust genesis. Catena, 4, 249-254. NANZYO, M., DAHLGREN, R. & SHOJI, S. 1993. Chemical characteristics of volcanic ash soils. In: SHOJI, S., NANZYO, M. & DAHLGREN, R. (eds)
VOLCANIC SOILS AND ARCHAEOLOGY Volcanic Ash Soils: Genesis, Properties and Utilization. Elsevier, Amsterdam, 145-187. NEALL, V. E. 1977. Genesis and weathering of Andosols in Taranaki. Soil Science, 123, 400-408. NlEUWENHUYSE, A. G., JONGMANS, A. G. & VAN
BREEMAN, N. 1993. Andisol formation in a Holocene beach ridge plain under the humid tropical climate of the Atlantic coast of Costa Rica. Geoderma, 57, 423-442. ORTON, G. J. 1996. Volcanic environments. In: READING, H. G. (ed.) Sedimentary Environments: Processes, Fades and Stratigraphy. Blackwell Science, Oxford, 485-567. PARFITT, R. L., RUSSELL, M. & ORBELL, G. E. 1983. Weathering sequence of soils from volcanic ash involving allophane and halloysite. Geoderma, 29, 41-57. PATON, T. R. 1974. Origin and terminology of gilgai in Australia. Geoderma, 11, 221-242. PILLANS, B. 1997. Soil development at snail's pace: evidence from a 6 Ma soil chronosequence on basalt in north Queensland, Australia. Geoderma, 80, 117-128. PLUNKET, P. & URUNUELA, G. 1999. The archaeology of a Plinian eruption of the Popocatepetl Volcano. This volume. POMEL, R.-S. 1983. L'evolution paleoclimatique de Santorin (mer Egee) au quaternaire recent: stratigraphie et paleosols intervolcaniques. Mkditerranee, 48, 27-37. PRENDEZ, M., HEVIA, J. & CARRASCO, M. A. 1994. Natural atmospheric pollution produced by the particulate matter ejected from the Hudson Volcano, Chile. Environmental Geochemistry and Health, special issue, 273-285. QUANTI, P., BALESDENT, J., BOULEAU, A., DELAUNE, M. & FELLER, C. 1991. Premiers stades d'alteration de ponces volcaniques en climate tropical humide (Montagne Pelee, Martinique). Geoderma, 50, 124-148. REES, J. D. 1979. Effects of the eruption of Paricutin volcano on landforms, vegetation and human occupancy. In: SHEETS, P. D. & GRAYSON, D. K. (eds) Volcanic Activity and Human Ecology, Academic Press, New York, 249-292. RENFREW, C. 1979. The eruption of Thera and Minoan Crete. In: SHEETS, P. D. & GRAYSON, D. K. (eds) Volcanic Activity and Human Ecology. Academic Press, New York, 565-585. ROSE, W. I., HAHN, G. A., DREXLER, J. W., MALINCONICO, M. L., PETERSON, P. S. & WUNDERMAN, R. L. 1981. Quaternary tephra of northern Central America. In: SELF, S. & SPARKS, R. S. J. (eds) Tephra Studies. NATO Advanced Study Series, Series C Vol. 75. D. Reidel, Dordrecht 193-211. ROUBY, D., SOURIOT, T., BRUN, J. P. & COBBOLD, P. R. 1996. Displacements, strains and rotations within the Afar depression (Djibouti) from restoration in map view. Tectonics, 15, 952-965. SANDOR, J. 1987. Initial investigation of soils in agricultural terraces in the Colca Valley, Peru. In: DENEVAN, W. M., MALHEWSON, K. & KNAPP, G. (eds) Pre-Hispanic Agricultural Fields in the
337
Andean Region. British Archaeological Research Series, 359, 163-192. SCOTT, W. E. 1989. Volcanic-hazard zonation and long-term forecasts. In: TILLING, R. I. (ed.) Short Course in Volcanology. American Geophysical Union, Washington, DC, 25-49. SHEETS, P. D. (ed.) 1985. Archaeology and Volcanism in Central America: the Zapotitan Valley of El Salvador. University of Texas Press, Austin. SHERMAN, G. D. & UEHARA, G. 1956. The weathering of olivine basalt in Hawaii and its pedogenic significance. Soil Science Society of America Proceedings, 20, 332-340. SHOJI, S., NANZYO, M. & DAHLGREN, R. 1993. Volcanic Ash Soils: Genesis, Properties and Utilization. Elsevier, Amsterdam. SIMONSON, R. W. & REIGER, S. 1967. Soils of the Andept suborder in Alaska. Soil Science Society of America Proceedings, 31, 692-699. SINGER, A. 1966. The mineralogy of the clay fraction from basaltic soils in the Galilee, Israel. Journal of Soil Science, 17, 136-147. 1978. The nature of basalt weathering in Israel. Soil Science, 128, 217-225. & BEN-DOR, E. 1987. Origin of red clay layers interbedded with basalts of the Golan Heights. Geoderma, 39, 293-306. SLATE, J. L., BULL, W. B., Ku, T.-L., SHAFIQULLAH, M., LYNCH, D. J. & HUANG, Y.-P. 1991. Soil-carbonate genesis in the Pinacate volcanic field, northwestern Sonora, Mexico. Quaternary Research, 35, 400-416. SMITH, J. 1957. A mineralogical study of weathering and soil formation from olivine basalt in Northern Ireland. Journal of Soil Science, 8, 255-239. SMITH, W. M. 1962. Weathering of some Scottish basic igneous rocks with reference to soil formation. Journal of Soil Science, 13, 202-215. SOIL SURVEY STAFF 1975. Soil Taxonomy: a Basic System of Soil Classification for Making and Interpreting Soil Surveys. US Department of Agriculture Natural Resources Conservation Service, Washington, DC. 1990. Keys to Soil Taxonomy, 4th edn US Department of Agriculture. Pocahontas Press, Blacksburg, VA. 1997. Keys to Soil Taxonomy, 7th edn US Department of Agriculture. Pocahontas Press, Blacksburg, VA. STEEN-MCINTYRE, V. 19810. Approximate dating of tephra. In: SELF, S. & SPARKS, R. S. J. (eds) Tephra Studies. D. Reidel, Dordrecht, 49-64. 198\b. Tephrochronology and its applications to problems in New-World archaeology. In: SELF, S. & SPARKS, R. S. J. (eds) Tephra Studies. D. Reidel, Dordrecht, 355-372. SULLIVAN, A. P. & DOWNUM, C. E. 1991. Aridity, activity, and volcanic ash agriculture: a study of short-term prehistoric cultural-ecological dynamics. World Archaeology, 22, 271-287. TAKAHASHI, T. & SHOJI, S. 1996. Active aluminium status in surface horizons showing continuous climosequence of volcanic ash-derived
338
P. JAMES, D. CHESTER & A. DUNCAN
soils in Towada District, Northeastern Japan. Soil Science and Plant Nutrition, 42, 113-120. , DAHLGREN, R. & VAN SUSTEREN, P. 1993. Clay mineralogy and chemistry of soils formed in volcanic materials in the xeric regime of northern California. Geoderma, 59, 131-150. TAN, K. H. 1984. Andosols. Van Nostrand Reinhold, New York. TEJEDO SALGUERO, M. L., QUANTIN, P. & FERNANDEZ-CALDAS, E. 1979. Sequence climatique des sols anciens de la region septentrionale de Tenerife (lies Canaries). 2e partie: typologie, pedogenese et classification. Cahiers ORSTOM, XVII, 185-193. THORARINSSON, S. 1981. The application of tephrochronology in Iceland. In: SELF, S. & SPARKS, R. S. J. (eds) Tephra Studies. D. Reidel, Dordrecht, 109-134. UGOLINI, F. C. & ZASOSKI, R. J. 1979. Soils derived from tephra. In: SHEETS, P. D. & GRAYSON, D. K. (eds) Volcanic Activity and Human Ecology. Academic Press, New York. UNGER-HAMILTON, R. 1989. The epi-palaeolithic of the southern Levant and the origins of cultivation. Current Anthropology, 30, 88-103. VEGA, O. T. 1980. Content and ratio of groups and forms of iron compounds in the volcanic soils of Costa Rica. Soviet Soil Science, 12, 545-553. VIDIC, N. J. & LOBNIK, F. 1997. Rates of soil development of the chronosequence in the Ljubljana Basin, Slovenia. Geoderma, 76, 35-64. VIOLANTE, P. & WILSON, M. J. 1983. Mineralogy of some Italian Andosols with special reference to the origin of the clay fraction. Geoderma, 29, 157-174. VITALIANO, C. J., TAYLOR, S. R., FARRAND, W. R. & JACOBSEN, T. W. 1981. Tephra layer in Franchthi Cave, Peloponnesos, Greece. In: SELF, S. & SPARKS, R. S. J. (eds) Tephra Studies. D. Reidel, Dordrecht, 373-379.
VUCETICH, C. G. 1968. Soil-age relationships for New Zealand based on tephrochronology. Transactions of the 9th International Congress of Soil Science, Adelaide, III, 121-130. WADA, K. 1977. Allophane and imogolite. In: DIXON, J. B. & WEED, S. B. (eds) Minerals in Soil Environments, American Society of Agronomy, Madison, WI, 603-638. 1985. The distinctive properties of Andosols. In: STEWART, B. A. (ed.) Advances in Soil Science 2. Springer, New York, 173-235. 1989. Allophane and imogolite. In: DIXON, J. B. & WEED, S. B. (eds) Minerals in Soil Environments. Soil Science Society of America, Madison, WI, 1051-1087. & AOMINE, S. 1973. Soil development on volcanic materials during the Quaternary. Soil Science, 116, 170-177. WASKLEWICZ, T. A. 1994. Importance of environment on the order of mineral weathering in olivine basalts, Hawaii. Earth Surface Processes and Landforms, 19, 715-734. WHITTAKER, R. J., PARTOMIHARDJO, T. & RISWAN, S. 1995. Surface and buried seed banks from Krakatau, Indonesia: implications for the sterilization hypothesis. Biotropica, 27, 346-354. WIELEMAKER, W. G. & WAKATSUKi, T. 1984. Properties, weathering and classification of some soils formed in peralkaline volcanic ash in Kenya. Geoderma, 32, 21-44. WILMHURST, J. M. & McGLONE, M. S. 1996. Forest disturbance in the central North Island, New Zealand, following the 1850BP Taupo eruption. Holocene, 6, 399-411. XAVIER DE FARIA, F. 1974. Os solos da Ilha do Fogo. Junta de Investiga?6es Cientificas do Ultramar, Lisbon, Estudos, Ensaios e Documentos, 129. YOUNG, A. 1976. Tropical Soils and Soil Survey. Cambridge University Press, Cambridge.
The use of volcaniclastic material in Roman hydraulic concretes: a brief review RUTH SIDDALL Department of Geological Sciences, University College London, Gower Street, London WC1E 6BT, UK (e-mail: [email protected]) Abstract: The realization that the addition of volcaniclastic material to hydrated lime cements produced a concrete that was not only waterproof but would also set under water revolutionized the building programmes of the Roman Empire. The material became known as pulvis Puteolis (dust of Puteoli) from the Latin name for modern Pozzuoli on the Bay of Naples. Pozzuoli itself is the root of 'pozzolana', meaning any material (in most natural cases, volcanic in origin) that is capable of producing a hydraulic (waterproof) set when combined with lime cement. A reaction between quicklime (CaO) and volcanic ash promotes the growth of phases that, rather than reverting to calcium carbonate on curing, as in nonhydraulic cements, produce a hard, watertight material. Although volcanic materials were the aggregates of choice in the building projects of Imperial Rome, being locally derived, abundant and relatively easily quarried, their hydraulic properties were soon fully realized, perhaps as a result of observations of naturally calcite-cemented scoria and ash flows in sea water or from experimental building. Away from the Imperial centre, terrestrial architecture involving waterproofing (baths and cisterns) effectively used waste ceramic sherds as a substitute for volcanic ash. However, for marine and riparian architecture, volcanically derived pozzolanic cements were the preferred material of use.
The Roman architect and engineer Vitruvius, writing in the first century BC (Granger 1931) described a powder that produced 'wonderful' results when added to a simple lime-water mix. This 'powder', composed of tuffs derived from the volcanic province of Campi Flegrei (Plegrean Fields) on the Bay of Naples, enabled mortars and concretes to set in the presence of sea water and, in addition, to produce stronger structures than those built with lime cement alone. The tuffs were quarried from the vicinity of the modern town of Pozzuoli. The Romans called this material pulvis puteolanis, dust of Puteoli, Puteoli being the Latin name for Pozzuoli, which in turn has given the modern term 'pozzolana', widely applied to all additives to cements that produce a hydraulic set. The unique properties of this material were probably discovered as the local tuffs, scoria and lavas were the aggregate of choice for the construction of the harbour and the port at Roman Puteoli. However, later they were possibly to become an important export from the area. Vitruvius wrote that 'the masonry which is to be in the sea must be constructed in this way. Earthy material is to be brought from the district which runs from Cumae to the promontory of Sorrento, and mixed in the mortar, two parts of it to one of lime'. The Roman's
knowledge of the volcanic activity associated with the Vesuvius-Campi Flegrei volcanic province was probably limited, despite the construction of heated saunas (sudatoria) in the area around Baiae and Puteoli, utilizing the naturally occurring hot springs (Vitruvius, De Architectura, II.6). Until the catastrophic AD 79 eruption of Vesuvius, which destroyed the towns of Herculaneum and Pompeii, there is no evidence that the Romans were aware that Vesuvius was an active volcano. Therefore, in the absence of an apparent volcanic connection, they ought perhaps be excused for believing that the tuffs around the Bay of Naples were unique in their origin and chemistry and therefore in the hydraulic properties that they imparted to concretes. The Romans are, rather unjustly, generally credited with the invention of concrete. More specifically, they were the first to develop the material for large-scale structures in the form of vaults and domes. However, the technology of calcining limestone to produce lime and subsequent rehydration to form, on drying, a synthetic rock has been understood for at least 9000 years. Lime-cement based materials have been recognized from as early as the Middle Eastern Pre-Pottery Neolithic B period (BarYosef 1983; Kingery et al 1988), thus calcination of limestone to produce lime was one of the
From. McGuiRE, W. G., GRIFFITHS, D. R., HANCOCK, P. L. & STEWART, I. S. (eds) The Archaeology of Geological Catastrophes. Geological Society, London, Special Publications, 171, 339-344. 1-86239-062-2/00/ $15.00 © The Geological Society of London 2000.
340
R. SIDDALL
earliest forms of pyrotechnology known to have been utilized by man. The Greeks were familiar with the technology of concrete and mortar construction and used the material routinely for laying floors and covering walls from at least the Early Bronze Age and as a waterproofing agent for cisterns, reservoirs and baths from a similar period. As a structural material, concrete was rarely used, although a Hellenistic cast concrete barrel vault is known from Upper Peirene at Corinth (Stillwell 1930). Chemistry and technology of classical hydraulic concretes For the purpose of this paper, and it is hoped more widely, the terms mortar or concrete are defined as being a mixture of slaked lime (called 'cement' once cured) plus an aggregate. The term aggregate is not composition specific. The subdivision between the terms mortar and concrete is based on the size of the aggregate incorporated into it. Mortars contain an aggregate with dimensions less than 5mm whereas concrete contains an aggregate with dimensions greater than 5 mm. The binding cement may also contain a fine sand grade aggregate (less than 1 mm dimensions). If this is used alone, without the addition of an aggregate, then the material may be termed plaster. Lime cement is manufactured by burning limestone cobbles in a kiln to temperatures around 900°C, thus removing carbon dioxide (CO2) from the calcium carbonate to leave calcium oxide (CaO) or lime. Kilns, with obvious similarities to those still used in rural Mediterranean communities, were described by the Roman authors Cato and Vitruvius, and were of the flare kiln type. Once charged with limestone rubble, the fire was lit and the kiln left burning for several days whereupon it was emptied of lime and recharged with the next load. Dissociation temperatures for the breakdown of calcium carbonate to calcium oxide of c. 900°C are required, and these are within the range of temperatures obtainable from natural, non-fossil fuel sources (Wingate 1985). Within the Mediterranean region, wood, or olive kernels and almond husks provide excellent and abundant fuels, and are still used in small-scale lime kilns (Adams 1994), which are left burning for around 3 days. It is the silica (SiO2) and alumina (A12O3) compounds within volcanic tuffs that react with lime (CaO) to produce a hydraulic set in concretes. Lime, when mixed with water, forms a paste that will on curing take in carbon dioxide
from the atmosphere and revert to a synthetic calcium carbonate. Hydraulic sets may be produced in two ways; either a hydraulic lime is produced by calcination of a argillaceous limestone producing a material not dissimilar to modern Portland Cement, or alternatively a pozzolana is added to the dry lime before the slaking process. It is the latter process that was adopted in antiquity. The set is produced with the addition of water (H; the abbreviations used here are those of cement chemistry), whereby available silica (S) and alumina (A) from the pozzolana will undergo complex reactions with the lime (C) to produce amorphous, low-solubility phases. Hydration of lime will produce the mineral portlandite Ca(OH)2, CH in the shorthand of cement chemistry. The presence of pozzolanaderived S and A will promote the formation of calcium silicates and aluminates, which hydrate to form an amorphous gel, known as C-S-H, and hydrogarnet, C3AH6, during the highly exothermic slaking reaction (see Gani (1997) and references therein). The resultant material is a strong, water-resistant and durable mortar, as is illustrated by the many examples of Roman architecture still standing. Pozzolanas in antiquity A 'pozzolana' is defined as a siliceous and/or aluminous substance that will, in the presence of water, combine with lime to form cementitious compounds (Allen 1992). Such material includes clays, which are rendered active by firing (fresh geological clays would absorb too much water during the curing process, resulting in spallation and ultimately cracking of the concrete), and other substances, including waste products from blast furnaces and even rice husk ash (see Hill et al (1992) and references therein). However, natural pozzolanas are derived primarily from volcanogenic products. Whether the Ancient Greeks used pozzolanas with lime cements specifically as a waterproofing agent is a subject of some debate. Certainly, suitable materials were available from major volcanic structures such as Santorini and Melos (Fig. 1), both of which are exploited today for pozzolanas. The Greeks were well aware of the properties of lime cement based concretes, and used them almost exclusively as waterproofing agents for baths, cisterns and flooring. Further research is necessary to evaluate intentional inclusion of pozzolanic material in Hellenistic and earlier cements, mortars and concretes. In the Hellenistic fountain houses of Corinth (Peirene, Hill 1964; Upper Peirene, Stillwell
VOLCANICLASTIC MATERIAL IN ROMAN HYDRAULIC CONCRETES
341
Fig. 1. (a) Ports using pozzolanas in concretes during the Roman period; (b) major sources of pozzolanas, in the forms of volcaniclastic deposits, in the Eastern Mediterranean.
1930), waterproofing was attained by 'burnishing', rubbing the surface of the 'leather hard' mortar with stones, to produce a hard skin. Presence of volcanogenic material either as a coarse aggregate or finely powdered in the matrix is not obvious, although chemical analyses of the plaster taken at the time of excavation defined the cement to have a high silica content. Efstathiadis (1978) claimed the presence of 'Santorini earth' in concrete lining a cistern in Cameiros on Rhodes, although it is not clear what methods were used for analysis of the cement. The intentional use of pozzolanas by the Classical and Hellenistic Greeks cannot be ruled out, although it appears that they failed to realize the adaptability of the material. The Romans almost certainly inherited the technology of cement manufacture from the Greeks, but adopted the material to their own uses and significantly altered and perfected the use of aggregates, including pozzolanas, producing through innovation an extremely versatile construction material. Thus an 'architectural revolution' in the use of lime cements and their derivatives, mortars and concretes, has lead to the (misleading) accreditation that the Romans were the inventors of 'concrete'. Vitruvius (De Architecture II.6) suggested that a powder that produced 'wonderful results' was to be found in 'the neighbourhood of Baiae and the municipalities in the vicinity of Mount Vesuvius'. He also believed that the formation of this substance was related to the 'fervent heats' of the area. More recently, the volcanic activity of the area has been fully described (for an overview, Lirer et al. 1987; Scandone et al. 1991). The Campi Flegrei Caldera is a Quaternary structure, which is surrounded by three other Quaternary centres, Ischia, Vesuvius and
Procida, providing abundant potential locales for Vitruvius' pulvis Puteoli. Thus the province of Imperial Rome was surrounded by Quaternary volcanoes of the Roman Comagmatic Province (Principe et al. 1987), with. Rome itself surrounded by the edifices Vulsini, Vico, Sabatini and Laziale. Consequently, building projects along the Tyrrhenian coast used volcanogenic pozzolanas as the aggregate of choice (see Roy & Langton 1982, 1989), ashes, tuffs and scoria being in abundance, local and easily quarried. It is also possible that the Roman engineers observed (naturally) calcite-cemented beach rock composed of scoria blocks and used this as inspiration for their own materials (Fig. 2). That the Roman lime-pozzolana cement would set under water was probably a chance discovery. Without access to modern strengthtesting technology it is likely that they failed to realize that a better set occurred under water than when exposed to air. We have no evidence that Romans allowed water to run over terrestrial architecture to aid in the curing process (Lechtman & Hobbs 1986). In the dome of the Pantheon in Rome, a further property of volcanogenic material was utilized: pumice was used as aggregate in the upper parts of the dome to create a strong but lightweight structure. As a result of the legacy of the building projects of Imperial Rome, it has often become reported in the literature, and become an assumed fact amongst classical archaeologists, that it is the presence of pozzolana that makes concrete 'Roman'. Away from the abundant sources of volcanic material this is apparently not the case. Roman construction in the provinces did not use natural pozzolanas in terrestrial architecture, preferring to use local aggregate and, when required, using a synthetic pozzolana
342
R. SIDDALL
Fig. 2. Scoria naturally cemented by calcite; beach rock from the Bay of Naples (sample courtesy of the Johnston-Lavis Collection, University College London).
in the form of crushed and powdered potsherds for waterproofing baths, cisterns and aqueducts. Roman architecture in Germany, utilized the local volcanigenic pozzolana, 'trass', notably in constructions in Cologne situated in the Rhine Graben, with access to abundant volcaniclastic products, but not in Trier near the Belgian border (Lamprecht 1993). Studies are in progress on the Roman building phases in the Athenian Agora and Corinth in, respectively, Attica and the northern Peloponnese, Greece. In both examples, volcanic pozzolanas are not in evidence, even in important water-carrying structures such as the Athenian Hadrian's Aqueduct (Leigh, in prep.). For marine architecture, it would appear that a hydraulic and therefore lime-pozzolana concrete is an essential feature. Concrete at the Imperial Roman Ports of Cosa, Ostia and at Puteoli certainly utilize local volcanic pozzolanas (Roy & Langton 1989). In the Roman provinces, where volcanigenic material may not occur locally, it was necessary to import aggregates and pozzolanas to provide the hydraulic set. Two examples of important provincial Roman harbours, Herod's Caesarea Maritima in Israel, and the Corinthian port of Cenchreai on the Saronic Gulf of eastern Greece, are briefly discussed here. Herod's harbour at Caesarea Maritima The positioning of the port of Herod's capital at Caesarea posed problems, as the coastline
of the Levant is particularly straight and has no natural indentations suitable for harbours. Therefore Herod organized a highly technical and magnificent building project to construct his harbour at Caesarea Maritima (Sebastos), importing Imperial Roman architectural styles and apparently, Italian architects and engineers. The technological advances made at the harbour have been described by Oleson & Branton (1992). The harbour moles are characterized by huge blocks of concrete, far larger than would be manageable from quarried stone. These were created by pouring concrete into wooden formwork, the forms floated out as rafts and then sunk to produce the moulds. The concrete was a pozzola-lime hydraulic set, and Oleson & Branton (1992) believe that the pozzolana was imported from quarries in the vicinity of Pozzuoli for the purpose. Their conclusions were reached after major and trace element analyses were conducted on volcanogenic fragments from the Herodean mortars and these were compared with material from local Palestinian volcanic deposits and those surrounding the Bay of Naples, and also material from Santorini. That Herod would go to the expense of importing shiploads of pozzolana from over 2000 km away is amazing enough in itself. It is believed that perhaps the imported Roman engineers preferred to work with materials that were familiar to them. At present, no other examples of trade in Italian pozzolanas are known during the Roman period, but the consequences of the discoveries at Caesarea Maritima are far reaching. Cenchreai, the eastern port of Corinth The harbours of Corinth were far less ambitious structures than that at Caesarea, designed rather to be functional than to impress. However, Corinth was none the less an important Mediterranean port. During the entirety of its history, Corinth has been an important city, sited as it is unique position on the 'cross-roads' of Greece, and Greece is the cross-roads of the Mediterranean. After the sack of Hellenistic Greece in 146 BC, the city lay in waste for a century before being resettled as a Roman colony and once more became a major trading post in the Mediterranean (Williams 1993). At the time, there being no sea-route across the Isthmus, Corinth had two ports, Lechaion on the Gulf of Corinth, and Cenchreai on the eastern Saronic Gulf. Excavations at Cenchreai have revealed submerged harbour moles and partly submerged buildings adjacent to the quayside. These are distinctive from other constructions of Roman
VOLCANICLASTIC MATERIAL IN ROMAN HYDRAULIC CONCRETES age in the Corinthia in that the stone and brickwork is bonded by a mortar that is pale purple and is packed with an aggregate of a porphyritic hornblende andesite with conspicuous plagioclase glomerocrysts and euhedral phenocrysts of hornblende. The groundmass has in hand specimen a 'red-purple' colour, which was distinctive enough to have been recognized by the excavators when they found concentrations of the same material in the Temple Cellar, used as a warehouse (Scranton et al. 1978). The material was recognized as being non-local, and its storage in the warehouse along with an abrasive for polishing, known as 'miltos', also nonlocal, suggests that it was imported. Miltos is composed dominantly of crushed volcanic glass and was known to have been imported from the island of Melos. Import of the concrete aggregate from Melos is also tentatively proposed. This material would have almost certainly produced a hydraulic set in the concrete, rendering it waterproof, and it appears it was specifically selected for this sea-coast architecture.
343
importance of volcanogenic materials in producing hydraulic sets, and certainly went to the trouble of importing volcaniclastic material when it was not locally available (or in the cases of Caesarea Maritima, even when local sources were present!), for the construction of harbour moles. The literature concerning Roman harbours is large, but scattered. The true scale of the trade in volcaniclastic material for hydraulic cements throughout the Mediterranean cannot be assessed without a systematic study of materials used in individual localities. This paper arises from research into classical concrete technology focused on the sites of ancient Corinth and the Athenian Agora, Greece. Funding has been received from the Weiner Laboratory and Corinth Excavations, and for both sites from the American School of Classical Studies at Athens. Individual thanks go to those who have provided impetus and inspirations for the project: the Directors of Excavations at Corinth, C. Williams (1966-1997) and G. Sanders (1997 to present), N. Bookidis and B. Gebhard; at the Athenian Agora, the Director of Excavations J. Camp and S. Leigh; at the Weiner Laboratory, S. Vaughan, S. Pike, K. To and M. Sakalis.
Discussion
References
The ability of volcaniclastic materials to produce a strong waterproof and water-resistant cement when mixed with lime was used with great success by Roman architects and engineers from at least the first century BC, and led to the material being used in spectacular building projects on land and for the construction of piers for bridges and harbours in aqueous environments. There is evidence that, although unaware of the direct petrogenesis of natural pozzolanas, the Romans were aware that they were to be found in volcanically active areas, and they certainly utilized other volcanic sources when local to their building projects (i.e. trass from the Rhine Graben). However, they were happy to substitute potsherds for small-scale and terrestrial waterproofing purposes. The Romans certainly were not the inventors of concrete, and were probably not even the inventors of lime-pozzolana hydraulic cements. They were, however, receptive and innovative enough to be the first to incorporate volcaniclastic material (ash, and crushed pumice, lava and scoria) into large-scale building projects, especially for use in marine architecture. It should be noted that the Romans used simple lime cement and lime-potsherd cements for terrestrial architecture; Roman cements, mortars and concretes, should not be defined as always containing a pozzolana. It is clear that Romans recognized the
ADAMS, J. P. 1994. Roman Building Materials and Techniques (FULFORD, M. transl.). Batsford, London. ALLEN, W. J. 1992. Locating reactive natural pozzolana. In: HILL, N., HOLMES, S. & MATHER, D. (eds) Lime and other Alternative Cements. Intermediate Tectnology, London, 64-72. BAR-YOSEF, 1983. The Natufian in the Southern Levant. In: CUYLER-YOUNG, T., SMITH, P. E. L. & MORTENSEN, P. (eds) The Hilly Flanks and Beyond: Studies in Ancient Oriental Civilisations. University of Chicago Oriental Institute, 36, 11-42. EFSTATHIADIS, E. 1978. Greek Concrete of Three Millennia. Research Centre of the Hellenistic Ministry of Public Works, Athens. GANI, M. S. J. 1997. Cement and Concrete. Chapman and Hall, London. GRANGER, F. (transl.) 1931. Vitruvius, De Architectura. Loeb Classical Library. HILL, B. H. 1964. Corinth 1: VI: The Springs, Peirene, Sacred Spring, Glauke. Excavations of the American School of Classical Studies at Athens. Princeton, NJ. HILL, N., HOLMES, S. & MATHER, D. (eds) 1992. Lime and other Alternative Cements. Intermediate Technology, London. KINGERY, W. D., VANDIVER, P. B. & PRICKETT, M. 1988. The beginnings of pyrotechnology, part II: production and use of lime and gypsum plaster in the pre-pottery Neolithic Near East. Journal of Field Archaeology, 15, 219-244. LAMPRECHT, H. O. 1993. Opus Caementitium. Bautechnik der Romer. Beton, Dusseldorf.
