Geomaterials in Cultural Heritage (Geological Society Special Publication No. 257)

Geomaterials in Cultural Heritage

The Geological Society of London Books Editorial Committee R. PANKHURST ( U K ) (CHIEF EDITOR)

Society Books Editors J. GREGORY (UK) J. GR1FF1THS (UK) J. HOWE ( U K ) P. LEAT ( U K ) N. ROBINS ( U K ) J. TURNER ( U K )

Society Books Advisors M. BROWN ( U S A ) R. GIERF- ( G e r m a n y ) J. GLUYAS ( U K ) D. STEAD ( C a n a d a ) R. STEPHENSON ( N e t h e r l a n d s ) S. TURNER ( A u s t r a l i a )

Geological Society books refereeing 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 Books Editorial 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 Book Editors 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 Journal of the Geological Socie~.'. The referees' forms and comments must be available to the Society's Book 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. More information about submitting a proposal and producing a book for the society can be found on its 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: MAGGETTI, M. & MESSIGA, B. (eds) 2006. Geomaterials in Cultural Heritage. Geological Society, London, Special Publications, 257. SHORTLAND, A. J., HOPE, C. A. & TITE, M. S. 2006. Cobalt blue painted pottery from 18th Dynasty Egypt. In: MAGGETTI, M. & MESSIGA, B. (eds) Geomaterials in Cultural Heritage. Geological Society, London, Special Publications, 257, 91 - 100.

GEOLOGICAL SOCIETY SPECIAL PUBLICATION NO. 257

Geomaterials in Cultural Heritage EDITED BY MARINO MAGGETTI University of Fribourg, Switzerland and BRUNO MESSIGA Universit~ degli Studi di Pavia, Italy

2006 Published by The Geological Society London

THE GEOLOGICAL SOCIETY The Geological Society of London (GSL) was founded in 1807. It is the oldest national geological society in the world and the largest in Europe. It was incorporated under Royal Charter in 1825 and is Registered Charity 210161. The Society is the UK national learned and professional society for geology with a worldwide Fellowship (FGS) of 9000. The Society has the power to confer Chartered status on suitably qualified Fellows, and about 2000 of the Fellowship carry the title (CGeol). Chartered Geologists may also obtain the equivalent European title, European Geologist (EurGeol). One fifth of the Society's fellowship resides outside the UK. To find out more about the Society, log on to www.geolsoc.org.uk. The Geological Society Publishing House (Bath, UK) produces the Society's international journals and books, and acts as European distributor for selected publications of the American Association of Petroleum Geologists (AAPG), the American Geological Institute (AGI), the Indonesian Petroleum Association (IPA), the Geological Society of America (GSA), the Society for Sedimentary Geology (SEPM) and the Geologists' Association (GA). Joint marketing agreements ensure that GSL Fellows may purchase these societies' publications at a discount. The Society's online bookshop (accessible from www.geolsoc.org.uk) offers secure book purchasing with your credit or debit card. To find out about joining the Society and benefiting from substantial discounts on publications of GSL and other societies worldwide, consult www.geolsoc.org.uk, or contact the Fellowship Department at: The Geological Society, Burlington House, Piccadilly, London W1J 0BG: Tel. +44 (0)20 7434 9944: Fax +44 (0)20 7439 8975; E-mail: enquiries @geolsoc.org.uk. For information about the Society's meetings, consult Events on www.geoisoc.org.uk. To find out more about the Society's Corporate Affiliates Scheme, write to [email protected].

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: www.geolsoc.org.uk/bookshop

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 2006. 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 WlP 9HE. Users registered with the Copyright Clearance Center, 27 Congress Street, Salem, MA 01970, USA: the item-fee code for this publication is 0305-8719/06/$15.00.

British Library Cataloguing in Publication Data A catalogue record for this book is available from the British Library. I S B N 1-86239-195-5 Typeset by Techset Composition, Salisbury, UK Printed by MPG Books Ltd, Bodmin, 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 bookstore @aapg.org India Affiliated East-West Press Private Ltd Marketing Division G-I/16 Ansari Road, Darya Ganj New Delhi I l0 002 India Orders: Tel. +91 11 2327-9113/2326-4180 Fax +91 il 2326-0538 E-mail [email protected] Japan Kanda Book Trading Company 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 E-mail geokanda@ ma.kcom.ne.jp

Contents

Preface

vii

MAGGETTI, M. Archaeometry: quo vadis?

SMITH, D. C. A review of the non-destructive identification of diverse geomaterials in the cultural heritage using different configurations of Raman spectroscopy

Pottery (BC) BASSO, E., BINDER, D., MESSIGA,B. & RICCARDI,M. P. The Neolithic pottery of Abri Pendimoun (Castellar, France): a petro-archaeometric study

33

LAVIANO,R. & MUNTONI,I. M. Provenance and technology of Apulian Neolithic pottery

49

MAGGETTI, M. & GALETTI,G. Late La Tbne pottery from western Switzerland: one regional or several local workshops?

63

MOMMSEN, H., BONANNO, A., CHETCUTI BONAVITA, K., KAKOULLI,I., MUSUMECI, M., SAGONA, C., SCHWEDT, A., VELLA, N. C. • ZACHARIAS,N. Characterization of Maltese pottery of the Late Neolithic, Bronze Age and Punic Period by neutron activation analysis

81

SHORTLAND,A. J., HOPE, C. A. & TITE, M. S. Cobalt blue painted pottery from 18th Dynasty Egypt

91

SHOVAL, S., BECK, P. & YADIN,E. The ceramic technology used in the manufacture of

101

Iron Age pottery from Galilee

SMITH, M. S. & TRINKLEY,M. B. Fibre-tempered pottery of the Stallings Island Culture from the Crescent site, Beaufort County, South Carolina: a mineralogical and petrographical study

119

Pottery (AD) BIANCHINI, G., MARROCCHINO,E., MORETTI, A. & VACCARO, C. Chemicalmineralogical characterization of historical bricks from Ferrara: an integrated bulk and micro-analytical approach

127

~OLAK, M., MAGGETTI,M. & GALETTI,G. Golden mica cooking pottery from

141

Grkeyfip (Manisa), Turkey

vi

CONTENTS

DELL'AQUILA, C., LAVIANO,R. & VURRO, F. Chemical and mineralogical investigations of majolicas (16th-19th centuries) from Laterza, southern Italy

151

VENDRELL-SAZ, M., MOLERA, J., ROQUI~,J. & PI~REZ-ARANTEGUI,J. Islamic and Hispano-Moresque (mfidejar) lead glazes in Spain: a technical approach

163

Glass

ARLETTI, R., CIARALLO, A., QUARTIERI, S., SABATINO, G. & VEZZALINI,G. Archaeometric analyses of game counters from Pompeii

175

ERAMO, G. Pre-industrial glassmaking in the Swiss Jura: the refractory earth for the glassworks of Derriere Sairoche (ct. Bern, 1699-1714)

187

FREESTONE, I. C. Glass production in Late Antiquity and the Early Islamic period: a geochemical perspective

201

MARCHESI, V., NEGRI, E., MESSIGA, B. & RICCARDI, M. P. Medieval stained glass windows from Pavia Carthusian monastery (northern Italy)

217

Stone

ANTONELLI, F., SANTI, P., RENZULLI,A. & BONAZZA, A. Petrographic features and thermal behaviour of the historically known 'pietra ollare' from the Italian Central Alps (Valchiavenna and Valmalenco)

229

BELLELLI, C., PEREYRA, F. X. & CARBALLIDO,M. Obsidian localization and circulation in northwestern Patagonia (Argentina): sources and archaeological record

241

D'AMIco, C. & STARNINI,E. Prehistoric polished stone artefacts in Italy: a petrographic and archaeological assessment

257

GANDIN, A., CAPEZZUOLI,E. & CIACCI,A. The stone of the inscribed Etruscan stelae

273

from the Valdelsa area (Siena, Italy)

MILLER, S., MCGIBBON, F. M., CALDWELL, D. H. & RUCKLEY, N. A. Geological tools to interpret Scottish medieval carved sculpture: combined petrological and magnetic susceptibility analysis

283

MORGENSTEIN, M. Geochemical and petrographic approaches to chert tool provenance studies: evidence from two western USA Holocene archaeological sites

307

QUARESIMA, R., GIAMPAOLO,C. SPERNANZONI, F. & VOLPE, R. Identification, characterization and weathering of the stones used in medieval religious architecture in L'Aquila (Italy)

323

Mortar

CARO, F., DI GIULIO, A. & MARMO, R. Textural analysis of ancient plasters and mortars: reliability of image analysis approaches

337

Index

347

Preface

The scientific study of monuments, as well as objects from excavations and museums, deals with their origin, technique, age and conservation. Such topics were addressed during the one-day topical symposium 'Geomaterials in Cultural Heritage' of the 32nd International Geological Congress held in Florence on 20-28 August 2005. We have edited this volume by assembling papers of participants of the Florence meeting, as well as invited contributions, to present a wide view of the interdisciplinary application of geoscience disciplines, and to reaffirm the important contribution of geosciences to solve problems concerning the study of complex materials such as minerals, rocks, glass, metals, mortar, plaster, slags and pottery. This interdisciplinary application of geosciences includes field geology, geophysics, microscopy, textural analysis, physical methods and geochemistry as fundamental support to disclose hidden information, retained by the ancient materials, such as the raw materials provenance, the firing technology, the ancient recipes and the alteration pathway. The volume

is dedicated to all scholars eager to undertake or to continue an exciting research activity. Many colleagues helped us in the review process and we thank C. D'Amico, F. Antonelli, M. Baxter, C. Belelli-Pereyra, G. Bianchini, G. Bigazzi, F. Car6, G. Eramo, I. Freestone, A. Gandin, K. Gherdan, B. Grob6ty, R. Heimann, A. Jornet, R. Laviano, L. Lazzarini, S. Miller, H. Mommsen, R. Quaresima, M. P. Riccardi, G. Schneider, V. Serneels, A. Shortland, S. Shoval, D. C. Smith, S. Smith, G. Thierrin-Michael, M. Tite, S. Trfimpler, M. Vendrell-Saz, G. Wagner and S. Wolf for their goodwill and rigorous review of the submissions. We acknowledge the efficient assistance and the exemplary editorial support of the Geological Society publishing staff (particularly Angharad Hills and Sally Oberst) and the remarkable technical help from Nicole Bruegger. Marino Maggetti Bruno Messiga

Archaeometry: quo

vadis?

MARINO MAGGETTI University of Fribourg, Department of Geosciences, Mineralogy and Petrography, Chemin du Musde 6, CH-1700 Fribourg, Switzerland (e-mail: [email protected])

Abstract: First, a brief overview of the tasks and the historical development of archaeometry will be given. Although archaeometry is generally doing well, a few issues currently faced by this discipline will be outlined. These include: (1) funding for projects and research positions; (2) the appeal of archaeometry to a new generation of academics; (3) the standard of publications; (4) the safeguarding of and the immediate access to scientific data.

Scientific study of raw materials and products used in prehistoric and historical time involves an interdisciplinary collaboration between archaeology, art history, preservation of the cultural heritage, ethnography and science. This area of research, in which these disciplines overlap, is known as archaeometry or archaeological sciences. The term geomaterials includes rocks, soils, mortars, pigments, ceramics, glass and slags. Scientific analysis of these objects aims at answering the following questions: (1) Where does the raw material come from? (2) Where was the object manufactured? (3) How was it manufactured (technique)? (4) What was its purpose (function)? (5) When was it manufactured (dating)? Scientific analysis should not limit itself to the qualitative and quantitative description of the 'chai'ne oprratoire'. Rather, it should approach these questions in a holistic manner. This involves the socio-cultural environment in which the artefact was manufactured (household, workshop, etc.), distributed and used. Collaboration with archaeologists and art historians needs to show how and why a particular technique was introduced or a specific manufacturing process used. It also needs to clarify the intention behind a certain function and the choice of a particular trading structure. In the field of preservation, material-specific properties of unweathered objects must be compared with their decay products so as to work out restoration concepts within a framework of interdisciplinary collaboration.

Methods and history Experimental methods used in the field of archaeometry have been described in a number of papers (Aitken 1961, 1985, 1990, 1998; Brothwell & Higgs 1968; Berger 1970;

Tite 1972; Fleming 1976; Hrouda 1978; Goffer 1980; Riederer 1981a, 1987; Matteini & Moles 1984; Cuomo di Caprio 1985; Mommsen 1986; Parks 1986; Wagner & Van den Haute 1992; Taylor & Aitken 1997; Wagner 1998; Ciliberto & Spoto 2000; Barclay 2001; Brothwell & Pollard 2001; Martini et al. 2004). Apart from a multitude of papers published in journals or books, there are many specifically geomaterial-related monographs, as well as proceedings from conferences. We shall name only a few of these and limit them to three domains as examples, because a complete listing would go beyond the scope of this Introduction. Ceramics. Shepard 1956; Matson 1965; Brill 1971; Picon 1973; Rye & Evans 1976; Peacock 1977, 1982; Winter 1978; Drmians d'Archimbaud & Picon 1980; Arnold 1981, 1985, 1993; Howard & Morris 1981; Hughes 1981; Thompson 1982; Rice 1984, 1987, 1997; Kingery 1985, 1986a, b; Laubenheimer 1985, 1992, 1998; Empereur & Garlan 1986; Jones 1986; Kingery & Vandiver 1986; Riederer 1987; Atasoy & Raby 1989; Lenoir et al. 1989; McGovern & Notis 1989; Middleton & Freestone 1991 ; Noll 1991; Wilson 1991; Li Jiazhi & Chen Xianqiu 1992; Mrry 1992; Neff 1992; Pollard 1992; Failla 1993; Burragato et al. 1994; Guo Jinkum 1995; Lindahl & Stilborg 1995; Vendrell-Saz et aL 1995; Vicenzini 1995; Whitbread 1995; Frontini & Grassi 1996; Cumberpatch & Blinkhorn 1997; Drmians d'Archimbaud 1997; Freestone & Gaimster 1997; Gaimster 1997; Gibson & Woods 1997; Lang & Middleton 1997; Santoro Bianchi & Fabbri 1997; Druc 1998; Fabbri & Lega 1999; Levi 1999; Ruf et al. 1999; Velde & Druc 1999; Wood 1999; Henderson 2000; Rosen 2000; Shortland 2001; Veeckman et al. 2002; D'Albis 2003;

From: MAGGETTI,M. & MESSIGA,B. (eds) 2006. Geomaterials in Cultural Heritage. Geological Society, London, Special Publications,257, 1-8. 0305-8719/06/$15.00 (c) The Geological Society of London 2006.

2

M. MAGGETI'I

D'Anna et al. 2003; Di Pierro et al. 2003; Keblow Bernsted 2003; Bargossi et al. 2004; Gurt i Esparraguera et al. 2005; Livingstone Smith et al. 2005. Glass. Lucas 1921; Caley 1962; Sayre 1964; Ankner 1965; Berger 1970; Oppenheim et al. 1970; Besborodov 1975; Newton & Davison 1978; Frank 1982; Olin & Franklin 1982; Wertime & Wertime 1982; Kazmarzyck & Hedges 1983; Lambert 1984; Bhardwaj 1987; Bimson & Freestone 1987; Riederer 1987; Henderson 1989, 2000; Brill & Martin 1991; Foy & Sennequier 1991; Mendera 1991; Tait 1991; Vandiver et al. 1992; Lilyquist & Brill 1993; Foy 1995, 2001; Hook & Gaimster 1995; Pollard & Heron 1996; Kingery & McCray 1998; Seibel 1998; Nenna 2002; Veeckman et al. 2002; Foy & Nenna 2003; Steppuhn 2003; Wedepohl 2003; Bargossi et al. 2004.

Gnoli 1971; Young 1973; Winkler 1973; Pensabene 1985, 1994, 1998; Sieveking & Hart 1986; Torrence 1986; Riederer 1987; Fant 1988; Herz & Waelkens 1988; Trou & Podany 1990; Borghini 1992; Moens et al. 1992; Bradley 1993; Klemm & Klemm 1993; Moorey 1994; Maniatis et al. 1995; Cunliff & Renfrew 1997; Shackley 1997; Schvoerer 1999; Henderson 2000; Roux 2000; De Nuccio & Ungaro 2002; Herrmann et al. 2002; Lazzarini 2002, 2004; Kardulias & Yerkes 2003; Poupard & Richard 2003; Bargossi et al. 2004. The first scientific analyses of ceramics, metals and pigments started early; that is, at the beginning of the 19th century (Riederer 198 lb, 1987; Maggetti 1990, 1994a). The foundation of specialist laboratories at museums and universities, such as the Chemisches Laboratorium der k6niglichen Museen zu Berlin (RathgenForschungslabor, 1888), as well as the Research Laboratory for Archaeology and the History of Arts (1955) at the University of Oxford, were milestones in the development of archaeometry. The number of similar institutions, active working groups and professional societies has increased ever since. The publication of several archaeometric journals was initiated, along with a great number of conferences. Obviously, archaeometry is an encouragingly vital discipline, but is it free of problems? Stone.

Problems faced by archaeometry Fundamental aspects of the status of archaeometry have been discussed extensively by Tite (1991, 2004) and Jones (2004). It is therefore unnecessary to further comment on them here.

However, it appears appropriate to take on some of the points raised by Widemann (1982), Fabbri (1992), Vidale (1992) and Maggetti (1994b). They deal with the funding of projects and research positions, the appeal of archaeometry to young scientists, the quality of scientific publications and the immediate and efficient access to research data. Funding

Although interdisciplinary research is up-to-date and highly praised by all entities, people working in this sector do indeed face difficulties. For instance, it is not easy to obtain funding, because problems concern historical disciplines, whereas answers and methods pertain to the sciences. In the quest for funds, one may find that a scientific body either rejects a project because questions are regarded as of historical nature, or it may pass it on to an arts or humanistic body, which in turn also declines the project, considering it to be of scientific nature. Citation index

It is becoming more and more common for universities, departments and scientists to be judged by the number of scientific papers being published in journals belonging to the citation index. Many archaeometric publications, however, do not appear in such journals, a fact that must have a detrimental effect on the career and reputation of the scientist concerned, if they do not already hold a position in archaeometry. On the other hand, the archaeometric results should also be published in archaeologically relevant journals or books, to strengthen interdisciplinary collaboration. As a result, young, enthusiastic scientists will be discouraged from pursuing a career in archaeometry. Stable research p o s i t i o n s

In addition to the problem mentioned above, there are far too few permanent posts for trained archaeometrists. It is understandable that in times when jobs are cut everywhere, scientific disciplines do not appear willing to redefine a vacancy as an interdisciplinary lecturing and research position. Because the questions pertain to the field of archaeology, art history and the preservation of ancient monuments, it should be up to these disciplines to safeguard or create the appropriate posts. Without such new positions it is impossible to retain the interest of young scientists or to motivate them to undertake research in archaeometry.

ARCHAEOMETRY: QUO VADIS? 'Hobby'

Unfortunately, there are far too many people doing archaeometric research as a 'hobby'. Many of these part-time archaeometrists are not familiar with archaeometric literature and reinvent the wheel, so to speak, i.e. they tackle questions that have been solved a long time ago. Often archaeologically irrelevant questions are investigated, insufficient numbers of samples are analysed and it can happen that 'poverty of measurements are covered up with sophisticated data processing' (Widemann 1982). Many such papers appear in unrefereed journals or books and escape the quality filter. Databases

Often, results of archaeometric studies will not be found in relevant journals, such as Archaeometry, Geoarchaeology, Journal of Archaeological Sciences, Revue d'Arch¢ombtrie, Journal o f Cultural Heritage and Marmora. Instead, they are published in books or journals that are difficult to access. Those who need to consider the citation index will be more likely to publish in journals of their specific discipline. In these journals, however, archaeometric contributions tend to disappear and colleagues cannot find them. It is therefore understandable that far too many good papers will hardly be read. As a result, analyses pertaining to problems that have already been investigated tend to be repeated, a fact that should be avoided considering the limited financial resources within the archaeometry community. In the field of ceramics, for instance, many working groups possess a large collection of chemical data and chemical reference groups, which, for the reasons mentioned above, are not readily accessible to everybody or will be lost once the group ceases its activity. Consequently, it is very important to make these treasures accessible to all people involved in archaeometry. These days, the internet provides a powerful instrument to display results periodically on the homepage of the relevant team, as is done by the Freiburg archaeometric working group (www.unifr.ch/geoscience/mineralogy/ archmet). However, this applies not only to ceramics, but also to the other geomaterials. As a result, one could avoid duplicate working and save money as well as time.

Final remarks The previous discussion has highlighted some of the problems faced by archaeometrists. What does the future of archaeometry look like? It is beyond any doubt that archaeometric research,

3

such as precise dating, is indispensable to the fields of archaeology and art history. Also, it is impossible to make restorations without the appropriate and relevant research being done. This presents a huge opportunity for archaeometry, because, owing to its impact on tourism, the study and preservation of our cultural heritage is likely to receive sufficient financial support from governments. However, it is necessary for research in all archaeometric sectors to focus on archaeometric competence centres that provide sufficient long-term employment, knowledge and sound technical apparatus. This is the only way in which young, well-trained scientists will be willing to commit themselves to the fascinating field of interdisciplinary research.

References AITKEN, M. J. 1961. Physics and Archaeology. Clarendon Press, Oxford. AITKEN, M. J. 1985. Thermoluminescence Dating. Academic Press, London. AITKEN, M. J. 1990. Science-based Dating in Archaeology'. Longman, Harlow. AtTKEN, M. J. 1998. An Introduction to Optical Dating. Oxford University Press, Oxford. ANKNER, D. 1965. Chemische und physikalische Untersuchungen an vor- und friihgeschichtlichen Gliisern. Technische Beitr~ige zur Arch~iologie, II. ARNOLD, D. (ed.) 1981. Studien zur altiigyptischen Keramik. Philipp von Zabem, Mainz. ARNOLD, D. 1985. Ceramic Theory and Cultural Process. Cambridge University Press, Cambridge. ARNOLD, D. 1993. Ecology and Ceramic Production in an Andean Community. Cambridge University Press, Cambridge. ATASOY, N. & RABY, J. (eds) 1989. Iznik. The Pottery of Ottoman Turkey. Alexandria Press, London. BARCLAY, K. 2001. Scientific Analysis of Archaeological Ceramics. A Handbook of Resources. Oxbow Books, Oxford. BARGOSSI, G. M., FRANZIN1,M. & MESSIGA, B. 2004. A showcase of the Italian research in applied petrology. Periodico di Mineralogia, 73, Special Issue 3. BERGER, R. (ed.) 1970. Scientific Methods in Medieval Archaeology. University of California Press, Berkeley. BESBORODOV, M. A. 1975. Chemie und Technologie der antiken und mittelalterlichen Gliiser. Ph. von Zabern Mainz. BHARDWAJ, H. C. (ed.) 1987. Archaeometry of glass. Proceedings of the Archaeometry Session of the XI International Congress on Glass, New Delhi. The Indian Ceramic Society, Calcutta. BIMSON, M. & FREESTONE,I. C. 1987. Early Vitreous Materials. British Museum Occasional Papers, 56. BORGHINk G. (ed.) 1992. Marmi antichi. Leonardo De Luca, Rome.

4

M. MAGGETI'I

BRADLEY, R. 1993. Interpreting the Axe Trade. Production and Exchange in Neolithic Britain. Cambridge University Press, Cambridge. GRILL, R. H. (ed.) 1971. Science and Archaeology. MIT Press, Cambridge, MA. GRILL, R. H. & MARTIN, J. H. (eds) 1991. Scientific research in early Chinese glass. In: Proceedings of the Archaeometr 3' of Glass Sessions of the 1984 International Symposium on Glass, Beijing, September 7, 1984 with Supplementary Papers. Coming Museum of Glass, Coming, NY. BROTHWELL, D. & HIGGS, E. 1968. Science in Archaeology. Thames and Hudson, London. BROTHWELL, D. R. & POLLARD, A. M. (eds) 2001. Handbook of Archaeological Sciences. Wiley, Chichester. BURRAGATO, E., GRUBESSI, O. & LAZZARINI,L. (eds) 1994. Ist European Workshop on Archaeological Ceramics 1991. Universit/a degli Studi 'La Sapienza', Rome. CALEY, E. R. 1962. Analysis of Ancient Glasses 17901957. Coming Museum of Glass, Coming, NY. CILIBERTO, E. ~; SPOTO, G. (eds) 2000. Moden7 Analytical Methods in Art and Archaeology. Wiley, New York. CUMBERPATCH, C. G. (~; BLINKHORN, P. W. (eds) 1997. Not so much a Pot, More a Way of Life. Oxbow Monograph, 83. CUNLIFF, B. 8£ RENFREW, C. (eds) 1997. Science and Stonehenge. Proceedings of the British Academy, 92. CUOMO DI CAPRIO, N. 1985./.z~ ceramica in archeologia. Antiche tecniche di lavorazione e moderni metodi d'indagine. L'Erma di Bretschneider, La Fenice, Rome. D'ALBlS, A. 2003. Traitd de la porcelaine de Sbvres. Faton, Dijon. D'ANNA, A., DESBAT, A., GARCIA, D., SCHMITT, A. & VERHAEGE, F. 2003. La Cdramique. La Poterie du Ndolithique aux Temps modernes. Errance, Paris. DE NuccIo M. & UNGARO, L. 2002. I n~lrmi colorati della Roma imperiale. Marsilio, Venice. DI~MIANS D'ARCHIMBAUD,G. (ed.) 1997. La Cdramique mddidvale en Mdditerrande. Actes du Vlbme Congrbs de I'AIECM2, Aix-en-Provence 1 3 - 1 8 novembre 1995. Narration, Aix-en-Provence. DI~MIANSD'ARCHIMBAUD,G. & PICON, M. (eds) 1980. La Cdramique mddidvale en Mdditerrande occidentale X d - X V d sikcle. Actes du Colloque international no 584, Valbonne 11 au 14 septembre 1978. CNRS, Paris. Dt PIERRO, S., SERNEELS, V. (~ MAGGETTI, M. (eds) 2003. Ceramic in the Society. Proceedings of the 6th European Meeting on Ancient Ceramics, Fribourg, Switzerland, 3 - 6 October 2001. Department of Geosciences, University of Fribourg. DRUC, I. C. 1998. Ceramic Production and Distribution in the Chavin Sphere of Influence (NorthCentral Andes). British Archaeological Reports, International Series, 731. EMPEREUR, J.-Y. & GARLAN, Y. (eds) 1986. Rechepz~hes sur les Amphores grecques. Actes du Colloque International organisd par le Centre National de la Recherche Scientifique, l'Universitd

de Rennes H et I'Ecole Fran9aise d'Athknes (Athbnes, 10-12 septembre 1984). Diffusion de Brocard, Paris. FABBRI, B. 1992. Archeometria: chi, come e perch,:. In: SANTORO BIANCHI, S., FAILLA, A. & GUERMANDI, M. P. (eds) Archeometria della Ceramica, Problemi di Metodo. Int. Centro Ceramico, Bologna, 127-129. FABBRI, B. & LEGA, A. M. (eds) 1999. I rivestimenti e le decorazioni nei manufatti antichi: quali opportunitgt per I'archeometria. Atti della 3a giornata di archeometria della ceramica. Faenza, 30 marzo 1999. University Press, Bologna. FAILLA, A. (ed.) 1993. Archeometria della ceramica. Problemi di Metodo. Atti 80 SlMCER-Simposio hlternazionale di ceramica, Rimini, 10-12 novembre 1992. Int. Centro Ceramico, Bologna. FANT, J. C. (ed.) 1988. Ancient Marble Quarrying and Trade. Papers from a Colloquium held at the Annual Meeting of the Archaeological Institute of America, San Antonio, Texas, December, 1986. British Archaeological Reports, International Series, 453. FLEMING, S. J. 1976. Authenticity in Art: the Scientific Detection of Forger3.'. Institute of Physics, London. FoY, D. (ed.) 1995. Le Verre de l'Antiquitd tardive et du haut Moven Age. Procddd de l'Association Franfaise pour l'Archgologie du Verre, Huitikme Rencontre. Mus~e arch~ologique d~partemental du Val d'Oise, Guiry-en-Vexin. FoY, D. 2001. Le Verre mddidval et son Artisanat en France mdditerrandenne. CNRS, Paris. FoY, D. & NENNA, M.-D. (eds) 2003. Ateliers de Verriers de l'Antiquitd b la Pdriode prd-industrielle. Actes de la 4kme Rencontre de l'Association Franfaise pour l'Archdologie du Verre, Rouen 1989. Association Franqaise pour l'Archrologie du Verre, Rouen. FoY, D. ~ SENNEQUIER, G. (eds) 1991. Echanges et Commerce du Verre dans le Monde antique. Actes du Colloque de I'AFA V, Aix-en-Provence et Marseille, 7 - 9 juin 2001. Monique Mergoil, Montagnac. FRANK, S. 1982. Glass and Archaeology. Academic Press, London. FREESTONE, I. & GAIMSTER, D. (eds) 1997. Pottery in the Making. World Ceramic Traditions. British Museum Press, London. FREESTONE, I. & JOHNS, C. (eds) 1980. Current Research in Ceramics: Thin Section Studies. British Museum Occasional Papers 32. FRONTINi, P. & GRASSI, M. T. (eds) 1996. Indagini Archeometriche Relative Alla Ceramica a Vernice nera: nuovi dati sulla Provenienza e la Diffusione. Atti del Seminario hzternazionale di Studio, Milano, 2 2 - 2 3 novembre 1996. New Press, Como. GAIMSTER, D. 1997. German Stoneware 1200-1900. British Museum Press, London. GmSON, A. & WOODS, A. 1997. Prehistoric Pottery. for the Archaeologist, 2nd edn. Leicester University Press, London. GNOLI, R. 1971. Marmora Romana. Elefante, Rome. GOFFER, Z. 1980. Archaeological Chemistry. Wiley, New York.