344
R. SIDDALL
LECHTMAN, H. N. & HOBBS, L. W. 1986. Roman concrete and the Roman architectural revolution. In: KINGERY, W. D. (ed.) Ceramics and Civilisation, Vol. HI; High Technology Ceramics: Past, Present and Future. 81-124. LIRER, L., LUONGO, G. & SCANDONE, R. 1987. On the volcanological evolution of Campi Flegrei. Eos Transactions, American Geophysical Union, 68, 226-234. OLESON, J. P. & BRANTON, G. 1992. The technology of King Herod's harbour. Journal of Roman Archaeology, Supplementary Series, 5, 49-67. PRINCIPE, C, Rosi, M., SANTACROCE, R. & SBRANA, A. 1987. Explanatory notes to the geological map. Quaderni de La Ricerca scientifica, 114, 11-51. ROY, D. M. & LANGTON, C. A. 1982. Longevity of borehole and shaft sealing materials: 2. Characterisation of cement-based ancient building materials. Report ONWI-202, Materials Research Laboratory, Pennsylvania State University, for the Office of Nuclear Waste Isolation, Battelle Memorial Institute, Columbus, OH. & 1989. Studies of ancient concrete as analogs of cementitious sealing materials for a repository in
tuff. Report LA-11527-MS, Materials Research Laboratory Pennsylvania State University at Los Alamos National Laboratory, for Civilian Radioactive Waste Management Program, US Department of Energy. SCANDONE, R., BELLUCCI, F. LIRER, L. & ROLANDI, G. 1991. The structure of the Campanian Plain and the activity of Neapolitan volcanoes. Journal of Volcanology and Geothermal Research, 48. SCRANTON, R., SHAW, J. W. & IBRAHIM, L. 1978. Kenchreai: Eastern Port of Corinth, I. Topography and Architecture. Excavations of the Universities of Chicago and Ohio for the American School of Classical Studies at Athens. Brill, Leiden. STILLWELL, R. 1930. Corinth 111:1 Acrocorinth: Upper Peirene. Excavations of the American School of Classical Studies at Athens. Princeton, NJ, 31-49. WILLIAMS, C. K. II 1993. Roman Corinth as a commercial center. Journal of Roman Archaeology, Supplementary Series, 8, 31-46. WINGATE, M. 1985. Small-scale Lime Burning, a Practical Introduction. Intermediate Technology, London.
Olmec stone sculpture: selection criteria for basalt PATRICK HUNT Stanford University, Humanities Program, Stanford, CA 94305, USA Abstract: The fact that Olmec monumental sculpture exclusively uses basaltic stone requires explanation. Although many possible stone selection criteria are potentially involved in Olmec culture for the deliberate choice of stone, it is unlikely that all or even many were considered when but one stone type (basalt) was chosen for such objects as colossal heads and monumental sculpture. It is suggested here that the consistent correlation between Olmec monumental sculpture and basalt is explicable, based on the deliberate criterion of stone selection for metaphysical reasons and by physical characterization: this mostly dark, vesicular stone was the most compelling reminder of underworld power expressed in volcanism and volcanic mountains, and was thereby appropriated for power in rulership of the Olmec culture.
Mesoamerican culture in the heartland of the isthmus has been justifiably famous for stone use, particularly sculpture, and the term 'mother culture' has been used by Covarrubias and others (Covarrubias 1957; Kubler 1962 (especially pp. 68-7 Iff), 1975, 1984; Coe 1981; Bernal 1969; de la Fuente 1973; Benson 1982, Chapter 2) to describe the Olmec culture from the Early to Late Formative periods (1500-100 BC), although some feel that seminality of Olmec cities is an exaggeration (SablofT 1989, Chapter 2). More than 20 sculpted Olmec colossal heads weighing as much as 201 and with a height of 3 m such as 'El Rey' at La Venta have been noted by researchers from Stirling onward (Stirling 1943), including the seriated studies of Clewlow (1974) and Milbrath (1979) (not even counting Olmecoid heads) and the research of Coe and Diehl (Coe & Diehl 1980; Coe 1981) from the sites of San Lorenzo, La Venta, Laguna de los Cerros and Tres Zapotes, as well as minor sites outside the heartland area (Fig. 1). Aspects such as the size of these colossal heads, probable monumental use, commonality of physical features and continuity of tradition over a millennium have all been broadly discussed in the literature, but the fact that all or nearly all of the identified heads (20+) and most other large-scale Olmec sculptures or objects greater than 50cm length (70+ monumental objects) are carved from basalt and related volcanic stone may be one of the most interesting facts about these colossal sculptures, which has not been noted outside a few studies (Williams & Heizer 1976; SablofT 1989, pp.35, 41; Hunt 1991, Chapter 6, 1994, p. 266) (Fig. 2). The
basalt was used for colossal heads (and no other type of stone other than volcanic material seems to have been used for colossal heads) and was also used for tenon sculpture, altars, stelae, zoomorphic figures, receptacles and vessels, slabs, columns, sarcophagi and platforms. This brief paper addresses some potential criteria of this deliberate choice of basaltic stone and suggests the attraction of the Olmec to this volcanic material in particular. Geological considerations That ancient cultures may have had criteria of stone selection presupposes some recognition of various stone typologies by such physical characteristics as colour and hardness (or, conversely, workability) among others. Although this may not be a nascent geological awareness approaching a scientific appraisal, none the less the fact that the colossal heads are uniformly of basalt is in itself a suggestion of deliberation, and possible rationales should be examined. Elsewhere, possible criteria of stone selection have been suggested as availability, accessibility, workability, durability, natural shaping or cleavage, aesthetic appeal and metaphysical associations, or combinations of these possible criteria (Gauri 1978; Hunt 1991, pp. 42-57, 1994, p. 266). As Williams & Heizer (1976, p. 4) first noted and Coe & Diehl (1980) discussed, the known provenance of basaltic or related volcanic stone for Olmec monuments can be traced to volcanic sources in the Tuxtlas Mountains. The sources
From: McGuiRE, W. G., GRIFFITHS, D. R., HANCOCK, P. L. & STEWART, I. S. (eds) The Archaeology of Geological Catastrophes. Geological Society, London, Special Publications, 171, 345-353. 1-86239-062-2/00/ $15.00 © The Geological Society of London 2000.
346
P. HUNT
Fig. 1. Olmec heartland with major sites and Tuxtlas range in the isthmus of Mexico.
are mostly known, notably Cerro Cintepec volcano for certain San Lorenzo and La Venta monuments, roughly 90km and 160km distant from the Tuxtlas range, respectively, and Cerro El Vigia volcano for Tres Zapotes monuments, roughly 30 km distant. Although these are hardly unusual distances from which to procure stone, and particularly proximal for Tres Zapotes because this mountain can be easily viewed from that site, the compelling singularity of volcanic stone type calls for discussion in part
because San Lorenzo and La Venta are not situated near volcanic stone regions but rather are in sedimentary regions. Although not all of the basaltic material used by the Olmec is uniform in petrology and mineral composition (ranging from pyroxene basalt, basaltic andesite, andesitic basalt, to pyroxeneolivine basaltic andesite and other related basalts and andesites (Williams & Heizer 1976; Hunt 1991, pp. 80-88)) or in colour (ranging from black, dark grey, dark green and dark brown to mixed hues of the others and sometimes weathered to light hues), it is mostly a dark stone with a vesicular texture, as is common to volcanic lava world wide, and could be thus recognizable by this combination of common physical features in dark vesicularity. Possible criteria of stone selection
Fig. 2. Olmec colossal head basalt sculpture.
Possible relevant factors influencing deliberate basalt choice for the Olmec could include a long-term local stoneworking tradition, availability and accessibility of this material, durability of basalt over time in a tropical environment when other local stone is unsuitable (mostly friable sandstone in both San Lorenzo and La Venta) or no other stone is
OLMEC STONE SCULPTURE available, colour or other aesthetic considerations, the natural rounded shaping and large size of volcanic ejecta or weathered shapes as amenable to further stonedressing, workability of basalt, metaphysical associations or considerations, or possible combinations of these and/ or other unknown reasons. The question must be asked whether it is truly a deliberate choice of basalt. Given the local jungle substratum of alluvial sandstone-derived geological matrix at San Lorenzo and La Venta and the alluvial plain of Tres Zapotes (as all three major sites are in riverine alluvial basins) and the fact that other, far more local sandstones or carbonates could have been more easily obtained and worked, the logical answer is most likely that deliberation was instrumental in basalt use. Whether basalt and volcanic stone in general was actually a knowledgeable choice is even more fascinating (as well as more challenging) and will be a primary focus of this discussion of the Olmec monumental stone sculptures.
347
A known stone source may also be considered important somewhat irrespective of its distance. If its working does not require adapting to new physical characteristics or learning new techniques of stonedressing, such a familiar stone source is likely to be utilized in some cultural continuity across many generations. Tradition therefore can be a factor in stone selection. On the other hand, what may begin as one criterion may eventually become a tradition in which it is not vital to maintain or even remember the original criterion of selection. This could be particularly true in cultures where tradition is more respected than practicality. This might also be true in a culture with a long history, and would be expected in a non-literate or theocratic culture. Whether or not these possibilities exist for Olmec culture cannot be proven at present, and further progress is difficult without a literary record or documented myth history.
Accessibility and availability
Familiarity with a particular stone may predetermine subsequent stone selection and continue a tradition already in place. What may not be easily discovered is why a tradition starts in the first place, although in the case of the original Olmec heartland around Lake Catemaco, itself a volcanic caldera and surrounded by newer volcanoes (Fig. 3), such early proximity to basalt may reinforce stone choices later when migration may remove a culture from such stone sources.
The proximity of Tres Zapotes to Cerro El Vigia at 30km (Fig. 4) could be natural evidence for the criterion of accessibility and availability (although they are not necessarily identical: available stone may be found locally in great volume but may be in an inaccessible context as a result of altitude, depth or some other feature). Conversely, recognition of volcanic stone as potential Stoneworking material and making its association with mountains but finding its local dearth on the alluvial plain may force the search for it to greater distances, which is still an inverse of availability as a criterion. The greater
Fig. 3. Lake Catemaco caldera with historical cinder cones (18th-century eruptions).
Fig. 4. Cerro El Vigia from near Tres Zapotes (c.20km).
Stoneworking tradition
348
P. HUNT
distance, however, of San Lorenzo and La Venta to Cerro Cintepec (90 and 160km, respectively) strongly suggests that availability and accessibility are not the primary criteria for stone selection in Olmec contexts. Furthermore, Williams and Heizer have also found Cerro El Vigia material in San Lorenzo (especially Monument 9), which is considerably further away, at a 190km distance, than even Cerro Cintepec as a source. Transport by water along the coast adjacent to the Tuxtlas range, up through the Coatzalcoacos and Papaloapan river watersheds along these waterways and possible canals has already been discussed (Williams & Heizer 1976; Coe & Diehl 1980; Miller 1986, p. 21; Sablof 1989, p. 41). Water transport can mitigate some of the difficulties inherent in the distance of land transport, thereby making inaccessibility a moot point as a possible criterion.
Durability In a tropical environment with high annual rainfall and subsequent high leaching and dissolution of sedimentary stone and carbonates in particular, the durability of one stone over others could be a vital characteristic for stone selection. The coefficients of thermal expansion and contraction of the more soluble limestone and sandstone are nearly double those of basalt and the dissolution potential increases with temperature (Blair 1955; Clark & Candle 1961). As Winkler has stated, 'carbonate solution in a tropical humid climate records much higher solution rates than in moderate humid areas' (Winkler 1975, p. 141). This is the tropical environment that the Veracruz and Tabasco states of Mexico in the Olmec heartland possess, where observation over a few hundred years in antiquity could have discouraged sculpture of local carbonates and sandstones. Studies also show that stone with igneous material such as basalt has a greater resistance to solubility than carbonates or most sandstones (Winkler 1975, pp. 46-47). Durability could also be expressed by its corollary hardness, and basalt is definitely harder as a rule than carbonate and nearly all sandstone (except metamorphosed sandstone, i.e. quartzite, which is not usually classified as sandstone), which can be calculated from relative Mohs hardness, compressive strength, Shore scleroscope and Schmidt impact hardness (Winkler 1975, pp. 34-37). Although basalt and related volcanic stone may be harder to work than most
sedimentary stone such as limestone and sandstone, it is also thus likelier to survive long term than these other local materials. Colour and aesthetic considerations Recognition of basaltic stone by colour range has been already briefly mentioned, and its general dark hues and vesicularity are nearly constant physical characteristics. Colour can be a distinctive aesthetic factor for stone selection, and other choices of stone for colour are well known in Mesoamerican cultures, with generic greenstones being the obvious example (Bishop et al. 1984; Hunt 1993) even when the stone type may be diverse. Greenstones ranging from light to dark green hues were valued highly, whether albitite, nephrite, jadeite, aventurine quartz, aragonite, serpentine, metasomatized basalt, basaltic andesite, green jasper quartzite or other materials. Olmec greenstone examples are seen in celts, were-jaguar masks or effigies such as the Kunz axe, the Offering 4 figures of La Venta and other objects, and the valuation of greenstone as a precious stone continues through nearly all other known Mesoamerican cultures (Kunz 1890; Washington 1922; Foshag 1957; Miller 1986; Harlow 1987; Sabloff 1989). Thus if greenstone can be a stone selection based on colour, it is not unlikely for basalt, typically dark grey to black and with noted vesicularity, to be a recognizable stone material for the sculptural medium. The only other known use of basalt (other than indicated previously) in Olmec culture are the hexagonal basalt prisms that are sometimes utilized alongside or instead of serpentine stelae, with serpentine also used mostly for small-scale celts alongside jadeite sculptures in the figures of Offering 4 or the large serpentine mosaic mask, both at La Venta. The primary exception to basalt for large sculpture is the Las Limas greenstone figure (although its exact stone type is unknown, as it may also be of basaltic material), which is again of a consistently used green-coloured material (Miller 1986, p. 30 and fig. 12). Thus stone recognition and physical characteristics can be easily deduced for the Olmec, which raises the question of why basalt might have been selected specifically for criteria other than accessibility, availability and durability. It is possible that some volcanic stone may have chosen for its green colour among other considerations, such as the distinctive weathered green Tres Zapotes basaltic andesite. Greenstone will be discussed again for metaphysical and spiritual associations.
OLMEC STONE SCULPTURE Natural shaping and cleavage Another possible criterion of stone selection could be the natural rounded shaping and large size of the volcanic ejecta (although 'bombs' begin at 64 mm, in some cases the weight exceeds hundreds of kilograms) (Bell & Wright 1985, p. 20; Le Maitre 1989, p. 53). Roundness could be conducive to choice as colossal head sculpture material (Fig. 5), especially when millennia of weathering by water action in stream beds has rounded these large boulders to cobble shapes, as can be observed in many riverine watercourses throughout the Cordilleras Volcanicas of Central Mexico (Hunt 1994). If stoneworking masons started with already rounded shapes, nature has done most of the work by weathering and erosion. Economy in stoneworking would be an asset with a natural rounded shape for colossal head material. Weathered shapes can be amenable to further stonedressing in that the cleavage of volcanic material is typically conchoidal. This can also ultimately favour development of rounded shapes, particularly when water erosion softens stone edges by tumbling large stones into cobbles in repeated flood seasons. This would again imply long-term observation by the Olmec and recognition of basaltic material even when not directly associated with contemporary volcanism. Depth of volcanic flows is also a potential factor in that extrusive volcanic deposition can
Fig. 5. Cerro El Vigia stream bed with large rounded basalt boulders (boulder size 1 m).
349
be massive, with thick, viscous flows of lowsilica basalt, which create the large mass of stone suitable for sculpture. This is true of igneous stone in general from magmatic melt sources, whereas sedimentary and even metamorphic stone can be deposited in much thinner bedding planes (although equally massive as well in other global contexts), which might limit the potential size and volume of stone for sculpture. Because lava is not usually deposited in bedding planes, its natural conchoidal fractures and cleavages can create massive potential blocks of stone. Workability On the other hand, silicate rock and igneous material are usually much harder to work for sculpture than are most carbonates (especially limestone and marble) and sedimentary sandstone. Workability here is defined as the ease or lack of resistance of a stone to modification, as reflected in relative hardness and predictability of fracture. The relative Mohs hardness of basalt at 5-6.5 was extrapolated by Winkler from the combination of silica and feldspar, whereas he placed the relative Mohs hardness of limestone as typically around three, and that of sandstone at 2-7 (the upper range being quartzite) depending on its adhesion, with anything above hardness six likely to be metamorphosed and therefore not true sandstone (Winkler 1975, p. 14, 31). Generally basalts are harder than and therefore less workable than limestones and other carbonates and most sandstone, which makes basalts and related volcanic rocks not a typical sculptural medium compared with these other stones (Hunt 1991, pp. 36, 41), as is borne out by the lower global quantity of volcanic stone in sculptural contexts relative to other more workable stone when both are available. Workability of basalt for sculpture is not therefore typically the highest criterion of stone selection (Winkler 1975, p. 30). Although all of the factors noted as potential criteria for stone selection for sculpture and monumental objects in Olmec culture (stoneworking tradition, accessibility, availability, aesthetic considerations, durability, natural shaping, and cleavage and workability), assuming there was not a lack of interest in intellectualizing the process, may have influenced Olmec choices, even in possible combinations, none appear entirely satisfactory given two deducible facts: the nearly unanimous choice of basalt for large-scale sculpture and monumental objects; the distance between find contexts in Olmec sites and geological sources for basalt
350
P. HUNT
when other materials are more local, available and workable. The highest priority selection criterion assumed here is covered in the following section. Metaphysical associations Metaphysical association is perhaps the most interesting possible criterion for stone selection of basalt in Olmec culture. It is also highly problematic in that it is an argument from silence, although any discussion of possible stone selection criteria may also be likewise impossible to prove in the case of the Olmec, as literary finds, which could document such criteria, are almost nonexistent. Preciousness and metaphysical associations or high valuation of one stone colour in the case of greenstone has already been asserted for Mesoamerican cultures beginning with Olmec use and continuing through Aztec culture (Rands 1965; Miller 1986, pp. 18, 29-30; Berrin & Pasztory 1993). The nature of basalt as volcanic stone may be the most important feature for Olmec culture. Several observations may support this hypothesis. First, Olmec mounds at the primary sites have been identified as possibly replicating volcanoes or cone volcanic peaks in several studies (Heizer 1968; Miller 1986, p. 24; Bernal-Garcia 1994, pp. 113-124), as Miller stated for La Venta: This impressive mound may have been intended to echo the shape of a Central American volcano' (Figs 6 and 7). Bernal-Garcia has also inferred that mountains are sacred and that they are supernatural precursors of all later Mesoamerican pyramids and synonymous with them. Bernal-Garcia also suggested mountains as being possible sources of divine and human speech (Bernal-Garcia 1994, pp. 113-114), which last idea may curiously evoke the thunderous sound of a volcanic eruption as a contemporary experience for the Olmec.
Fig. 6. La Venta mound, model (after Sabloff 1989).
Fig. 7. La Venta mound (c. 22m height).
Second, it is also a feature of the primary Olmec sites of San Lorenzo, La Venta and Tres Zapotes that certain high volcanoes of the Tuxtlas range are or were visible from each site. This is certainly true of Tres Zapotes, with Cerro El Vigia only 30 km distant, and was likely to be true of San Lorenzo and La Venta before the advent of petrochemical pollution from the Coatzalcoacos refineries. Coe alluded to this visibility in his fieldwork at San Lorenzo (Coe & Diehl 1980) and Bernal-Garcia also suggested that having 'a mountain nearby was of utmost importance for the well-being of the whole settlement, particularly for the rulers' (BernalGarcia 1994, p. 115). Third, other seminal Mesoamerican comparanda have been reinforced by Bernal-Garcia, in showing the Maya cosmology of the Popol Vuh where 'Zipacna the mountain dragon turns to stone' (Bernal-Garcia 1994, p. 117; quoting Tedlock, 1985, pp.98, 182) and in quoting Nuttall's Peabody Museum 1926 papers from colonial reports that the Teotihuacan hill of Cerro Gordo was originally named Tenon or 'mother of stone' (Nuttall 1926), which is doubly interesting in that Cerro Gordo is a volcanic peak and that lava as volcanic stone would literally be generated there. That this could be inferred from historic association with volcanic mountains seems very likely in an active volcano-rich region. Fourth, Bernal-Garcia stated that 'Stone, being the mountain's main substance, made these [rulers'] thrones "small mountains"', and also that 'as the ruler emerged from the entrails of the mountain, he carried the principal emblem of rulership, namely the divine ancestor, the baby jaguar' (Bernal-Garcia 1994, p. 117). Again, the 'mountains' probably meant those in the Olmec heartland and the only 'mountains' from which they sculpted stone were the Tuxtlas range, all volcanic in source. This fact is not insignificant and reinforces that basalt as recognizable 'mountain' stone is the stone deliberately chosen to represent special monuments in a consistent
OLMEC STONE SCULPTURE material through the entire Olmec history. The observations of Bernal-Garcia and others in these first four lines of evidence directly relate to the final line of evidence suggested below. Fifth, and perhaps most important, as awed witnesses of historical volcanic activity, which can be easily documented between 1800 and 100BC in the Tuxtlas range from the Early through Late Formative periods (Gill 1981; Sheets 1983; Hasenaka & Carmichael 1987; Luhr & Prestegaard 1989), the Olmec would have been able to mark the enormous natural power of volcanism. Immediately local to the Olmec heartland, some of the new cinder cones around Lake Catemaco are as historically recent as the 18th century. Other ancient Mesoamerican cultures were also eyewitnesses to volcanism and its destructive power, including Cuiculco, the earlier rival state of Teotihuacan in the Valley of Mexico, which was wiped out by volcanic activity in the Late Formative period around 150 AD, and other cultures in Central America (Sheets 1983; Sabloff 1989, p. 61). The hot molten lava in contact with organic material would enflame anything combustible in its path. More important perhaps, the lava flowing from the underworld, a locus of Olmec power (Reilly 1994), could undergo a 'divine' transformation from molten and mobile translucent flow to cold dark stone within a few days. Lava 'from the entrails of the mountains' and
Fig. 8. Active volcanism (Vesuvius, Italy).
351
Fig. 9. Cooling lava (Kilauea, Hawaii).
the 'mountain dragon turning to stone' (BernalGarcia 1994, p. 117) can be a frighteningly powerful force in nature. If the Olmec abandoned this original heartland of the Tuxtlas near Lake Catemaco because of volcanic activity (the speech and authority of the mountains being too powerful for them to live safely in adjacent valleys), they could still want the reminder and association with these basaltic volcanoes. It is likely that the metaphysical appropriation of this power of the mountain would be the aim and prerogative of the Olmec ruler (Figs 8 and 9). Thus sculpting monuments and objects from this basalt would be a most emphatic method of maintaining power from the mountains, whose reminders would also be present in the site mounds of Olmec communities at San Lorenzo, La Venta and Tres Zapotes. The volcanic stone tradition is even continued in Olmecoid sites in the volcanic highland regions of Guatemala, particularly around Lake Atitlan as another seminal Mesoamerican area (Miller 1986, p. 38) where Late Formative colossal heads, smaller but still carved from volcanic stone, have been found near El Baul and Kaminaljuyu. Not only is this in a volcanic region as well, as mentioned, but even the visibility of volcanic cones here may have in some way influenced site choice or at least ensured that the stone would be recognizable as volcanic to the stoneworkers or those for whom they carved the heads. Thus, stone selection could have been a process of physical characterization and metaphysical requisites met together in volcanic stone.
352
P. HUNT
Conclusion Although many possible stone selection criteria are potentially involved in Olmec culture for the deliberate choice of basaltic stone, it is unlikely that all or even many were operative when but one primary stone material was chosen for monumental sculpture. In conclusion, it is suggested here that the consistent connection between Olmec monumental sculpture and basalt is explicable. It could be based on the deliberate criterion of stone selection for metaphysical reasons and by physical characterization: this mostly dark, vesicular stone was the most compelling reminder of underworld power expressed in volcanism and volcanic mountains, and was thereby appropriated for power in rulership and, perhaps by apotropaic appeasement, in ensuring continuity of Olmec culture. Reference BELL, P. & WRIGHT, D. 1985. Rocks and Minerals. Macmillan Field Guides. Collier, New York, p. 20. BENSON, E. (ed.) 1982. The Olmec and their Neighbors. Dumbarton Oaks, Washington, DC. BERNAL, I. 1969. The Olmec World. University of California, Berkeley. BERNAL-GARCIA, M. E. 1994. Tzatza: the sacred mountain and the Olmec ruler. Seventh Palenque Round Table. Precolumbian Art Research Institute, San Francisco, CA, 113-124. BERRIN, K. & PASZTORY, E. 1993. Teotihuacan: Art from the City of the Gods. Thames and Hudson, New York. BISHOP, R. L., SAYRE, E. V. & VAN ZELST, L. 1984. Characterization of Mesoamerican jade. In: VAN ZELST, L. (ed.) Applications of Science in Examination of Works of Art. Museum of Fine Art, Boston, MA. BLAIR, B. E. 1955. Physical Properties of Mine Rock, Part IV. US Bureau of Mines, Investigation Report 5244. CLARK, G. B. & CANDLE, R. D. 1961. Geologic structure stability and deep protection construction. Air Force Special Weapons Center, Project 1080, Technical Documentary Report AFS WC-61-93. CLEWLOW, C. W. 1974. A Stylistic and Chronological Study of Olmec Monumental Sculpture. University of California Research Facility, Berkeley, 19. COE, M. 1981. San Lorenzo Tenochtitlan. In: BRICKER, V. & SABLOFF, J. (eds) Supplement to the Handbook of Middle American Indians, Vol. 1. University of Texas, Austin. & DIEHL, R. 1980. In the Land of the Olmec, Vols 1 and 2. University of Texas, Austin. COVARRUBIAS, M. 1957. Indian Art of Mexico and Central America. Knopf, New York. DE LA FUENTE, B. 1973. Escultura monumental Olmeca. Institute de Investigationes Esteticas, Universidad Autonoma de Mexico, Mexico City.
FOSHAG, W. F. 1957. Mineralogical Studies on Guatemalan Jade. Smithsonian Miscellaneous Collections, Washington, DC, 25. GAURI, K. L. 1978. The preservation of stone. Scientific American, 238(6), 126-137. GILL, J. B. 1981. Orogenic Andesites. Springer, Berlin. HARLOW, G. E. 1987. Jadeites, albitites and related rocks from the Motagua Fault Zone, Guatemala. Abstracts of the Geological Society of America Annual Meeting. HASENAKA, T. & CARMICHAEL, I. S. E. 1987. The cinder cones of Michoacan-Guanajuato, Central Mexico: petrology and chemistry. Journal of Petrology, 28(2), 241-269. HEIZER, R. 1968. New observations on La Venta. In: BENSON, E. (ed.) Dumbarton Oaks Conference on the Olmec. Washington, DC, 9-40. HUNT, P. N. 1991. Provenance, weathering and technology of selected archaeological basalts and andesites. PhD dissertation, Institute of Archaeology, University College, London. 1993. Teotihuacan Exhibition Greenstones. A report for the Fine Arts Museums of San Francisco. 1994. Maya and Olmec stone weathering contrasts: limestone and basalt contexts. Seventh Palenque Round Table. Precolumbian Art Research Society, San Francisco, CA, 261-267. KUBLER, G. 1962. Art and Architecture of Ancient America, 2nd and 3rd edns. Pelican, Harmondsworth, 68-7Iff. KUNZ, G. F. 1890. Gems and Precious Stones of North America. Scientific Publishing, New York. LE MAITRE, R. W. (ed.) 1989. A Classification of Igneous Rocks and Glossary of Terms. Blackwell Scientific, Oxford. LUHR, J. F. & PRESTEGAARD, K. L. 1989. Caldera formation of Volcan Colima, Mexico, by a large Holocene volcanic debris avalanche. Journal of Volcanology and Geothermal Research, 36(1). MILBRATH, S. 1979. A Study of Olmec Sculptural Chronology. Dumbarton Oaks Studies in Precolumbian Art and Archaeology, 23. MILLER., M. E. 1986. The Art of Mesoamerica: from Olmec to Aztec. Thames and Hudson, London. NUTTALL, Z. 1926. Official Reports of the Towns of Tequizistlan. Tepechpan, Acolman and San Juan Teotihuacan sent by Francisco de Castaneda to His Majesty Philip II, and the Council of the Indies, in 1580. Papers of the Peabody Museum of American Archaeology and Ethnology, 11(2). RANDS, R. L. 1965. Jades from the Maya lowlands. Handbook of Middle American Indians, Vol. 3. University of Texas, Austin. REILLY, F. K. 1994. Enclosed ritual spaces and the watery underworld in Formative Period architecture: new observations on the function of La Venta Complex A. Seventh Palenque Round Table. Precolumbian Art Research Institute, San Francisco, CA, 125-137. SABLOFF, J. 1989. Cities of Ancient Mexico. Thames and Hudson, New York. SHEETS, P. D. 1983. Archaeology and Volcanism in Central America: the Zapotitan Valley of El Salvador. University of Texas, Austin.