ARCHAEOMETRY: QUO VADIS? Guo JINKUM (ed.) 1995. Science and Technology of Ancient Ceramics 3. Proceedings of the International Symposium on Ancient Ceramics (ISAC'95), Shanghai. Shanghai Science and Technology Press, Shanghai. GURT I ESPARRAGUERA,J. M., BUXEDAI GARRIGOS,J. & CAU ONTIVEROS, M. A. (eds) 2005. LRCW, Late Roman Coarse Wares, Cooking Wares and Amphorae in the Mediterranean. Archaeology and Archaeometry. British Archaeological Reports, International Series, 1340. HENDERSON, J. (ed.) 1989. Scientific Analysis in Archaeology and its Interpretation. Oxford University Committee on Archaeology Monograph, 19. University of California, Los Angeles, Institute of Archaeology Research Tools, 5. HENDERSON, J. 2000. The Science and Archaeology of Materials. An Investigation of Inorganic Materials. Routledge, London. HERRMANN, J. J., HERZ, N. & NEWMANN, R. (eds) 2002. Interdisciplinary Studies on Ancient Stones. Proceedings of the 5th ASMOSIA Conference, Museum of Fine Arts, Boston, June 11-15, 1998. Archetype, London. HERZ, N. & WAELKENS, M. (eds) 1988. Classical Marble: Geochemistry, Technology, Trade. NATO ASI Series E, 153. HOOK, D. R. & GAIMSTER, D. R. M. (eds) 1995. Trade and Discovery. The Scientific Study of Artefacts from Post-Medieval Europe and Beyond. British Museum Occasional Papers, 109. HOWARD, H. & MORRIS, E. (eds) 1981. Production and Distribution: a Ceramic Viewpoint. British Archaeological Reports, 120. HROUDA, B. (ed.) 1978. Methoden der Archiiologie. Eine Einfiihrung in ihre naturwissenschaftlichen Techniken. Beck, Munich. HUGHES, M. J. (ed.) 1981. Scientific Studies in Ancient Ceramics. British Museum Occasional Papers, 19. JONES, A. 2004. Archaeometry and materiality: materials-based analysis in theory and practice. Archaeometry, 46(3), 327-338. JONES, R. E. 1986. Greek and Cypriot Pottery. A Review of Scientific Studies. British School at Athens, Fitch Laboratory Occasional Papers, 1. KARDULIAS, P. N. & YERKES, R. W. (eds) 2003. Written in Stone: the Multiple Dimensions of Lithic Analysis. Lexington Books, Lexington, MA. KAZMARZYCK, A. & HEDGES, R. E. M. 1983. Ancient Egyptian Faience. Aris and Phillips, Warminster. KEBLOW BERNSTED, A. M. 2003. Early Islamic Pottery. Materials and Techniques. Archetype, London. KINGEt~Y, W. D. (ed.) 1985. Ancient Technology to Modern Science. Ceramics and Civilisation, Vol. L American Ceramic Society, Columbus, OH. KINGERY, W. D. (ed.) 1986a. Technology and Style. Ceramics and Civilisation, Vol. II. American Ceramic Society, Columbus, OH. KINGERY, W. D. (ed.) 1986b. High Technology Ceramics. Past, Present and Future. Ceramics and Civilisation, VoL IlL American Ceramic Society, Westerville, OH.

5

KINGERY, W. D. & MCCRAY, P. (eds) 1998. The Prehistory and History of Glassmaking Technology. Ceramics and Civilization, Vol. VIII. American Ceramic Society, Columbus, OH. KINGERY, W. D. & VANDIVER, P. 1986. Ceramic Masterpieces--Art, Structure and Technology. Free Press, New York. KLEMM, D. & KLEMM, R. 1993. Steine und Steinbriiche im alten Aegypten. Springer, Berlin. LAMBERT, J. B. (ed.) 1984. Archaeological Chemistry III. American Chemical Society. Advances in Chemistry Series, 205. LANG, J. & MIDDLETON, A. (eds) 1997. Radiography of Cultural Material. Butterworth Heinemann, Oxford. LAUBENHEIMER, F. 1985. La Production des Amphores en Gaule Narbonnaise. Annales Litteraires de l'Universit~ de Besanqon, 327. LAUBENHEIMER, F. (ed.) 1992. Les Amphores en Gaule. Production et Circulation. Table Ronde lnternationale Metz, 4 - 6 octobre 1990. Annales Litt~raires de l'Universit6 de Besanqon, 474. LAUBENHEIMER, F. (ed.) 1998. Les Amphores en Gaule. Production et Circulation. Les Belles Lettres, Paris. LAZZARINI, L. (ed.) 2002. Interdisciplinary Studies on Ancient Stone. Proceedings of the Sixth International Conference (ASMOSIA VI), Venice, 1518 June 2000, Bottega d'Erasmo Aldo Ausilio, Padora. LAZZARINI, L. (ed.) 2004. Pietre e marmi antichi. Natura, caratterizzazione, origine, storia d'uso, diffusione, collezionismo. CEDAM, Padova. LENOIR, M., MANACORDA, O. & PANELLA, C. (eds) 1989. Amphores romaines et histoire ~conomique. Dix ans de recherche. Actes du colloque de Sienne (22-24 mai 1986) organis~ par l'Universith degli studi di Siena, l'Universitgt degli Studi di Roma-La Sapienza, le Centre national de la recherche scientifique (RCP 403) et l'Ecole franfaise de Rome. Ecole Fran~aise de Rome, Rome. LEVI, S. T. (ed.) 1999: Produzione e circolazione della ceramica nella Sibaritide protostorica. I. Impasto e doll Prima di Sibri 1. Insegna del Giglio, Firenze. LI JIAZHI & CHEN XIANQIU (eds) 1992. Science and Technology of Ancient Ceramics. Proceedings of the 1989 International Symposium, Shanghai. Shanghai Science and Technology Press, Shanghai. LILYQUIST, C. & BR1LL, R. H. (eds) 1993. Studies in Early Egyptian Glass. Metropolitan Museum of Art, New York. LINDAHL, A. & STILBORG, O. (eds) 1995. The Aim of Laboratory Analyses of Ceramics in Archaeology. Kungliche Vitterhets Historic och Antivitets Akademien Konferenser, 34. LIVINGSTONE SMITH, A., BOSQUET, D. & MARTINEAU R. (eds) 2005. Pottery Manufacturing Processes: Reconstitution and Interpretation. Acts of the XIVth UISPP Congress, University of Likge, Belgium, 2 - 8 September 2001. British Archaeological Reports, International Series, 1349. LUCAS, A. 1921. Ancient Egyptian Materials and Industries. Edward Arnold, London.

6

M. MAGGETTI

MAGGETTI, M. 1990. I1 contributo delle analisi chimiche alia conoscenza delle ceramiche antiche. In: MANNONI, T. ~ MOLINARI, A. (eds) Scienze in Archeologia. Insegna del Giglio, Firenze, 65-88. MAGGETT[, M. 1994a. Mineralogical and petrographical methods for the study of ancient pottery, h~: BURRAGATO, F., GRUBESSI, O. & LAZZARINI, L. (eds) Proceedings of the 1st European Workshop on Archaeological Ceramics. Dipartments Scienze della Terra, Universitg degli Studi di Roma 'La Sapienza', 23-35. MAGGETTI, M. 1994b. Is there still a future for archaeometry at universities? In: OLCESE, G. (ed.) Ceramica Romana e Archeometria: lo Stato degli studi. Insegna del Giglio, Firenze, 15-22. MANIAT1S, Y., HERZ, N. & BASIAKOS, Y. (eds) 1995. The Study of Marble and other Stones used in Antiquity. Transactions of the 3rd ASMOSIA Conference. Archetype, London. MARTINI, M., MILAZZO, M. & PIACENTINI, M. (eds) 2004. Physics Methods in A rcheometo'. Proceedings o f the International School o f Physics 'Enrico Fermi', Varenna on Lake Como. Villa Monastero, 1 7 - 2 7 June 2003. IOS Press, Amsterdam. MATSON, F. R. 1965. Ceramics and Man. Methuen, London. MATTEINI, M. & MOLES, A. 1984. Scienza e Restauro, Metodi di Indagine. Nardini, Firenze. McGOVERN, P. E. & NOTIS, M. D. (eds) 1989. Crosscraft and Cross-cultural Interactions in Ceramics. Ceramics and Civilization, Vol. IV. American Ceramic Society, Columbus, OH. MENDERA, M. 1991. Archeologia e storia della produzione del vetro preindustriale. Giglio, Firenze. MERY, S. (ed.) 1992. Sciences de la Terre et Cdramiques archdologiques. Expdrimentations, applications. Documents et travaux de l'institut g6ologique Albert-de-Lapparent, 16. MIDDLETON, A. & FREESTONE, I. C. (eds) 1991. Recent Developments in Ceramic Petrology. British Museum Occasional Papers, 81. MOENS, L., DE PAEPE, P. & WAELKENS, M. 1992. Multidisciplinary Research and Interlaboratorv Cooperation: Keys to a Successful Provenance Determination of White Marble. Acta Archaeologica Lovanensia Monographiae, 4. MOMMSEN, H. 1986. Archiiometrie. Teubner Studienbiicher. Teubner, Stuttgart. MOOREY, P. R. S. 1994. Ancient Mesopotamian Materials and Industries. Clarendon Press, Oxford. NEFF, H. (ed.) 1992. Chemical Characterization o f Ceramic Pastes in Archaeology. Monographs in World Archaeology, 7. NENNA, M. D. (ed.) 2002. La Route du Verre. Ateliers primaires et secondaires du Second Milldnaire av. J.-C. au Moyen Age. Travaux de la maison de l'orient M~diterran~en, 22. NEWTON, R. G. & DAVlSON, S. 1978. Conservation of Glass. Butterworth, London.

NOLL, W. 1991. Alte Keramiken und ihre Pigmente. Studien zu Material und Technologie. Schweizerbart, Stuttgart. OLIN, J. S. & FRANKLIN,A. D. (eds) 1982. Archaeological Ceramics. Smithsonian Institution Press, Washington, DC. OPPENHEIM, A. L., BRILL, R. H., BARAG, D. & VON SALDERN, A. (eds) 1970. Glass and Glassmaking in Ancient Mesopotamia. Coming Museum of Glass, Coming, NY. PARKS, P. A. 1986. Current Scientific Techniques in Archaeology. Seminar Press, London. PEACOCK, D. P. S. (ed.) 1977. Pottery and Early Commerce. Characterization and Trade in Roman and Later Ceramics. Academic Press, London. PEACOCK, D. P. S. 1982. Pottery in the Roman World. Longman, Harlow. PENSABENE, P. (ed.) 1985. Marmi antichi. Problemi d'impiego, di restauro e d'identificazione. "L'Erma" di Bretschneider, Rome. PENSABENE, P. 1994. Le vie del marmo. Itinerari osticusi 7, Rome. PENSABENE, P. (ed.) 1998. Marmi antichi II. Cave e tecnica di lavorazione. Provenienze e distribuzione. 'L'Erma' di Bretschneider, Rome. PICON, M. 1973. Introduction b l'Etude technique des Cdramiques sigilldes de Lezoux. Centre de Recherches Techniques Gallo-Romaines, Lyon. POLLARD, A. M. (ed.) 1992. New Developments in Archaeological Science. Proceedings of the British Academy, 77. POLLARD, A. M. & HERON, C. 1996. Archaeological Chemistry. Royal Society of Chemistry, Cambridge. POUPARD, L. & RICHARD, A. (eds) 2003. Marbres en Franche-Comtd. Actes des journdes d'dtudes Besanfon, 10-12 juin 1999. ASPRODIC, Besanqon. RICE, P. M. (ed.) 1984. Pots and Potters. Current Approaches in Ceramic Archaeology. Institute of Archaeology, University of California, Los Angeles, Monograph, XXIV. RICE, P. M. 1987. Potter3' Analysis: a Sourcebook. University of Chicago Press, Chicago, IL. RICE, P. M. (ed.) 1997. The Prehistory & History of Ceramic Kilns. Ceramics and Civilization, Vol. VII. American Ceramic Society, Westerville, OH. RIEDERER, J. 1981a. Kunstwerke chemisch betrachtet. Materialien. Analysen, Altersbestimmung. Springer, Berlin. RIEDERER, J. 1981b. Zum gegenw~irtigen Stand der naturwissenschaftlichen Untersuchung antiker Keramik. In: ARNOLD, D. (ed.) Studien zur altiig3ptischen Keramik. Ph. yon Zabern, Mainz. RIEDERER, J. 1987. Archiiologie und Chemie-Einblicke in die Vergangenheit. RathgenForschungslabor, Berlin. ROSEN, J. 2000./_zl Manufacture de Meillonnas (Ain). Etude d'une Fabrique de Cdramique rdgionale. 1760-1870 Temps modernes 2. Monique Mergoil, Montagnac. Roux, V. (ed.) 2000. Cornaline de l'Inde. Des Pratiques techniques de Cambay aux Technosystkmes

ARCHAEOMETRY: QUO VADIS? de I'Indus. Maison des Sciences de l'Homme, Paris. RUF, M. A., MAGGETTI, M. & VENDRELL-SAZ, M. (eds) 1999. Archaeological and Archaeometric Studies. Proceedings of the 4th European Meeting on Ancient Ceramics, Andorra 1995. Govern d'Andorra, Ministeri de Turismo i Cultura, Andorra. RYE, O. S. & EVANS, C. 1976. Traditional Pottery Techniques of Pakistan: Field and Laboratory Studies. Smithsonian Contributions to Anthropology, 21. SANTORO BIANCHI, S. & FABBRI, B. (eds) 1997. Il Contributo delle Analisi Archeometriche alldo Studio delle Ceramiche grezze e Comuni. I11rapporto forma/funzione/impasto. Atti della l a Giornata di Archeometria della Ceramica, Bologna, 28 febbraio 1997. University Press, Bologna. SAYRE, E. V. 1964. Some Ancient Glass Specimens with Compositions of Particular Archaeological Significance. Brookhaven National Laboratory, Brookhaven, NY. SCHVOERER, M. (ed.) 1999. Archdomatdriaux. Marbres et autres Roches. Actes de la IVb Confdrence internationale ASMOSIA IV, France, Bordeaux, Talence, 9-13 octobre 1995. Presses Universitaires de Bordeaux, Bordeaux. SEIBEL, F. 1998. Technologie und Fertigungstechniken rOmischer Glashiitten am Beispiel der Ausgrabungen im Hambacher Forst. Aktualistische Vergleiche und Modelle. Galda & Wilch, Berlin. SHACKLEY, M. S. (ed.) 1997. Method and Theory in Archaeological Volcanic Glass Studies. Advances in Archaeological and Museum Studies. Plenum, New York. SHEPARD, A. O. 1956. Ceramics for the Archeologist. Carnegie Institution of Washington. Washington, DC. SHORTLAND, A. J. (ed.) 2001. The Social Context of Technological Change. Oxbow Books, Oxford. SIEVEKING, G. & HART, M. B. (eds) 1986. The Scientific Study of Flint and Chert. Proceedings of the Fourth International Flint Symposium held at Brighton Polytechnic, 10-15 April 1983. Cambridge University Press, Cambridge. STEPPUHN, P. (ed.) 2003. Glashiitten im Gespriich. Berichte und Materialien vom 2. lnternationalen Symposium zur archiiologischen Erforschung mittelalterlicher und friihneuzeitlicher Glashiitten Europas. Schmidt-R6mhild, Ltibeck. TAIT, H. (ed.) 1991. Five Thousand Years of Glass. British Museum Press, London. TAYLOR, R. E. & AITKEN, M. J. (eds) 1997. Chronometric Dating in Archaeology. Plenum, New York. THOMPSON, I. 1982. Grog-Tempered 'Belgic' Pottery of South-Eastern England. British Archaeological Reports, 108. TITE, M. 1972. Methods of Physical Examination in Archaeology. Seminar Press, New York. TITE, M. S. 1991. Archaeological science--past achievements and future prospects. Archaeometry, 33(2), 139-151.