OLMEC STONE SCULPTURE STIRLING, M. 1943. Stone Monuments of Southern Mexico. Smithsonian Insitution,Washington, DC, Bureau of American Ethnology, Bulletin, 138. TEDLOCK, D. (transl.) 1985. Popol Vuh. Simon and Schuster, New York. WASHINGTON, H. S. 1922. The jades of Middle America. Proceedings of the National Academy of Science, 8, 319-326.
353
WILLIAMS, H. & HEIZER, R. 1976. Sources of rock used in Olmec monuments. Sources of Stone used in Prehistoric Mesoamerican Sites. Contributions of the University of California Archaeological Research Facility, 1, 1965. WINKLER, E. M. 1975. Stone: Properties, Durability in Man's Environment. Springer, Berlin.
This page intentionally left blank
Seismic and volcanic hazards affecting the vulnerability of the Sana'a area of Yemen RICHARD HUGHES1 & ADRIAN COLLINGS2 1
Earthquake Engineering Field Investigation Team (EEFIT), Institute of Structural Engineers, London SW1X 8BM, UK 2 Ove Amp & Partners, Birmingham B16 8NH, UK
Abstract: The City of Sana'a is a World Heritage Site, renowned for its ancient tall and ornate domestic dwellings. Following a moderate, but locally destructive seismic event near Sana'a (Dhamar) in 1982 (Ms 5.5-6.0), the seismic vulnerability of the city and its hinterland have been reviewed, using the extensive records of the Arab chroniclers, going back over 1500 years. A hazard study indicated that Yemen is a country of moderate seismicity rather than one of low seismicity as previously considered, as records indicated that moderate-sized earthquakes (Ms 4.0-6.0), have occurred regularly over the last 1000-1500 years. It was determined that Sana'a is in a location where a damaging event could statistically occur every 200 years or so. Using a seismic character similar to that of the 1982 Dhamar event (shallow hypocentre, 40s duration, peak ground acceleration 0.3g, and Ms 5.5-6.0), the vulnerability of Sana'a was determined from condition surveys of some 25% of the historical houses, which proved remarkable levels of active structural distress and fabric decay. It was estimated that nearly 68% of the building stock in Sana'a would be severely damaged to totally damaged beyond repair by a similar earthquake, causing a significant loss of life and destruction of a World Heritage Site. Strategies and techniques for reducing the vulnerability of the urban fabric of Sana'a, ranging from simple upgrading within the owners' means to significant engineering measures, sympathetic to the historical and archaeological fabric of the buildings, are considered. Also considered are volcanic hazards, as records indicate that closely related to the seismic activity in Yemen are occasional, generally small-scale, volcanic eruptions. Several eruptions are indicated to have occurred in the Sana'a area over the last 1500 years, and the area is vulnerable to future eruptions. An event in about AD 200 is indicated by archaeological work to have overwhelmed several religious sites 20 km north of Sana'a. The first part of this paper is based on the findings of a preliminary study of earthquake hazard that was undertaken in response to a locally damaging earthquake in 1982 (Ove Arup 1990). The report included a review of volcanic activity and hazard, as it was noted that seismic and volcanic activity are closely related in Yemen. The second part of this paper develops an approach for assessing the structural archaeology and the vulnerability to future earthquakes of the existing building stock in the Sana'a area, following on from the observations made by the preliminary study of earthquake hazard.
Geological background Yemen is located on the western part of the Arabian Shield. The Cenozoic and Quaternary geological history has been influenced by the
development of the adjacent Red Sea and Gulf of Aden basins and the eventual establishment of sea-floor spreading (Coleman 1993). The geological evolution of the Red Sea Basin has taken place in two phases. The earlier phase during Oligocene time involved extension accompanied by the voluminous extrusion of alkali basalts mainly on the Arabian side of the developing Red Sea Basin. After relative quiescence in Miocene time, extension commenced again in Pliocene time, with sea-floor spreading becoming established along the axis of much of the Red Sea Basin and the Gulf of Aden, together with renewed alkali basalt volcanism on the western part of the Arabian Shield (Chiesa et al. 1983; Menzies 1996). The tectonic background of the region is summarized in Fig. 1, which is based on studies by Manelli et al. (1991) and Coleman (1993). It is noted that the establishment of sea-floor spreading in the Red Sea and the Gulf of Aden
From: McGuiRE, W. G., GRIFFITHS, D. R., HANCOCK, P. L. & STEWART, I. S. (eds) The Archaeology of Geological Catastrophes. Geological Society, London, Special Publications, 171, 355-372. 1-86239-062-2/00/ $15.00 © The Geological Society of London 2000.
356
R. HUGHES & A. COLLINGS
Fig. 1. Tectonic map (after Manelli et al. (1991) and Coleman (1993)).
THE SANA'A AREA OF YEMEN does not completely connect through the triple junction with the East African rift system (Coleman 1993). Instead, Quaternary extensional processes appear to be distributed across a broad area of the adjacent western Arabian Shield, and the Afar area of Ethiopia, rather than being focused on a discrete zone of sea-floor spreading (Davidson et al. 1994). Much of the Yemen is affected by this diffuse zone of crustal stretching, which is accompanied by seismicity and volcanicity (Gettings et al. 1986). Impact of volcanic activity The Quaternary volcanic activity across Yemen has generally comprised relatively small-scale monogenic basaltic eruptions from fissures, reflecting dyke emplacement in the upper crust. Some rhyolitic activity has occurred as well. The areas affected are scattered and widespread across the broad zone of crustal stretching (see Fig. 2, which is based on the study by Grolier & Overstreet (1978)). A notable feature of many of the basaltic eruptions is the initial vigorous volatile phase, forming a tuff ring or cinder cone. This is usually followed by a lava flow as the more degassed aspect of the magma pulse emerges, typically breaching the tuff or cinder cone. Numerous examples of this type of activity can be seen just north of Sana'a, in an area known as the Harra of Arhab (Neumann Van Padang 1963). The Arab chroniclers provide a 1500 year record of natural events (Ambraseys et al. 1994). Also proto-historic, but unrecorded volcanic activity can be demonstrated in several areas by the observation of lava flows disrupting sites of neolithic occupation (Camp et al. 1987). A total of 20 volcanic eruptions have affected Yemen (and the adjacent Red Sea) in protohistoric and historic time, and four of these have occurred within 30km of Sana'a (see Table 1). On the basis of these data, the northern part of the greater Sana'a area would be vulnerable to a renewal of volcanicity on the Harra of Arhab just to the north. Although the area affected by the lava flow and tephra fall would probably be relatively localized, the impact on the local population of an immediately adjacent volcanic eruption could be significant, particularly if there were no civil defence plans in place to respond to such an event. The civil defence aspects of volcanic activity adjacent to urban areas have been considered by Scandon et al. (1993) and Lirer et al. (1997) (in the context of the Naples area); more general guidelines have been outlined by UNDRO (1982).
357
It is observed that there is also a significant possibility of a future volcanic eruption in the Dhamar area some 100km south of Sana'a (see Fig. 2) (Plafker et al. 1987). An eruption in the third century AD on the Harra of Arhab just north of Sana'a is indicated to have overwhelmed a Himyarite settlement (Neumann Van Padang 1963). The area was excavated by German archaeologists in the 1920s, with the main finds now held by the National Museum in Sana'a. Inscriptions found by the excavations related to the existence in the area of an important temple to the moon god Talah Riyam dating from the third century BC. However, no remains of the temple were found, and it was concluded that the temple had been destroyed by the lava flow from Djebel Zebib. The adjacent temple of the sun goddess Dhat Bagan was also excavated, and this was found to be burnt, probably as a result of the same adjacent volcanic activity. Many of the stones from this temple can be found in present-day buildings in the area (Hamalainen 1995). Because of the extensional tectonic environment, volcanic and seismic activity in Yemen are closely related (Plafker et al. 1987; Coleman 1993). During the 16th and 17th centuries there was much seismic activity on the Harra of Arhab. Some of the events were in swarms, which culminated in volcanic activity (see Table 1). It is possible that dyke emplacement was taking place, allowing and possibly causing, stress release in the upper crust, with perhaps only a small proportion of the emplaced magma finding an expression at the surface, similar to other rifting situations, e.g. Iceland (Bjornsson et al. 1978).
Dhamar area The 1982 earthquake affected an area just north of Dhamar (see Fig. 1), close to the location of a reported small-scale volcanic eruption in 1937 (Neumann Van Padang 1963). It is possible that the 1937 eruption at Dhamar may have been the result of volatile release following a sub-surface basaltic dyke intrusion. The entry in the Catalogue of Active Volcanoes of the World listing contains few details of the event. Plafker et al. (1987) in their review of the 1982 Yemen earthquake made reference to the 1937 event, but there are no specific data to suggest that it was an explosive eruption of felsic composition. Nevertheless, it is noted that the Quaternary volcanicity of the region, although predominantly basaltic, does involve some more
358
R. HUGHES & A. COLLINGS
Fig. 2. Volcanic activity in Yemen and surrounding areas.
differentiated products (Neumann Van Padang 1963), so it is conceivable that the 1937 event involved an evolved magma. The epicentre of the 1982 earthquake was only 30km northeast of the location of the 1937
eruption. It is not surprising, therefore, that following the earthquake the local population feared there would be a volcanic eruption. Indeed, some of the features of the 1982 earthquake indicated that it was of volcano-tectonic
THE SANA'A AREA OF YEMEN Table 1. Summary of recorded volcanic eruption affecting
359
Yemen
Date
Name
Location
Details
Unknown 1200BC
Djebel el Marha Unknown Djebel Zebib Kaulet Hattab Dubbi (Ethiopia) Es Sawad Dubbi (Ethiopia) Zubayr Jebel Zubayr Jebel Zubayr Jebel Unknown (Ethiopia?) Djebel Zebib? Unnamed Zuqar Island Djebel en Nar? Teyr Djebel Zubayr Jebel Teyr Djebel Teyr Djebel Zubayr Jebel Dubbi (Ethiopia) Dubbi (Ethiopia) Teyr Djebel Teyr Djebel Zubayr Jebel Harras of Dhamar
15°28N, 044°22E 15°43N, 044°78E 15°60N, 044° 12E 15°63N, 044°08E 13°58N, 041°81E 13°58N, 046°12E 13°58N, 041°81E 15°08N, 042° 17E 15°08N, 042° 17E 15°08N, 042° 17E 15°60N, 044° 10E 15°70N, 043°80E 14°OON, 042°70E 13°33N, 034°73E 15°70N, 041°74E 15°08N, 042° 17E 15°70N, 041°74E 15°70N, 041°74E 15°08N, 042° 17E 13°58N, 041°81E 13°58N, 041°81E 15°70N, 041°74E 15°70N, 041°74E 15°08N, 042° 17E 14°50N, 044°50E
75 km E of Sana'a 10km S of Sana'a 25 km NNW of Sana' a 30 km NNW of Sana' a N of Gulf of Aden Red Sea Red Sea Red Sea
AD 200 AD 500
1203 1253 1400 1433 1434 1435 1444 1583 1637 1679 1788-1789 1796 1824 1832 1845 1846 1861 1863 1863 1884 1914 1937
25 km NNW of Sana'a 60 km NW of Sana'a Red Sea SW Yemen Red Sea Red Sea Red Sea Red Sea Red Sea Red Sea Red Sea Red Sea 100km SSE of Sana'a
From Neumann Van Padang (1963), Simkin el al (1981), Ambraseys & Melville (1983) and Ambraseys et al. (1994).
origin (Coleman 1993). This relates to a prolonged period of aftershocks, and the observation of extensional ground cracks, some of which exhibited continued dilation for several weeks after the main shock. It was concluded by George Plafker and his coworkers, who studied the 1982 event (Plafker et al. 1987), that it could have been caused by the upward movement of magma below the Earth's surface. Seismic hazard assessment A preliminary study of earthquake hazard in Yemen (Ove Arup 1990) was undertaken after the 1982 event to consider the seismicity of the region using the records of the Arab chroniclers, together with recent instrument data. This information indicated that moderate-sized earthquakes with surface-wave magnitude 4.0-6.0 have occurred regularly over the past thousand years, as noted by Ambraseys & Melville (1983). The preliminary study noted that the instrument data alone are insufficient to adequately define a return frequency. The indicated return frequency (see Fig. 3) is for the region west of
45°E, which includes the capital city and World Heritage Site, Sana'a. On the basis of these return frequency data, a damaging event could statistically occur within the area considered every 200 years. It was concluded by the preliminary study of earthquake hazard in North Yemen that the region west of 45°E is an area of moderate seismicity, rather than an area of low seismicity as had previously been assessed (Ove Arup 1990). This has implications for the Sana'a site, and a consideration of the structural archaeology of the vernacular architecture is outlined below. Structural archaeology related to the 1982 North Yemen earthquake
Introduction - earthquake damage surveys During the last two decades various UK universities and the Earthquake Engineering Field Investigation Team (EEFIT) have systematically evaluated earthquake damage in many countries. One must thank Nicholas Ambraseys, Ian Davis and Robin Spence for the vision behind
360
R. HUGHES & A. COLLINGS
Fig. 3. Earthquake magnitude-frequency recurrence plot.
such fieldwork. As Table 2 shows, the evaluations have had a well-appreciated humanitarian aim. 'Earthquakes and Archaeology' is just one aspect, which, as this conference demonstrates, has an increasing value. In many EEFIT surveys non-engineered and historical buildings have been inspected, from what can be classed as a 'structural archaeological' approach. This is because such buildings are old and full of interesting features, and have often survived many previous disasters. In addition, the building stock has an overall value as an 'upstanding' archaeological resource. In financial terms, the buildings play a significant role in the economy, especially in a tourismdriven market. Historical buildings help to define a culture and give it coherency. For these reasons, UNESCO regards restoration of earthquake-damaged historic buildings as a key element in recovery. Surveys, in particular those of the research-oriented type carried out by EEFIT, make a significant contribution to archaeological interests. In such surveys,
researchers see and thoroughly document parts of a building that are normally not visible, as in an archaeological excavation, and account for how the resource is formed and survives Hughes I990a&b).
The 1982 Yemen earthquake and its effects on traditional buildings A significantly damaging earthquake occurred on 13 December 1982 in the Dhamar Province of the North Yemen (Coburn & Hughes 1983). The epicentre was about 150 km south of Sana'a, the capital city. The epicentral intensity was estimated to be 7S IX MSK and the surface-wave magnitude was calculated to be Ms 5.5-6.0. Extensive damage resulted, over an area of some 6000 km2, with an estimated 'total' destruction of 40 villages (see Fig. 4). Approximately 25000 houses were destroyed and 18 000 badly damaged. More than 300000 people were made homeless and 4000 killed. In global terms this was not a
THE SANA'A AREA OF YEMEN Table 2. Reasons for earthquake surveys Reasons for a building survey by an earthquake management team (1) To define economic losses to building stock and related infrastructure (2) To estimate the need for local, national and international reconstruction aid (3) To locate buildings that are safe, can be reinstated or have to be demolished (4) To understand weaknesses in construction and performance so non-damaged existing buildings of the same type can be appropriately strengthened (5) To document historical buildings and damage to them and their contents (6) To assess the effectiveness of emergency procedures and disaster management (7) To design new earthquake-resisting buildings that utilize traditional materials and construction methods Typical reasons for an EEFIT survey (1) To determine pre-earthquake building fabric and structural condition (2) To determine the damage from the main shock and successive aftershocks (3) To define building failure mechanisms and adverse siting conditions (4) To determine damage resulting from rescue and demolition activities (5) To quantify future demolition, shoring and repair (6) To aid the understanding of mortality and morbidity of the building occupants (7) To aid the introduction of better design and construction processes of new buildings (8) To help detect developing stresses and strains that may result in future damage (9) To protect historical buildings from further damage and to give emergency protection
large magnitude earthquake, and if located in California or Japan, with their different housing stocks, would have probably resulted in only low damage and injury levels. Most structures were built with stone rubble and mud mortar, which was put together well in urban centres and very roughly in remote mountain villages. Ashlar-faced structures were more frequent in the larger settlements but most village chiefs had such grander edifices. In the main Dhamar wadi many villages were built in soil (adobe or pise). Fired brick and concreteframed houses and shops were relatively rare and very new; such buildings were restricted to sites near the main road traversing south through the province. The high damage levels were attributed, by the authors, to many factors including the following: (1) The shallow earthquake hypocentre with high-frequency ground motion matching
361
that of the predominantly masonry structures. There was then a considerable period of large-magnitude aftershocks. These were personally experienced and one can vouch jfor their size, cause of secondary damage, and widespread alarm! (2) Local ground amplification as a result of rock structure and topography. Damage was particularly severe in villages founded in deep colluvial soils, on rock-capped plateaux, on cliff edges, embedded in steepsided scree slopes, and along mountain crests. The complexity of damage distribution should be noted from Fig. 4 as it has a major bearing on the difficulty of interpreting epicentral locations from limited archaeological sampling of historical events. (3) Poor condition of the structures, often with superstructure overloading of shallow foundations, use of soil mortars, and many phases of alterations and facades. Walls were found to have loose cores, with external rubble skins replaced by unbonded ashlar facings. (4) Pre-existing structural defects and local failures; these were often the result of phased abandonment and reoccupation. There was a noticeable residual occupancy by old people, who were unable to maintain the fabric; many house owners reporting damage were found to be including defects that were clearly very old (cracks were found to be stuffed with old bits of paper and cloth, etc.). On the basis of systematic data collection, the most significant damage characteristics were determined (see Fig. 5a & b and Table 3). One of the main findings of the survey was that the collapse of roofs and floors occurred after failure of only one load-bearing wall, and by this time several non-load-bearing walls had normally failed. Damage was also precipitated by many small details of construction including the following: (1) foundations set in soil with variable engineering qualities (half on rock outcrops and half on soil, or partially on concretionary bands in the plateaux soils); (2) abutting walls without connection keying; (3) short lintel bear-ings in walls to either side of openings; (4) lack of wall plates to seat beam ends; (5) rows of windows forming horizontal zones of weakness. Within the earthquake-affected area wood was a major building component, predominantly for roof and ceiling beams. The wood, in a most artistic way, is usually left in its contorted tree form and then plastered to give an abstract
Fig. 4. Damage distribution of the 1982 Dhamar earthquake.
THE SANA'A AREA OF YEMEN
363
Fig. 5a. Photograph illustrating severe damage in Dhamar following the 1982 earthquake.
Fig. 5b. Photograph illustrating moderate damage in Dhamar following the 1982 earthquake.
morphology. Straight rectangular sectioned beams are rare. The use of well-engineered timber for earthquake-resisting wall ties (cators in Pakistan and hatils in Turkey) is evidenced only in a limited number of Dhamar town structures, and is associated with the large prestigious and historical structures. Wooden ties are of a small dimension, no more than 40 mm x 80 mm cross-section. The ties are at only one or two levels, marking storey heights, and although appearing to form complete ring beams are usually only on the front fa9ade. They are not repeated on the inner face nor tied
through the wall thickness. It is possible that in such cases the timbers are associated with a later phase of external facade replacement, along with coloured ashlar stones now so typical of 'modern' Yemen. Occasionally, timbers are found in the walls of village structures but only in short lengths, in the following situations: (1) across cracks, as a vernacular tie to stitch the crack together and reduce further crack movement; here, anchoring the timber is difficult and one suspects that the cause of cracking is hardly ever identified and rectified; (2) at locations where arch and column loads are brought down onto a wall, and inserted below to help distribute the load and resist punching effects; (3) in mosques where there are broad areas of arches and columns, and here the wood is used beneath the column bases as cushion a and for tensile ties across the springing of the arches; these traditions are seen throughout the Islamic world. It is possible that the use of wood, because of its tensile capabilities, has in Yemen been copied from Turkish traditions going back to at least the seventh millennium BC (it should be noted that the country was once part of the Ottoman Empire). It is also possible that a greater use of
Table 3. Dhamar earthquake damage Damage type (see % seen in Figs 6 and 8 for further highly details) damaged structures
% seen in light to moderately damaged structures
Corner wedge failure 46 Wall to wall separation 62 Skin splitting 58 Mid wall 32 Internal wall and floor 100 separation
35 71 21 14 100
364
R. HUGHES & A. COLLINGS
timber wall strapping was not possible because of its scarcity in a desert environment and because the available timber is too distorted for appropriate use. Much of the timber for beams has traditionally been brought in from East Africa. Wood is an expensive commodity and still the main source of fuel. Given short memories of disasters, structural tying for earthquake purposes is not perceived as essential but rather is part of facade decoration: you can do without a tie but not a floor or ceiling.
However, one success of the structural archaeological approach to damage assessment is that it showed that the ties significantly contributed to lessening the earthquake impact in the town of Dhamar. The continued use of the technique has therefore been encouraged, with some technical modifications, and is taught in the Yemen construction training and rebuilding programmes. Since 1981 the timber wall strengthening technique has been utilized in several other earthquake-prone countries.
Table 4. The problems and causes of major building and decay defects recorded in Salvitelle, March 1981 (1) Cracking of the external wall The use of random rubble stonework; poorly staggered joint system; decay of the stone and soft mortar; salt efflorescing at wall base; rain penetration into the wall fabric; missing render; house built on unstable slope; previously cracked wall, often badly repaired, often infilled with loose decayed stone; floor joists at too close centres causing a line of weakness; collapsing joists acting as levers so forcing off portions of the wall; repair of the wall with too good a new method so creating a 'hard spot'; rebuilding of ground-floor wall in modern materials or construction of additional floors; collapse of stone and foundations into rock-cut cellars; vertical load placed on metal tie rods (2) Bowing of external load-bearing walls Wall too thin for its height; differential expansion of rotting mortar; rotting floor joists; separation between inside and outside wall faces and internal loose rubble; lack of keying between cross walls, party walls and main external walls (3) Cracking and collapse of walls above windows and doors Rotten wooden lintels; lintels not the full width of the wall; ends of the lintels poorly embedded into the wall; horizontal or shallow arched voussoirs; door and window jambs not tied back into wall fabric; new windows constructed through older walls (4) Failure of the corners of the building Inadequate corner quoins; foundation failure; thermal expansion of long walls; lack of ring tie at first-floor level; inserted metal tie rods in one direction only or at one level only (5) Zone of weakness near chimney Chimney flue built inside the wall thickness; chimney structure not built on adequate foundations; differential thermal expansion-contraction; absorption of corrosive gases and acids produced by fires (6) Cracking of the inside wall face often seen as settlement Horizontal slot cut into wall face for the insertion of a new floor structure; the insertion of service pipes and cables into the wall face; roof and floor bearing on inside face of wall where it is rubble filled (7) Movement of internal walls The use of lightweight walls such as hollow box tiles for load bearing; lack of foundations for load-bearing walls; use of heavy shelving (8) Falling masonry and external fixtures Decayed stonework and mortar; wall leaning; loose render; weakly connected architectural mouldings; pantiles not permanently fixed to roof batons; large stones placed on the edge of roof to stop high winds from lifting the pantiles; poorly maintained gutters and down-pipes; roof joists not fixed to wall plate (9) Collapsing floor joists and floor Inadequate depth of embedment of joist into the wall; rotting end of the joist and wall plate; no wall plate; replacement floor joist bearing onto vertical mortar joint; modern concrete or tiled floor increasing floor loading; lack of bond of floor into wall; bulging walls removing joist bearing; heavy oak planking; heavy lath and mortar ceiling inadequately fixed to joists; too heavy loads placed on the floors (10) Collapsing roof Absence of truss system - no horizontal member in tension; low angle of pitch so small wall displacement allows the ridge to collapse downwards; use of unsquared timbers often still with bark on them; inadequate joint system at ridge; rotting or absent wall plate; lack of tie to wall plate; weight of pantiles (and patched areas) too much for the supporting structure; no ring beaming at roof level to stop walls from being displaced outwards
THE SANA'A AREA OF YEMEN
Comparative damage mechanisms from the 1980 Italian earthquake The Dhamar earthquake damage survey focused on a few key failure mechanisms: those that had a fundamental role in structural performance, and that could be assessed within the time frame of the fieldwork and without disruption to or interference in inhabitants' pressing needs. The data should be compared with those obtained after the November 1980 Italian ('Basilicata') disaster, where the approach aimed at a general examination, to develop a building failure model. Table 4 shows how much information can be gleaned from damaged structures, suggesting a future application when interpreting excavated archaeological remains. From the above observations it was possible to define a model to show progressive failure of a structure (see Figs 6 and 8): (1) development of diagonal cracking and the displacement of the wall corner; (2) diagonal cracking and the displacement of the end wall; (3) development of vertical cracks; (4) dropping of poorly supported
365
wall above windows and doors; (5) falling chimney; (6) collapse of a portion of the roof; (7) collapse of a weak area of wall often associated with the position of a floor joist; (8) collapse of a central section of wall, for example, around windows and above large cracks. The value of earthquake damage surveys to archaeology Earthquakes, such as those of Yemen, Italy and Turkey, where the author has made observations, offer the opportunity to examine the way buildings become archaeological sites; but in a dramatically speeded-up way compared with normal collapse, decay and burial processes. Also, parts of a structure not often seen are available for inspection. Immediately after an earthquake there is still a substantial vertical dimension that is absent in the buried domain. Clues are therefore provided that can aid the 'digging' archaeologist in understanding the earthquake process and the products discovered.
Fig. 6. Deformation mechanisms that result in total collapse. Structure prior to earthquake: (1) wall bursting into multiple units, mostly outward, occasionally inward. (2) Collapse as a single unit, mostly outward, occasionally inward. (3) Downward collapse of roof. Collapse of floor joists normally from one end. (4) Foundation failure. (5) Wall falls as two units, down and outwards. (6) Wall splits and mostly falls or slumps outwards. (7) Top of wall pushed outward by falling roof. (8) Top of wall pushed outward by falling joist. (9) Decayed material falls out of base of wall. (10) Displaced block of masonry falls vertically. (11) Disintegration into a pile of rubble. (12) Collapsing floor joist, which, by a pivoting action, forces off a portion of wall above the socket. (13) Outward collapse as multiple small units but leaving a stub in situ wall below.
366
R. HUGHES & A. COLLINGS
Fig. 7. The historical buildings of Sana'a.
Indeed, excavation methods can be better designed to respond to anticipated features. It may then be possible to separate out real destruction components from those resulting from knock-on effects or straightforward site formation processes. The problems of identifying real earthquake damage have been addressed by a few researchers (Stiros & Jones 1996) and a discussion of their observations is presented later on in this paper.
The character and product of building collapse is of fundamental importance to archaeological concerns, as it is these elements that are generally preserved below ground. Some personal observations of earthquake damage sites are as follows. (1) In lightly to moderately damaged areas, collapse debris is reused very quickly and route-ways are re-established. Buildings
Fig. 8. Some major mechanisms resulting in partial collapse. (1) Development of diagonal cracking and the displacement of the wall corner. (2) Diagonal cracking and the displacement of the end wall. (3) Development of vertical cracks. (4) Dropping of poorly supported wall above window and doors. (5) Falling chimney. (6) Collapse of a portion of the roof. (7) Collapse of a weak area of wall often associated with the position of a floor joist. (8) Collapse of a central section of wall, for example, around windows and above large cracks.
THE SANA'A AREA OF YEMEN
(2)
(3)
(4)
(5)
(6)
(7)
(8)
(9)
and areas that are abandoned become sites of informal use. Here it is difficult to unravel the earthquake scenario, but such uses and new wastes become archaeologically informative about the community under new pressures. There is very little evidence of foundation failure resulting from activation of preexisting defects or related to new soil conditions, for example, landslides and liquefaction. It is very rare to recover information that shows fabric crushing in foundations resulting from out-of-plane swaying and one-sided overloading. In open environments, there is a greater spread of debris from non-load-bearing elements: there is no vertical load to provide restraints. Stone walls collapse more as piles of rubble than as whole elements; rubble and ashlar walls shake apart as granular masses. In open environments the debris tends to collapse to a natural angle of repose. Where 'bursting' occurs building elements can be thrown considerable distances, and here it is difficult to model the original wall morphology. In confined environments rubble piles often form up to the base of failure and may be in a very unstable state. Here, most valuably, the debris protects the lower intact element. Wall toppling with surviving intact structural stratigraphy occurs with a slow failure. Such a collapse is usually initiated by a distinctive horizontal line of weakness; for example, long-term undercutting on mud brick walls, and where a wall has a significant ductility and a long resonant period, can result in this type of toppling. The structural debris is mixed with mortar and renders, and these move down through the open cavities. This further disseminates the materials from the source position. The materials of internal wall tend to be of a finer 'grain' (more render or smaller stones). Core structures tend to stand to a higher level than externally. Floor failure tends to be from one side and lateral sliding of debris occurs. Rescue of victims, possessions and animals significantly disturbs the pattern of collapse. Wood is recovered for new use. Stone is quickly sorted out, as anything 'worked' has a market value even if it has to be transported. Sites tend to be reused quickly, as foundations and stubs of walls are a resource and save on new works. The style of
367
superstructure varies in response to fashion. Reuse of foundation and therefore building footprints tends to result in building conservatism. (10) Given social and economic constraints after an earthquake, and the pressure to reuse materials on the original site plan, there has been traditionally little innovation in building technology. Engineering improvements tend to relate to the more public and prestigious buildings, and here imported materials and skills are most applicable. In the ordinary stock the same construction mistakes occur, resulting in a vulnerable situation soon matching the previous one. There is very little to see, in Darwinistic terms, of applying little tricks of the trade learnt from surviving buildings. For this reason it is possible to extrapolate from standing 'archaeological structures' to the buried archaeological resource. (11) There are cases where site burial occurs rapidly, and this results in a potentially good preservation of the vertical dimension of structures: (1) where structures are substantially terraced back into the hillside (Yemen 1982); where structures are founded deep into bedrock (Noto, Sicily 1695); where mud and rock flows are triggered in mountainous regions (frequently in the Hindu Kush); this is not always the case, and stone debris flows can be destructive (Yungay, Peru); in areas where topographic changes have changed drainage in flood plains (Moenjodaro(?)); where structures that have earth walls and earth roofs have effectively buried structures in a soil formation and the material has no reuse value (Turkey 1985); at coastal locations where liquefaction and 'block' submergence occurs (Kingston, Jamaica). This formation of 'new archaeological sites' is now dramatically reducing. Damage to an ordinary building stock now results in the abandonment of traditional building practices in favour of imposed styles, new materials and new engineering. In this 'new age' process the affected communities have little say, and the results are often socially and economically untenable. The Yemen earthquake resulted in a substantial amount of international aid. Both technical help and substantial financial resources were provided. An example of a good project was the Yemen builder training programme, which encouraged the retention of local traditions with
368
R. HUGHES & A. COLLINGS
appropriate improvement of existing building technologies. Perhaps a poor contribution was the building of 'western' blocks of flats, which were badly built, and have little relationship to local society and lifestyles or occupations. It is also clear that the fabric of destroyed villages is bulldozed away to allow for new forms of development or to remove its visual evidence. In so doing, future archaeological resources are substantially lost without documentation. Earthquake damage reported by other researchers To compare the above observations with how earthquake damage has been previously assessed and evaluated, a review has been undertaken of papers at a key conference (Stiros & Jones 1996). Attention is drawn to two of the presentations (Guidoboni 1996; Stiros 1996) that very much accord with this paper. Some 19 papers were presented that contain reasonably site-specific information. Generally, very few data were presented to support interpretations that the encountered building features resulted from earthquakes rather than from general decay and collapse. There appears to be a jumping to conclusions, as the authors know that the eastern Mediterranean is seismically active. Strong cases for earthquake damage were expressed in six papers that referred to skeletons under rubble. There were four cases of crushed artefacts and three of wall lateral displacements, where an earthquake cause can also be supported. Only one paper identified a fault rupture through man-made features. Five papers identified coastal readjustment at archaeological sites and three cases of column drum rotation were discussed as evidence of destructive events. Earthquakes are identified on the basis of wall collapses in seven papers, large-scale destruction in one paper, and large-scale rebuilding in another. Generally, the excavators had difficulties in resource examination, as a result of later rebuilding processes and robbing of building materials. However, it would seem that the excavation processes were not geared to the examination of collapse debris for the identification of failure mechanisms; such a procedure is promoted by this author. Rubble has been a material to be removed as quickly as possible to discover the occupation horizons. At the conference other features said to be evidence of earthquakes were foundation cracking (one example noted), wall bowing and tilting (two examples), and burnt roofs (four examples). Such features could as easily be related
to geotechnical siting factors, accidents and responses to economic considerations. One of the most pertinent presentations at the conference was by Stiros (Stiros 1996). The main thrust of that study was to describe how structural behaviour relates to form, orientation, materials and condition. These factors dictate the structural coherency, resonant frequency and modes of dynamic failure. Sequences of failure were noted in relation to main shocks and aftershocks, with damage triggered at weak points and where there have been accumulative effects (resulting from earlier earthquakes and other causes of structural distress). This paper, a somewhat critical appraisal of the other papers, noted the difficulties of identifying earthquake damage in the archaeological domain. This is due to dating uncertainties, the quality of archaeological observations, and similar modes of collapse being hard to differentiate. To aid future work, deformation mechanisms of arches were discussed along with earthquake performance characteristics of masonry walls, arches, vaults, domes and long walls. The Sana'a earthquake study One activity resulting from the Dhamar earthquake was a 1990 UNDP-UNESCO-sponsored hazard-vulnerability study for the World Heritage Site of Sana'a, the medieval capital city of former North Yemen (Hughes 19900 &b). The Arup hazard study has been described above. The focus of the vulnerability study has been related to defining global structural conditions of the magnificent tower houses and what would happen to them in a future earthquake. Although it was found that many types of decay were specifically related to the unique form and history of a structure, there were some elements of structural distress affecting the building stock as a whole (see Table 5). As a result of the distress, many buildings have been and still are being abandoned. Active Table 5. Existing feature defects* Feature
Slight problemL
Moderal;e Severe problem problem
Building tilt Fa9ade lean Fa9ade bulge Corner failure Mid-wall failure General fabric decay
1120 437 1792 530 559 2536
829 481 1708 530 488 1006
Numbers of buildings affected.
536 202 769 32 143 483
THE SANA'A AREA OF YEMEN tilting of the tall structures is not rare but common! Here, the study was able to specifically account for the cause: the over-watering of the traditional bustans, the market gardens spread throughout the various districts of the medieval city. The water is causing the ground to soften and this process is expanding beyond the boundaries of the bustans and below the surrounding buildings. The result is that the soils have a variable bearing capacity and the foundations settle differentially. As the buildings then tilt, the other defects are induced. What is remarkable is that the defects are similar to those present in Dhamar before the earthquake and then induced by the ground motions there. Clearly, a dynamic motion is capable of generating similar failure mechanisms independent of the rate of activity: the earthquake produces rapid effects whereas structural readjustment to tilting produces slow ones; the ultimate stresses and stress paths are much the same. External timber ring ties are even rarer in Sana'a than in Dhamar, and were noted on only a few buildings. This may be a reflection of the considerable age of most structures. Many ties may have been replaced as they may not have appeared to serve a functional purpose, because there has never been a perceived earthquake hazard. Given the enormous number of buildings with structural problems and the limited number with timber ties, it has not been possible to analyse the benefits or detrimental effects the ties may have had. Other inputs to the UNESCO campaign have been related to conservation of a series of old important buildings, through bilateral aid ventures. UNESCO started by conserving two structures, as demonstrations. Later there was a refocus to create a 'network' of structures that could be conserved and reused for interrelated heritage and commercial themes. However, the attention given to some 40 structures has left some 6960 others outside the main thrust of the campaign. It is to be noted that the setting, geology and tectonic history of Sana'a are very similar to those of the Dhamar region. The con-
369
siderable number of structures actively deforming, most with significant levels of fabric decay and with many pressures for modernization, also mirrors the pre-earthquake character of Dhamar. In Sana'a, perhaps because of its importance and wealth, there is more evidence of increasing earthquake problems. The study by Amp was able to show that Yemen was a country of moderate seismicity, rather than one of low seismicity as previously determined (Ove Arup 1990). More than 30 historical events, resulting in reported damage, were defined compared with six generally noted by the inhabitants before the study started. Also, it was determined that Sana'a was in a location where a damaging event could statistically occur every 200 years. It would be anticipated, given the geological structures of this part of Yemen, that an earthquake here would possibly have a similar seismic character to the Dhamar one of 1982: shallow hypocentre, with duration of more than half a minute, peak ground acceleration of 0.3g, and magnitude Ms 5.5-6.0. The vulnerability modelling was undertaken utilizing the damage data from the Dhamar earthquake and from the engineering structural studies of Sana'a. The aim was to predict likely damage from an earthquake similar to that at Dhamar, and which therefore could be considered within the UNESCO framework for international conservation support. The design event for the study was of Ms 5.2 with an epicentre 10 km away from Sana'a. The study excluded for a consideration of damage the modern suburbs and outlying villages. The modelling results are as tabulated in Table 6. Table 6 predicts substantial damage for what would be a relatively small event. However, it must be noted that this is superimposed on a pre-earthquake condition with extremely high levels of structural distress. The conclusion is that a small magnitude shallow event not too far from Sana'a would be catastrophic for the World Heritage Site. It would also have a significant impact on the national economy, as Sana'a is now far larger than the medieval city and most national administrative functions are
Table 6. Whole building damage
Pre-earthquake* Assumed maximum attenuationf Assumed minimum attenuationf
No damage
Slight damage
Moderate damage
Severe damage
Collapsed
770* 280 0
4760 3850 1120
1330 840 1120
70 1120 1120
70 910 3640
* Numbers of affected buildings based on extrapolation from Dhamar damage data. t Based on the attenuation-damage relationship in the Dhamar earthquake.
370
R. HUGHES & A. COLLINGS
now based there. In an earthquake disaster the protection and rehabilitation of historical structures would undoubtedly be a low priority. Clearly, there is a justification for the whole building stock of old Sana'a being upgraded against earthquakes and also for improving their ordinary working condition. The proposed strategy for this was developed on the understanding that it is difficult to persuade owners, occupiers, the local authority and international aid agencies to insist on earthquake mitigation (after all, the earthquake may not come for 500 years, and the 'unbelievers' say 'if at all'). On the contrary, upgrading the structures for social reasons is plausible given the status of Sana'a as a capital city and a World Heritage Site. The study proposed the following levels of improvements: (1) owner: at no cost with better use of local materials; (2) owner or craftsmen: low cost, use of higher-quality local material; (3) owner or group construction: moderate cost with purchased building materials; (4) contractor or specialist construction: high cost with special materials and equipment, and with specialist inputs (e.g. by historical building conservators). Within each option a series of actions were defined, in response to the type of details noted in Table 4. Options (1) and (2) were the most commonly applied in the past. The interventions allow for a continued evolution of local traditions and a local way to keep, in an unregulated way, structural archaeological features. They allow for repair and replacement of decayed timber elements such as ring ties and roof beams and hence encourage the two sets of elements to be tied together. Options (1), (2) and (3) allow for personal choice and commitment, and matching limited financial resources. The application of any single action from any of the options results in reduced vulnerability, but spread over time with a progressive and affordable improvement. This longterm expectation clearly has to be planned for. Option (3) presents the greatest risk to indigenous architectural traditions, as alien materials and techniques are imported and not sensitively used by building contractors. The use of chemical wood preservatives is to be encouraged and because of the widespread application could be affordable by bulk buying by the city administration. The wood preservatives could even be sold on in a subsidized way. Options (3) and (4) allow for the introduction of strengthening elements. With the addition of further timber ties and cross-crack wooden splices the local tradition continues. The effects are charming and the conservation skill base is promoted.
Option (4) is basically that being applied to the conservation of UNESCO-defined 'very important' historical monuments. Here, one expectation is to maintain the authenticity 'as found'. It was considered critical that option (4) should only be undertaken after a sophisticated analysis, as the introduction of new materials and structural elements can change the quality or value of the heritage resource. The new element can also change the level of building stiffness, perhaps detrimentally in terms of earthquake vulnerability. (Such an intervention strategy has been undertaken on one project, and has been criticized as an 'overkill' approach, as significant amounts of historical fabric were removed. Reinforced concrete ring beams were introduced at many levels and walls were internally stitched with steel rods glued in place.) Such vastly expensive techniques are not a sustainable intervention for the whole of Sana'a. Furthermore, they are not reversible and dramatically reduce ductility. Conclusions The study of recent earthquake-damaged buildings permits an improved understanding of archaeological remains, for structural remnants that have slowly decayed or have been purposely pulled down, and specifically for remains thought to result from a long-forgotten earthquake disaster. In summary, earthquake damage is very complex, as a result of the following factors: building- and site-specific factors; direct and indirect knock-on effects, e.g. falling of one masonry element onto another; secondary effects, e.g. artefact recovery, rescue, or material recycling; demolition for safety; site clearance for personal gain (one of many reasons for artificially increasing damage statistics!). Many characteristics of a building and building collapse are not appreciated from archaeological excavation processes, which focus on below-ground details where only foundations, stubs of wall and a base of disturbed rubble spread survive. It will generally be difficult to recognize the presence in the archaeological domain of superstructure details, timber horizontal ties and windows, for example. Archaeological excavations tend to pass through masses of seemingly random rubble very quickly and with a coarse removal technique. Information in the rubble, when seen in the light of earthquake collapse mechanisms, is potentially revealing about superstructures. On the other hand, foundations
THE SANA'A AREA OF YEMEN found in excavations after an earthquake may show evidence of decay, and perhaps even a fault rupture. Understanding an earthquake from a single archaeological excavation will often be of limited value, but may help with dating the event and showing how a specific structure collapsed. Area assessment is required for understanding seismological characteristics: magnitude, intensity distribution, and location of epicentre. The time of day and the season in which an earthquake occurs can also have a significant effect on the scale and character of the disaster, but which it is hard to take into account later on. For example, most fires occur if the earthquake coincides with cooking time, and fewer people are killed if men and women are working in the fields and children are playing in open yards. Artefact and skeletal assemblages may also change with the season and time of day at which the earthquake struck. Rapid 'knock-on' natural burial processes commonly result in surface geomorphological features, and from these it can be possible to predict archaeological sites. Such features can be evident, along with disrupted agricultural land patterns, for centuries afterwards. People in Sana'a, as elsewhere, live in buildings where there can be many defects, and these are mended only when serious; a somewhat fatalistic attitude. Normally, features of distress still allow the building to be serviceable. Where the damage results from an earthquake this may become a location of slow decay or a trigger point for damage in a future earthquake. Where there are pre-existing problems in a building stock the character of destruction will be always hard to attribute to an earthquake and identify in an archaeological excavation. For example, in Sana'a the problems resulting from building and foundation soil softening (which appears to be of epidemic scale throughout the Middle East) would probably not be identified. After abandonment the soils would dry out and differential foundation settling would not be easily seen. In other words, some factors may continue to change after a disaster and are lost without excavations throughout the region. Clearly, the best archaeological 'potential' will remain with sites where there is immediate and complete abandonment (and therefore no material recycling), and rapid burial. The sudden burial of internal features and artefacts can enhance archaeological data recovery. Some items are left that would be otherwise removed upon buildings being abandoned, and others that if left exposed would decay and scatter. It is the rich assemblage of artefacts that is per-
371
haps the best test for the identification of an earthquake, as illustrated in several of the conference papers. However, it is to be noted that multi-period occupation sites also offer an archaeological potential to identify reoccurrence patterns. References AMBRASEYS, N. N. & MELVILLE, C. P. 1983. Seismicity of Yemen. Nature, 303(5915), 321-323. , & ADAMS, R. P. 1994. The Seismicity of Egypt, Arabia and the Red Sea. Cambridge University Press, Cambridge. BJORNSSON, A., JOHNSEN, G., SIGURDSSON, S., THORBERGSSON, G. & TRYGGVASON, E. 1978. Rifting of the plate boundary in North Iceland 1975-1978. Nordic Volcanological Institute, Reykjavik, Iceland, Report 7807. CAMP, V. E., HOOPER, P. R., ROOBOL, M. J. & WHITE, D. L. 1987. The Madinah eruption, Saudi Arabia: magma mixing and simultaneous extrusion of three basaltic chemical types. Bulletin of Volcanology, 49, 489-508. CHIESA, S., LA VOLPE, L., LIRER, L. & ORSI, G. 1983. Geology of the Dhamar-Rada volcanic field, Yemen Arab Republic. Neues Jahrbuch fur Mineralogie, Geologic und Palaeontologie, Monatshefte, (8), 481-494. COBURN, A. & HUGHES, R. 1983. Dhamar Province earthquake - 13th December 1982. Preliminary Report to the Central Planning Office Joint Relief Committee. Martin Centre for Architectural Research, Cambridge University. COLEMAN, R. G. 1993. Geologic Evolution of the Red Sea. Clarendon Press, Oxford. DAVIDSON, L, AL-KADASI, M., AL-KHIRBASH, S. et al. 1994. Geological evolution of the SE margin of the Red Sea. Geological Society of America Bulletin, 106, 1474-1493. GETTINGS, M. E., BLANK, H. R., MOONEY, W. D. & HEALY, J. H. 1986. Crustal structure of south western Saudi Arabia. Journal of Geophysical Research, 91, 6491-6512. GROLIER, M. J. & OVERSTREET, W. C. 1978. Geologic map of Yemen Arab Republic (Sana'a). US Geological Survey Miscellaneous Investigation Series, Map I-1143-B, scale 1: 500000. GUIDOBONI, E. 1996. Archaeology and historical seismology: the need for collaboration in the Mediterranean area. In: STIROS, S. & JONES, R. E. (eds) Archaeoseismology. IGME and British School at Athens, Fitch Laboratory Occasional Paper, 7, 7-13. HAMALAINEN, P. 1995. Yemen, Lonely Planet Travel Survival Kit. Lonely Planet, London. HUGHES, R. 1990^. North Yemen. Preliminary Study of Earthquake Risks to Old Sana'a and Strategies for Vulnerability Reduction. UNESCO, Paris. 19906. Structural and Material Problems of Old Sana'a. Mission Report. UNESCO, Paris. LIRER, L., MURNO, R., POSTIGLIONE, I., VINCI, A. & VITELLI, L. 1997. The AD 79 eruption as a future
372
R. HUGHES & A. COLLINGS
explosive scenario in the Vesuvian area: evaluation of associated risk. Bulletin of Volcanology, 59, 112-124. MANELLI, P., CAPALDI, G., CHIESA, S.etal.\99\. Magmatism in northern part of Yemen from Oligocene to present. Tectonophysics, 198, 181-202. MENZIES, M. 1996. Geology of Yemen. Unpublished report. Royal Holloway College, London. NEUMANN VAN PADANG, M. 1963. Arabia and the Indian Ocean. Catalogue of Active Volcanoes of the World, Part 16. International Association of Volcanology, Institute di Geologia Applicata, Fac. di Ingengneria, Rome. OVE ARUP 1990. Preliminary Study of Earthquake Hazard in North Yemen. Internal Report. Ove Arup, and Partners, London. PLAFKER, G., AGAR, R., ASKER, A. H. & HANIF, M. 1987. Surface effects and tectonic setting of the 13 December 1982 North Yemen earthquake. Bulletin of the Seismological Society of America, 77(6), 2018-2037.
SCANDON, R., ARGANESE, G. & GALDI, F. 1993. The evaluation of volcanic risk in the Vesuvian area. Journal of Volcanology and Geothermal Research, 53,263-271. SlMKIN, T., SlEBERT, L., McCLELLAND, L., BRIDGE, D.,
NEWHALL, C. & LATTER, J. H. 1981. Volcanoes of the World. Hutchinson Ross Publishing Company, Stroudsberg, PA. STIROS, S. 1996. Identification of earthquakes from archaeological data: methodology, criteria and limitations. In: STIROS, S. & JONES, R. E. (eds) Archaeoseismology. IGME & British School at Athens, Fitch Laboratory Occasional Paper, 7, 129-152. & JONES, R. E. (eds) 1996. Archaeoseismology. IGME & British School at Athens, Fitch Laboratory Occasional Paper, 7. UNDRO 1982. Volcanic Emergency Management. United Nations Environment Programme, 1987 Environmental Data Report. Blackwell, Oxford.
Archaeological, geomorphological and geological evidence for a major earthquake at Sagalassos (SW Turkey) around the middle of the seventh century AD MARC WAELKENS1, MANUEL SINTUBIN2, PHILIPPE MUCHEZ3 & ETIENNE PAULISSEN4 1
Department of Archaeology, Blijde Inkomststraat 21, B-3000 Leuven, Belgium Laboratorium voor Algemene Geologic, Redingenstraat 16, B-3000 Leuven, Belgium 3 Fysico-chemische geologic, Celestijnenlaan 200C, B-3001 Heverlee, Belgium (e-mail: [email protected]) 4 Geomorfologie en Regionale Geografie, Redingenstraat 16, B-3000 Leuven, Belgium
2
Abstract: The ancient city of Sagalassos, located 7km north of the ilce (centre of an administrative unit) of Aglasun in the Turkish province of Burdur, was an important centre during the Roman period. The collapse and subsequent repairs of buildings may suggest that at least four distinct earthquakes struck this city. These could have occurred in the second half of the first century AD, the middle of the third century AD, the first quarter of the sixth century AD and the middle of he seventh century AD. The youngest of these four earthquakes could have been so destructive that the city was abandoned. The age of this earthquake is based on coins and pottery dated to the middle of the seventh century AD, overlain by the collapsed monuments. Additional evidence is provided by the fractures in a baked mosaic floor belonging to a library, set on fire in the third quarter of the fourth century AD and filled in with waste material and earth during the fire. A directional analysis of the fractures in the mosaic floor of the Library, on the pavement of the Upper Agora and on the Theatre steps gives some insight into the local stress field associated with this earthquake. The palaeostress field, inferred from this analysis, is similar to that created by a transtensional strike-slip activity on an E-W-trending tear fault, situated just north of the city. Although it cannot be proven that the earthquake was related to activity along this fault, the magnitude of the destruction suggests an epicentre in the proximity of the city.
The Pisidian city of Sagalassos is located 7 km north of the ilfe of Aglasun in the Turkish province of Burdur and 10km south of Isparta (Fig. 1). The Aglasun Daglan form the mountain chain between Isparta and Sagalassos. Remains of Sagalassos, which was an important centre during the Roman period (Waelkens 19930), are scattered over several natural terraces at an altitude between 1450 and 1750m (Paulissen et al. 1993) (Fig. 2). The site may have been temporarily occupied in the Neolithic period (Waelkens et al. 1997), but must have been settled during the middle Bronze Age at the latest (16th century BC). During the Hellenistic period it rose in importance, and in Roman Imperial times it became the wealthiest city of Pisidia (Western Taurus, Waelkens 19930). Perhaps after a brief decline in the first half of the third century AD, it maintained its prosperity until the early sixth century AD. The site was
suddenly abandoned shortly after the middle of the seventh century AD. Inhabitants partly moved to nearby Aglasun (Fig. 1), where the bishopric of Sagalassos continued to exist until the middle Byzantine period (Waelkens 1993a). In this paper we would like to discuss, based on archaeological evidence found in collapsed buildings, that possibly four distinct earthquakes (in the second half of the first century AD, the middle of the third century AD, the first quarter of the sixth century AD and the middle of the seventh century AD) struck the ancient city. It will, moreover, be shown that the youngest one could have been responsible for such extensive destruction that the city was abandoned. For comparison, known earthquakes in southwestern Turkey, which occurred within the same time period as those at Sagalassos, are briefly discussed. In the geological approach we focus on fracture patterns observed at three
From: McGuiRE, W. G., GRIFFITHS, D. R., HANCOCK, P. L. & STEWART, I. S. (eds) The Archaeology of Geological Catastrophes. Geological Society, London, Special Publications, 171, 373-383. 1-86239-062-2/00/ $15.00 © The Geological Society of London 2000.
374
M. WAELKENS ET AL.
Fig. 1. (a) Simplified map of the 'Isparta Angle' with indication of the main tectonic domains (after Brunn et al. 1970; Glover & Robertson 1998); (b) geological map of the Sagalassos area.
EARTHQUAKES AT SAGALASSOS (SW TURKEY)
375
Fig. 2. Overview map of the ancient city of Sagalassos.
building sites (Neon Library, Upper Agora and Theatre; Fig. 2). The directional analysis, performed on the fracture patterns observed, especially takes into account fractures crosscutting several building units (mosaic blocks or pavement blocks), to exclude as much as possible the influence of the geometry of the building units. Geological setting Technically, the region around Sagalassos is situated near the apex of the so-called Isparta Angle (Fig. 1; Dilek & Rowland 1993). The triangular-shaped configuration results from the complex Mesozoic-Tertiary convergence history in the Eastern Mediterranean, involving a large number of limestone platforms and interjacent small oceanic basins (Poisson et al. 1975; Robertson 1993). The Bey Daglan limestone platform (Fig. 1) is considered to be parautochthonous regarding the Antalya Nappes, which were emplaced onto this limestone platform in latest Cretaceous-Paleocene time (Glover & Robertson 1998), and regarding the Sultan Dag-Beys.ehir and Lycian Nappe Complex, which were thrust onto this limestone platform during late Eocene and late Miocene
times, respectively (Fig. 1). The emplacement of the Lycian Nappe Complex (e.g. Collins & Robertson 1997) induced the development of a foreland basin, the Aksu Basin. The late Miocene to Plio-Pleistocene evolution of the Aksu Basin was determined by a change in the regional stress field from a N-S extension in late Miocene time to a NE-SW extension in late Pliocene time (Glover & Robertson 1998). This resulted in a change of reactivation of preexisting N-S- to NNW-SSE-oriented crustal weaknesses, which probably originated during the Mesozoic rifting. The late Miocene reactivation resulted in a transtensional basin development. During late Pliocene time a half-graben configuration occurred in a pure extensional regime. Examples in the area of such long-lived crustal weaknesses are the Kirkkavak fault, bordering the Aksu Basin to the east, and the Kovada Graben (Fig. 1). The area around Sagalassos is situated in the frontal parts of the Lycian Nappe Complex (Fig. 1). This stack of thrust sheets consists of limestone nappes, which were derived from the Domuz Dag limestone platform and were thrust on top of ophiolites. This ophiolite originated from the oceanic crust between the allochthonous and parautochthonous platforms and was
376
M. WAELKENS ET AL.
thrust on top of its own flysch (Hayward 1984). The flysch lies immediately on the parautochthonous limestones. The thrust sheets are limited by weakly S-dipping thrust faults, suggesting gravitational gliding as a mechanism of emplacement of these thrust sheets into the flysch basin. The city itself (Fig. 1) was built on the ophiolites and on isolated hills derived from the overlying limestone thrust sheet, which dominates the city to the north and forms the Aglasun Daglan (summits at about 2000 m altitude). Some of the limestone hills have a tectonic origin and others represent megaslumps. On the mountain slope several indications for rotational slumping have been mapped (Verstraeten et al. 2000). These slumps displace not only limestone blocks but also slope material. Just north of the city (c. 700 m), the basal thrust fault of the Aglasun Daglan is displaced by an approximately E-W-trending, sub-vertical tear fault (Yalcinkaya 1983), which is also apparent as a prominent E-W-running scarp in the topography (c. 5 km long). This tear fault is characterized by the occurrence of metre-thick breccias, oriented parallel to the topographic scarp. This subvertical fault plane may have acted as a zone of weakness along which megaslumping occurred. Geomorphological setting Slope processes have been and are still active on the mountain slopes around Sagalassos (Paulissen et al 1993; Poesen et al 1995; Covers & Poesen 1998; Librecht et al 2000; Verstraeten et al 2000). Creep, gully erosion, rock falls, debris flows, debris avalanches and even an earth flow have been recognized in the area investigated. Fossil slope processes are indicated by the presence of block slopes, a block stream, various types of scree deposits and giant mass movements. On the slopes above the city we did not observe devastating slope processes such as debris avalanches, earth flows or block streams. It should be stressed that the site of Sagalassos (2.5km by 1.2km) is far too large to be devastated by a single event of slope processes. Therefore, we will discuss in detail the geomorphological setting and the significance of slope processes and deposits for the constructions on which directional analyses have been performed. The greatest distance between these constructions is c. 350m, nearly parallel to the mountain slope. The greatest height difference is 55m (between 1525m and 1580m altitude). The Neon Library (Fig. 2) is situated at c. 1550m on the foot slope of the mountain scarp and could therefore directly be threatened
by slope processes. On the mountain slope above the Library, the exposure of the limestone substrate is dominant with block slopes reworked by debris flows. From a geomorphological point of view, however, the site of the Library is well chosen, as it is situated in a depression just downslope of a limestone cliff of about 20m height. This cliff has been important enough to protect the Library from geomorphological events originating from the mountains. The cliff diverts the slope processes above the Library towards two channels, situated to the west and east of the Library. The channel-cliff-channel succession results in the preferential deposition of slope materials and in the formation of a depression just downslope of the limestone cliff and between the talus cones. The latter are situated at the outlet of the channels towards the east and the west. The excavations revealed that, after its destruction, the Library ruins have been filled artificially with dump materials and that very little material eroded from the mountains has been deposited in the depression. The foundation of the Library is situated directly on the ophiolites, which crop out around the Library. The top of the ophiolites is flat and forms a terrace. In the excavated area, no deformation structures or creep phenomena have been observed on top of the ophiolites around the Library. A borehole through the mosaic floor in the Library has shown that it has not been paved directly on the ophiolites, but that it was situated on an artificial bed of sandy material of c. 1 m thickness. This filling could have been subject to differential setting; however, its importance was less than 10cm. The fracture system in the mosaic is not related to this differential setting. Although the building is situated on the foot slope of a mountain scarp, the local situation protected the Library from slope processes. Mass movements, even slow ones, and deformation structures have not been recognized on top of the substrate or on the steps in front of the Library. The Upper Agora is situated at about 1525 m and its northern limit occurs at c. 100m from the foot of the mountains. It has been built on a volcanic tuff. The mountain slopes above the Upper Agora are nearly entirely covered with slope debris forming block slopes. On these block slopes, debris flows with a length of several metres are active. Debris avalanches are lacking. The Upper Agora was totally free of slope deposits and was largely uncovered. The western and northern edges of the Agora were covered with collapsed building materials. The excavations have revealed that natural slope deposits were lacking.