7

TITE, M. S. 2004. Archaeometry--an overview. In: MARTINI, M., MILAZZOM. & PIACENTINI,M. (eds) Physics Methods in Archaeometry. IOS Press, Amsterdam, 347-356. TORRENCE, R. 1986. Production and Exchange of Stone Tools. Cambridge University Press, Cambridge. TROU, M. & PODANY,J. (eds) 1990. Marble. Art Historical and Scientific Perspectives on Ancient Sculpture. Papers Delivered at a Symposium Organized by the Department of Antiquities and Antiques Conservation and Held at the J. Paul Getty Museum April 28-30, 1988. J. Paul Getty Museum, Malibu, CA. VANDIVER, P. B., DRUZIK, J. R., WHEELER, G. S. & FREESTONE, [. C. (eds) 1992. Materials in A n and Archaeology II1. Materials Research Society Symposium Proceedings, 267. VEECKMAN, J., JENNINGS, S., DUMORTIER, C., WHITEHOUSE, D. & VERHAEGHE, F. (eds) 2002. Majolica and Glass from Italy to Antwerp and Beyond. The Transfer of Technology in the 16thearly 17th Century. Stad Antwerpen, Antwerp. VELDE, B. & DRUC, I. C. 1999. Archaeologial Ceramic Materials. Springer, Berlin. VENDRELL-SAZ, M., PRADELL, T. & MOLERA I MAITE GALERA, J. (eds.) 1995. Estudis sobre cer&mica antiga. Studies on Ancient Ceramics. In: Proceedings of the 2nd European Meeting on Ancient Ceramics, Barcelona, 18-21 November 1993. VICENZINI, P. (ed.) 1995. The Ceramics Cultural Heritage. In: Proceedings of the International Symposium The Ceramics Heritage of the 8th CIMTEC-World Ceramics Congress and Forum on New Materials, Florence, Italy, June 28-July 2, 1994. Monographs in Materials and Society, 2. VIDALE, M. 1992. I givani siberiani. Archaeometria della ceramica antica e il problema delle nuove figure professionali. In: SANTORO BIANCHI, S., FA1LLA, A. & GUERMANDI,M. P. (eds).Archeometria della Ceramica, Problemi di Metodo. Int. Centro Ceramico, Bologna, 131 - 150. WAGNER, G. A. 1998. Age Determination of Young Rocks and Artefacts--Physical and Chemical Clocks in Quaternary Geology and Archaeology. Springer, Berlin. WAGNER, G. A. • VAN DEN HAUTE, P. 1992. Fission Track Dating. Enke, Stuttgart. WEDEPOHL, K. H. 2003. Glas in Antike und Mittelalter. Geschichte eines Werkstoffes. Schweizerbart, Stuttgart. WERTIME, T. A. & WERTIME, S. F. (eds) 1982. Early Pyrotechnology: the Evolution of the First Fireusing Industries. Smithsonian Institution Press, Washington, DC. WHITBREAD, I. K. 1995. Greek Transport Amphorae. A Petrological and Archaeological Study. British School at Athens, Fitch Laboratory Occasional Papers, 4. WIDEMANN, F. 1982. Why is archaeometry so boring for archaeologists? In: OLIN, J. & FRANKLIN, F. (eds)Archaeological Ceramics.

8

M. MAGGE'Iq'I

Smithsonian Institution Press, Washington, DC, 29-34. WILSON, T. (ed.) 1991. Italian Renaissance Potterr: Papers Written in Association with a Colloquium at the British Museum. British Museum Publications, London. WINKLER, E. M. 1973. Stone. Properties, Durability in Man's Environment. Springer, Berlin.

WINTER, A. 1978. Die antike Glanztonkeramik. Praktische Versuche. Philipp von Zabern, Mainz. WOOD, N. 1999. Chinese Glazes. Their Origins, Chemistry and Recreation. A&C Black, London. YOUNG, W. J. (ed.) 1973. Applications of Science in Examination of Works of Art. Proceedings of the Seminar, 15-19 June 1970. Museum of Fine Arts, Boston, MA.

A review of the non-destructive identification of diverse geomaterials in the cultural heritage using different configurations of Raman spectroscopy D A V I D C. SMITH Museum National d'Histoire Naturelle, Laboratoire LEME, USM0205, 61 Rue Buffon, 75005 Paris, France (e-mail: [email protected]) Abstract: Non-destructive Raman microscopy (RM) applied to geomaterials in the cultural

heritage is reviewed by means of explaining selected examples representative of the different kinds of geomaterials that can be characterized and of the different kinds of analytical configuration that can be employed. To explain the versatility and considerable analytical potential of RM that result from its unique combination of capabilities, the first sections summarize the theory and practice of the method and its advantages and disadvantages. The most modern configurations (mobile RM (MRM) and ultra-mobile RM; micromapping and imaging; telescopy) are described. Applications in the new age of 'don't move it, don't even touch it' archaeometry have previously been classified into 10 domains, seven of which concern geomaterials: gems; rocks; ceramics; corroded metals; coloured vitreous materials; and mineral pigments on an inorganic or organic substrate. The representative examples here include all these domains and cover the time range from Prehistoric through Egyptian, Roman, Meso-American, Medieval, Chinese, Renaissance and Mogul cultures to modern colouring of glass and a contemporaneous simulation of submarine archaeology.

The analysis of geomaterials in the cultural heritage, to clarify the nature of the material employed, evaluate possible provenances, detect treatments or to recognize fakes, calls for a variety of techniques, depending upon the type of material available and the kind of information sought. Raman microscopy (RM) (one kind of Raman spectroscopy (RS)) has become an important technique in archaeometric studies in archaeology and art history since about 1996, and the pseudo-acronym 'ARCHAEORAMAN' was coined by Smith & Edwards (1998) to summarize this wide field of research activity. More recently the term 'mobile Raman microscopy' (MRM) (Smith 1999) was employed to analyse art works in situ inside museums by taking the laboratory to the object, rather than the object to the laboratory as in conventional 'immobile Raman microscopy' (IRM). Subsequently, the possibility of using MRM for subaquatic archaeology was evaluated positively (Smith 2003), and more recently Raman micromapping has been used to clarify the microstructural mineralogy of artworks (Smith 2004a) or of rocks susceptible to be the provenance thereof (Smith 2004b,c). The most recent development in RS is telescopy (Sharma et al. 2002, 2003)

for very remote studies (such as planetology); this approach has not yet been applied to archaeology, but it could be useful for analysing gemstones in shop windows from across the street, which brings us into the domain of 'Raman spying' (Smith 2005a), and 21st-century social science, which will not be pursued here. Future developments will no doubt soon include synthetic vocal replies for automated analysis (Smith 2005a). In 1986, during a review of RM applications to mineralogy in general, Smith (1987) argued that RM should be of great value to archaeometry, but no significant studies were known to the geological community at that time, except for some pioneering studies on gemstones and their microinclusions (Drlr-Dubois et al. 1981a,b, 1986a,b). In fact, some chemists and physicists had already begun analysing artworks (Delhaye et al. 1985; Guineau 1987), but only pigments, and only publishing in journals in fields other than geology or mineralogy, especially chemistry or art history; furthermore, they generally avoided mineralogical terminology by using chemical names such as mercury sulphide or colour names such as vermilion instead of mineral names such as cinnabar. At the

From: MAGGETTI,M. & MESSlOA,B. (eds) 2006. Geomaterials in Cultural Heritage. Geological Society, London, Special Publications, 257, 9-32. 0305-8719/06/$15.00 ~:) The Geological Society of London 2006.

iPrehistoric i

i

!Roman

Ukraine

water

4 x micro vertical immobile in lab x

5 x micro vertical immobile in lab x air ~

6 x micro vertical immobile in lab x air x

~

r/"c3

, __

7 x micro vertical immobile in lab x air x

8 x micro vertical immobile in lab x air

9 x micro vertical immobile in lab x air '

10 x nlcro ~ imTdoite in lab x air trap

11 x moo ~ irm'ebile in lab x air trap ~ |

12 Illl x rricro ~ m3bile in situ fibres ar x

~ /

....

i

i i

Barbet (1999)

i .

icorr~ed ic°ins (Fe' i& °ther !Cu' Zn Pb iartefacts in ;i

Smith (2001, 2005b)

(2001), Bouchard & Sn~th /2005b)

I

~

o ~

i I

~

(1999b); (2004b); (2004a) Bouchard & Smth Smith (2005c) /2002/

iglass

ipgrnentson i or n stained!

i

ia

_

'in

i axehead,

! Meso-

iA~ncan

(1°J99a); Robin (2001b, Gendron Vemioles Smith & (1997) 2003) (1997a), (1997) Bouchard Smith /2000al /2005cl Notes to tabular part: x denotes "no"; bold type in the configurations highlights unusual features.

references

! i

ivrmo,

iI~

i i i

',METALLOIRAMAN

- -

i

i

....................................................... ! .....

i

ievaluation IofRMo n igemstones ~underwater i

'PETRO.

iRoman iintaglios' .France

',~A~N

!GEMI~ RAI~N

(1999, 2900); Smth /2005at

!l~ished

am~c~

.

.......................................................................................... i..................................... I. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . i lCONO-

i~s, Ro~, i

[pigments in !pigmantsin i

i RAMAN

3 x micro vertical immobile in lab x air x

iFRESCO-

2 yes micro vertical immobile in lab x air

1 yes micro vertical immobile in lab x air x

micro-extraction macro/micro horizontal/vertical mobile/immobile in situ/in lab optical fibre head under micro-mapping

i

(2001a, 2005a); Srffth elal, /2003a/

~

i

,

13 x x x mobile in situ til~'es air

Snith (2002); Srffth t2005a/

L

14 x x x mobile in situ fibres air

i

blue

17 x x x rn:Yoile in situ fibres air x

a ~oodm slalue,

!i i

i

16 x x x m:3bile in situ fit:~es gl~ss x

-

'

! ~

!

' ,

.- _

'~ ........................ JR/IM~

(2001a,b) S r ~ & references Lmblanchel (unpub. data/

. . . . . (2003b); F~ndeau ~?th (2001); (2005a) Srdth /20(35a/

!

i~"Y

i

~................................. }.............................................

cave,

Ii~a i

18 x r r i ~ x macro/rricro x ~ ~ u/~ra-].l~l,f rn~nn'doile in situ in situ/in lab all in one opticalfibrelead air under x nicro-map~ng i I:~ristmc i iPg~ , ~

~'-- ....................... ~ ........................~ i n c ~ i n irnarl~ MIdl:le/~ , . , ~ i s o r ~ i stone J

15 x x x rrd)ile in situ fibres air x

NON-DESTRUCTIVE RAMAN SPECTROSCOPY international GEORAMAN-1996 conference in Nantes an attempt was made to bring ARCHAEORAMAN topics to the attention of the geological community and since then contributions on archaeology and art history became significant at every G E O R A M A N meeting (1999 in Valladolid; 2002 in Prague; 2004 in Honolulu) (see table 8 of Smith & Carabatos-Nedrlec (2001) for a list of archaeological or art historical topics presented at these meetings). Another series of international congresses on nondestructive analysis in the cultural heritage brought in RM at Antwerp in 2002, and this continued at Lecce in 2005. The meetings of ICAM (International Congress on Applied Mineralogy), IRUG (InfraRed users Group), GFSV (Groupe Franqais de la Spectroscopie Vibrationelle), and GMPCA (Groupe des M~thodes Pluridisciplinaires Contribuant h l'Archrologie), and others, have started to include RM, as have other more archaeological meetings (e.g. Smith et al. 2000). A separate series of international congresses on exclusively 'Raman Spectroscopy applied to Archaeology and Art History' ('ArtRaman') was started in London in 2001 and continued in Ghent in 2003 and in Paris in 2005. The literature on A R C H A E O R A M A N has thus increased enormously in a decade, but it is dissipated amongst journals in many disciplines. This paper cannot review all the literature; it thus focuses on explaining why RM is so useful and describes a series of examples of studies by the author's research group that are in two ways representative: of the different kinds of geomaterials that can be analysed, and of the different kinds of analytical configuration that can be employed (Fig. 1).

What is Raman spectroscopy? RS is an optical, hence physical, technique by which the wavelength of light is modified by interactions with interatomic vibrations

11

(e.g. Smith & Carabatos-Nrdelec 2001; Nasdala et al. 2004). The modified light is called Raman diffused light according to the 'Raman effect' discovered by Sir Chandrasekhara Venkata Raman in 1928, for which he received the Nobel Prize for Physics. Thus the technique does not analyse a single atom, as do a great number of chemical analytical techniques such as X-ray fluorescence, as at least two atoms are required. The vibrational energies involved are the same as those in infrared (IR) spectroscopy, such that the two techniques are often considered similar. They are indeed complementary, but are not really similar, because in IR spectroscopy photons are absorbed or reflected according to the various vibrational energies, whereas in RS, incoming photons lose some energy, which leaves a vibration mode more excited, and hence the outcoming photons have lost some energy, i.e. they have a higher wavelength, and hence a lower wavenumber (the reciprocal of wavelength) (Fig. 2). This is called Raman Stokes scattering. Raman AntiStokes scattering also occurs whereby a vibration mode gives up some energy to become less excited and the outcoming photons have gained energy, i.e. they have a lower wavelength, and hence a higher wavenumber; this effect is weaker and will be ignored here. Thus with Raman Stokes scattering a single kind of interatomic vibration causes a shift of the wavenumber of the incoming exciting light, usually from a laser (although Raman used sunlight) and necessarily monochromatic. The exciting wavelength (e.g. 514.5nm from an Ar + green laser or 632.8 nm from a H e - N e red laser) is placed at zero cm - l on the relative wavenumber scale such that the Raman band created occurs at a characteristic Raman shift (e.g. 465 cm -1 from the major vibration of quartz). Raman shifts are conventionally plotted as being positive, as a shift is an amount without direction, but in reality it should be plotted as - 4 6 5 cm -1, as

Fig. 1. Representative examples of ARCHAEORAMAN studies on geomaterials: configurations, domains and images. Tabular part: configurations listed horizontally; domains listed diagonally; examples placed in the appropriate case. Arrowed superimposed images demonstrate the following selected cases. (a) Raman spectra from the Meso-American stone axe-head in eclogite; from top to bottom: titanite, garnet, clinoamphibole, clinopyroxene (modified after Smith & Gendron 1997a). (b) A Domitian denier silver alloy coin with cuprite corrosion (modified after Bouchard & Smith 2005b). (e) Raman spectra of microcline under air, distilled water and water badly contaminated by animal or vegetable debris as a simulation of subaquatic archaeology (modified after Smith 2003). (d) A Meso-American corroded metal axe-head (modified after Bouchard & Smith 2005b). (e) An Egyptian inscribed commemorative scarab in polycrystalline enstatite established by Raman mapping with a RENISHAW® Invia® spectrometer (modified after Smith 2004a). (f) A Chinese sculptured pendant in jadeite-jade (modified after Smith 2005a). (g) A Medieval cloisonnr-gold style fibula encrusted with garnets (photo D. C. Smith'S). (h) A Teotihuac~in sculptured mask in marble with the DILOR® LabRaman® horizontal microscope (modified after Nasdala et al. 2004). (i) A Florentine table in stone marquetry being analysed vertically with a KAISER® Holoprobe® remote head through the thick protective plate glass (invisible here) (modified after Smith 2005a).

D. C. S M I T H elastic

inelastic

inelastic

scattering

scattering

scattering

Rayleigh diffusion

Raman Stokes diffusion

Raman Anti-Stokes diffusion

Ee + E i +E,

vibrational

virtual state

E~+E,

Ee

excited state

Es

~round state

energy levels

extra relations

input (i) energy

E,

E,

Ei

frequency

v, = E , / h

v, = E i / h

vi = E i ] h

= h. v i = c / Li

wavelength

k~ = c /v,

~.~= c / v ,

k, = c / v ,

= h . c / Ei

wavenumber (absolute) wavenumber (relative)

W, = 1 / k,

W i = 1 / k~ W, set at zero

W, = 1 / k i W, set at zero

= v, / c

change to sample change to light

none

(EcEg) gained

(E~-Eg) lost

none

(Ee-Eg) lost

(Ee-Eg) gained

output (o~ energy

Eo = Ei

Eo = Ei - ( E e-E~)

E o = E, + ( E e - E 0

frequency

vo = E o / h

vo= E o/ h

vo = E o / h

wavelength wavenumber (absolute)

k o = c / Vo Wo= I /~o

~.o = c / v o W o = ! /ko

ko = c / v o Wo= 1 /k o higher

comparison: input to output

none

lower

wavenumber (relative)

Wr = W, - W o

Wr = W o - W 1

Wr= W 0- W

= Raman shift

zero

negative

positive

c = speed o f light; h = Planck's constant Fig. 2. S c h e m e o f the different ways in which inter-atomic vibrational energy levels give rise to three types o f light scattering (diffusion): Rayleigh, R a m a n Stokes, R a m a n Anti-Stokes.

t h e a b s o l u t e w a v e n u m b e r is l o w e r t h a n t h a t o f t h e e x c i t i n g l i n e ( F i g . 2). A n i m p o r t a n t p o i n t is that Raman shifts are constant for any wavel e n g t h o f t h e e x c i t i n g l a s e r as t h e s h i f t s a r e f i x e d r e l a t i v e to t h a t w a v e l e n g t h a n d a r e l i n e a r in c m - ~ ; v e r y f e w e x c e p t i o n s to this r u l e o c c u r (e.g. t h e D b a n d o f g r a p h i t e ) . As there are many different kinds of v i b r a t i o n a l s y m m e t r y , e a c h w i t h its o w n e n e r g y

l e v e l (e.g. s y m m e t r i c s t r e t c h i n g , a n t i - s y m m e t r i c stretching, deformation, bending, rocking, w a g g i n g , t w i s t i n g ) , a n d all o f this f o r e a c h k i n d of combination of chemical elements (depending upon the Raman 'selection rules', which depend upon the crystal or molecular symmetry and also u p o n the n u m b e r o f c h e m i c a l e l e m e n t s present), there are several distinct Raman bands created (which occasionally overlap) such that

NON-DESTRUCTIVE RAMAN SPECTROSCOPY a spectrum is obtained (where the ordinate shows photon intensity, and the abscissa shows the wavenumber) (Fig. l a and c). According to these rules, some materials give only one band (e.g. diamond), simple carbonates and sulphates give fewer than 10 bands, silicates such as garnet give about 20, and more complex silicates such as micas and amphiboles may give more. Organic molecules may give rise to hundreds of bands. Spectra are variably plotted with the zero at the left or the right, but the zero is never plotted as this is where Rayleigh scattering occurs; this involves the restitution of the exciting light with the same wavelength (Fig. 2) (physically not the same, but effectively the same as simple reflection). Rayleigh scattering is very approximately 1012 times more efficient than Raman scattering and this important fact has several consequences: (1) a Raman spectrum cannot show the intensity at 0 c m - i as it would plot somewhere in interplanetary space; (2) it would burn the detector, or create a plasma from it, and has to be filtered out; (3) a 'Rayleigh tail' occurs in the 10100 cm -1 spectral range where the Rayleigh scattering intensity decreases to zero; (4) only about one photon in several billion incoming photons is subject to the Raman effect, so the development of RM necessitated strong laser sources and powerful detectors of very weak signals as well as coupling to a microscope (Dhamelincourt & Bisson 1977); (5) commonly 1 - 1 0 0 mW power is used to analyse a 1 Ixm sized portion of a sample or an art object; if the same power per ixm2 were applied over a 1 m 2 surface it would need 1012 times more power, i.e. 1 - 1 0 0 GW, which brings us to the scale of several nuclear power stations (and this ignores the third dimension and another 106). Thus we are dealing with an extremely powerful energy applied to an extremely small location to detect an extremely weak effect. It is important to appreciate that the intensity of a Raman band of a crystal depends, often strongly, on the orientation of its crystal symmetry with respect to the polarization of the laser (compare X-ray diffraction) such that in certain situations a Raman band may disappear completely. If it is not possible to rotate either the art work or the RM, one can introduce a half-wave plate and rotate it to see the missing band (Smith 1996). There are basically two ways of using RS. One approach uses RS to satisfy the chemist's, physicist's or mineral physicist's need to try to predict and calculate Raman phenomena and to extract thermodynamical data, often by measuring Raman spectra at high or low

13

temperature (T) and/or high or low pressure (P); this is not discussed further here. The second is to use 'Raman spectral fingerprinting' (Dhamelincourt & Bisson 1977; Smith 1987) to identify mineral or molecular species, as different species cannot give the same spectrum and the same species will always give the same spectrum (at the same P - T, if there are no differences in chemical composition, crystal structural order, etc.). This of course requires spectral databases; several now exist, but all are limited in scope (see White 1975; Griffith 1987; Guineau 1987; Pinet et al. 1992; Bell et al. 1997; Burgio & Clark 2001; Bouchard & Smith 2003, 2005a) and numerous others are in preparation as every Raman research group builds its own.

Why has Raman microscopy become so polyvalent and powerful? This is principally because of its great versatility owing to its unique combination of capabilities, as follows. (1) It characterizes simultaneously the physical structure and the chemical composition of an unknown species by comparison of its Raman spectrum with reference spectra (compare IR and X-ray diffraction (XRD)). This is extremely useful for distinguishing polymorphs such as quartz-moganitetridymite-cristobalite-coesite (SiO2), aragonitecalcite (CaCO3), sanidine-orthoclase-microcline (KA1SiO3), rutile-anatase-brookite (TiO2), etc., which cannot be done with any purely chemical technique. (2) It can do this with inorganic or organic material in different states or forms, such as crystalline, molecular, glassy or amorphous; whether solid, powdery, suspended, plastic, vitreous, liquid or gaseous; and whether pure or mixed. Apparently only IR can also do this. Mixed phases, such as in a pigment or in sub-micronsized mineral intergrowths in a rock, gem or ceramic, are commonly encountered in archaeometry. (3) The analysed volume may be on a micrometre scale, from about 0.5 p~m to about 50 l~m in surface diameter, commonly 1 - 2 txm, but the analysed object may have any size and different parts thereof may be systematically analysed. IR and XRD cannot do this except with a synchrotron (which must be the least mobile analytical apparatus). (4) No sample preparation whatsoever is required (no extracting, drilling, scraping, sawing, cutting, grinding, polishing, liquefaction, gasification, etc., nor a vacuum chamber,

14

D.C. SMITH

KBr pelleting, or other kind of processing) as the method is non-destructive; with an appropriate reflection configuration IR can also be non-destructive. This non-destructive property is true as long as one maintains a laser power sufficiently low to avoid damage; if, unfortunately, this is not achieved then a micron-sized volume of the analysed object may be 'burned' or otherwise disintegrated, but fortunately this will be invisible to the naked eye and harmless to most materials such as gemstones, although it could become dangerous for inflammable materials such as the paper of a priceless ancient book. (5) RM can provide micro-mapping or microimagery of textures of intergrown phases, of chemically zoned crystals or of physically deformed crystals. Other techniques can map structures, but IR is on a larger scale. (6) With the use of mobile optical fibres one can analyse any part of an artefact (including re-entrant angles such as under the arm of a statue or gemstones mounted inside a crown). (7) MRM may be carried out almost anywhere, such as in situ inside a museum display cabinet, a conservation or storage building, or on an archaeological site. (8) One can identify a phase under another transparent one, such as microinclusions inside a mineral, as well as pigments under glass, gems under plastic, or statues under water, so that submarine archaeology by MRM has become possible (Smith 2003). (9) One can obtain semi-quantitative chemical analysis of mineral solid-solutions by RM for example, by using the RAMANITA method devised by Smith & Pinet (1989), calibrated by Pinet & Smith (1993, 1994) and updated by Smith (2002b, 2004d, 2005b). The method is based on the time-consuming calibration of wavenumber shifts along each binary join (if natural or synthetic samples are available) and then within various choices of multivariant chemical space. All these possibilities and developments led Smith (2002a) to declare that 'The new age of "don't move it, don't even touch it" archaeometry has now arrived to allow remote non-destructive characterisation in all the domains of ARCHAEORAMAN and in situ almost anywhere'.

What disadvantages exist with Raman microscopy? As with all analytical techniques there are some disadvantages with RM, but they are small in number compared with the advantages. Very

few minerals give no Raman band at all because they have a high symmetry and a low number of different atoms in the unit cell (e.g. halite (NaCI)). Most pure metals give no Raman signal, partly for the preceding reason, and partly because of their high reflectivity; on the other hand, as soon as a metal is corroded to form oxides, hydroxides, carbonates, sulphates, chlorides, etc., RM works very well. Opaque or semi-opaque minerals absorb too much light and give either no Raman signal or a very weak one; manganese oxyhydroxides are a good example as they have been difficult to recognize in pigments; however, with more recent instrumentation one can now obtain Raman spectra from many of these phases (Ospitali & Smith 2005). Some materials are rather photosensitive and need low laser power to avoid instantaneous dehydration (e.g. iron hydroxides and lead hydroxides). The detector picks up not only the Raman signal but also various kinds of 'parasite' signals, such as laser lines from the laser source that have not been sufficiently well filtered, cosmic rays, daylight, incandescent room light, Hg and Ne emissions in common neon 'fluorescent tube' lamps, photoluminescence (PL) from chemical impurities in the sample or in the optical trajectory (e.g. the infamous 843 c m - I band from the Olympus × 50 objective), or fluorescence. These sources can be attenuated by laser filters, by reducing daylight or room light, or by changing the exciting laser wavelength such that photoluminescence lines occur elsewhere in the spectrum. Background fluorescence, which gives a very high baseline that partially or totally obscures the Raman spectrum, is no doubt the worst problem, but its true cause is not always obvious. It is known that it can come from electronic transitions in imperfectly crystallized minerals, from some nanocrystalline materials such as clays, and from mixed organic materials (living or dead). If the parasite does not interfere in the same spectral range as relevant Raman bands then the problem is avoided. Waiting a few minutes before acquiring spectra usually reduces the baseline, perhaps as a result of some annealing by heating. Analysing under water is beneficial (Smith et al. 1999a; Smith 2003). Changing the exciting laser wavelength often (but not always) creates drastic improvements. Pulsing the laser is an excellent antidote but it is not easy to acquire the necessary configuration. Interchanging a troublesome optical component (e.g. filter, mirror, objective) in the RM system with one of a different kind will cure the problem in some cases.

NON-DESTRUCTIVE RAMAN SPECTROSCOPY Raman spectra frequently need some amount of spectral treatment if we are to be able to exploit the data by spectral fingerprinting (on the other hand, treatments are usually avoided for thermodynamical studies as one must not modify the raw data upon which certain calculations are based). First, the 'baseline correction' tries to make the background line horizontal, regardless of the cause of it not being flat (fluorescence, luminescence); this procedure can dramatically increase signal-to-noise visibility. Subtracting an oblique straight line is acceptable if the baseline has a sub-linear steep slope, but often it is necessary to subtract a polynomial 'best line' curve calculated from selected landmarks on a distinctly curved baseline. Automatic correction can be disastrous as the computer program may confuse wide Raman bands with an undulating baseline. More than a × 2 polynomial can produce major distortions and, in any case, it is not necessary to achieve a perfectly flat baseline. Second, one may eliminate known parasite peaks or known detector defects by 'rubbing out' with the computer mouse instead of a piece of rubber. An automatic 'peak elimination' procedure may be useful for eliminating narrow cosmic rays that are distinctly narrower than Raman bands, but it needs to be used with care. Third, 'smoothing' by averaging all intensities over a selected small wavenumber zone is very useful to make real Raman bands more visible by eliminating the basic zigzags of the irreducible background flutter, but must not be done over zones too wide or real Raman bands will become too diluted in intensity or separate nearby bands (doublets) may become fused together. With these three treatments one can frequently transform apparently hopeless spectra into perfectly exploitable ones, and this is because the basic information exists in the raw spectrum and it just needs to be rendered visible. A variety of more sophisticated computerized techniques exist, such as spectral combination, peak-fitting, Fourier transforms and 3D-plotting, but they will not be dealt with further here.

Classifications of Raman microscopic studies of the cultural heritage To demonstrate applications of RM to the cultural heritage it is convenient to classify the examples according to some criteria. Here the cultural period (Prehistoric, Roman, Medieval, Renaissance, etc.) is not used as this paper is more mineralogical-technological than archaeological. The type of material analysed can be a

15

Table 1. The 10 domains of ARCHAEORAMAN, updated from Smith (1999, 2002a) (1) GEMMORAMAN from 'gems': gemstones (rough, cut or mounted), cameos, corals, intaglios, jewellery, collection stones, etc. (2) CERAMIRAMAN from 'ceramics': brick, china, earthenware, faience, glass, porcelain, pottery, slags, tiles, other vitrified minerals, etc. (3) PETRORAMAN from 'petros' for rocks: axeheads, building columns, ceremonial stones, inlaid rock, millstones, mosaics, necklaces, sculptures, etc. (4) METALLORAMAN from 'corroded metals': corroded bracelets, coins, cutlery, necklaces, statues, swords, tools, etc. (5) RESINORAMAN from 'resin' as an example of a non-cellular organic material composed of only a few different molecules or of amorphous hydrocarbons without a growth texture: amber, glue, gum, oil, putty, wax, bitumen, lignite, coal, etc. (6) TISSUERAMAN from 'tissue' as an example of cellular organic molecules or biominerals with a growth texture: bone, claw, cotton, feather, fur, hair, hoof, horn, ivory, leather, linen, nail, papyrus, parchment, silk, skin, teeth, wool, wood, etc. (7) FRESCORAMAN from 'fresco' as an example of pigments/inks/dyes on or in an inorganic substrate: brick, ceramic, plaster, stone, stucco, etc. (8) ICONORAMAN from 'icon' as an example of pigments/inks/dyes on or in an organic substrate: bone, canvas, paper, skin, textile, wood, etc. (9) VITRORAMAN from the 'vitreous' state: pigments on or in enamel, glass or glaze, etc. (10) ENVIRORAMAN from 'environmental' deterioration of any of these materials by climate, burial or immersion: original materials, corrosive agents involved, intermediate and final products

useful criterion, and this was used by Smith & Edwards (1998) as there may be different analytical protocols for different materials; Table 1 lists the 10 domains of research activity as updated by Smith (2002a). The analytical configuration employed is also relevant (macro or micro; vertical or horizontal microscope; optical fibres or not; mobile or immobile; in situ or in a laboratory; under air, glass, mineral, or water). Figure 1 plots the seven domains relevant to geomaterials against combinations of analytical configurations and lists the studies (by the author's research group) that are mentioned here as being representative of research in A R C H A E O R A M A N in general.