EARTHQUAKES AT SAGALASSOS (SW TURKEY) The Theatre, dominating Sagalassos in the east, occurs on the transition between a gentle slope (15% inclination) and a plateau at an altitude of 1580m. The entire building is constructed on ophiolites. Most of the Theatre is built in a depression, so that the top of the monument is only a few metres above the surface of the adjacent plateau, where the in situ ophiolites occur near the surface and slope deposits are lacking. Only the northwestern part of the Theatre is entirely constructed in open air, with the building foundation near the original topographic surface. To the north, the Theatre is separated from the mountain front by a broad shallow depression. This depression protects it from eventual slope processes generated from the mountains to the north. In conclusion, the further discussed destruction of the buildings seems to be unrelated to slope processes or local differences within the buildings. Archaeological evidence for the age of major earthquakes Several observations suggest that an earthquake could have caused extensive damage at the site
377
some time before the second century AD. Several buildings (Fig. 2), such as a late Hellenistic fountain house, the peripteros of the Augustan Temple of Apollo Klarios and a Tiberian gateway, giving access to the Lower Agora of the city, had to be repaired during the first half of the second century AD (Waelkens 1993#, c). Tilting and directional collapse should not be treated as unequivocal evidence of an ancient seismic disaster, as they may also be the result of poor construction, soil creep and slides on natural or artificial slopes, or other adverse geotechnical effects (Karcz & Kafri 1978). However, it is unlikely that this was the case at Sagalassos during this period. The three buildings discussed were very well built ashlar structures, of which at least one (the peripteros of the Augustan Temple of Apollo Klarios) stood on a solid limestone bedrock. In the late Hellenistic fountain house, the movement seems to have been from the northwest to the southeast (Fig. 3), as the backwall of the west portico apparently collapsed under the pressure of the backfill material. At the same time, the fountain courtyard, originally open to the south, was closed off by a wall. Its purpose seems to have been to provide additional stability to the
Fig. 3. Maps showing the topography and the direction of collapse of the constructions.
378
M. WAELKENS ET AL.
structure by resisting the pressure of earth behind the back walls of the side porticoes and to act as a wedge-like structure between them. In the Tiberian gateway, the original capitals of the four central columns were completely replaced by new ones, whereas the lateral capitals, belonging to the more solid side wings of the structure, were kept in place. In the entablature elements, new clamp and dowel holes were cut and some pieces needed repair after cracking. This shows that at least the central part of the building has been affected by rock movements, which damaged the upper structure of the central part of the building. The reason for this is not directly evident, but poor construction or mass flows can be excluded. The prolific building programmes of the city during the Julio-Claudian period (25 BC-AD 56) also came to an end rather suddenly around the middle of the first century AD (Waelkens 1993a). Building activities resumed only in the late first and early second century AD, especially during the reign of Trajan (AD 97-118; e.g. construction of the Odeion). Even in this period, the new monumental architecture was characterized by its plain, undecorated mouldings, and sharply contrasted with the rich Julio-Claudian and Hadrianic buildings of the city (Vandeput 1997). All these elements suggest that an earthquake could have caused damage to the city around the middle of the first century AD. It can be compared with the earthquake (70 = VIII) that destroyed Apamea (Dinar), located 50km to the northwest of Sagalassos, in AD 53 (Ergin et al 1967, p. 12; Guidoboni 1994, p. 90). In fact, the nature of the damage, which could not be due to a landslide or poor construction, could correspond well to that of a mild earthquake at Sagalassos (70 < VIII). A strong seismic event would have caused the Tiberian gateway to collapse completely. In 1994, a not too strong earthquake caused considerable damage in Dinar. During this earthquake, cracks developed in the building walls at Aglasun. A second earthquake may have occurred in the first half of the third century. The Neon Library, originally built shortly after AD 120 and repaired near the beginning of the third century AD, received a completely new mosaic pavement and facade during the third quarter of the fourth century AD. This may have taken place during the reign of the emperor Julian (AD 361-363), as part of the new building programme meant to restore the old pagan culture (Waelkens 1993&; Waelkens et al. 1997). During this intervention the old facade and the front part of the two lateral walls were completely rebuilt by using older material. This suggests that the front part
of the building must have collapsed or that it had been damaged beyond repair between a first series of repairs at the beginning of the third century, affecting the side walls (Waelkens 19936), and the reign of Julian. In addition, an interruption in new building projects in the city took place starting from the second quarter of the third century AD. Although the reason for these observations may be multiple, it is interesting to recall that Mitchell (1995, pp.63, 67, 137, 157) described severe damage at the forum and in particular of the basilica at Kremna, c. 25 km south of Sagalassos, which can be dated around AD 220-230. Mitchell (1995) has attributed the destruction at Kremna to an earthquake. If such an event took place, it must have been felt at Sagalassos as well and could have caused the damage observed. In any case, the effects certainly were not massive and the collapse of the library facade may also have been due to other unknown reasons. Nearly three centuries later, an earthquake (/0 = VI) with its epicentre to the north of Antalya (c. 100 km to the south of Sagalassos, Fig. 1), occurred in AD 518 (Ergin et al. 1967, p. 14). The collapse of some very light buildings, such as the Augustan propylon of the Doric Temple, the Tiberian gateway on the Lower Agora and the portico on the west side of this square, have previously been attributed to this earthquake (Waelkens et al 1997). The propylon seems to have collapsed in a northern direction (Fig. 3). The collapsed elements were covered by material of the first half of the sixth century AD. The Tiberian gateway did not collapse completely. Its west wing remained standing, but its central part and eastern wing were completely dismantled. Some of their cornices were reused as steps in the monumental stairway constructed above the old approach. Sherds in the foundation upon which the new steps were built indicate a construction date during the first half of the sixth century AD. The observation that during the same period, dated by pottery in the fill that covered the pipes, the water of the late Hellenistic fountain house was captured and directly distributed over the city by means of terracotta pipes suggests that the city started to suffer from water shortage. This water shortage may have been due to the destruction of the aqueducts by the earthquake (Waelkens 1993c). In the area to the north and northwest of the Doric Temple, a large dump, containing many coins, sherds, smaller stones and mortar fragments, including stucco, was piled up against the fortification walls (early fifth century AD), so that the latter completely lost their defensive function. The material inside this dump suggests
EARTHQUAKES AT SAGALASSOS (SW TURKEY) that this operation, which looks like the clearing of debris from within the city, again took place during the first half of the sixth century AD. The area inside these walls and the artisanal area to the east of it were also raised and completely transformed as part of the same event. In the central part of the city, a private house, at present being excavated, was also destroyed and rebuilt during the same period. Recently, the excavations revealed that the western facade of the Roman bath building had been rebuilt at that time as well (Waelkens et al 1997). All excavation areas in the city thus show damage and subsequent repairs dated by sherds and coins, during the first half of the sixth century AD. Their distribution all over the city seems too widespread to attribute them to anything else but a seismic event. Several archaeological observations indicate that the city was abandoned around the middle of the seventh century AD, and that most buildings collapsed in this time period. First, Sagalassos red slip wares were mass-produced and traded without interruption from the beginning of the Imperial period until the middle of the seventh century AD (Degeest 1996, p. 311). No pottery can be dated after this time. Second, the most recent coins found in a stratigraphical context are three coins from the reign of Constans II (AD 641-668). They were found in collapsed debris on the Upper Agora, on the sixth-century stairway leading to the Lower Agora and inside simple dwellings built inside the western portico of the Lower Agora. Many coins dated to the reign of Heraclius (AD 610640) are present in the same debris (Scheers 1995, 1997). Third, inside the temenos of the Doric Temple, in the artisanal area to the north of the Upper Agora, and inside the western portico of the Lower Agora, primitive houses and workshops collapsed on top of household and storage vessels, dated around the middle of the seventh century AD. This indicates that when the buildings collapsed, they were still in use. The late Hellenistic fountain house, the NW Heroon and the surrounding buildings, the small sides of the Doric Temple, and the west facade of the Roman baths equally fell down on strata with sixth- and earlier seventhcentury pottery and coins. Other buildings of which the destruction cannot yet be documented stratigraphically may also have been fallen down during this period. On the Lower Agora, shortly after the destruction of the monuments around it, a rather primitive graveyard was arranged in the destruction debris. The few coins and other datable objects point towards a date around or shortly after the middle of the
379
Fig. 4. Photograph showing a collapsed wall of the Upper Agora. The collapse was to the southeast and occurred in the middle of the seventh century.
seventh century AD, providing a terminus ad quern (Waelkens et al 1997). The collapse of all the monuments mentioned above was consistently towards the southeast (Figs 3 and 4). The observation that the collapsed monuments immediately overlie strata with sixth-century and early seventh-century material (without any other slope deposit or soil horizon) and that all the buildings collapsed towards the southeast indicates that this severe destruction of the city was probably due to a major earthquake around the middle of the seventh century AD. The magnitude of the damage suggests that the epicentre of the earthquake may have been situated in the proximity of the city. Other evidence for this earthquake is less direct. The Neon Library had been destroyed and set on fire shortly after the death of Julian (AD 363), in any case during the second half of the fourth century AD. At that time, the inhabitants smashed the central polychrome mosaic panel of the fourth century AD, as it represented a pagan scene involving the goddess Thetis. The rest of the floor in black and white tesserae was not touched and is even well preserved after the building had been set on fire and filled in with earth and industrial waste (Waelkens 19936). The material of the fill suggests that the event occurred shortly after Julian's death. According to the restorers of the mosaic, the fire had firmly baked the tesserae to their mortar support, so that they remained virtually intact and were better protected against damage. Now, the floor contains numerous fractures, many of which continue on the lateral walls of the building (Fig. 5). The absence of displacements along the fractures related to the smashing of the central polychrome panel indicates that the fractures post-date this destruction and are younger than
380
M. WAELKENS ET AL.
Fig. 5. Photograph of the fracture patterns in the Fig. 6. Photograph of fractures in the Upper Agora mosaic floor (a) and in the walls of the Neon Library floor (a) and in the Theatre steps (b) (hammer is (b) (pencil is 14cm long). 30cm long). AD 363. Also, the cross-cutting relation of the fractures with the tesserae and their mortar support can best be explained by fracturing after baking of the mosaic (Kokten, pers. comm.). Hierapolis in Phrygia, c. 200 km further west, seems to have been destroyed by an earthquake during the early seventh century AD. As the latest coin evidence from that destruction dates to AD 602-603 (Guidoboni 1994, pp. 350-351), this cannot have been the same catastrophe as the one that struck Sagalassos. However, the catastrophic earthquake at Sagalassos may correspond to another earthquake described in the Byzantine area between AD 641 and 668 (Guidoboni 1994, p. 238), i.e. during the reign of Constans II, to which period belong the latest coins found in the debris. Unfortunately, the epicentre of this seismic catastrophe is not specified in the Byzantine sources.
Geological evidences A directional analysis has been performed on the fracture patterns observed in the mosaic floor of the Neon Library, on the pavement
of the Upper Agora and on the Theatre steps (Figs 5 and 6). These three constructions are partly built on different rocks and the slopes in the immediate vicinity of the constructions do not have the same orientation. The mosaic floor of the Library shows extensive fracturing (Fig. 5), which can clearly be followed in the walls (Fig. 5). Because of the dimensions of the mosaic blocks, their influence on the geometry of the fracture pattern can be neglected. The observed fracture pattern, based on 108 orientation data, is rather consistent (confidence angle 6.9°). The strike of the dominant fracture population ranges from N80E to E-W (Fig. 7). A mean strike of N87E has been calculated. Secondary populations have strikes of N60-65E and N80-85W. The N87E and the N60-65E fracture sets form a stepwise pattern (Fig. 5). The fracture system on both floor and walls shows a normal dip-slip movement towards the south. The Upper Agora floor consists of metresized, rectangular limestone pavement blocks, basically organized in a N10W-N80E pattern. The orientation distribution of the fractures measured (183 data, Fig. 6) shows a rather
EARTHQUAKES AT SAGALASSOS (SW TURKEY)
Fig. 7. Rose diagrams of the fractures measured in the Library floor (a), the Upper Agora floor (b) and the Theatre steps (c). The length of the rose petals is proportional to the frequency (expressed in per cent) of fracture orientation in 5° intervals.
dispersed image (confidence angle 30.5°, Fig. 7). The two directions of the pavement pattern are dominantly present. Although primarily fractures cross-cutting several pavement blocks were measured, the pavement pattern still biased the orientation distribution of the fractures. The most prominent fracture population, however, has a N50-55E orientation. Another distinctive population has a N50-55W orientation. The orientation of the blocks forming the Theatre steps is variable (Fig. 6). Two dominant fracture populations are observed (71 orientation data, confidence angle 15.3°): a N75-80W and a N45-50E population (Fig. 7). These trends are independent of the orientation of the steps or the local topography.
381
The fracture patterns show a consistent directional image. In the Neon Library and in the Upper Agora floor, the N80E to E-W direction is predominantly present, although in the latter case it coincides with the orientation of the pavement blocks. In the Library floor and walls a normal dip-slip movement to the south has been observed. Based on this fracture pattern a NNW-SSE-oriented extensional stress direction (cr3) can be suggested (Fig. 7). The orientation of both other principal stress directions cannot be constrained solely based on this extensional fracture pattern. It should therefore be taken into account that this fracture system does not necessarily have a tectonic origin but may have been induced by slope processes related to slumping, which, however, does not exclude that these processes were triggered by seismic activity. In the Theatre steps the N45-50E direction shows a conjugate relationship with a N75-80W direction. Such a conjugate system would imply a N75E-oriented compressional principal stress direction (ai), a sub-vertical intermediate principal stress direction (0-2) and a N15W-oriented extensional principal stress direction (0-3) (Fig. 7). This stress field is, moreover, in accordance with the normal dip-slip activity observed on the dominant fracture system in the Library and Upper Agora floors. This conjugate fracture system with a subvertical, intermediate principal stress direction clearly has a tectonic origin. On the basis of the fracture patterns observed at the various building sites, a consistent palaeostress field could be derived. The latter corresponds to a stress field creating transtensional strike-slip activity on the E-W-trending tear fault to the north of the city (Fig. 8). Although
Fig. 8. Synthetic map with indication of the sinistral tear fault, displacing the thrust front of the Aglasun Daglarl Limestone nappe. Indication of stress field inferred from directional analysis of fracture systems.
382
M. WAELKENS ET AL.
the fracture system observed in the three constructions could have formed during a single event, a multiple origin in the same stress field cannot be excluded. The fracture development post-dates the destruction of the Library and the smashing of the Library mosaic floor during the third quarter of the fourth century AD. On the basis of filling material, immediately underlying the collapsed buildings, the fatal earthquake occurred around the middle of the seventh century AD. The magnitude of the destruction was enormous, suggesting an epicentre in the proximity of the city. The extent of destruction, together with the geological evidence, indicates that an epicentre on the nearby tear fault is possible. However, no studies have yet been undertaken to demonstrate eventual coseismic displacements on this fault, so other epicentres remain equally possible.
Discussion and conclusion Archaeological evidence suggests that two and possibly four major earthquakes struck the city of Sagalassos in classical times. The youngest one, dated around the middle of the seventh century AD, caused such extensive damage that no trouble was taken to clear the debris from the marketplace, which henceforth served as a burial site for the last inhabitants of the city (Waelkens et al. 1997). Buildings show an overall preferential collapse towards the north or southeast, suggesting that earthquakes may be related to two different seismic systems. From the fracture sets observed in the Library and Upper Agora floor and in the Theatre steps, a stress field is inferred that could be created by sinistral transtensional strike-slip activity on a tear fault. Although it cannot be proven that the earthquake was related to activity of this fault, the magnitude of the destruction suggests an epicentre in the proximity of the city. From this study it can be concluded that taking into account the specific geometric characteristics of the building material, buildings on archaeological sites may serve as 'strain markers' in an attempt to determine local stress fields associated with earthquakes. In a first stage, a relationship is assumed between earthquake-related damage on buildings and geological stress regimes. It is, however, clear that, in a later stage, only a complete geological survey of the active and palaeostress fields will allow the confirmation of the current interpretations. We would like to thank P. L. Hancock, S. H. Stiros and an anonymous referee for the constructive review
of the manuscript. This research has been funded by the Belgian Programme on Interuniversity Poles of Attraction (IUAP 28), initiated by the Belgian State Prime Minister's Office, by the Federal Services for Scientific, Technical and Cultural Affairs, by a Concerted action (GOA 97/2) and the Belgian Fund for Collective Fundamental Research (FKFO). The scientific responsibility remains with the authors. Important financial support came also from the Research Council of the Katholieke Universiteit Leuven (OT 89/8), the National Bank of Belgium, the L. Baert-HorTman Fund, the family L. LambertsVan Assche (Marlux), the ASLK/CGER Bank, the BACOB Bank, the ABB Insurance Company, Rotary (Zeebrugge-Oosthoek), Solvay and Agfa-Gevaert Films. P.M. and M.S. are respectively senior research associate and postdoctoral fellow of the Fund of Scientific Research of Flanders (Belgium).
References BRUNN, J., GRACIANSKY, P., GUTNIC, M., JUTEAU, T., MONOD, O. & POISSON, A. 1970. Structures majeures et correlations stratigraphiques dans les Taurides occidentales. Bulletin de la Societe Geologique de France, 12, 515-556. COLLINS, A. S. & ROBERTSON, A. H. F. 1997. Lycian melange, southwestern Turkey: an emplaced Late Cretaceous accretionary complex. Geology, 25, 255-258. DEGEEST, R. 1996. The common wares of Roman Sagalassos from the first to the seventh century AD. Ph D thesis, Katholieke Universiteit Leuven. DILEK, Y. & ROWLAND, J. C. 1993. Evolution of a conjugate passive margin pair in Mesozoic southern Turkey. Tectonics, 12, 954-970. ERGIN, K., GUCLU, U. & Uz, Z. 1967. A Catalog of Earthquakes for Turkey and Surrounding Areas (11 AD to 1964AD). Maden Fakiilteri Ofset Baski Atolyesi, Istanbul. GLOVER, C. P. & ROBERTSON, A. H. F. 1998. Role of regional extension and uplift in the Plio-Pleistocene evolution of the Aksu Basin, SW Turkey. Journal of the Geological Society, London, 155, 365-387. GOVERS, G. & POESEN, J. 1998. Field experiments on the transport of rock fragments by animal trampling on scree slopes. Geomorphology, 23, 193-203. GUIDOBONI, E. 1994. Catalogue of Ancient Earthquakes in the Mediterranean Area up to the 10th Century. Instituto Nazionale di Geofisica, Rome. HAYWARD, A. B. 1984. Miocene clastic sedimentation related to the emplacement of the Lycian Nappes and the Antalya Complex, SW Turkey. In: DIXON, J. E. & ROBERTSON, A. H. F. (eds) The Geological Evolution of the Eastern Mediterranean. Geological Society, London, Special Publications, 17, 287-301. KARCZ, I. & KAFRI, U. 1978. Evaluation of supposed archaeoseismic damage in Israel. Journal of Archaeological Science, 5, 237-253.
EARTHQUAKES AT SAGALASSOS (SW TURKEY) LIBRECHT, I., PAULISSEN, E., VERSTRAETEN, G. & WAELKENS, M. 2000. Implications of environmental changes on slope evolution near Sagalassos. In: WAELKENS, M. & LOOTS, L. (eds) Sagalassos V. Report on the Survey and Excavation Campaigns of 1996 and 1997. Acta Archaeologica Lovaniensia Monographiae, 10, in press. MITCHELL, S. 1995. Cremna in Pisidia. An Ancient City in Peace and in War. Duckworth with the Classical Press of Wales, London. PAULISSEN, E., POESEN, J., GOVERS, G. & DE PLOEY, J. 1993. The physical environment at Sagalassos (Western Taurus, Turkey). A reconnaissance survey. In: WAELKENS, M. & POBLOME, J. (eds) Sagalassos II. Report on the Third Excavation Campaign of 1992. Acta Archaeologica Lovaniensia Monographiae, 6, 229-249. POESEN, J., GOVERS, G., PAULISSEN, E. & VANDAELE, K. 1995. A geomorphological evaluation of erosion risk at Sagalassos (Western Taurus, Turkey). A reconnaissance survey. In: WAELKENS, M. & POBLOME, J. (eds) Sagalassos III. Report on the Fourth Excavation Campaign of 1993. Acta Archaeologica Lovaniensia Monographiae, 7, 341-355. POISSON, A., AKAY, E., DUMONT, J. & UYSAL, S. 1975. The Isparta angle: a Mesozoic paleorift in the Western Taurides. In: TEKELI, O. & GONCUOGCY, C. (eds) Geology of the Taurus belt. Proceedings of the International Symposium on the Geology of the Taurus Belt, MTA, Ankara, 11-26. ROBERTSON, A. H. F. 1993. Mesozoic-Tertiary sedimentary and tectonic evolution of Neotethyan carbonate platforms, margins and small ocean basins in the Antalya complex, southwest Turkey. In: FROSTICK, L. E. & STEEL, R. J. (eds) Tectonic Controls and Signatures in Sedimentary Successions. International Association of Sedimentologists, Special Publications, 20, 415-465. SCHEERS, S. 1995. Catalogue of the coins found in 1993. In: WAELKENS, M. & POBLOME, J. (eds) Sagalassos III. Report on the Fourth Excavation
383
Campaigns of 1993. Acta Archaeologica Lovaniensia Monographiae, 7, 307-326. 1997. Coins found during 1994 and 1995. In: WAELKENS, M. & POBLOME, J. (eds) Sagalassos IV. Report on the Fourth and Fifth Excavation Campaigns of 1994 and 1995. Acta Archaeologica Lovaniensia Monographiae, 9, 315-342. VANDEPUT, L. 1997. The architectural decoration in Roman Asia Minor. Sagalassos: a case study. In: WAELKENS, M. (ed.) Studies in Eastern Mediterranean Archaeology 1. Brepols, Turnhout. VERSTRAETEN, G., LIBRECHT, L, PAULISSEN, E. & WAELKENS, M. 2000. Limestone platforms around Sagalassos as the result of giant mass movements. In: WAELKENS, M. & LOOTS, L. (eds) Sagalassos V. Report on the Survey and Excavation Campaigns of 1996 and 1997. Acta Archaeologica Lovaniensia Monographiae, 10, in press. WAELKENS, M. 19930. Sagalassos. History and archaeology. In: WAELKENS, M. (ed.) Sagalassos I. First General Report on the Survey (1986-1989) and Excavations (1990-1991). Acta Archaeologica Lovaniensia Monographiae, 5, 37-82. 1993/7. The 1992 season at Sagalassos. A preliminary report. In: WAELKENS, M. & POBLOME, J. (eds) Sagalassos II. Report on the Third Excavation Campaign of 1992. Acta Archaeologica Lovaniensia Monographiae, 6, 9-42. 1993<:. The excavations of a late Hellenistic fountain house and its surroundings (Site N). An interim report. In: WAELKENS, M. & POBLOME, J. (eds) Sagalassos II. Report on the Third Excavation Campaign of 1992. Acta Archaeologica Lovaniensia Monographiae, 5, 37-82. , VERMEERSCH, P.-M., PAULISSEN, E. et al. 1997. The 1994 and 1995 excavation seasons at Sagalassos. In: WAELKENS, M. & POBLOME, J. (eds) Sagalassos IV. Report on the Fourth and Fifth Excavation Campaigns of 1994 and 1995. Acta Archaeologica Lovaniensia Monographiae, 9, 103-216. YALCINKAYA, S. 1983. Geological map of Isparta-M25d.l, scale 1/25.000. MTA, Ankara.
This page intentionally left blank
Fault pattern of Nisyros Island volcano (Aegean Sea, Greece): structural, coastal and archaeological evidence STATHIS C. STIROS Department of Civil Engineering, University of Patra, Pair a 26500, Greece (e-mail: [email protected]) Abstract: There has been much debate about the fault pattern of Nisyros Island at the southeastern edge of the Aegean volcanic arc. The small active volcanic island, less than 200-100 ka old, is dominated by a well-developed caldera and by post-caldera domes that are less than 25 ka old and up to 600m high. Detailed mapping of the tectono-volcanic features of Nisyros have revealed that faults have a clear radial pattern, and they are more abundant in the northwest of the island, where volcanic domes are also concentrated. In contrast to previous speculations that certain major faults control the tectono-volcanic development of Nisyros, this paper argues that the radial fracture pattern is expected where faulting is a secondary effect of volcanic doming. Structural, coastal and archaeological evidence supports this contention. Most faults have a short fault length and a 'scissors-type' geometry typical of magma or salt ascent dynamics, and variable throws that are too high to reflect simply tectonic effects. Elevated coastal marine fossils (vermetids) in the northwest of the island indicate rates of uplift too high to be explained solely by differential fault movements. In the same area, a fault that borders the fortifications of a fourth-century BC castle is inferred to be the source of seismic damage effects observed in its ramparts and responsible for the near-total destruction of its westernmost fortifications. Together, the evidence suggests that localized high rates of uplift, faulting and tilting reflect tectonovolcanic deformation effects, and remind us that associated fault activity is likely to have constituted a major threat for the island since antiquity.
Nisyros Island is an active volcano at the northeastern edge of the Aegean volcanic arc (Fyticas et al. 1985), and one of the three volcanoes in this region (the others being Santorini and Methana, Fig. 1) for which an eruption during the last 2000 years is documented. It is a small island, 8 km in diameter, with a broadly conical form truncated by a well-developed caldera of 4km width (Fig. 2). Craters of recent hydrothermal eruptions can be observed inside this caldera (Figs 3 and 4); the most recent ones occurred during the paroxysmal period of 1873 (Di Paola 1974; Marini et al. 1993). The surrounding area is very technically active, and has been affected by numerous strong earthquakes during the historical period. The last major event, of magnitude 6.6, occurred in 1933 and killed about 200 people; several villages in Nisyros and nearby Kos Island were totally destroyed and some were abandoned (Papazachos & Papazachou 1989). Several other seismic sequences of smaller magnitudes (<5) have struck Nisyros in the last few decades (for instance, in 1953, 1968 and 1970). Some of these
seismic sequences have been explained as volcanic tremors (e.g. Bornovas 1953), but no clear evidence has yet been presented to support this view (Kallergis & Morphis 1968; Stiros & Vougioukalakis 1996). Seismic activity resumed recently, and for about 2 years earthquake swarms struck both Nisyros and the wider area, although with only minor destructive effects. One of these earthquake swarms occurred in spring 1997 and was associated with the activation of ground fissure of 200m length cutting houses and roads in Mandraki, the island's capital. This fissure follows the Langadha Fault, a structure that is discussed later in this paper. Although it is not clear which specific earthquake event(s) the fissure was associated with, some workers regard it as a seismic fault (that is, a non-volcanic or a non-surficial effect), and at the same time envisage the recent seismicity as an expression of magma movement at depth (c. 150 km) (D. Papanikolaou, interviewed in daily Greek newspaper Ta Nea, 4 September 1997). In contrast, Stiros & Vougioukalakis (1996) claimed
From: McGuiRE, W. G., GRIFFITHS, D. R., HANCOCK, P. L. & STEWART, I. S. (eds) The Archaeology of Geological Catastrophes. Geological Society, London, Special Publications, 171, 385-397. 1-86239-062-2/00/ $15.00 © The Geological Society of London 2000.