16

D.C. SMITH

RM analysis of pigments, whether inorganic or organic materials on inorganic (FRESCORAMAN) or organic (ICONORAMAN) substrates has dominated ARCHAEORAMAN from the early works of Delhaye et al. (1985) and Guineau (1987) to the production of minicatalogues of Raman spectra of pigments (Bell et al. 1997; Burgio & Clark 2001), and from applications to prehistoric rock art (e.g. Bouchard 1998, 2001; Edwards et al. 1998; Smith et aL 1999a,b; Smith & Bouchard 2000a) via Roman art (e.g. Smith & Barbet 1999) through various periods of the last millennium (e.g. Rull-Perez et al. 1999; Withnall 1999; Rull-Perez 2001) to modem art (e.g. Vandenabeele et al. 2000). The biomaterials domains RESINORAMAN and TISSUERAMAN have been mainly limited to specialists in biology and/or organic chemistry (e.g. the early works of Edwards et ai. 1996a,b,c; Brody et al. 1998). Turning to geomaterials, the earliest known work was on GEMMORAMAN (D~lr-Dubois et al. 1978). The advantages for gemmology are considerable, as RM can be employed for several different purposes: to verify the nature of the gemstone itself, to examine for treatments (e.g. heating, resin impregnation, pigmentation), to explore solid or fluid microinclusions, or to detect synthetic and imitation stones. Certain aspects of gemmology have been studied in detail by RM by Lasnier (1989) and Maestrati (1989), and the first catalogue of the Raman spectra of gemstones was published by Pinet et al. (1992); more recent studies have been made notably by Coupry & Brissaud (1996), Schmetzer et al. (1997), Smith & Robin (1997), Smith & Bouchard (2000b), Kiefert et al. (2001) and Smith (2001a, 2005a). Apart from extremely few early works (Coupry et al. 1993; Macquet 1994; Wang et al. 1995), 1997 saw the effective beginning of RM studies in the remaining four geomaterial domains, in particular: (1) PETRORAMAN of jade and eclogite by Smith & Gendron (1997a,b) or of sculptured polished ceremonial rocks by Smith & Bouchard (2000b) and Smith (2005a); (2) CERAMIRAMAN of vitrified forts by Smith & Vernioles (1997), of the minerals constituting pottery by Fry et al. (1998) or of the pigments in glazes by Colomban & Treppoz (2001), Colomban et al. (2001) and Liem et al. (2000, 2002); (3) METALLORAMAN on corroded metal coins and various archaeological metals (Fig. l b and d) by McCann et al. (1999), Bouchard & Smith (2000a,b, 2001, 2005a,b), Bouchard (2001), Di Lonardo et al. (2002), Frost et al. (2002a,b),

Smith & Bouchard (2002) and Martens et al. (2003); (4) VITRORAMAN on the minerals colouring stained glass by Edwards & Tait (1998), Smith et al. (1999c) and Bouchard & Smith (2002). ENVIRORAMAN studies are less common (e.g. Seaward & Edwards 1998). The RM spectral catalogues of Bouchard & Smith (2003, 2005a) included minerals of relevance to prehistoric paintings, corroded metals and stained glass. Probably at least 80% of all ARCHAEORAMAN publications to date concern pigments. Apparently over 90% of all RM analysts are physicists or chemists, which is logical given the physico-chemical basis of the technique. Thus, like botanists and zoologists, geologists of one kind or another (e.g. crystal chemists, mineralogists or petrographers) engaged in archaeometry via RM make up a very small community worldwide. However, each specialist brings his own particular competence and, similar to the need for an experienced botanist to identify a kind of tree, geologists are clearly necessary when studying natural rock artefacts from the cultural heritage (and solid-solutions, microinclusions, transformations, etc. in their constituent minerals, and their possible provenance in one or other geological unit). It was argued by Smith & Edwards (1998) that ARCHAEORAMAN studies really require three co-authors, a spectroscopist for the analysis, a natural scientist for the species of the natural raw material, and a social scientist for the artefact (form and cultural context). Individual scientists can often manage to adequately cover two of these disciplines, but to cover all three properly (or all five if one separates geology, botany and zoology) would be utopia, surely requiring a born-again Leonardo da Vinci. The following sections, organized by analytical configuration, focus on the geomaterials applications listed in Table 1.

Representative examples of RM applications I m m o b i l e a n a l y s i s in a l a b o r a t o r y : u n d e r air with a vertical microscope

This is the standard method of performing archaeometry with RM, either by placing on the microscope stage micro-samples extracted from a cultural item (i.e. not strictly non-destructive in this case) or by placing the whole item under the microscope if it is small enough to squeeze between the objective and the stage, or by taking away the stage. Methods.

NON-DESTRUCTIVE RAMAN SPECTROSCOPY

i

Wavenumber (cm -I)

500

1000

~

1500

Wavenumber ( cm- I )

400

17

600

800

1000

(c)

....

oo

,ooo

'z°° (d)

Fig. 3. Raman spectra of selected subjects. (a) Raman spectra of pigments from the Roman tomb at Kertsch (Ukraine): red minium (BJMI55yy, bottom left); blue cuprorivaite (BHCV03zz, top left); black carbon (BHCA21hh, right) (modified from Smith & Barbet 1999). Int, intensity. Eight-digit codenames are the computer spectra filenames. (b) Raman spectra of minerals from the new type of jadeite-jade from Guatemala: from top to bottom: jadeite alone (AHCP03 mm); jadeite + quartz (key peak at 468 cm- 1, AJQZ05 mm); jadeite + rutile (key peak at 445 cm- ~, AGCP22 mm); jadeite + titanite (key peak at 543 cm- 1, AHUN 16 ram). Some of the key peaks of jadeite are present in all spectra: 203, 373,698, 986, 1039 cm -1 (modified from Gendron et al. 2002). (e) Raman spectra of Cu-hydroxysulphates, from top to bottom: archaeological brochantite (DGCU 17je); standard brochantite (BSCUO6je); archaeological antlerite (CRCU08je); standard antlerite (BOCU08je). Some bands are at the same wavenumber in all spectra but there are significant shifts between the two species, notably the SO]- vibration just below 1000 cm-l (modified from Bouchard & Smith 2005b). (d) Raman spectra of the interior of two modem glasses: colourless 'verre cord616' (top, BUVE071f) showing intense bands revealing a high Na content (573 cm-i) and a tectosilicate Si-O arrangement (1100 cm- 1); red 'verre antique' (bottom, BQCO04jv) dominated by the bands at 195 cm-1 (CdSe) & 288 c m - l (CdS) characteristic of CdSo.g5Seo.55(modified from Bouchard & Smith 2005b).

A D I L O R ® X Y ® spectrometer belonging to the M u s e u m National d'Histoire Naturelle (MNHN) was employed.

Pigments: Roman wall-paintings. Black, red and blue are the major colours in decorations on a wall-painted Roman tomb at Kertsch, Ukraine. Micro-samples more or less invisible to the naked eye were extracted by the archaeologist A. Barbet and submitted to RM examination. It was easy to focus the 1 - 2 tzm diameter laser beam onto any selected mineral grain or part of a composite micro-assemblage to determine its mineral constitution (Smith & Barbet 1999). In this way it was found that the black is semi-amorphous carbon (C) (Fig. 3a);

this is a very c o m m o n phase in all cultures (often called 'carbon black', but such varietal names are not always used with precision) and it was probably the first pigment ever used by mankind. The blue pigmentation derived from cuprorivaite (CaCuSi4Olo) (Fig. 3a), which is the key constituent in the pigment called 'Egyptian Blue' and which was widely used in the Roman Empire. The red turned out to be m i n i u m (Pb304) (Fig. 3a); although k n o w n elsewhere in the R o m a n Empire it was not previously k n o w n as far NE as Kertsch.

Pigments: Prehistoric cave paintings. Although RM work on pigments had begun in the mid1980s, it was not until the late 1990s that R M

18

D.C. SMITH

analysis of Prehistoric pigments from surface rock art (Edwards et al. 1998; Smith et al. 1999b) or cave wall-paintings (Smith et al. 1999a; Smith & Bouchard 2000a; Bouchard 2001) was attempted. Prehistoric pigments are, in general, far more difficult to determine than pigments from historical times. This is not only because they tend to give an enormous fluorescence but also because the three most common phases used, other than carbon black, each have an additional problem. Thus yellow goethite (a-FeO(OH)) rapidly dehydrates to form red hematite (a-Fe203) even at very low laser power; red hematite strongly absorbs a green laser beam, overheats and decomposes into a black dot that might contain magnetite (Fe304); black MnxOvOHz phases absorb so much light that they give particularly bad Raman spectra. In the case of the limestone caves Pergouset, Les Merveilles and Les Fieux, in the Quercy district, Lot, France, it was possible to identify on various drawings (lines, dots, negative hands, etc.) predominant hematite with minor goethite in the red colours, and carbon in most black parts. Some other black parts were not of carbon and did not give a Raman signal until the micro-fragments were covered with water to keep them cool (Smith et aL 1999a). The Raman signal obtained resembled that of bixbyite (Mn203), a rare species in nature. This raised the question of the possible creation of bixbyite by heating some other MnxOyOHz phase, either by prehistoric man or by the laser beam during the analysis. Using more recent Raman apparatus, better spectra from some MnxOyOH: phases have been obtained both from samples in the MNHN mineral collection (Ospitali & Smith 2005) and from other limestone caves in Quercy (Roucadour and Combe N~gre 1) (Ospitali et al. 2005). Thus it is now easier to distinguish carbon from MnxOyOHz, which helps enormously in deciding which drawings to sacrifice for carbon isotope dating. A spectrum of an interesting orange microphase was obtained at Pergouset, which is neither goethite nor hematite because of a strong band at precisely 400 cm -j that lies between the values for well-crystallized goethite or wellcrystallized hematite; it was called 'disordered goethite' as it shared several bands with goethite (Smith et al. 1999a) and was probably created by prehistoric man heating 'yellow ochre' (a mixture coloured by goethite). G e m s t o n e s : R o m a n intaglios. Gemstone identification is one of the applications where RM excels. Three small intaglios were excavated

from a Roman site at Lut~ce (Paris) by the archaeologist S. Robin. When they were studied on a microscope stage it was rapidly established by RM that they were all composed of quartz (SiO2) (Smith & Robin 1997). The texture under the microscope indicated polycrystalline quartz, i.e. chalcedony, but this mineral has a great number of varieties. Two intaglios were apple-green in colour and it was first thought that they were of chrysoprase, a variety coloured green by Ni. Subsequently, some other green chalcedonies in other rocks were shown to be green because of Cr and have been called Cr-onyx. Because RM does not detect trace elements, as about 1 atomic % of an element is necessary to create a detectable spectral difference, the naming of the mineral variety of these intaglios could not be established with confidence, but the mineral species was unequivocal. One of them had a small mineral inclusion, which turned out to be zircon (ZrSiO4). The third intaglio was metallic blue under reflected light but bordeaux red under transmitted light; RM showed that this was also of quartz; its variety name could be jasper or sard.

Rocks: M e s o - A m e r i c a n axe. A Meso-American polished axe-head from Cozumel Island, Mexico, now in the collection of the Musre de l'Homme, Paris (Gendron 1998), had previously been classified as a 'greenstone', which literally means a green rock that has not been identified. This one contained at least two reddish minerals as well as two greenish minerals. With RM four kinds of Raman spectra were obtained and identified as clinopyroxene ((Na,Ca)(A1, Fe3+,Mg,Fe2+)Si206) (green), 3clinoam~hibole ((D,K,Na)(Na,Ca)z(A1,Fe ,Mg,Fe )5 (Si,AI)sOzz(OH)2) (darker green), garnet ((Mg,Mn,Fe~+,Ca)3(A1,Cr,Fe3+)2Si3012) (red) and titanite (CaTiSiOs) (brown) (Fig. la) (Smith & Gendron 1997a). The positions of the T - O - T bands of the clinopyroxene and the SiO4 bands of the garnet implied considerable proportions of respectively jadeite (NaA1Si2Or) and pyrope (Mg3A12Si30~2) in solid-solution, based on the semi-quantitative analytical method RAMANITA (mentioned above) (Smith 2005b). These two species indicated an eclogite, a rock type in which clinoamphibole and titanite often occur (Smith 1988). The kinds of clinoamphibole cannot be established as there are over 50 amphibole end-members and relatively few published data on their Raman spectra. Eclogite does not occur geologically on Cozumel Island, thus proving its

NON-DESTRUCTIVE RAMAN SPECTROSCOPY transport from afar, possibly from Guatemala (McBirney et al. 1987). Rocks: jades. A second axe head, from Guatemala but of uncertain provenance, was shown to be a true jadeite-jade by comparison with the Raman spectrum of a Burmese jadeite-jade (Smith & Gendron 1997a). Indeed, RM is undoubtedly the best technique for rapidly and non-destructively distinguishing the three types of jade: jadeite-jade (clinopyroxene); nephrite jade (clinoamphibole close to the tremoliteactinolite series (Ca2(Mg,Fe)5(Si)8022(OH)2)) and 'tourist jade' (anything else) (Smith 2005c). Thanks to RM, a fiver pebble subsequently collected by the archaeologist F. Gendron was shown to be a new sub-type of jade (quartzjadeitite) composed also of rutile (TiO2) and titanite (CaTiSiOs) (Fig. 3b) formed at higher pressure than usual Meso-American jade (albite-jadeitite) (Smith & Gendron 1997b; Gendron et al. 2002) from a new locality, on the south side of the Motagua River Valley, whereas all previous findings of geological jade had come from the north side (Harlow 1994). The light greyish-green 'type' jadeite in the MNHN mineral collection, which is itself a Neolithic jade axe whose provenance was most probably in the Western Italian Alps, as well as a strong green 'chromo-jadeite' from Burma both gave typical spectra of jadeite (NaA1Si206) with > 9 0 mol% Jd characterized by the S i - O Si Raman band at 701 -t- 2 c m - l (Gendron et al. 2002; Smith 2005c). The singlet (OH) Raman vibration of nephrite at c. 3673 cm-1 is very useful proof of the presence of nephrite jade, when found in addition to the lower wavenumber of the S i - O - S i stretching vibration close to 675 c m - l , which is much lower than that in jadeite. The nephrite jade nature of a series of artefacts, mainly polished flat bracelets or rings, but also some geological source rocks, all from China, was analysed by Smith et al. (2003b). Many were found to be of nephrite, but one of the six source rocks was a serpentine (Mg3SizOs(OH)4),and three artefacts were not nephrite but either calcite or quartz. A few darker artefacts revealed only a weak band at about 675 cm-~ suggestive of nephrite. Two probable tourist jades from SE Asia were also examined: a supposed sculptured 'lilac jade' was only quartz with a colour between that of amethyst and 'rose quartz', and a green and white bracelet of supposed jade turned out to be of calcite (Smith 2005a). Ceramics: vitrified wall. Enigmatic vitrified forts are known throughout the c. 1000 BC to

19

c. AD 1000 Celtic world from Portugal to Sweden, and especially in Ireland and Scotland (Ralston 1983; Buchsenschutz et al. 1998; Kresten et al. 1998). They have in common the fact that stone building blocks at the lower levels are often found to have been fused together by melting. Whether fused for defence, by attack or for religious reasons, a second major archaeological problem is to elucidate how such high temperatures were achieved, and over long surfaces (e.g. 100 m) and sometimes several centimetres depth. A few fragments of vitrified wall were collected by the archaeologist J. Vernioles from the vitrified base of the frequently rebuilt fort at St. Suzanne, Mayenne, France. Amongst glass, some crystals were shown by RM to be of e~-cristobalite (SiO2), which is supposed to require a temperature of 1470 °C if created by cooling from [3-cristobalite (Smith & Vernioles 1997). There is considerable doubt over the real temperature achieved, as the literature on this topic is poor and sometimes contradictory, and the presence of other elements such as A1 or Na could reduce this temperature; furthermore, polymorphic and order-disorder phenomena are also relevant, as metastable forms of c~- and [3-tridymite and or- and [3-cristobalite can exist. Nevertheless, the temperature must have been high (at some other localities quartzite has been melted (P. Kresten, pers. comm.) and pure quartz melts at 1713 °C). This enigma, strangely unheard of by many archaeologists, is likely to remain a mystery for some time. Chemists have confirmed that wood smouldering during rain could produce gaseous unsaturated hydrocarbons (e.g. acetylene), which could migrate and burst into flame at extremely high temperature, but the energy available would not be sufficient to penetrate deep into the rock wall. Accumulated lightning strikes over a few millennia provide an alternative possible explanation, otherwise it might be necessary to invoke UFOs (unidentified flying objects)! Interestingly, identical Raman spectra were obtained (Smith & Vernioles 1997) from c~-cristobalite in glass in 'Libyan Desert Glass', which is believed to have been formed by some kind of extra terrestrial impact event. Corroded metals: copper coins. Coins constitute one of the most obvious kinds of metal artefact of the cultural heritage and their size is ideal for being placed under a fixed microscope objective. Coins from different periods and composed of various metals (Fe, Cu, Zn, Pb, Ag, A1, Ni, Sn) were thus examined by Bouchard (2001) and Bouchard & Smith (2001, 2005b). As mentioned above, the pure metal or even many alloys do not

20

D.C. SMITH

give a Raman signal, but their corrosion products do. Care must be exercised in interpretation, as the metal in an identified corrosion product may not be a major constituent of the original coin because of 'preferential corrosion'; thus Cu salts are often found on Ag coins that contain a small amount of Cu (Fig. l b). Hence the main purpose is to recognize the kind of corrosion process that has occurred, so as to help restorers and curators decide on the appropriate method to treat and conserve the coins (or tools, weapons or statues, etc.) (Fig. ld). Concerning copper, the products observed by RM on coins and other artefacts of various ages included Cu-oxides (cuprite (Cu20), tenorite (CuO)), Cu-hydroxycarbonates (azurite (Cu3(CO3)2(OH)2), malachite (Cu2CO3(OH)2)), Cu-hydroxychlorides (atacamite (Cu2CI(OH)3), clinoatacamite (Cu2CI(OH)3)) and Cuhydroxysulphates (antlerite (Cu3SO4(OH)4), brochantite (Cu4SO4(OH)6))(Fig. 3c). Particular attention was paid to the RM distinction of the Cu-hydroxychlorides by Bouchard (2001) and Bouchard & Smith (2005b), as clinoatacamite has only recently been recognized by the International Mineralogical Association (IMA) (Jambor et al. 1996) and in earlier works this mineral species may have inadvertently been called paratacamite, which is a C u - Z n solidsolution ((Cu,Zn)2CI(OH)3)). Corroded metals: lead plates and an iron ingot. A fragment of a strongly corroded Roman sarcophagus in lead is archived in the MNHN mineral collection and is labelled 'cotunnite', which thus indicates corrosion by chloride. A R M study found no cotunnite (PbC12) nor any other chloride, but only a mixture of several Pbhydroxycarbonates: plumbonacrite (Pblo(CO3)6 O(OH)6 ), hydrocerussite (Pb3(CO3)z(OH)2) and cerussite (PbCO3) (Bouchard 2001; Bouchard & Smith 2005b). Another similar plate revealed only the two oxides litharge (PbO) and massicot (PbO), hence only the valency Pb 2+ (as no minium (Pb304) or plattnerite (PbO2) was detected) and no chloride, carbonate or hydroxide. A Roman ingot brought up from a shipwreck at Sainte-Marie-de-la-Mer off the French coast was examined (Bouchard 2001; Bouchard & Smith 2005b). The minerals found on the highly corroded surface included the Fe-oxide maghemite (~/-Fe203) and the Fe-oxyhydroxides akaganrite, goethite and lepidocrocite (all (FeO(OH))). There were also RM spectral indications of the presence of the ion FeCI42-, and it is known that in akaganrite some (OH)may be replaced by C I - , especially in marine environments (Arnould-Pernot et al. 1994).