386
S. C. STIROS
Fig. 1. Map showing the location of the volcanoes of Methana, Melos, Thera (Santorini) and Nisyros, which define the Quaternary Aegean volcanic arc. A, Athens.
that the recent seismicity, at least that between August 1995 and June 1996, is scattered over a very broad area and there is no evidence that it is directly associated with magma tic processes. Regardless of its origin, seismic activity on Nisyros causes unrest and even panic among a local population that still retains fresh memories of the destructive earthquake of 1933; the ruins of a village abandoned after this earthquake are one of the island's tourist attractions. Hence, discussions and debate on the nature of the recent fissures and of the wider fault pattern on the island have an important social dimension. Recent volcanic history of Nisyros The volcanic geology of Nisyros consists of a lower succession of lava flows, and pyroclastic layers deposited above them. The oldest of these volcanic rocks are pillow lavas and hyaloclastites that are younger than 300-200 kaBP and that are exposed at the northwestern corner of the island (beneath the Speliani Monastery hill at Mandraki). The pre-volcanic basement of Mesozoic and Neogene sediments is not exposed on Nisyros (Di Paola 1974; Vougioukalakis 1993). The island itself represents the emergent portion of an andesitic composite volcano that
has built up over the last 100-150 ka (see Fig. 2) (Vougioukalakis 1993). Cone-building activity associated with relatively deep (13-27 km) magma chambers characterizes the first cycle of the geological history of the island. The second cycle started possibly around 25-30 ka ago and is characterized by intense explosive activity and shallow (<6km depth) magma chambers. As a result of this activity, the volcano was truncated by a summit caldera of 4km width and the upper pumice levels were deposited at Nisyros and nearby Yali. The northwestern part of the Nisyros caldera was subsequently filled by a series of dacitic-ryodacitic domes up to 600m high. Following this last phase of major activity at Nisyros volcano, the subsequent (<20kaBP) geological history of the island has been characterized by generally weak volcanic activity. Volcanic necks, fumaroles and craters, most concentrated in the southeastern part of the crater (at an altitude of 110-120m), are mainly the products of hydrothermal eruptions (Vougioukalakis 1993). Their history is imperfectly known, but the most recent craters are associated with two periods of volcanic 'paroxysms' in modern times (Di Paola 1974; Marini et al. 1993). In particular, between 1871 and 1873 (Marini et al. 1993) and in 1887, intense volcanic
FAULT PATTERN OF NISYROS VOLCANO
387
Fig. 2. Topography and bathymetry of Nisyros. Contours at 100m intervals. The caldera rim is after Vougioukalakis (1993).
activity was responsible for the formation of craters in the central part of the caldera (craters named Polyvotis and Alexandros (or Phlegethro) in 1873, and Mikros Polyvotis in 1887 (Vougioukalakis 1998)) (Fig. 4). There was also a possible minor eruption between 1912 and 1916 (Kallergis & Morphis 1968), and some researchers argue that a 15th century description of the island testifies to an eruption c. AD 1422 (Vougioukalakis 1998). Furthermore, Vougioukalakis (1998) has argued that excellent geomorphological preservation of the Stephanos crater (Figs 3 and 4), which is 27 m deep and cut in very soft and easily erodable rocks, indicates an eruption not older than 3000-4000 years ago. Some of the historical eruptions caused ground fissuring (Marini et al. 1993; Vougioukalakis
1998) but there is no information on other types of ground deformation, such as coastal elevation changes, accompanying the volcanic events. St Seymour & Vlassopoulos (1989) argued that Nisyros volcano is currently dormant but that it may possess a high potential for future explosive activity, indicated by its high pre-eruptive volatile contents and the high apparent viscosity of recent eruptive products.
Fault pattern of Nisyros The fault pattern at Nisyros has been studied in recent years, mainly as part of projects related to exploitation of the geothermal field of the island.
388
S. C. STIROS
Fig. 3. The Nisyros caldera (a) and the Stephanos crater (b) within it. According to Vougioukalakis (1993), the state of preservation of Stephanos crater (also located in Fig. 4) indicates an eruption not more than 3-4 ka ago.
FAULT PATTERN OF NISYROS VOLCANO
389
Fig. 4. Main tectono-volcanic structures of Nisyros: rim of caldera collapse (bold lines with ticks), fault pattern (lines occasionally with ticks on the downthrown side), domes (closed curve with five radial curves) and craters of phreatic explosions (•). Based on Vougioukalakis (1993, 1998). S and AP, Stephanos and AlexandrosPolyvotis craters of hydrothermal eruptions; P and L, Palaiaokastro and Langadha Faults.
However, results of the various studies are conflicting, and the geometry and nature of the island's fault pattern remains controversial.
Previous interpretations Microtectonic analysis of fault-slip directions at about 25 sites led Simaiakis (1992) to conclude that faults on Nisyros could be grouped into two main systems of normal faults, one set trending
N50-70E and a second set trending N150-170E. The corresponding 0-3 axis was estimated to strike N15W, and it was concluded that faults striking N60-70E are the dominant controls on the tectono-volcanic evolution of the island. A similar conclusion had earlier been arrived at by Nakamura & Uyeda (1980) who, on the basis of data available at that time (Di Paola 1974), proposed a linear zone of monogenetic vents and argued that a main polygenetic vent connected with a deep magma source may be located in the
390
S. C. STIROS
centre of the caldera. According to these workers, the NE-SW trend of the monogenetic crater zone may indicate the direction of maximum horizontal compression (cr#max) of the ambient stress field (aligned N60E), following the reasoning of Nakamura (1977). A contrasting view, proposed by Papanikolaou et al (1991) on the basis of criteria such as fault length, morphological expression and separation of geological units, recognized four main radial fault sets of different trends. Vougioukalakis (1993) recognized three main fault directions (NE-SW, NWSE and E-W) suggestive of an extensional stress field aligned N-S to NW-SE. Vougioukalakis (1993) also suggested that magma ascent was related to dykes trending E-W to ENE-WSW.
An alternative interpretation An alternative explanation for the fault pattern on Nisyros is based on structural data collected in a detailed study of the volcanic stratigraphy and evolution of the island (Vougioukalakis 1993). The resulting structural map, shown in a simplified form in Fig. 4, is the product of many years of systematic fieldwork by various workers and of remote sensing studies. Furthermore, it is representative of the actual distribution of faults, and thus does not reflect a spatial bias in the collection of field data (i.e. only a few faults mapped in the nearly inaccessible northeastern part of the island). From this map, as well as from the topographic and bathymetric map of the island (Fig. 2), it is evident that faults in Nisyros have the following characteristics: (1) They are short, about 2km long, and their traces are lost both close to the coast, or occasionally close to the 100km bathymetric contour, and beneath recent material of the caldera. (2) They are generally of 'scissors-type' geometry in that their throw is greatest close to the caldera rim (up to 100-150m) but decreases rapidly away from it, and in most cases disappears close to the coast. However, scissor-type faults with a different sense of displacement gradient (polarity) mapped as well, and two of these, the Palaiokastro and Langadha Faults, are examined in more detail below. (3) They are arranged in a radial pattern, with their frequency being much higher in the northwestern part of the island where the largest and most numerous domes also are found. It is argued here that the radial pattern of scissors-type faults, and their concentration in
the northwest of the island, can be most easily explained if faulting is regarded as a secondary effect of volcanic doming, i.e. a result of strain induced by vertical pressure (magma ascent). In the following section, faults in this northwestern domain, the Mandraki area, are considered in more detail.
Faulting and uplift in the Mandraki area: archaeological and geomorphological evidence In the northwestern corner of Nisyros, next to Mandraki village, faults seem to deviate from the general pattern described above, that is, they are not radial to the caldera and although they are of scissors-type, their throw increases towards the coast. Archaeological and geomorphological evidence is here used to shed light on the nature and kinematics of two of these faults: the Palaiokastro and Langahda Faults.
Palaiokastro Fault: archaeological evidence of activity Palaiokastro hill, south of Mandraki, is crowned by the ruins of an ancient Greek (Classical) castle built of well-hewn blocks of basaltic andesite (Fig. 5a). The construction appears to have been completely neglected in the archaeological literature, but from building style it seems to date to the fourth century BC (R. Tommlinson, pers. comm.). A southern line of fortifications comprises a wall of 3.5m thickness with rectangular ramparts on its external side and bulky staircases on its eastern side. These ancient remains are in an excellent degree of preservation in the southeast, close to the castle gate, where the topography is relatively gentle. In contrast, they are very poorly preserved further west, where the walls are built along the crest of an escarpment (Fig. 5b) whose height increases seaward up to 110m (Fig. 6). In addition to this change in the general state of preservation, the degree of damage similarly varies. Close to the gate, at the well-preserved southeastern corner of the castle, both the main wall and the first two ramparts of the castle show no signs of fracture (Fig. 5a). The situation changes at the third rampart (counting from east to west; rampart C in Fig. 7), where a slight southward tilting and contrasting building styles (Fig. 8) are suggestive of damage and subsequent partial repair, possibly in the Classical period. This rampart is situated on a 2m scarp.
FAULT PATTERN OF NISYROS VOLCANO
391
Fig. 5a. View of the southeast corner of the Palaiokastro fortifications, made of well-hewn blocks of basaltic andesite. Rampart A, seen on the left, is coincident with the trace of the Palaiokastro Fault, which runs along the south line of fortifications. The throw of the fault is small close to this point but increases abruptly to the west (seaward). (b)
Fig. 5b. The topographic depression, controlled by the Palaiokastro fault.
Signs of damage indicative of dynamic seismic effects (i.e. opening of joints as a result of seismic vibrations (see Stiros 1996)) exist at the nearby staircase in the inner part of the castle, as well as in the well-preserved east line of fortifications. Damage becomes more intense at the fourth rampart (rampart D in Fig. 7), which
bears clear signs of repairs and exhibits a vertical fissure that has caused a horizontal and vertical offset of 15-20 cm (Fig. 9). This rampart is slightly backtilted northwards whereas the fifth rampart (E in Fig. 7) is tilted slightly southward. However, this last rampart is badly preserved, and further west, where the escarpment is
392
S. C. STIROS
fault with a 'scissors-type' geometry, as its topographic expression is nearly zero at the eastern corner of the southern line of fortifications but increases westward until, close to the coast, it has a throw of c. 110m. Because the degree of preservation of the fortifications decreases with increasing fault throw, fault slip appears to be a likely cause of the pattern and style of damage (and repair) observed in ramparts D and E, and in the near-total destruction of fortifications in the southwest. It is likely that erosional instability along the fault scarp would have added to the effects induced by fault slip. However, although the trace of the Palaiokastro Fault is slightly concave, there are no signs of features that would explain the structure as a landslide or other local instability effect (e.g. an arcuate morphology, bulges in the lower part of the downthrown block testifying to rotational movement, transverse scarps, etc.).
Langadha Fault
Fig. 6. Topography in the area of the Palaiokastro and Langadha Faults (dotted lines). A bold line indicates the best preserved ruins of the Palaiokastro castle. Partly based on the 1: 5000 Hellenic Military Geography Survey topographic map. An arrow points to the site of Fig. 11.
several tens of metres high, only poor signs of fortifications are preserved. The escarpment along the southern line of fortifications can be identified with one of the smaller faults of Nisyros, the Palaiokastro Fault (Fig. 6). The 300-250 m long fault is a normal
North of Palaiokastro hill, a hill of 35m height close to Mandraki village is the site of the Speliani Monastery and the ruins of a Medieval castle. The hill is bounded to the east by a normal fault of 1-2 km length, the Langadha Fault (Figs 6 and 10). This fault appears to have a scissors-type geometry, with greatest topographic relief (30m) close to the coast (in the vicinity of Mandraki), and is one if the few nonradial faults on the island. In spring 1997, during a period of seismic unrest in the surrounding area, a fissure about 200m long was observed cutting paved road surfaces and certain houses along the approximate trace of the Langadha Fault. Although this fissure appeared to be a reactivation of the
Fig. 7. Schematic plan of the south line of fortifications at Palaiokastro hill, based on interpretation of air photos and field observations. Light shading indicates badly preserved ruins. The trace of the Palaiokastro fault is also shown.
FAULT PATTERN OF NISYROS VOLCANO
393
Fig. 8. Changes in the building style (at least four types of building style are visible) probably indicate seismic destructions and repairs of ramparts C (in the foreground) and D (in the background).
Langadha Fault, trenches excavated to reinforce damaged houses showed that, instead, it was likely to have been caused by a differential response to seismic shaking of the two different lithologies across the fault (alluvium and coastal conglomerates in the hanging wall and pillow lavas in the footwall) (Stiros 1997).
Recent coastal uplift Together, the Palaiokastro and Langadha Faults seem to define a wedge in which the oldest volcanic rocks of Nisyros are uplifted and exposed. At Kochlaki Beach, located within this faultbounded wedge and immediately north of the Palaiokastro Fault, traces of coastal marine fossil vermetids (Dendopoma petraeum and Vermetus triqueteur) can be found encrusting pyroclastic pinnacles up to elevations of at least 3 m above sea level (Fig. 11). Similar vermetid encrustations are found 200 m south of the Palaiokastro Fault, in the hanging wall, but at much lower elevations (about 1 m above sea level). The fossil vermetids correspond to species characteristic of the infralittoral zone, whose upper limit defines a line that corresponds to the 'Biological Mean Sea Level' (BMSL) (Laborel & Laborel-Deguen 1994). Above this lies the mid-
littoral zone, an area of intense bio-mechanical erosion in which, if isolated vermets are found, they will be quickly destroyed. The fragile nature of these rocks and fossils (the fossils can be readily removed by hand), viewed in the context of their exposure to a high-energy wavedominated coast, indicates that these fossils are very 'fresh', up to a few thousand years old. Furthermore, the identification of their elevated position above BMSL suggests that rapid relative coastal uplift permitted these fossils to quickly rise above the midlittoral zone (see Laborel & Laborel-Deguen 1994). In summary, the observed fossils in northwestern Nisyros suggest a series of recent (probably up to a few thousand years old) abrupt coastal uplifts with a cumulative amplitude of 3 m in the footwall of the Palaiokastro Fault, and 1 m in its opposing hanging wall. Thus, relative land uplift rates of the order of 1 mm per year are inferred. Discussion The main concern of previous investigators, inspired from regions free of tectono-magmatic processes, was to identify the main faults of fault directions based on morphological and statistical criteria. This study argues that this approach
394
S. C. STIROS
Fig. 9. Close-up view of the offset (c. 20cm in both horizontal and vertical sense) at rampart D. View to the west.
is not appropriate for Nisyros. One of the reasons is that the fault pattern shows characteristics atypical of tectonic faulting: fault lengths are short (2-4 km) but even faults of 300m length like the Palaiokastro Fault attain significant throws (up to 110m). By contrast, field and experimental studies of volcanic and salt-dome dynamics indicate that a nearly radial pattern of fractures, mostly extensional, dominates (Komuro 1987; Marti et al 1994). In the case of Nisyros Island, known to be a product of recent doming and caldera collapse, such a radial pattern of faulting would be expected. However, as faults of volcanic origin are likely to have taken advantage of pre-existing tectonic discontinuities, the observed deformation is likely to reflect a combination of tectonic and tectonovolcanic effects, with the latter dominating. There are several reasons why a solely tectonic origin for faults in Nisyros ought to be excluded, as follows:
(1) Scale of deformation. Because the island is built of fairly young rocks (mostly <100ka old) there was little time for tectonic displacements of the scale observed to develop. Yet observed displacements involve throws of the order of 150 m on faults 2-3 km long and coastal uplift rates of 1 mm per year on faults up to a few hundred metres long. (2) Distributed deformation. High rates of deformation are, in principle, confined to the dominant faults in a region. In the case of Nisyros, an island 8 km wide, it is difficult to explain the observed offsets in terms of tectonic slip distributed across numerous very small faults of different trends. (3) Evidence of recent doming. The highest postcaldera domes, up to 600m high, and the oldest exposed volcanic formations are concentrated in the northwestern part of the island. This could indicate recent doming of the entire northwestern part of the
FAULT PATTERN OF NISYROS VOLCANO
395
Fig. 10. View of the Langadha Fault, the throw of which is c. 30 m at the coast and diminishes towards the caldera. The fault bounds the Speliani Monastery hill, where the oldest volcanic rocks of the island (pillow lavas and hyaloclastites) are exposed. Ruins of a Crusaders' castle on the hill are visible at the bottom right of the photo, and the Palaiokastro hill is shown in the background (top right). View is towards the southeast. island, possibly followed by the generation of a second system of radial fractures. (4) Geophysical evidence. A detailed study of the magnetic field of the island revealed no major magnetic anomalies related to specific major faults (Thanassoulas & Xanthopoulos 1989). Consequently, there is no evidence for 'major' deep-seated faults of preferred orientations controlling the magmatic evolution of the island. It can be inferred, therefore, that recent doming is associated with a magma chamber at relatively shallow depth. (5) Lack of parallels. In various parts of the Aegean region, two or even more fault trends have been recognized, and indeed the large-scale structure of the central Aegean appears to be controlled by intersecting NW- and NE-trending fault systems (Stiros 1991) (Fig. 1). However, only in Nisyros has a radial pattern of small faults been recognized. Thus, it can be concluded that the style of surface faulting in Nisyros, particularly the radial arrangement and scissors-type geometry, is indicative of tectono-volcanic rather than tectonic deformation, as may be expected in a
young and active volcanic environment. This result is in agreement with experiments showing that doming produces not only radial fractures nucleating at a well-defined central point and propagating outwards, but also fractures that are not perfectly radial (Marti et al. 1994). Therefore, non-radial fractures such the Palaiokastro and Langadha Faults are probably due to the deviations in the dome geometry, to conjugate sets of extensional shear fractures or to pre-existing discontinuities. Implications for the volcanic risk in Nisyros A first attempt at assessing the volcanic risk in Greece was made in 1986, and the Nisyros volcano was classified as a volcano of medium risk. A multidisciplinary surveillance network, somewhat similar to that operational in Santorini (Casale et al. 1998) was proposed for the island (Delibasis et al. 1986). Currently, however, the only noticeable progress on this matter has been a systematic study of coastal changes (unpublished data), of phreatic eruptions and of seismic crises (Vougioukalakis, unpublished data). Nevertheless, this reveals that, in association with magmatic activity, fault activity is
396
S. C. STIROS
Fig. 11. Pyroclastic pinnacles along the south edge of Kochlaki beach (a), in the footwall of the Palaiokastro Fault, are encrusted with fossil vermetid encrustations (b) up to heights of at least 3 m above the present-day sea level. Because these rocks are very fragile and easily credible, the preservation of these marine fossils in them indicates a recent (probably only few centuries old) rapid relative sea level drop. In the background of Fig. 1 la is Speliani Monastery hill and the ruins of a Medieval castle. For location see Fig. 6.
FAULT PATTERN OF NISYROS VOLCANO
likely to have constituted a major threat for the island since antiquity. This paper greatly benefited from reviews by G. Ernst and I. Stewart. Information on the volcanic stratigraphy by G. Vougioukalakis and on the ancient castle by G. Chartophilis, former guardian of the archaeological site, and R. Tommlinson are greatly acknowledged.
References BORNOVAS, J. 1953. On the earthquakes of Nisyros of January 1953. Institute of Geology and Subsurface Research (now Institute of Geology and Mineral Exploration, IGME), unpublished report E334 [in Greek]. CASALE, R., FYTICAS, M., SIGVALDASSON, G. & VOUGIOUKALAKIS, G. (eds) 1998. The European Laboratory Volcanoes. European Commission, Volcanic Risk, EUR 18161 EN. DELIBASIS, N., LEVENTAKIS, G., PAPADOPOULOS, G., PAPPIS, I., STIROS, S. & FYTICAS, M. 1986. Study of volcanic risk in Greece. Unpublished report, Organisation for Antiseismic Planning and Protection, Athens [in Greek]. Di PAOLA, G. 1974. Volcanology and petrology of Nisyros island (Dodecanese, Greece). Bulletin of Volcanology, 38, 944-987. FYTICAS, M., INNOCENTI, F., MANETTI, P., PECCERILLO, A. & VILLARI, L. 1985. Tertiary to Quaternary evolution of volcanism in the Aegean region. In: DIXON, J. & ROBERTSON, A. (eds) The Geological Evolution of the Eastern Mediterranean. Geological Society, London, Special Publication, 17, 687-689. KALLERGIS, G. & MORPHIS, A. 1968. The earthquakes of November 1968 at Kos-Nisyros. Institute of Geology and Mineral Exploration (IGME), Athens, unpublished report El590. KOMURO, H. 1987. Experiments on cauldron formation: a polygonal cauldron and ring fractures. Journal of Volcanology and Geothermal Research, 31, 139-149. LABOREL, J. & LABOREL-DEGUEN, F. 1994. Biological indicators of relative sea level variations and of co-seismic displacements in the Mediterranean region. Journal of Coastal Research, 10, 395-415. MARINI, L., PRINCIPE, C, CHIODINI, G., CIONI, R., FYTICAS, M. & MARINELLI, G. 1993. Hydrothermal eruptions of Nisyros (Dodecanese, Greece). Past events and present hazard. Journal of Volcanology and Geothermal Research, 56, 71-94.
397
MARTI, J., ABLAY, G., REDSHAW, L. & SPARKS, S. 1994. Experimental studies of collapse calderas. Journal of the Geological Society, London, 151, 919-929. NAKUMURA, K. 1977. Volcanoes as possible indicators of tectonic stress orientation - principle and proposal. Journal of Volcanology and Geothermal Research, 2, 1-16. & UYEDA, S. 1980. Stress gradient in arc-backarc regions and plate subduction. Journal of Geophysical Research, 85, 6419-6428. PAPANIKOLAOU, D., LEKKAS, E. & SAKELLARIOU, D. 1991. Geological structure and evolution of Nisyros volcano. Bulletin of the Geological Society of Greece, 25, 405-419. PAPAZACHOS, B. & PAPAZACHOU, C. 1989. Earthquakes in Greece. Zitis, Thessaloniki. ST SEYMOUR, K. & VLASSOPOULOS, D. 1989. The potential for future explosive volcanism associated with dome growth at Nisyros, Aegean volcanic arc, Greece. Journal of Volcanology and Geothermal Research, 37, 351-364. SIMAIAKIS, K. 1992. Neotectonic study of Nisyros. Institute of Geology and Mineral Exploration (IGME), Athens, unpublished report, 6843. STIROS, S. 1991. Heat flow and thermal structure of the Aegean Sea and the Southern Balkans. In: CERMAK, V. & RYBACH, E. (eds) Terrestrial Heat Flow and Lithosphere Structure. Springer, Berlin, 395-416. 1996. Identification of earthquakes from archaeological data: methodology, criteria and limitations. In: STIROS, S. & JONES, R. E. (eds) Archaeoseismology. Fitch Laboratory Occasional Paper,?, 129-152. 1997. 'Unusual' post-seismic repairs of traditional houses in Nisyros. Information Bulletin, Technical Chamber of Greece, 1982, 28 and 30 [in Greek]. & VOUGIOUKALAKIS, G. 1996. The 1970, Yali (SE edge of the Aegean volcanic arc) earthquake swarm: surface faulting associated with a small earthquake. Annales Tectonicae, 10, 20-30. THANASSOULAS, C. & XANTHOPOULOS, N. 1989. Geophysical study of Nisyros Island (gravimetry— magnetics). Institute of Geology and Mineral Exploration (IGME), Athens, unpublished report, 5966. VOUGIOUKALAKIS, G. 1993. Volcanic stratigraphy and evolution of Nisyros Island. Bulletin of the Geological Society of Greece, 28, 239-258. 1998. In the Blue Volcanoes; Nisyros. Nisyros Council [in Greek].
This page intentionally left blank
The geological origins of the oracle at Delphi, Greece J. Z. DE BOER1 & J. R. HALE2 1
Department of Earth & Environmental Sciences, Wesley an University, Middletown, Connecticut, USA ' College of Arts & Sciences, University of Louisville, Louisville, Kentucky, USA Abstract: Ancient authors from Plato to Pausanias have left descriptions of Delphi's oracle and its mantic sessions. The latter were interpreted as events in which the Pythia (priestess) placed herself on a tripod over a cleft (fissure) in the ground below the Apollo temple. Here she inhaled a vapour rising from the cleft, and became inspired with the power of prophecy. French archaeologists who excavated the oracle site at the turn of the century reported no evidence of either fissures or gaseous emissions and concluded that the ancient accounts were myths. As a result, modern classical scholars and many archaeologists reject the ancient testimonies concerning the mantic sessions and their geological origin. However, the geological conditions at the oracle site do not a priori exclude the early accounts. A major WNW-ESE fault zone and a minor swarm of NNW-SSE fractures intersect below the site. These intersection(s) provided pathways for rising ground water, including a spring below the Apollo temple. The faults broke through a bituminous limestone formation at relatively shallow depth. Hydrocarbon gases that originated in this formation may have escaped during and after seismo-tectonic events. Such gases can induce mild narcotic effects. It is highly probable therefore that the Pythia's inspiration resulted from the inhalation of light hydrocarbon gases, which rose along a fissure (fracture) in the adyton below the Apollo temple.
The ancient Greeks believed that the power of the Delphic oracle derived from the geological setting of the sanctuary. According to a number of Greek and Roman authorities, the women who spoke the prophecies at Delphi sat on a tripod that spanned a cleft or fissure in the rock within the temple of Apollo. Vapours rose from this chasm into the inner sanctum or adyton, where they intoxicated the preistess and inspired her prophecies. The ancient testimonies have been challenged during the past half-century, as modern archaeologists failed to locate any cleft or source of vapours within the foundations of the ruined temple and therefore concluded that the ancient sources must have been in error. However, a recent geological study of the sanctuary and adjacent areas has shown that the preconditions for the emission of intoxicating fumes are indeed present at Delphi. These findings suggest that Greek and Roman texts may preserve an accurate record of ancient geological events and phenomena. Bronze Age Greeks established a religious centre at Delphi in central Greece before 1200 BC, but the famous oracle can be traced back with certainty only as far as the eighth century BC. Tradition named Gaia (Earth) followed by Poseidon (Earthshaker) as gods originally worshipped at Delphi, but they were succeeded by Apollo, god of prophecy (Parke 1956). The oracle of
Apollo at Delphi became the most prestigious in the Mediterranean. Gifts from those who consulted the oracle made the mountain village of Delphi one of the richest sites in Greece. Down to the end of the fifth century BC the oracle was consulted before new colonies were founded, wars were declared, or changes of government were made. The woman who served as prophet, known as the Pythia, held her mantic or prophetic sessions only nine times during the year, on the seventh day after each new moon except during the cold, rainy months of winter. The historian Diodorus of Sicily (c. 90-20 BC) recorded a local legend that attributed the discovery of the Delphic chasm to goats grazing on the limestone slopes before the site was settled: 'The herdsman in charge of the goats marvelled at the strange phenomenon and having approached the chasm and peeped down it to discover what it was, had the same experience as the goats, for the goats began to act like beings possessed and the goatherd also began to foretell future events. After this as the report spread among the people of the vicinity concerning the experience of those who approached the chasm, an increasing number of persons visited the place and, as they all tested it because of its miraculous character, whosoever approached the spot became inspired. For these reasons the oracle came to be regarded as a
From: McGuiRE, W. G., GRIFFITHS, D. R., HANCOCK, P. L. & STEWART, I. S. (eds) The Archaeology of Geological Catastrophes. Geological Society, London, Special Publications, 171, 399-412. 1-86239-062-2/00/ $15.00 © The Geological Society of London 2000.