Stained glass: experimental, modem and archaeological. Glass can be coloured in various ways. The colour may derive from a single chemical element dissolved in trace amounts inside the glass, in which case there is no longer any crystalline mineral phase left to provide a Raman spectrum. Alternatively, there may be micro- or nano-crystalline inclusions, which can provide a Raman spectrum. However, the most common situation in stained glass in church windows (apart from unheated superficial paint) is coloured reaction products formed after pigment minerals (with or without fluxes such as minium and silica to produce a P b - S i - O glass) had been spread on the surface of the glass and then heated; this produces several distinct phenomena: dissolution of some original material into the glass; migration of certain elements from the glass onto the surface (especially alkalis and alkaline earths); intercrystalline reaction between the applied pigments with or without involvement of the glass; relict original pigment; or glass that did not react at all. A project involving the study of commercially available mineral pigments (whose precise chemical composition is not provided by the manufacturers), experimentation to create stained glass and to study the reaction products, and analysis of real archaeological stained glass from the 13th to 20th centuries was described by Smith et al. (1999c) and Bouchard (2001). The experimentation showed that there is not so much chemical reaction between the original glass and the mixture placed on top as multiple reactions within the mixture. Blue stain caused by superficial cobalt aluminate 'smalt' or 'cobalt blue' (CoO.nAl203) was easily recognized by characteristic strong Raman bands along with relict initial corundum (A1203). In a green superficial experimental stain on glass the complex Raman spectrum revealed a considerable number of intermixed phases: principally blue smalt with orange crocoite (PbCrO4) to create the average green colour by 'colour subtraction' (i.e. the opposite of the 'colour addition' rules that apply to RGB computer screens) along with minor relict green eskolaite (Cr203) and red minium (Pb304), which had created the crocoite by the oxidizing reaction 6Cr203 + 4 P b 3 0 4 + 702 = 12PbCrO4. A modern commercial deep red glass gave an interesting strong Raman spectrum from the interior of the glass (Fig. 3d); it was possible to identify bands of CdS and CdSe typical of a CdS-CdSe solidsolution (Bouchard 2001), and even deduce the S/(S + Se) proportion to be about 45 atomic % on the basis of a Raman shift calibration made by Schreder & Kiefer (2001).

NON-DESTRUCTIVE RAMAN SPECTROSCOPY The most common mineral pigment found in real archaeological stained glass from earlier periods is hematite (e.g. 13th century from Mans; 16th-17th century from the Mus6e Carnavalet, Paris; 18th-19tb centuries from Strasbourg), but 19th-century glass from Mans and Strasbourg revealed respectively smalt and crocoite. Minerals created by environmental alteration of stained glass included calcite (CaCO3) and gypsum (CaSO4.2H20), and in contact with Pb structural supports a mixture of lead carbonates was found (Bouchard 2001).

immobile analysis in a laboratory: under air with micro-mapping Micro-inclusions in Guatemalan jade. The rutile-quartz-jadeitite from Guatemala mentioned above was examined by Raman micro-mapping with a RENISHAW ® INVIA ® spectrometer to gain more information on the nature of the quartz-jadeite contacts (Smith 2004b,c, 2005d). It was already established from Raman point analysis that the Jd content of the clinopyroxene is highest in the grain cores (c. 95 mol%) and diminishes sharply at the grain boundaries (sometimes 1-2mo1% lower, sometimes 1 0 20 mol% lower), and that the quartz occurs as

(a)

I~

21

micron-sized inclusions in the clinopyroxene grain cores (Gendron et al. 2002). Micromapping of a 50 txm x 90 ~ m surface with a motorized step of 0.4 Ixm acquired over 20 000 complete spectra overnight. These data were then treated and presented in different ways; for example, the integrated area of the main band of quartz (Fig. 4a) or of the T - O - T band of the clinopyroxene was used to reveal the distribution of the presence and absence of the quartz microinclusions, and the Raman wavenumber shift of the T - O - T band was used to reveal the tool% Jd distribution in detail (Fig. 4b). The latter map summarizes the collision of the North American Plate with the Caribbean Plate, subduction and exhumation, all in a 50 ~ m x 90 Ixm surface.

Crystal orientation in an Egyptian scarab. An inscribed Ancient Egyptian commemorative scarab was supposed to be made of enstatite ((Mg,Fe)SiO3 with Mg > Fe) (Fig. le). Despite a strong fluorescence, possibly due to patina formed over several millennia, it was possible to confirm from the Raman spectra obtained by placing the scarab on a Raman microscope stage that it does contain enstatite, and so far no other mineral has been found except minor

(b)

E~

-j -~z:7

"<.~.."~

-~'o

X

40

Fig. 4. Raman micromaps of the quartz-jadeitite RET27 from Guatemala. (a) 3D plot of the 2D spatial distribution of the intensity (signal-to-baseline) of quartz microinclusions over a 95p~m x 50 p~m part of the thin section involving three jadeite grains: they occur only in the core of the larger grain (centre right). (b) 3D plot of the same XY surface as in (a) but with the wavenumber position of the Si-O-Si Raman band around 700 cm- ~plotted increasing upwards (more Jd upwards). This single 3D map summarizes a complex geodynamical history of tectonic plate collision and subduction that created the high-pressure jadeite + quartz assemblage (the 'hills'), and the subsequent exhumation during which element migration allowed Ca + Mg to enter during retrograde depressurization metamorphism and replace Na + A1, thus reducing the Jd content towards the grain boundaries (the 'valleys').

22

D.C. SMITH

iron oxides. As the patina prevented a good visual observation, it remained possible that the scarab was made of one monocrystal (with minor iron oxide inclusions) or of millions of sub-millimetric crystals as in the case of jadeite-jade. A short Raman map was made and the relative intensity of two bands around 1000cm -1 was examined, as this parameter would vary with crystal orientation under the polarized laser beam. The bands were found to vary in relative intensity, which confirms that their orientation varies and hence that the scarab is polycrystalline (Smith 2004a).

Mobile analysis in situ inside a museum: with a horizontal microscope Methods. A DILOR ® LabRam ® spectrometer equipped with a prototype horizontal microscope was carried by four people into the Musre de l'Homme in Paris in an to attempt to identify various Meso-American sculptured rocks. This LabRam also had a vertical microscope, which allowed the identification of the natural pigment indigo in blue paint on an Aztec whistle in 'terra cotta' (Smith 2000), as this item was small enough to be placed on the microscope stage. Most of the other artefacts were much larger. They were also too heavy, some up to c. 60 kg, to place on a special moving platform designed to place objects in front of the horizontal objective. The objects were thus placed on top of various supports, usually strong wooden boxes, and various other pieces of wood, metal or plastic were used to support the artefact in the desired position at which the focused laser beam arrived exactly on that part of the artefact to be analysed. A knob on the horizontal objective permitted focusing movement in the z direction, but the xy positions could not be moved. An extra computer screen visualized a magnified image of the artefact and also the precise position of the laser beam. The only problem encountered was when it was necessary to move the artefact only a few microns so as to be able to analyse a particular micron-sized crystal. The 'handyman' solution was to slightly wobble the heavy artefact; when it settled down again after a few moments, the laser beam was never in exactly the same spot; after several attempts the laser beam fell upon either by good luck the desired crystal or one that was of the same species. These analyses are considered as being in situ in the sense that they took place inside the conservation room of the objects, which therefore did not have to leave their room but only be moved a few metres; also,

gloves were worn to avoid any surface damage, so that the artefacts were effectively not touched by hand. The configuration proved excellent for the purpose and produced data on many objects, a few of which are mentioned below. Rocks: Teotihuacrn, Ta~'no, Totonac and Aztec. A Teotihuacan mask (Fig. l h) labelled 'marble' was composed of layers of greenish-white and greenish-grey colour with parts polished to create angular patterns (glyphs); this artefact rapidly revealed calcite (CaCO3) in both colour zones. A large heavy well-polished Tamo object in a horseshoe shape and labelled 'rock' also yielded only calcite. (This object has been referred to as a 'ceinture/joug' (belt or yoke in English), although this shape has also been considered as a form upon which leather was worked to make large objects, such as a saddle. However, the quality of the finish implies a ceremonial usage.) These data suggested that the rock in each case was composed essentially of calcite (Smith 1999, 2005a), probably marble, the compact dense metamorphosed form of chalk or limestone. A strangely shaped three-pointed doubleheaded sculpture from the Ta'ino culture labelled 'rock' was mostly white or grey with some very dark parts. This object revealed spectra of calcite accompanied by a Raman band at 1006 cm-1, compatible with gypsum, but one band is not enough for a definitive identification. Spectra in the dark parts yielded the four Raman bands at 145, 395, 513 and 635 cm - t characteristic of anatase (Smith 1999, 2005a). Anatase is one of the polymorphs of (TiO2) (compare rutile and brookite) but it is usually brilliant white in colour (it is often used as a white pigment) although it can be light grey. It would appear that the anatase occurs as inclusions in the dark parts, which were not identified in the short time available for this analytical operation. The combination of anatase + calcite or of anatase + gypsum is not common in nature and hence no name was attributed to this enigmatic rock. Another 'ceinture/joug', but from the Taino culture, was reddish in colour and labelled 'rock'. It revealed the presence of quartz and of albite (NaAISi308) (Smith 1999, 2005a). Several types of igneous, metamorphic or sedimentary rocks may contain the association quartz + albite, so no name was given, but at least these mineral species were easily identified. A third 'ceinture/joug', Totonac this time, but dark green and labelled 'diorite', revealed the presence of clinopyroxene close to diopside (CaMgSi206) and plagioclase close to

NON-DESTRUCTIVE RAMAN SPECTROSCOPY labradorite ((Na, Ca)(A1, Si)408) (Smith 1999, 2005a). This association is typical of the dark igneous rock gabbro, in which these two mineral species are essential, but they can also occur in the metamorphic rock granulite. The two mineral species essential to a diorite (amphibole + andesine) were not observed. It was thus deduced that this artefact was probably made of gabbro. Diopside was also discovered in a dark reddish 'feathered serpent' from the Aztec culture labelled 'red porphyry', but diopside is a typical ferromagnesian mineral from a basic rock and is not a typical mineral of red porphyry, an acid igneous rock. The Raman analysis provided support for the idea that it also is a gabbro, as the reddish colour can be explained by a superficial pigment (see below).

Mineral pigments: Aztec and Timshian. The above-mentioned 'feathered serpent' has a surface that is dark red and smooth. Several spectra revealed hematite. A small Aztec reddish sitting statue labelled 'andesite' again yielded analyses of hematite (Smith 1999, 2000). Hematite is a mineral not normally found abundantly in either 'red porphyry' or andesite. However, the ease of finding hematite implied that hematite had been painted on the object as a pigment that masks the real colour of the rock (it is recognized by ethnologists that hematite was sometimes painted on Meso-American ceremonial works). A more obvious pigment is the red colour of the ears and lips of a Timshian mask sculptured out of a heavy bluish rock. This quickly proved to be cinnabar (HgS) (Smith 1999, 2000). Monocrystals: Aztec. Several small clear transparent carved objects had been labelled 'quartz' or 'calcite'. These could have been verified by traditional mineralogical methods but it was convenient to pass them rapidly under the laser beam. Most identifications were rapidly confirmed, but in a few cases a 'calcite' turned out to be quartz and a 'quartz' turned out to be calcite. The famous life-size Aztec skull said to be in 'rock crystal' proved indeed to be in quartz (Smith 1999; Smith & CarabatosN6delec 2001). Mobile analysis in situ inside a museum: under air with optical fibres Methods. A KAISER ® Holoprobe ® mobile Raman microscope was carried by one man into the treasure vault of the Gallery of

23

Mineralogy at the MNHN in 2000. A remote head was suspended from a tube on a tripod and in most cases orientated such that the laser beam was focused downwards onto an artefact placed on a table (see Fig. 5b). Occasionally it was more convenient to direct the laser horizontally (see Fig. 5a) but any orientation was possible as an optical fibre carried the incident laser and directed the Raman signal back along an adjacent fibre to the spectrometer box (containing the laser source, spectrometer, detectors, etc.) placed on a trolley on the floor to facilitate moving around the Gallery. The technique was impeccable and gave good results on many stones.

Sculptured rocks. Several Chinese jade artefacts in the gallery were examined to confirm their jade nature as well as the type of jade. A polychrome white + chromium-green pendant with the colour and texture typical of Burmese jade sculptured in China, a much darker greento-black pendant whose jadeite nature was less obvious (Fig. l f), and a homogeneous pale grey-green sculptured buckle were all proven to be made of jadeite (Smith 2001a, 2005a). A Chinese grasshopper cage, a large Chinese cup studded with other encrusted gemstones (Fig. 5a), and a Mogul dagger head mentioned below encrusted with diamonds, emeralds and 'rubies' were all proven to be made of nephrite (Smith 2001a, 2005a). Gemstones: Mogul. The Mogul dagger head from NW India (Fig. 5b) is encrusted with emeralds surrounded by a string of diamonds, and rubies are neatly dispersed throughout the artefact. All the emeralds and diamonds were rapidly shown to be correctly identified. However, only about half of the rubies were red corundum (A1203); the others, of the same colour, were found to be red spinel ((Mg,Fe)AI204) (Smith 2005a). This may not indicate any kind of fraud or even of error as the term 'ruby' in certain languages is synonymous with both corundum and spinel. Gemstones: Navaratna. An armband and a bracelet encrusted with the specific nine gemstones (including diamond in the centre) of the Navaratna legend (India) were also examined (Rondeau & Smith 2002). The correct mineral varieties and species were found in eight gems in the bracelet (species in square brackets): coral (red) [calcite] (CaCO3); emerald (green) [beryl] (Be3AlzSi6018); (colourless) [diamond] (C); (red) [grossular] (Ca3A12Si3OI2); pearl (white) [aragonite] (CaCO3); ruby (red)

24

D.C. SMITH

Fig. 5. Photographs of different MRM configurations. (a) Treasure Vault, MNHN, Paris, 2000. A KAISER® Holoprobe ® remote head connected by optical fibres to the spectrometer box on a trolley below the table. A Chinese nephrite jade cup; the laser beam, coming horizontally from the objective suspended from the tripod, can be seen through the jade (photo modified after Smith 2005a). (b) Treasure Vault, MNHN, Paris, 2000. The KAISER® Holoprobe ® remote head as in (a). A Mogul inlaid jade dagger handle encrusted with rubies, spinels, emeralds and small diamonds analysed vertically (photo modified after Smith 2005a). (c) Louvre Paris, 2000. The KAISER® Holoprobe ® remote head as in (a), but with the spectrometer in another room. An Oceanian 3 m high wooden statute painted blue. Noteworthy features are the laser impact spot and that the optical fibre cable is 100 m long (~hoto modified after Smith 2005a). (d) Roucadour cave, Quercy, France, 2004. A DELTA NU ® Inspector Raman with the entire MRM system inside the small black box, oriented here to examine prehistoric pigments on an inclined wall (photo D. C. Smith' ).