400
J. Z. DE BOER & J. R. HALE
marvel and to be considered the prophecy-giving shrine of Earth' (Diodorus Siculus, 1952). The geographer Strabo (c. 64 BC-AD 25) also wrote a description of the oracle that was instituted at Delphi, and the geological conditions associated with it: 'They say that the seat of the oracle is a cave that is hollowed out deep down in the earth, with a rather narrow mouth, from which arises breath (pneuma) that inspires a divine frenzy; and that over the mouth is placed a high tripod, mounting which the pythian priestess receives the breath and then utters oracles in both verse and prose' (Strabo 1927). Many other writers referred to the chasm and vapour, including Cicero and Pliny the Elder. One source added an additional element to the geological background of the Apollo temple and its surroundings: 'They say that the water of Cassotis spring plunges underground and in the adyton (inner sanctum) makes the women prophetic' (Pausanias 1935). Pausanias was a travel writer and religious scholar of the second century AD, who described the site and the monuments of Delphi, including the Cassotis spring on the slope above the temple of Apollo. The tradition of a cavern or narrow cave-mouth linked to the oracle of Apollo goes back much earlier, however, to the poets Pindar and Aeschylus in the fifth century BC. Even the philosopher Plato (c. 429-347 BC) made a cryptic reference to 'the rock' as the source of prophetic power at Delphi: 'It was when they were mad that the prophetess at Delphi and the priestesses at Dodona achieved so much for which both states and individuals in Greece are thankful; when sane they did little or nothing . . . The authorities of the temple of Zeus at Dodona say that the first prophetic utterances came from an oak tree. In fact the people of those days were content in their simplicity to listen to the oak tree and the rock, provided these spoke the truth' (Plato 1989). The most valuable ancient source on the Delphic oracle is the Greek philosopher and essayist Plutarch (c. AD 50-120), who actually served at one time as priest of Apollo in the temple at Delphi. Plutarch was thus not only an eyewitness of the Pythia's performance, but also an insider familiar with the oracle's operations and traditions. The story of the goatherd discovering the chasm was known to him, and he actually cites the name of the man involved, a certain Koretas. Plutarch also knew that the Pythia drank water from the spring before each prophetic session. Shortly before Plutarch arrived at Delphi an extraordinary incident had occurred, which was reported to him by the temple officials. Plutarch
relates the story to make the point that the mind of the prophet must be in a receptive state for the vapour (pneuma) to have its proper effect. A deputation from abroad had arrived at Delphi to consult the oracle, and the priests in their eagerness to please these visitors ignored the ill omens and compelled the Pythia to prophesy: 'She went down into the oracle unwillingly, they say, and half-heartedly; and at her first responses it was at once plain from the harshness of her voice that she was not responding properly; she was like a labouring ship and was filled with a mighty and baleful spirit. Finally she became hysterical and with a frightful shriek rushed towards the exit and threw herself down, with the result that not only the members of the deputation fled, but also the oracle-interpreter Nicander and those holy men that were present. However, after a little, they went in and took her up, still conscious; and she lived on for a few days' (Plutarch 1936). In old age, Plutarch wrote an essay known as De defectu oraculorum (On the obsolescence of the oracles) in which he speculates on the reasons for the weakening of the prophetic power at Delphi. His explanation is expressed in almost purely physical and geological terms. First, he considers how veins of silver, copper and asbestos may disappear when they have been exhausted through mining operations. Then he suggests that this disappearance may serve as an analogy for the disappearance of the vapour: 'Plainly the same sober opinion is to be held regarding the spirits (pneumatd) that inspire prophecy; the power that they possess is not everlasting and ageless, but is subject to changes. For excessive rains most likely extinguish them, and they probably are dispersed by thunderbolts, and especially, when the earth is shaken beneath by an earthquake and suffers subsidence and ruinous confusion in its depths, the exhalations shift their site or find completely blind outlets, as in this place [Delphi] they say that there are still traces of that great earthquake which overthrew the city' (Plutarch 1936). In spite of these changes, Plutarch had himself experienced an unusual phenomenon in the adyton of the Apollo temple, which he offered as evidence that the exhalations still emerged at the site of the oracle although they fluctuated in intensity: 'I think, then, that the exhalation is not in the same state all the time, but that it has recurrent periods of weakness and strength. Of the proof on which I depend I have as witnesses many foreigners and all the officials and servants at the shrine. It is a fact that the room in which they seat those who would
GEOLOGICAL ORIGINS OF THE DELPHIC ORACLE consult the god is filled, not frequently or with any regularity, but as it may chance from time to time, with a delightful fragrance coming on a current of air which bears it towards the worshippers, as if its source were in the holy of holies (adyton)' (Plutarch, Moralia 1936). Not one ancient source denied the existence of the chasm and vapour at Delphi, and thanks to Plutarch the statements of those writers who had not actually visited Delphi themselves are supported by the first-hand evidence of an important official within the temple hierarchy. Yet the ancient testimony concerning the geological origins of the Delphic oracle has been almost universally rejected by modern classical scholars and archaeologists. One influential monograph on Delphi states that 'geologically it is quite impossible at Delphi where the limestone and schist could not have emitted a gas with any intoxicating properties' (Parke & Wormell 1956, p. 22). Another declares that 'a close study of all reliable evidence for Delphic mantic procedures reveals no chasm or vapors, no frenzy of the Pythia, no incoherent cries interpreted by priests' (Fontenrose 1978, p. 10); an opinion echoed by Morgan (1990) and Maass (1993). The primary source of the current scholarly opinions is a book on Delphic prophetic procedure by Amandry (1950), who cited the failure of archaeologists to discover a chasm under the Apollo temple. In addition, from a geological standpoint, he noted that the area around Delphi was not volcanic, and declared that it was therefore incapable of producing mephitic vapours (Amandry 1950, pp. 196-230). This negative view gained widespread acceptance, despite the fact that the French excavators had not actually reached bedrock under the foundations of the temple (Darcque 1991, pp. 689-690). Amandry also failed to consider the possibility of non-volcanic sources of vapours. A recent geological study of a temple of Apollo in Turkey has reopened the question of the credibility of ancient writers such as Strabo and Pausanias. In examining the site of an ancient temple at Hierapolis (Pamukkale), a US team was able to demonstrate that Strabo's description of a chasm filled with misty, toxic vapour was in fact accurate in every detail (Cross & Aaronson 1988). The deadly vapour at the Hierapolis temple of Apollo was identified as carbon dioxide along with a fine aerosol of water from a hot spring. In view of the positive correlation established in this case between modern geological findings and the account of the ancient Greek geographer Strabo, who also reported the presence of a chasm and vapour at Delphi, it seemed worth while to undertake a fresh study
401
of the geological setting and the geophysical phenomena associated with the temple of Apollo at Delphi. The geological setting: new data and interpretations
Regional tectonic setting From a geological viewpoint, the Delphi oracle is located on the northern flank of a major crustal structure, the Korinth rift zone (Fig. 1). This tectonic entity consists of a WNW-ESE trending axial zone of subsidence (the Korinth Gulf) and adjacent elevated blocks which rise in staircase fashion and reach maximal heights of respectively 2378 and 2458m on the summits of Mount Killini (northern Pelopponesos) and Mount Parnassos. The width of the Korinth rift zone increases from west to east. Between the above-mentioned mountains the rift zone is about 70 km wide, of which about half is submerged. The maximal depth of the Gulf is about 400m but the presence of a thick layer of sediments indicates that the crustal blocks underlying these relatively recently accumulated deposits sank to considerably deeper levels. The rift zone's topography is mainly the result of crustal extension in a NNE-SSW direction (Fig. 1). Application of these forces led to crustal thinning, which was followed by regional arching, axial subsidence and pervasive fracturing. The majority of fractures trend WNW-ESE. Those in the northern flank dip predominantly southward (40-70°); those in the southern flank northward (40-70°), forming conjugate sets. Individual fault lengths vary from about 10 to 40km. Motion along the fractures involved downward slip of rock units above the fault plane (hanging block) and contemporary upward motion of units below this plane (foot block). During major seismo-tectonic events, which usually followed long periods (50-150 years) of gradually increasing strain, hanging blocks would slip downward over distances varying from 50 to 150cm and foot blocks rose by isostatic compensation over distances of 5-15 cm (Jackson et al. 1982). Energy released during such tectonic events caused both earthquakes and frictional heating of rock units adjacent to the fault. The geological process that gave rise to the Korinth rift zone is believed to have been initiated in early Pliocene time (c. 5 Ma ago). In preceding geological times a crustal mass that used to be a huge submarine limestone plateau was thrust westward (and up) out of the subtropical
Fig. 1. Central Greece, with the Korinth (KG) and Evian Gulfs (riftsystems). (A) Athens; (AE) Aegina volcanic complex; (B) Boura; (D) Delphi; (H) Helice; (Li) Likades volcanic complex; (M) Megalopolis; (Mi) Milos volcanic complex; (P) Perachora peninsula; (Po) Poros volcanic complex; (S) Sparta; (Z) Zakinthos island. Arrows indicate principal directions of crustal extention.
Fig. 2. Topography of the Delphi region and principal tectonic elements. (W) western and (E) eastern exposures of Delphi faultzone. Dots indicate principal course of Pleistos river.
GEOLOGICAL ORIGINS OF THE DELPHIC ORACLE
403
waters of the Tethys (palaeo-Mediterranean sea) in which it had slowly accumulated. About 5 Ma years ago regional stress fields changed and this mass was broken up into long crustal slivers, which moved sub-vertically with respect to each other and formed the Evian and Korinth rift zones.
Stratigraphy One of the most detailed geological studies of the oracle site and adjacent mountains was carried out by Birot (1959). Five years later Aronis & Panayotides (1964) published the geological map of the Delphi Quadrangle. More recently, an excellent geological analysis of the Delphi region was published by Pechoux (1992). Aronis & Panayotides concluded that the basic geological stratigraphy of the Delphi region is relatively simple. The oldest (and lowermost) rock unit is composed of rather massive (locally bedded), white, microcrystalline limestones of Tithonian to Cenomanian age (152-90 Ma). This up to 400m thick unit is overlain by rather compact and microcrystalline limestones of Senonian to Turonian age (90-84 Ma). Limestones in the lower part of this 80-100m thick unit are bituminous. Locally, the bituminous content of this unit can reach 20% (Aronis & Panayotides 1964). The limestone plateau, to which these Mesozoic formations once belonged, stretched from Iran to Italy. Bituminous units provided much of the oil found now in structural traps throughout the Near East and Arabia. Overlying these limestone formations is a sedimentary unit of Paleocene age (66-58 Ma), which is composed of reddish brown and yellowish brown shales, sandstone and conglomerates (cumulatively referred to as flysch). These terrigenous deposits, which reflect regional uplift and erosion, are found mainly in valleys such as that of the Pleistos river.
Local tectonic setting The topography of the Delphi region sensu lato is shown in on Fig. 2. Two tectonic lineaments stand out: WNW-ESE and NNW-SSE trending fault zones, which intersect below the oracle site. These lineaments also clearly show on the satellite image (Fig. 3). The image further shows several WNW-ESE trending fault zones intersecting the Pangalos peninsula, reflecting a downward (to the Gulf) stepping series of parallel faults.
Fig. 3. Satellite photo of Delphi region. Arrows indicate trends of principal tectonic lineaments.
WNW-ESE fault: evidence and interpretation. The oracle site is located on or in limestone debris that has accumulated predominantly in the last million years (the Pleistocene period) at the foot of the Phaedriades cliffs, and on yellowish brown shales of the flysch formation. The latter Paleogene deposits lie in contact with the much older Tithonian-Cenomanian limestone formation, which dominates the cliffs behind (to the north of) the oracle. Such juxtaposition of younger and older geological formations is possible only if the former moved down and/or the latter rose with respect to the other. This motion occurred along a major WNW-ESE trending fracture referred to as the Delphi fault zone (Fig. 2). Because of an increase in the tilt angle of some flysch layers near the fault zone, Birot (1959) assumed this fault to dip northward in the area of the oracle. However, the phenomenon he observed is due to reverse drag along a listric (curved) fault zone that clearly dips southward. Pechoux (1992) showed this southward dipping fault on all his cross-sections and in addition found evidence that the fault intersects most if not all glacial and interglacial Pleistocene deposits. Our study supports Pechoux's observations and indicates that the eastern segment of the Delphi fault zone experienced down-dip slip motion after deposition of the most recent (Wiirm) glacial deposits and overlying soils. This indicates that the fault zone should be classified as active, and it is at present in repose. The Delphi fault zone is well exposed both to the east (1.5 km) and west (2.5 km) of the oracle site (Fig. 4a and b). Fault scarps can be traced virtually to the oracle's borders, where the fault
404
J. Z. DE BOER & J. R. HALE
Fig. 4a. Exposure of curvilinear Delphi fault west of Delphi.
Fig. 4b. Exposure of curvilinear Delphi fault east of Delphi. Note abrupt discontinuation of Pleistocene and Holocene sediments against faultplane.
zone disappears below debris from the Phaedriades and the oracle's ruins. The exposed fault segments are curvilinear, which is characteristic for most normal faults intersecting limestone formations in central Greece (de Boer 1992). Figure 5 shows a plot of normals to different fault-plane segments and their variable trends.
Also shown are the slip directions of the latest (youngest) tectonic event along the fault segments. It is obvious that fault attitudes and slip directions are identical and that the fault segments exposed east and west of the oracle site form part of a single through-going fault zone. Cumulative dip-slip along the Delphi fault is
Fig. 5. Sterographic plots of western (W) and eastern (E) exposures of Delphi fault. Dots represent lower hemisphere plots of normals to faultplane segments. Squares indicate slip directions. Curved lines are plots of faultplane segments with maximal variation from common trend.
GEOLOGICAL ORIGINS OF THE DELPHIC ORACLE estimated to amount to several hundred metres (Pleistos block down and Phaedriades block up), and to have occurred by relatively small (50150cm) offsets over millions of years. NNW-SSE faults: evidence and interpretation. During fieldwork in 1981, de Boer noticed the presence of a major NNW-SSE fault zone in the limestone plateau south of the Pleistos valley (Figs 2 and 3). Upon crossing the latter valley the fault appears to splay into several NNWSSE trending faults that intersect the oracle site. Birot (1959) mapped two of these faults along the eastern and western boundaries of the oracle site sensu lato. The most important in this fault swarm follows the Delphusa stream. Figure 6 shows a plot of NNW-SSE faults (attitudes and slip directions) that intersect the limestone cliffs at the oracle site. The faults clearly form conjugate sets, dipping both ENE and WSW. Their slip directions indicate downward motion of the hanging blocks. Such fault sets result from ENE-WSW crustal extention (de Boer 1992). Evidence for their age is sparse. They do not clearly intersect glacial and post-glacial deposits as does the Delphi fault. However, an interesting offset occurs in the northwestern stand of the Stadium (F, in Fig. 7). The northeastern segment of its foot wall has moved about 14cm downward and 40cm southward with respect to its southwestern section. Shear fractures in the stone building blocks close to this offset trend NNW-SSE and dip predominantly ENE. Such deformation could have been caused by rather localized oblique slip along a buried NNW-SSE fault and suggests a young age.
405
The best information concerning the age of the NNW-SSE faults, however, is provided by seismo-tectonic events elsewhere. Sparta was destroyed in 464 BC by earthquakes related to dip-slip along a NNW-SSE trending fault separating this town from the Taygetos Range, and recently (1986) Kalamata was severely damaged by quakes originating along a NNWSSE fault on the west side of the Taygetos mountains (Fig. 1). The tectonic forces responsible for such faulting are extensional and trend ENE-WSW. The NNW-SSE faults therefore should also be considered active although at present in dormant state. Present-day crustal extension in a NNE-SSW direction dominates from the northern Pelopponesos northward, but ENE-WSW crustal extension predominates in the southern Pelopponesos (de Boer 1992). Delphi is located in a crustal zone where NNE-SSW crustal extension predominates. The ENE-WSW extensional stress, however, is present but plays a minor role and is responsible for dip-slip along NNW-SSE fractures. Presence of springs The tectonic data show that the Delphi oracle site is located at the intersection of a major WNW-ESE, southward dipping fault zone and a swarm of minor NNW-SSE trending, predominantly east dipping fractures. Intersections of 'young' faults provide subvertical pathways through which ground water and/or gases can rise to the surface. Although little water surfaces at present at the oracle site,
Fig. 6. Stereographic plot of NW trending fracture in limestone in eastern (E) and western (W) part of oracle site. Dots represent lower hemisphere plots of normals to the fractures. Squares indicate slip directions. Curved lines are plots of mean faultplanes.
406
J. Z. DE BOER & J. R. HALE
Fig. 7. Location of springs on the oracle site from Muller (1992). (1) Kerna spring; (2) unnamed springhouse; (3) the well of the Muses; (4) location of thick travertine crusts on Iskhegaon wall (see Fig. 9). (F) Site of offset in stadium wall. Full lines, possible extent of NW (NNW) faults. natural and anthropogenic evidence exists for at least six springs (Muller 1992). Their sites occur in two groups (Fig. 7). The three springs in the eastern group are aligned in a NW-SE direction. A fourth spring located below the Apollo temple (Hansen 1992) plots on the same alignment. These springs are, from NW to SE, the Kerna (1), which emerges from below a large mass of cemented limestone fragments; an unnamed spring (2), which must have fed the spring house northeast of the theatre; and the well of the Muses (3), directly south of the Apollo temple. The spring (4) below the Apollo temple emerges from what appears to be bedrock below the southeastern terrace of the Apollo temple (Hansen 1992, section 4) (Fig. 8). This spring, or a spring in close proximity to it, probably represents the (original) Cassotis well from which the Pythia drank before her mantic sessions. Additional evidence for the emergence of significant
volumes of ground water along this lineament is provided by relatively thick encrustations of travertine (Fig. 9), which were deposited on the massive retaining wall (Iskhegaon) northwest of the Apollo temple (Fig. 8). These travertine crusts are clearly the result of a decrease in the solubility of calcium during cooling of ground waters, which splashed across the wall when earlier outlets further downhill (i.e. below the temple) were blocked. (This travertine deposit, which formed in situ, should not be confused with the travertine (poros) building blocks brought from the Kastri mill site near the Kephalovrisi spring.) The presence of NNW-SSE faults and the NW-SE arrangement of springs and/or sites of ground-water emergence is too coincidental to be ignored. The waters of these springs probably accumulated on the plateau of Livadi, drained through fractures in the Phaedriades limestone
GEOLOGICAL ORIGINS OF THE DELPHIC ORACLE
407
Fig. 8. Apollo temple by Hansen (1992). (3) The well of the Muses; (4) spring below temple. (F) Fractures in temple floor limestone blocks (see Fig. 10), and direction of responsible stress (arrow).
massif, accumulated in the Delphi fault zone and rose along its intersection with a NNW-SSE fault, following this structure downhill.
Possible presence of gases in spring water (s) Could gases have risen with the waters of the springs surfacing on the oracle site? Previous researchers are correct in stating that Delphi is not located in a volcanic region, hence no volcanic gases would or could be expected to have risen below the temple (Fig. 1). However, faults provide pathways for gases in both volcanic and non-volcanic regions. Dominant among the gases that surface in the eastern Mediterranean and Near East are carbon oxides and hydrocarbons. The limestones below the oracle site contained (and possibly still hold) hydrocarbon gases, which formed during burial and diagenesis of its biogenic constituents. It is reasonable to assume that these gases rose along fault zones, especially in periods after seismic and tectonic agitation. Oil and gas seeps along faults are common in the Near East, but also in the eastern Mediterranean and Greece. Tar pits occur on the Greek
island of Zakinthos and in the Peloponnesos (Megalopolis) (Fig. 1). Herodotus (c. 485-425 BC) wrote: 'I myself have seen something similar in Zacynthus, where pitch is fetched up from the water in a lake. There are a number of lakes - or ponds - in Zacynthus, of which the largest measures seventy feet each way and has a depth of two fathoms. The process is to tie a branch of myrtle on to the end of a pole, which is thrust down to the bottom of this pond; the pitch sticks to the myrtle, and is thus brought to the surface. It smells like bitumen, but in all other respects it is better than the pitch of Pieria. It is then poured into a trench near the pond, and when a good quantity has been collected, it is removed from the trench and transferred to jars. Anything that falls into this pond, passes underground and comes up again in the sea, a good half mile distant' (Herodotus 1954). If oil and hydrocarbon gases emerged along NNW-SSE trending fault zones on Zakinthos (and near Megalopolis) the same phenomena could have occurred in Delphi, given the similarities in geological stratigraphies and structure. To understand what gases could have risen from the bituminous limestone below the oracle site, reference should be made to active gas seeps
408
J. Z. DE BOER & J. R. HALE
Fig. 9. Travertine deposits on Iskhegaon wall. Note stalagtites and limestone block lodged in travertine after its decollement.
in the Gulf of Mexico. These gases originated in bituminous limestones and rose along faults in the form of gas hydrates. The geological and tectonic setting in this region is thus similar to that in Delphi except that the emissions occur at shallow depths below sea level. Gas hydrate concentrations in the Gulf of Mexico act as a kind of pressure relief system, alternately checking and releasing the flow of hydrocarbons along localized vents (MacDonald et al. 1994). The accumulated volumes escape when the plug of gas hydrate either dislodges because of excess buoyant force or disassociates because the water temperature rises above the limit of hydrate stability (MacDonald et al. 1994). Seismic activity would or could trigger the former activity. Gas release from a vent studied in detail was intermittent, and occurred only during periods of relatively elevated water temperature. With respect to this behaviour it is of interest to note that the mantic sessions at the Delphi oracle were never held during the winter months, when Apollo was believed to have gone north to the land of the Hyperboreans. This suggests that gas emissions at Delphi may have diminished during the colder periods when much of the water had accumulated on Parnassos as snow and ice, and ground-water temperatures were relatively low.
As the temperature of the ground water rose during spring, more of the gas it had incorporated was released. The gases emitted by the Gulf of Mexico vent are composed of methane (88% CH4), ethane (c.8% C2H6), propane (c.2% C3H8), rc-butane (c. 1%) and traces of iso-butane, iso-pentane and w-pentane. Methane, ethane and butane are colourless and odourless, and can produce (mildly) narcotic effects if inhaled in sufficient concentrations (Windholz 1983). No mention was made of the possible presence of ethylene (C2H4). This gas should be emitted by frictional heating of the bituminous limestone during seismo-tectonic activity. Ethylene is colourless and smells somewhat sweet. This gas was used by surgeons for anaesthesia around the turn of the century. When inhaled, induction is rapid, and not disagreeable mental clouding supervenes (Goodman & Gilman 1996). Assuming the adyton below the Apollo temple to have been a small, rather confined space, such gases if rising in relatively small volumes, but over extended periods of time, may have been able to accumulate in concentrations sufficiently high to induce mild narcotic effects. Of interest is the placing of the Omphalus next to the Tripod on which the Pythia sat. Could it have covered an opening in the fissure, holding back the gases to provide higher concentrations during monthly mantic sessions? Bituminous limestones also contain hydrogen sulphide (H2S). This gas is colourless, but smells like rotten eggs. During seismic agitation of an area containing gasenriched limestone, this is probably the gas that is emitted earliest, because it will not form gas hydrates. When Apollo occupied Delphi's oracle he had to slay a female serpent. Her name Python was bestowed on the site because he left her corpse to 'rot'. This myth suggests an early phase of H2S gas emission, which probably followed a seismo-tectonic event along the Delphi fault zone. Analysis of a piece of travertine collected from the Delphi fault zone north of the village (outside of the oracle site), showed traces of sulphur, indicating that sulphuric compounds did indeed rise along the fault. More analyses, however, are needed, especially of the travertine on the Iskhegaon wall. Could seismo-tectonic events have triggered intermittent release of hydrocarbon gases? Plutarch drew the conclusion that the 'exhalation is not in the same state all the time, but that it has recurrent periods of weakness and strength'.
GEOLOGICAL ORIGINS OF THE DELPHIC ORACLE Plutarch's observation can be explained by shortterm fluctuations in gas content during variations in ground-water temperature and/or dilution by heavy rains (increased ground-water flow). To continue gas emission over centuries it is necessary to evoke intermittent seismic agitation of its source. Plutarch's observation that earthquakes influenced the exhalations, clearly indicates that seismic activity played a major role in varying the volume (and site) of gas emissions. Because of changes in the solubility of calcium in enriched ground waters that percolate through fractures and karst from the heights of the Phaedriades, spaces in the fault zone(s) will be slowly but inexorably filled with calcite. To reopen such pathways and increase both porosity and permeability, brecciation is needed, and such a process commonly results from motion along the fault(s). The Delphi fault is clearly an old structure, which possibly dates back to the early phases of subsidence in the Korinth rift zone. Strain release along this fault must have occurred intermittently, but information on such events is sparse. Sometime in the Bronze Age a goatherd, Koretas, detected the presence of intoxicating fumes, when he noticed the aberrant behaviour of his goats. He may have witnessed seismic activity and found a fissure that had resulted from renewed tectonic activity. The earthquakes may have resulted from strain release (slip) along the Delphi fault zone and may have been responsible for a new phase in the release of gases from the underlying bituminous limestone formation. Neeft (1981) believes that an earthquake destroyed the main part of Delphi's settlement around 730 BC. Delphi's period of highest prestige and wealth started around this time and lasted to the end of the fifth century. Activity along the Delphi fault in 730 BC reopened older fissure(s) and caused renewed emission of gases. Although it probably took only a few years before most of the H2S had reached the surface, gas hydrate emissions lasted much longer. Gases activated by the 730 BC event, however, could not have continued to surface for the 12 centuries during which the oracle 'spoke'. Renewed releases of the hydrocarbon gases must therefore have occurred when the Delphi fault was reactivated, but also when motion occurred along faults elsewhere in the western segment of the Korinth rift zone and strong earthquakes passed through the region. The temple complex at Delphi was severely damaged in 373 BC, the same year in which Boura and Helice, small towns along the southern coast of the Gulf of Korinth, were utterly destroyed. These towns were located on the
409
northern (hanging) block of a WNW-ESE trending fault zone, which had developed in the southern flank of the Korinth rift zone (Fig. 1). Downward slip of the hanging block presumably submerged part of Helice (Mouyaris et al. 1992; Lekkas et al 1996) and destruction was completed by liquefaction of sediments and tsunamis. Contemporaneous destruction of the early temple complex at Delphi may have been due to severe shaking by seismic waves emanating from the reactivated Helice fault and/or related falls of rock masses shaken loose from the Phaedriades. However, it is equally possible that strain release along a major fault in the southern flank of the Korinth rift zone increased strain in and caused subsequent slip along the Delphi fault in the northern flank. An example of such a process was provided in 1861-1862. Motion along the Helice fault occurred on 28 December 1861. Surface rupture resulted, indicating downward slip of the northern (hanging) block. Damage as a result of liquefaction, tsunamis and structural collapse was considerable. A second strong earthquake followed on 1 January 1862. Its epicentre was located below the northern shore of the Korinth rift zone. Khoury et al. (1982) believe that slip along a southward dipping fault was responsible for this event. The seismo-tectonic sequence thus appears to have started with strain release along a north dipping normal fault and was followed by strain release along its conjugate, a south dipping fault on the opposite side of the rift zone (Gulf of Korinth). A similar earthquake sequence occurred in 1981 in the eastern segment of the Korinth rift zone (Alkionides Gulf). On 24 February, this region experienced an earthquake with magnitude Ms 6.7. This quake was followed by a major aftershock (Ms 6.4) 5 h later. Elastic strain release responsible for the seismicity and numerous minor aftershocks was due to dip-slip along an ENE-WSW trending north dipping normal fault. The reactivated fault segment ripped upward from a depth of about 10 km and offset the surface on the Perachora peninsula (Fig. 1). The principal quake was felt over a region of about 250 000 km2 and was responsible for generating intensity VIII effects within an area of 1400km2. Ten days later (4 March 1981) an earthquake with MS 6.3 shook the same region. Its epicentre, however, was located further north (and east) than that of the previous event. It and a series of aftershocks were generated along an ENEWSW fault zone that dips south and is located along the northern flank of the Korinth rift zone. The motion allowed for further subsidence of the crustal block in the axial zone of the
410
J. Z. DE BOER & J. R. HALE
rift. Strain release along the north dipping only. Ambraseys (1996) mentioned the occurfault plane thus added strain to the south dip- rence of earthquakes near Delphi with Ms > 6.0 ping fault(s) and triggered its subsequent reacti- in 1580, 1769, and 1870 (two quakes). McKenzie vation. This pattern appears to be rather (1978), Tselentis & Makropoulos (1986), and common for the tectonic evolution of a rift Hatzfeld et al. (1990) described events that zone and probably was repeated many times in shook the region in 1965 and 1970. Time spans Greece's past. between these quakes suggest a recurrence interDestruction of the temple complex at Delphi in val that varies widely. The Helice fault zone, the 373 BC may therefore have been due to actual slip principal fracture on the opposite side of the rift along the Delphi fault, which followed strain zone from Delphi, was reactivated in 373 BC and release along the Helice fault zone. Unfortu- AD 23. No data are available for the interval nately, little historical evidence exists for the between the latter date and the early 18th cen373 BC events at Delphi. Information that does tury, but between 1748 and 1995 reactivation exist is indirect. Agathon and his brothers (acting occurred five times (in 1748, 1817, 1861, 1888 as consultants for the Italian towns of Thourion and 1995) (Lekkas et al. 1996). Recurrence and Tarenton) requested that their privileges periodicity in this interval thus varied from 27 for free access to the oracle be renewed. Their to 107 years. Tectonic activity along the Delphi segment was inscribed on a stele, which suggests fault probably occurred with similar frequency. that the Alcmeonides temple was destroyed and During each of such events new pathways were had not yet been rebuilt. The request and created and additional volumes of gas may have permission thus were incised on a marble stele, been squeezed from the underlying limestone rather than on the temple walls, and could be formation(s). used for reference in the interval of several References to historical landslides may have decades separating destruction and rebuilding of been indicative of offsite seismic activity. The the temple. Persians in 480 BC, the Phocians in 354-352 BC, and the Gauls in 279-278 BC were thwarted in their attempts to ransack the temple complex because of rockfalls. Pechoux (1992) wrote: Recurrence interval of seismogenic events 'Time and again earthquakes had rumbled here, Delphi's earthquake record is very incomplete frightening away the plundering Persians, and and reliable evidence exists for the last century a century later the plundering Phocians, and a century later the plundering Gauls; it was the God protecting his shrine'. The oracle's ruins contain evidence for at least one such event, which caused significant damage. The limestone blocks that form the temple floor
Fig. 10. Conjugate set of shear fractures in limestone block of temple floor (see Fig. 8).
Fig. 11. Stereographic plot of shear fractures in limestone blocks of the temple floor. Dots represent normals to fractures and curved lines the mean fracture trends. Arrow indicates direction of principal stress component responsible for deformation.
GEOLOGICAL ORIGINS OF THE DELPHIC ORACLE
411
enough to close it. The spring waters subsequently had no further outlet below the temple and were forced to emerge along a fault segment further uphill (to the north-northwest). After surfacing, these waters ran downhill and splashed across the Iskhegaon wall, leaving the thick coating of travertine (Fig. 9). This new spring (Fig. 7, no. 2) was also 'silenced' in due time. Very little water emerges at present from the Kerna spring. In 1980 the remains of its spring house were severely damaged by ofTsite seismic activity, and flow appears to have been reduced even more. Conclusions
Fig. 12. Collapsed segment of SW part of the Apollo temple, which is also the area with the greatest density of shear fractures (Fig. 11). Collapse probably resulted from same stress field as that represented by the fractures and indicates the former presence of a space (adyton?) below the temple floor.
are locally intersected by a conjugate set of shear fractures (Fig. 10). The responsible force came from the SW (Fig. 11) and appears to have caused collapse of the southwestern temple section (Fig. 12). Such collapse can have occurred only if there was sufficient space below the temple floor. It appears possible therefore that this was the site of the adyton, the space into which the Pythia retreated for her mantic sessions. This earthquake may have originated in the same fault zone (Patras-Nafpaktos area) as the one that occurred in AD 551. The latter quake was felt in all of central Greece, and caused major damage throughout the Korinth rift zone. Gregor of Nazians (AD 329-390) told Emperor Julian that the Pythia spoke no more and that the Kastalian spring (Cassotis?) had been silenced. It appears logical to assume that the earthquake that did so much damage to the southwestern part of the Temple was also responsible for 'silencing' the spring. The principal reason for such assumption is that the seismic waves originated in the southwest and travelled at right angles to the (NNW-SSE trending?) fissure in the adyton. They appear to have been strong
The following conclusions can be drawn from geological observations on and near the oracle site at Delphi: (1) The Delphi oracle site is intersected by a major WNW-ESE normal fault and a conjugate swarm of NNW-SSE trending fractures. (2) Several springs emerged along the NNW fractures and one (possibly the ancient Cassotis) flowed below the Apollo temple. (3) The oracle site is underlain by Paleogene calcareous shales, which in turn overlie a bituminous limestone formation of late Cretaceous age. Hydrocarbon gases emitted from the latter probably emerged locally along the NNW-SSE fractures. (4) Seismic waves associated with an earthquake that most probably originated in the Korinth rift zone between Itea and Diakofto virtually closed a fissure(s) below the temple, caused local collapse of the temple floor, and resulted in the re-emergence of spring waters further uphill.
Note added in proof Chemical analyses of water samples and travertine deposits in the adyton (carried out in 1989) have shown that the springs on site have in the past and continue at present to emit small volumes of hydrocarbon gases (methane, ethane and ethylene). Light doses of ethylene were used during the early 20th century as a surgical anaesthetic. References AMANDRY, P. 1950. La Mantique apolliniene a Delphes. Boccard, Paris. AMBRASEYS, N. N. 1996. Material for the investigation of the seismicity of Central Greece. In: STIROS, S.
412
J. Z. DE BOER & J. R. HALE
& JONES, R. E. (eds) Archaeoseismology. British School at Athens, Occasional Paper, 7, 23-36. ARONIS, G. & PANAYOTIDES, G. 1964. Geologic Map of the Delphoi Quadrangle. Institute for Geology and Subsurface Research, Athens. BIROT, P. 1959. Geomorphologie de la region de Delphes. Bulletin de Correspondance Hellenique, 83, 258-274. CROSS, T. M. & AARONSON, S. 1988. The vapours of one entrance to Hades. Antiquity, 62, 88-89. DARCQUE, P. 1991. Temple d'Apollon: fouille. Bulletin de Correspondance Hellenique, 115, 689-690. DE BOER, J. Z. 1992. Dilational fractures in the Corinth and Evian riftzones of Greece; their geometrical relation and tectonic significance in the deformational process of normal faulting. Annales Tectonicae, 6, 41-61. DiOCORUS SICUIAJS 1952. The Library of History, Harvard University Press, Cambridge, 308-311. FONTENROSE, J. 1978. The Delphic Oracle. Berkeley, Univ. of California. GOODMAN, L. S. & GILMAN, A. 1996. The Pharmacological Basis of Therapeutics. Macmillan, London, 74-75. HANSEN, E. 1992. Autour du Temple d'Apollon. In: BOMMELAER, J.-F. (ed.) Delphes. Travaux du centre de recherche sur le proche-Orient et la Grece antiques, 12, 125-149. HATZFELD, D., PEDOTTI, G., HATZIDIMITRION, P. & MAKROPOULOS, K. 1990. The strain pattern in the western Hellenic are deduced from a micro earthquake survey. Geophysical Journal, 101, 181-202. HERODOTUS 1954. The Histories, Penguin, London. JACKSON, J., CAGNEPAIN, J., HOUSEMAN, G., KING, G. C. P., PAPADIMITRION, P., SOUFLERIS, C. & VIRIEUX, J. 1982. Seismicity, normal faulting and the geomorphological development of the Gulf of Corinth (Greece). Earth and Planetary Science Letters, 57, 377-397. KHOURY, S. G., TILFORD, N. R., CHANDRA, U. & AMICK, D. 1982. Some problems in the construction and interpretation of isoseismal maps. Report 4 Ebasco Services Screens Boro North Carolina, 27-49. LEKKAS, E., Lozios, S., SKOURTSOS, E. & KRAMIS, H. 1996. Liquefaction, ground fissures and coastline change during the Egio earthquake (15 June,
1955; central western Greece). Terra Nova, 8, 648-654. MAASS, M. 1993. Das antike Delphi: Orakel, Schatze und Monumente. Wissenschaftliche Buchgesellschaft, Darmstadt. MACDONALD, I. R, GUINASSO, N. L., SASSEN, R., BROOKS, J. M., LEE, L. & SCOTT, K. T. 1994. Gas hydrate that breaches the seafloor on the continental slope of the Gulf of Mexico. Geology, 22, 699-702. MCKENZIE, D. P. 1978. Active tectonics of the AlpineHimalayan belt: the Aegean sea and surrounding range. Geophysical Journal of the Royal Astronomical Society, 55, 217-254. MORGAN, C. 1990. Athletes and Oracles. Cambridge University Press, Cambridge. MOUYARIS, N., PAPASTAMATIOU, D. & VITA-FINZI, C. 1992. The Helice fault? Terra Nova, 4, 124-129. MULLER, S. 1992. Delphes Mycenienne : un reexamen du site dans son contexte regional. In: BOMMELAER, J.-F. (ed.) Delphes. Travaux du centre de recherche sur le proche-Orient et la Grece antique, 12, 67-83. PAUSANIAS 1935. Description of Greece, Vol. 4, Harvard University Press, Cambridge, 510-511. PARKE, H. W. 1939. A History of the Delphic Oracle. Blackwell, Oxford. & WORMELL, D. 1956. The Delphic Oracle. Blackwell, Oxford. PECHOUX, P. Y. 1992. Aux origines des paysage de Delphes. In: BOMMELAER, J.-F. (ed.) Delphes. Travaux du centre de recherche sur le procheOrient et la Grece antique, 12, 13-38. PLATO 1989. Collected Dialogues, Princeton University Press, Princeton, 491-520. PLUTARCH 1936. Moralia, Vol. 5, Harvard University Press, Cambridge, 498-499. STABO 1927. The Geography, Vol. 4, Putnam's Sons, New York, 352-353. TSELENTIS, G. A. & MAKROPOULOS, K. C. 1986. Rates of crustal deformation in the Gulf of Corinth (central Greece) as determined from seismicity. Tectonophysics, 124, 55-66. WINDHOLZ, M. 1983. The Merck Index. Encyclopedia of Chemicals, Drugs and Biologicals. Merck, White House Station, N.J.
Index
Page numbers in italics refer to Figures and Tables acid rain/fog 267, 270-73, 275, 277, 307-14 Aegean 3, 25-30, 81-93, 95-6, 99-102, 385-97 aerial photography 146 Aeschylus 400 Aetna 183-4 Africa, North 28 agates 21 agriculture 202, 242; see also subsistence Akrotiri, Thera 71-9, 81-3, 87, 105, 112-19, 123, 128-31, 138-9 Alai, Greece 37-9, 40-42 Alaska 101, 245-65, 323 Aleutian volcanic arc 245, 264 algal blooms 102 Ammianus Marcellinus 99-100 amphibole 170, 251 andesite 17, 18-19, 22, 211, 220, 253, 343, 346, 348, 386 andesitic soils 320, 329, 333 animals, effects of tephra 262-3, 265 Appian 186 Appolonius of Rhodes 183 Apulia, Italy 159-74 Arab chroniclers 357, 359 Arabian shield 355-7 aragonite 192 archaeological assessments, Katmai Park 255 archaeological GIS, Pompeii area 143-58 archaeological sites, newly formed 365-7, 370-71 archaeoseismology 25-30, 45-70 architecture, Mexico 198-203 ARCVIEW 3.0 152 Argolid 29-30, 330 Aristotle 184 artefacts 15-22, 202, 330, 237-9, 241 Atalanti, Greece 25-6, 33-44 Athens, earthquake activity 26-7 augite 167, 327 Australia, soils 324, 326, 327, 328 Aztecs 195, 350 Baru volcano, Panama 226-7 basalt 19, 205-23, 345-52, 357-9 basaltic soils 320, 325-9, 333 biological effects of volcanism 260-64, 269, 270-73, 277, 307, 313, 332-3 biostratigraphy 330 bituminous limestones, Delphi 403, 407-8, 409 bloodstone 21 bog-oak record 268 bone objects 211 breccias 106, 126, 131, 133, 376 Britain, acid fog 313; see also Scotland bronze 91-2, 182 Bronze Age 29-30, 73-6, 268, 269, 340; see also Minoan; Vesuvius, Avellino eruption Brooks River, Alaska 245-65 building stone 1-13, 18-19, 194 buildings, as earthquake evidence 360-67, 373-81, 390-93
burials 73-6, 169, 211, 221 Byzantine 5, 9, 29, 85, 380 Caesarea Maritima 342, 343 calcium 230 Cameiros, Rhodes 341 Cameroon 323, 324, 325 Campania, Italy 168, 169-70, 174, 184 Campi Flegrei, Italy 192, 193, 339 Canary Islands 281-92, 302, 304, 324, 328-9 Cape Verde Islands 281-2, 292-303 carvings 17, 18, 19 Catal Hiiyuk, Anatolia 18, 81 Cato 340 caves, in ash 288 cements, hydraulic 19, 339-43 Cenchreai, Corinth 342-3 ceramics see pottery chalcedony 21 chert, nodular 233, 235 Chiginagak Volcano, Alaska 253 chronology Alaskan tephra 245-65 earthquake events 33, 36-42 Etna eruptions 185-7 Italian Bronze Age 174 Minoan 27, 81, 174 and territorial dynamics 51 volcanism 227-32, 240-41, 268, 307 see also dating chronostratigraphy 226 chryselephantine 89 Cicero 400 cinnabar 20 Classical mythology 180-85 Classical period responses to volcanic events 89-93, 98-100, 179-87 Classical texts 33-44, 61-2, 399-411; see also historical accounts; literary evidence classifications 51, 56-7, 156, 317, 319, 324 Claudian 185 clay deposits, Akrotiri 72-8 clay mineralogy, volcanic soils 320-22, 325-9 clays, firing 340 climate, and volcanic soils 320-29 climatic anomalies 85 climatic change 268, 269-70, 330 clinopyroxene 163, 170, 251, 325 coastal uplift 393 coins, Sagalassos 378, 379, 380 concrete, hydraulic 339-43 Cook Island, Polynesia 324, 327 copper 20, 192, 400 Corinth 28-9, 30, 340, 341, 342-3 cornelian (carnelian) 21 Cornell Halae and East Lokris Project (CHELP) 39 Cosa, Italy 342 Costa Rica 226-7, 240, 243, 321, 323, 324, 325 creation myths, Mexico 195-6 Crete 27, 29, 47, 59, 81-93, 95-102 crisis behaviour, Minoan Crete 85-93
414
INDEX
crystal concentration method (CCM) 166-7, 173 cults/ritual 16, 21-2, 89-93, 98-100, 180-82, 211, 221 Cumbre Viejo volcano, La Palma 281-92, 303 Cyprus 28 dacite 106, 253, 320, 329, 386 dating Akrotiri 75, 78-9 Popocatepetl eruptions 196-8, 223 radiometric 283, 285-8, 330, 331; see also radiocarbon dating Sagalassos 378, 379, 380 using soils 329-332 using volcanism 16, 21-2 Xitle eruption 211, 212, 214, 217-23 see also chronology; luminescence Delphi, Greece 13, 399-411 Demetrius of Callatis 36 Denizli basin, Turkey 3-7 devitrification 251 Dhamar earthquake (1982) 357-65 digital elevation model (DEM) 152-3 Dio 186 Diodorus Siculus 36-7, 42, 44, 183, 185, 399-400 disaster events 85 Djibouti 324, 326 drainage system, Akrotiri 79 dyke emplacement 282-3, 357, 390 earthquakes 11, 28, 33, 36, 40, 193 archaeological evidence 11-13, 25-30, 33, 37-42, 45-70, 373, 377-80 damage surveys 365-7 and faulting 1-13, 33-44, 376, 380-82 hazard assessment 47, 355-71 preceding eruptions 123, 138, 139; see also Minoan repeat times 45-63 seismic indicators 56-61 territorial evidence 47-57, 65-70, 375-7, 380-82 see also archaeoseismology ecofact studies 243 Ecuador 321 El Salvador 226-7, 332 Elba 193 electron microscopy 127, 170, 229-30, 247-9 Empedocles 182-3 energy dispersive spectrometry (EDS) 170 Epirus 30 erosion 78, 332, 333-4; see also weathering Etruscans 180-82 Europe, effects of AD 1783 Laki fissure eruption 270-73, 307, 311, 313, 314 extensional tectonics 3-5 eyewitness accounts, line-of-sight 292, 299-303 famine 85, 174, 275-7 farming, see agriculture faulting 1-13, 33-44, 376, 380-82, 385-97, 401-5, 409-11 fertility, volcanic soils 20 fire 21, 180-84, 198-9 fish, effects of tephra 262-3 fissures 7, 8, 11-13, 312, 313, 385-6, 387, 399-401; see also Laki
flint 211 floors, fracture patterns 376, 378, 379-82 Fogo, Cape Verde Islands 281-2, 292-303 foraminifera, Akrotiri 76-8 forests 241, 242, 268, 269 fossils 76-8, 192, 393 fracture patterns, floors 376, 378, 379-82 France 190, 192, 313 garnet 170 gas hydrates, Gulf of Mexico 408 gases, Delphi 399-401, 407-11 Geikie, Sir Archibald 185 geochemistry, tephra 230 geographical information system (GIS) 143-58 geographical reconstruction 71-9 geomorphology 375-6, 390-93 geothermal activity 312, 313, see also hydrothermal activity Germany 190, 192, 307-12 gold 20, 89 Gortyn, Crete 29 grain size 131, 163-5, 170, 171, 172, 211, 330 Greece earthquake evidence 25-30, 33-44 volcanic soils 324, 326, 332, 333, 334 Greece, Classical Atalanti region 33-42 cement and concrete 19, 340, 341 Delphi oracle 399-411 Straits of Messina 49, 51, 57-63, 63-67 and volcanism 180, 182-4 see also Classical texts greenstones 348, 350 Griggs Volcano, Alaska 251-2, 253 ground-penetrating radar (GPR) 105-19, 198 Guatemala 351 hauyne 171 Hawaii 18, 282, 323, 325, 327, 333 hazard studies 368-70, 395-7 Hekla 268, 269, 270 Heraclitus 182, 183 Herculaneum 133-6, 137, 138-9, 143, 147, 184, 339 Herodotus 6 Hesiod 180 Hierapolis (Pamukkale) 1-13, 380, 401, 440 historic buildings surveys 360-67 historical accounts 283, 289-92, 296-304, 307-14; see also Classical texts; literary evidence hoarding 86, 91-2 holy springs 5, 13, 405-8, 411 hot springs 5, 7, 8, 13, 193, 284, 312, 339 housing, Tetimpa, Mexico 198-203 human sacrifice 99 human skull offering 202 human victims of earthquakes 367, 371 human victims of eruptions 136, 137, 138, 202 Hungary, Lake Balaton 190 hydraulic concretes 339-43 hydrogen sulphide 20, 408, 409 hydrothermal activity 7, 8, 9, 20-21, 386; see also hot springs
INDEX Iceland 190, 192, 267-77, 330, 357 ignimbrites 73, 76, 79, 106, 113-17, 128, 131, 321 ilmenite 251 Ilopango volcano, San Salvador 227, 332 Incas 17, 18-19 Indonesia 85, 330 Ireland 268, 326-7 iron 230 irrigation channels, petrified 8-9 Ischia, Italy 190, 192, 193, 341 Israel 27-8, 324, 329 Italy 131-9, 143-58, 159-74 1980 (Basilicata) earthquake 363-5 archaeoseismology 45-70 earthquake damage surveys 363-5, 366 Johnston-La vis Collection 189-94 see also specific regions ivory 89 jade 211 Japan 321, 323, 331, 332 Java 19 Jericho 27-8 Jerome 61-2 Johnston-Lavis, Henry James 189-94 Julius Caesar 185, 187 Katmai, Alaska 101, 245-65 Kenya 323 Klithi, Epirus 30 Krakatoa 101, 332 Kynos, Greece 26, 40 Kyparissi (Opus), Greece 25-6, 36, 40-42 lahars 198 Laki fissure eruption (AD 1783) 269, 270, 276, 277, 307,311,313 Lamia, Greece 26, 37, 43 land-use patterns 235-7, 239, 241, 242 landscape archaeology 49, 55-6, 243 lapilli 143, 161, 196-203, 245, 249, 291, 293, 298, 301 laser diffraction 330 Laurion, Greece 191-2 lava, uses see volcanic products lava flows 209, 211, 349, 351, 357-9, 386 lava soils 20, 198, 317-19, 324-34 lead 191-2 Lerna, Greece 30 leucite 171 Libanius 61 lichens 325 lime, calcination 339-40 limestones 375-6, 403, 407-8, 409 line-of-sight, eyewitness accounts 292, 299-303 Lipari Islands 190, 192, 193 literary evidence 45-70, 179-87, 92, 189-90, 193; see also Classical texts; historical accounts Lucretius 182, 183, 184, 186 luminescence 217-20, 221-3, 329 magic 182 magnetic susceptibility 220 magnetite 220, 251, 327 manganese 249, 251
415
mapping archaeological sites 145-6 geological 190, 193, 285-8 volcanic deposits 51, 105-19, 229 Marcellinus 186 Martial 143, 184 Martinique 321 Mayans 195, 227, 350 Melos, Greece 343 Messina 45, 55, 57, 60-69 Straits of 47-70 metaphysical associations, stone selection 350-52 Mexico Barcena, AD 1952 eruption 101 Cuicuilco pyramid 205-23 gas hydrates in Gulf 408 Olmec stone sculpture 17, 345-52 Popocatepetl 195-203, 223 Sonora Desert 324, 326, 328 Teotihuacan 196, 223, 350 use of obsidian 18, 212 Xitle volcano 205-23 mica 170 Midea, Greece 29, 30 miltos 343 mineralization 20-21 mineralogy 76, 191-3, 320-22, 325-9 Minoan earthquakes 27, 83, 87, 92, 124, 128-9, 138, 139 volcanic deposits, GPR mapping 105-19 see also Crete; Santorini mobility patterns 237, 239, 241, 242 modelling 27, 84, 268 Monte Somma, Italy 189, 190, 192 Montserrat 333 mortars, hydraulic 339-43 mosaic floor, fracture patterns 376, 378, 379-82 Mt Etna 179-87, 190, 192, 193, 271 soils 324, 325, 329, 333 Mt Pinatubo, 1991 eruption 187 Mt St Helens eruptions 85, 139, 261 Mt Vulture 168, 171,323 Myceneans 29-30, 95, 96 mythology 180-85, 195-6 Naples area (Johnston-Lavis Collection) 189-94 Neolithic 18, 37, 41, 167, 170 nepheline 163, 170, 171 Netherlands, acid fog 270-72, 313 New Zealand 321, 323, 324, 330 Nicaragua 327 NisyrosSl, 385-97 Northern Ireland 326-7 Novarupta eruption (AD 1912) 101, 251-2, 260-64 obsidian 18, 21, 22, 202, 211, 232-7, 242 oceanic island volcanoes 281-304 olivine 209, 325, 327, 346 Olmec stone sculpture 17, 345-52 Olympiodorus 186 onyx 211 ophiolites 375-7 optical microscopy 169 optically stimulated luminescence (OSL) 217-20,221-3
416 Opus (Kyparissi), Greece 25-6, 36, 40-42 Orosius 186 orthopyroxene 251 Ostia, Italy 26, 342 Ovid 184-5 palaeoecology 269 Palaeolithic rock shelter 30 palaeomagnetic determination 330 palaeosols 329, 330, 331 palaeostress fields 381-2, 390 palaeotopography mapping 105-6, 112-19 palynology 243 Pamukkale (Hierapolis) 1-13, 380, 440 Panama 226-7 Papua New Guinea 225-43, 323, 327 Parmenides 183 particle size, see grain size Pausanius 399, 400, 401 Pella, Macedonia 29 Peloponnese, Greece 29-30 Pergamon 29 Peru 330, 333, 367 petroglyphs 288 petrology, basalts 209-11, 212 Phira quarry, Thera 105, 109-12, 125 photogeological interpretation 285, 287-8 photographic collection 190, 192 phreatic activity 126 phreatomagnetic activity 126, 127, 137-8, 161 phreatomagnetic ash 288, 293 Phrygian marble 9 Pindar 180, 185, 186, 400 pisolite 161 plagioclase 170, 251, 325, 343 plate tectonics, Red Sea Basin 355-7 Plato 184, 400 Plinian eruptions 184;, see also Mt Pinatubo; Popocatepetl; Santorini; Vesuvius plinthite 326 Pliny the Elder 182, 184, 185, 400 Pliny the Younger 132, 138, 184 Plutarch 400 pollution 85, 101-2, 269, see also acid rain Pompeii 30, 46, 81, 143-58; see also Vesuvius Ponza Islands 190 population dynamics 55, 65-6, 70, 87 portlandite 340 pottery Akrotiri 72-8, 81-3 Apulian Bronze Age 159, 167-74 Cuicuilco 214, 217 Guanche 288 Lapita 239, 241,242 Minoan Crete 81, 83, 86, 90-91, 95-102 Neolithic 167, 170 pyroclastic temper 159, 167-74 Sagalassos 378, 379 pozzolanas 19, 339-43 Pozzuoli (Puteoli) 19, 193, 194, 339, 342 Procida, Italy 192, 341 pumice 19, 71-2, 76, 81, 85, 90, 100, 101, 125-39, 159-74, 196-203, 341, 343, 386 pyramids, Mesoamerican 205-23, 350
INDEX pyroclastic deposits 19, 105-19, 123-39, 143, 161-7, 198, 245, 339-43; see also tephra pyroxene 167, 346 quarry, Phira (Thera) 105, 109-12, 125 quarrying, obsidian and chert 233-5, 237, 238 quartz/quartzite 348 radar (GPR) 105-19, 198 radio-carbon dating Akrotiri 75, 79 Canary Islands 287 Mexico 199, 212, 214, 217, 221 Papua New Guinea 227-9, 231, 240 volcanic soils 329, 331 radiometric dating 283, 285-8, 330, 331 Ras Shamra, Levant 27 realgar 20 Red Sea Basin 355-7 red tides 102 Reggio Calabria, Italy 50-51, 55-63, 69-70 rescue archaeology 51, 59 resistivity surveys 198 rhyolite251, 253, 357-9 rhyolitic soils 320, 328, 333 rift zones 282-3, 401-3, 409-10, 411 risk assessment 395-7 ritual objects, Cuicuilco 211, 221 ritual/cults 16, 21-2, 89-93, 98-100, 180-82, 211, 221 Roccamonfina, Italy 192 rock carvings 17, 18 rock specimen collection 191-3 Roman period building construction, Hierapolis 1-13 concretes 19, 339-43 inscriptions 60-61 North Africa sites 28 responses to volcanism 180-86 Sagalassos, earthquake evidence 373-82 Straits of Messina 49, 51, 57-67 Rome, earthquake activity 26-7 Sagalassos, earthquake evidence 373-82 St Vincent 321 salmon, effects of tephra 262-3 San Salvador 226-7, 332 Sana'a area, Yemen 355-71 sanidine 161, 163, 170, 171 Santorini (Thera) 330, 340 AD 725 eruption 85 Minoan eruption 81-93, 105-19, 123, 128-31, 138-9 pre-Minoan geography 71-9, 105-6, 117-19, 128-9 satellite imagery 152, 153, 403 scanning electron microscopy (SEM) 127, 170, 229-30 scapolite 163, 170, 171 scoria 293, 298, 299, 341, 343 Scotland 190, 192, 267-77, 327 sea level drop, and fault pattern 393 seismic archaeology see archaeoseismology seismicity see earthquakes Seneca 184 Senegal 326 settlements, patterns 50, 51-5, 60, 63-67, 85-7, 198-203, 237-43, 267-77
INDEX sherds, see pottery shrines 198-203, 350 Siberia, lava soils 327 Sicily 45-70, 179-87, 366 silver 20, 191-2, 400 slope processes 376-7, 381 smithsonite 192 social organization, and volcanism 225-7, 237-43 soils 20, 194, 198, 241, 274-5, 317-34, 368, 371 Solfatara, Italy 190, 192 Sparta 27, 28 springs 5, 7, 8, 13, 193, 284, 312, 339, 405-8, 411 Stabiae 143 Statius 143, 184 stone, uses 1-13, 15-22, 194, 230, 232-9, 241, 243, 345-52 Strabo 36-7, 41, 42, 43-4, 179, 184, 400, 401 stratigraphy 105, 107, 109, 112-19, 212-17, 226, 403 stress fields 381-2, 390 stromatolites 71, 73, 74 Stromboli 193, 271 structural archaeology 360-71, 373-82, 390-93 subsistence 85, 235-7, 239, 241, 242, 267-77, 274-7 Suetonius 186 symbolism, and volcanism 16, 21-2 Tacitus 138, 184 Tenerife281, 282, 328-9 tephra basaltic, Xitle 205, 208-11, 214 composition analyses 247-51 deposit thickness calculation 172-4 effect on Scottish uplands 267-77 electron microscope studies 247-9 fallout volume calculation 166-7 from Santorini Minoan eruption 83-4, 85, 96, 99, 101-2 similarity coefficient 249 soils 198, 241, 317-24, 329-34 tephrachronology 225-6, 227, 245-65, 330 tephrastratigraphy 225—6, 227-32 Tertiary volcanics 190, 192, 312 Thebes, Greece 30 Thera see Akrotiri; Santorini thermokalite 192 thermoluminescence 329 Thucydides 36-7, 41, 42, 43, 185, 186 Tiryns, Greece 29, 30 titanium 249, 251 tools, stone 17-18, 230, 232-9, 241, 243 topography database 152-3 trachyte 320
417
trass 342, 343 travertines 1-13, 406, 408 trees 241, 242, 268, 269 triangulated irregular network (TIN) 153 Troy 27 tsunamis 36, 43-4, 78, 84-5, 91, 100 Turkey 1-13, 18, 27, 29, 81, 365, 367, 373-82 UNESCO 360, 367-70 uplands abandonment 268, 269 USA 85, 139, 261, 193, 321, 326, 329, 331; see also Alaska; Hawaii Vanuatu 321 vegetation 241, 242, 263-4, 265, 268, 269 vent complexes 281-304 vermetids 393 Vesuvius 179, 190, 332, 341 AD 79 "Pompeii" eruption 131-9, 143, 184, 339 Avellino eruption 81, 133, 159-74 geographical information system 143-58 Johnston-Lavis Collection 189-94 Virgil 180-81, 182, 184, 185, 186 Vitruvius 8, 182, 339, 340, 341 volcanic aerosols 267, 270-73, 275, 277 volcanic ash see pyroclastic deposits; tephra volcanic events 15-16, 165-6, 225-43, 307-14, 357 Classical period responses 89-93, 98-100, 179-87, 390-92 Olmec response 351 precursory activity 123, 126-8, 134, 139, 313 see also specific volcanoes volcanic glass 18, 343 analysis 170-71 comparisons 247-60 refractive indices 229-30 in soils 320, 325, 327 see also tephrachronology volcanic products 15-22, 339-43, 345-52 volcanic soils see soils volcanic stratigraphy 105, 107, 109, 112-19, 212-17 volcanic winter 85, 269 volcano worship 201, 350 volcanology (Johnston-Lavis Collection) 189-94 Vulcano, Italy 193, 271 weathering 20, 22, 320, 325-9, 331, 349 X-ray techniques 169 Yemen, seismic and volcanic hazards 355-71 zinc 192