[corundum] (A1203); sapphire (blue) [corundum] (A1203); (yellow) [topazl (A12SiO4(OH,F)2)). H o w e v e r , the expected 'cat's eye' [chrysoberyl] (BeA1204) is in fact a 'cat's eye' quartz (SiO2). In the Navaratna armband, w h i c h has three rows each of three large gemstones, the central supposed d i a m o n d is in fact a zircon, the expected ruby is a spinel, the expected chrysoberyl is again 'cat's eye' quartz and the last stone is yellow sapphire [corundum]. The last three stones could easily be due to genuine misidentifications at the historical time of mounting, but the central zircon has a noticeably lower-quality cut and is rather small for the space available; it is thus very likely

that at some stage in the history of the armband the original stone was replaced by a zircon.

Gemstones." Medieval cloisonn~ gold. Earl~ in 2001 a similar remote head, a D I L O R ~ 'Superhead '® plus a spectrometer were carried into the Mus6e des Antiquit6s Nationales in St G e r m a i n - e n - L a y e near Paris. The same kind of configuration was e m p l o y e d to analyse numerous stones encrusted in cloisonn6-gold style Medieval jewellery. Most stones were red and all but one were shown to be garnet; all green stones were glass. The garnets in Medieval cloisonn6-gold j e w e l l e r y from Vicq, France (Fig. l g), were e x a m i n e d in a routine fashion by

NON-DESTRUCTIVE RAMAN SPECTROSCOPY rapidly placing each crystal beneath the laser beam, one after the other without the need to verify the laser focusing (Smith & P6rin 2003). They were all almandine-pyrope solid-solutions, which have been called 'rhodolite' ((Mg,FeZ+)3AlzSi3Oi2), although this is not an official IMA term. Such compositions were already known, but by non-mobile analytical techniques, e.g. by XRF (Greiff 1998), energydispersive SEM (Quast & Schltisser 2000) and by Proton Induced X-ray Emission spectroscopy (PIXE) (Calligaro et al. 2002). Most analyses at Vicq were closer to almandine but in two bird's eye fibulae made of silver instead of gold the stone was closer to pyrope. Their chemical distinction was possible by employing the RAMANITA method of semi-quantitative analysis (see above) (Smith 2005b). The same method was employed on other Medieval cloisonne-gold jewellery from Brut, North Ossetia, Russian Federation, and similar almandine-pyrope solid-solutions were found (Smith et al. 2003b; Smith 2005a). However, some garnets were found to be very rich in andradite (Ca3Fe3+Si3Ol2); as far as is known andradite has never been recorded before in this kind of archaeological material.

Mineral pigments: Oceanian statue. The above-mentioned KAISER ® Holoprobe ® mobile Raman microscope was also carried into the Louvre in 2000 to examine some pigmented artworks. Of particular interest was a 3 m high wooden statue from Oceania painted light blue. It was easy to position the remote head on the tripod just in front of the statue and direct the laser at any angle onto the statue (Fig. 5c). In this analytical operation a 100m optical fibre was used and the mobile Raman microscope's spectrometer was placed in another room of the museum. Mobile telephones were used to communicate between the person adjusting the focusing and the person controlling the computer, so this is an excellent example of 'remote' analysis (Smith 2001a, 2005a). Very little time was left for this analysis before having to evacuate the building at closing time, and the neon lights of the room in addition to considerable daylight increased the background so much that only one poor spectrum could be collected. However, after extended spectral treatment it was just possible to see a Raman band at the characteristic wavenumber for lazurite ((Na,Ca)4_8(A16Si6024)(S,SO4,C1) 1- 2)- It is highly unlikely that the expensive rock 'lapis lazuli' was ground up to be used as a pigment; it is more likely that this statue was painted in

25

the early 20th century after lazurite had become a synthetic pigment ('ultramarine'). Mobile analysis in situ inside a museum: under glass with optical fibres Methods. RM has the great advantage over many other analytical techniques that it can analyse through transparent media such as glass, mineral or plastic. It is thanks to this capacity that micro-inclusions within minerals, and especially within gemstones, can be determined precisely, as mentioned in the introduction. To test the MRM method through thick plate glass, the KAISER Holoprobe was used in 2000. The usual tripod with the remote head suspended below it was simply placed on the protective plate glass, as were the computer, keyboard and mouse. The laser beam was focused through the glass onto the mineral below, which reduced enormously the intensity of the Raman spectrum of the glass itself. When relatively significant Raman bands occurred from the glass (because of a weak Raman signal from the mineral below and hence longer counting times) they were in different spectral zones from the Raman bands of the mineral of interest, and were wider and hence distinct, so that the glass was not a problem at all. Gemstones encrusted in stone marquetry: Florentine tables. Three 17th-century tables with various stones encrusted into black or white marble are exhibited in the Treasure Vault of the Gallery of Mineralogy at the MNHN. They are covered by heavy 1.6 mm thick plate glass, which was not removed. The stones had been inlaid in delicate designs of flowers, fruit, birds and insects. The laser beam was focused on the mineral surface, and as it is flat, it was sufficient to merely slide the tripod over the table and the computer screen gave the Raman spectrum of the mineral under the laser (Fig. l i). In this way it was possible to confirm, for example, the quartz nature of purple crystals composing grapes, quartz or calcite in many designs, quartz in the thorax of a green insect suspected to be a mineral other than quartz, and dolomite (CaMg(CO3)2) in yellowish-white 'lily-of-the-valley' flower petals that could have been in calcite, quartzite or even ivory (Smith & Rondeau 2001; Smith 2002a,c). Blue flowers in lazurite were recognized by their luminescence spectrum. The most exciting discovery concerned the red pips of a pomegranate design that were presumed to be of garnet; in fact, some were indeed of garnet but others were of ruby. Thus a very precious stone had been used

26

D.C. SMITH

and one may wonder if the manufacturer had realized that he was mixing two different mineral species in a single kind of design (compare the Mogul dagger head mentioned above). This short study demonstrated very well that analysing under thick glass is perfectly feasible and indeed rather easy, as exploitable spectra were obtained in a reasonable time. Raman analysis under water Methods. As RM can analyse under glass, it can also analyse under water. The world's lakes and oceans are littered with sunken ships still with their cargos, often at shallow depths, and numerous cities from Alexandria to Zeugma have been submerged. Of course, if sufficient finance is available one could build a special submarine and take an entire mobile Raman microscope to great depths and make analyses though a window, i.e. through both glass and water, and this has been done as deep as 3600 m (Brewer et al. 2002). However, for much shallower depths a simpler solution was proposed (Smith 2001b): to take a tripod (or rather a more robust stand) to the sea floor and have a diver focus the laser beam onto part of an artefact of interest. This is possible as there is no electricity at the remote end of the optical fibres. Thus the spectrometer could be in a boat just above. In this way, one could verify supposed beryl as the eyes of an Egyptian statue in granite, supposed lazurite as the toenails, and even the kind(s) of feldspar in the granite itself. However, before attempting to mount an expedition, it was important to evaluate if dirty water would cause too much fluorescence and thus hamper if not prohibit the exercise. Simulation with gemstones. Three gemstones chosen for their Raman response (strong, zircon (ZrSiO4); medium, microcline (KA1Si3Os); weak, sodalite (NasA16Si6024C12)) were placed in a plastic bowl and successively analysed under air, pure distilled water, and various impure waters. These were concocted to create colour, solute and/or suspensions by initially making saturated sugar or salt solutions and then using pure red wine (Smith 2001b). These results were very positive, as pure water actually increased the Raman intensities (perhaps because of cooling), the solutes did not make the spectra any worse, and although the wine considerably reduced the Raman signal because of absorption by the colour and the suspensions, simple spectral treatment made it possible to recognize all of the Raman bands except for those of deep blue sodalite. To simulate natural impure

waters more realistically, a second set of experiments was carried out with very impure water full of fish dejections or rotten vegetation (Smith 2003) (Fig. lc). The former increased the baseline but did not reduce the Raman signal, whereas the latter reduced the Raman signal but did not increase the baseline. Thus the relevant Raman bands were still visible and after spectral treatment there was effectively no loss of information. Hence even if lake or ocean water does inhibit subaquatic archaeometry by MRM at certain natural sites of concentrated impurities and at places polluted by human activities, in general terms there are no physical, chemical, botanical or zoological obstacles. Ultra-mobile analysis in situ inside a museum with a hand-held Raman microscope Methods. A very recent development is an even more extreme miniaturization of an MRM apparatus, which allows it to be even lighter in weight and easier to manipulate in any direction. The entire DELTA NU ® 'Inspector Raman ~ can be held in one hand, except for the mini-computer, which can be operated with the other hand; it can run on batteries and hence be used effectively anywhere in air (N.B. there is electricity inside this particular 'remote head', as it also contains the laser source, the detector and the spectrometer). An optional special plastic nozzle fitted over the objective gives exactly the fight focusing distance so that for artefacts that may be touched, such as tough gemstones, it is sufficient to place the nozzle on the gem and literally 'shoot' the laser from the 'gun'. For less tough materials the nozzle can be quickly taken off, and for analysing mixtures such as pigments a special video attachment with triple illumination gives an enlarged view of the grains so that the precise zone to be analysed can be chosen. Of course, with all these advantages there must be some disadvantages; these are principally a lower spatial resolution, c. 50 p~m minimum compared with c. 1 ~m for a high-resolution mobile Raman microscope, and a lower spectral resolution such that the bands are wider and their centres less precise. An Inspector Raman was tried in 2004 on the same Florentine tables as mentioned above (Smith & Ospitali 2005), and it was also carried down the Roucadour cave, Quercy, France, in 2004 to examine some other Prehistoric paintings (Smith et al. unpub, data). The first data were disappointing because of several desired features not being available in the supplied software (these problems are being remedied at the time of writing), but the

NON-DESTRUCTIVE RAMAN SPECTROSCOPY apparatus excelled in its mobility, especially during the short speleological expedition, and it fully merits the term ultra-mobile (Fig. 5d).

Discussion and conclusions There can be no doubt that archaeometry in general has gained enormously by the development of a wide range of analytical techniques, especially non-destructive ones. It has been demonstrated here that RM and MRM are powerful. It has also been mentioned that more A R C H A E O R A M A N research has been made on pigments than on all the other domains put together; this is not because minerals are illsuited to RM analysis (they are generally wellsuited but give less powerful signals than do organic liquids). Most of the world's 'ramanists' are not mineralogists or geologists, but physicists or chemists who of course are very familiar with their own disciplines, but not with natural minerals, rocks and geological maps. It would be useful if more archaeologically orientated mineralogists or geologists turned to RM. As in medicine, in archaeometry one needs specialists of various techniques as well as generalists. A visiting generalist doctor needs a portable case of small apparatus adaptable to as many situations as possible. A mobile archaeometrist should also have a portable case containing at least one small polyvalent apparatus: can any other technique challenge ultra-MRM here? It has already been mentioned that future activities may include 'Raman spying' using telescopy (Smith 2005a, thanks to Sharma et al. 2002, 2003), and a voice-controlled MRM apparatus where one asks the microscope for the identification of a mineral and the microscope speaks the answer (Smith 2005a). In the last decade other research groups have been busy designing a miniature mobile Raman microscope for planetary exploration (e.g. Wang et al. 1996). One idea was to place a mobile Raman microscope inside a Coca-Cola ® tin; in fact, the Inspector Raman is already down to the size of just four such tins. There are always physical limits that prevent technology going further than a certain degree of development, but often ingenious ways can be devised to bypass such problems. Given the pace of innovation, many of the technological revolutions of the last few decades were unthinkable a decade ahead. There are only two more decades left until the centenary of the discovery of the 'Raman effect' by Sir Chandrasekhara Venkata Raman in 1928. By then there will have been other spectacular inventions, and one may wonder if in 2028 a mobile Raman microscope may resemble a present-day USB memory key!

27

The author is pleased to acknowledge the help, during various stages of his projects mentioned above, kindly provided by museum or university staff, especially A. Barbet, M. Bouchard, C. Carabatos-Nedrlec, P.-J. Chiappero, J.-M. Fourcault, F. Gendron, E. Gonthier, M. Kazanski, D. LEvine, M. Lorblanchet, C. Naffah, F. Ospitali, P. Prrin, M. Pinet, S. Robin, B. Rondeau, H.-J. Schubnel, F. Valet and J. D. Vernioles, and also the representatives of four manufacturers, M. Belleil (RENISHAW), K. Carron (DELTA NU), B. Lenain (KAISER), and S. Morel and J. Oswalt (DILOR/JOBIN-YVON/ HORIBA).

References ARNOULD-PERNOT, P., FORRII~RES,C., MICHEL, H. & WEBER, B. 1994. Optimisation d'un traitement de dfchloruration d'objets ferreux par plasma d'hydrogbne. Studies in Conservation, 39, 232-240. BELL, I. M., CLARK, R. J. H. & GIBBS, P. J. 1997. Raman spectroscopic library of natural and synthetic pigments (pre-~1850 AD). Spectrochimica Acta, Part A, 53, 2159-2179. BOUCHARD,M., 1998. Etude des pigments de grottes orn(es du Pal(olithique par Microscopie Raman. Diplome de DEA, IPH, Musrum National d'Histoire Naturelle, Paris. BOUCHARD,M. 2001. Evaluation des capacit~s de la microscopie Raman dans la caractrrisation min~ralogique et physico-chimique de mat~riaux arch~ologiques: m~taux, vitraux et pigments. Thbse de Doctorat, Musrum National d'Histoire Naturelle, Paris. BOUCHARD, M. & SMITH, D. C. 2000a. METALLORAMAN. Arch~om~trie, Dossiers de l'Arch~ologie, 253, 60-61. BOUCHARD, M. & SMITH, D. C. 2000b. New data for the application of Raman microscopy to archaeological corroded metallic objects. ICORS'2000, Beijing, China, 1140-1141. BOUCHARD, M. & SMITH, D. C. 2001. Evaluating Raman microscopy for the non-destructive archaeometry of corroded coins: a powerful technique for conservation studies. Asian Chemical Letters, 5(3), 157-170. BOUCHARD, M. & SMITH, D. C. 2002. Archaeological and experimental stained glass: a non-destructive Raman microscopic (RM) study. 'ART 2002', 7th International Conference on Non-destructive Testing and Microanalysis for the Diagnostics and Conservation of the Cultural and Environmental Heritage, University of Antwerp, June 2002, Abstracts volume, 152. BOUCHARD,M. & SMITH, D. C. 2003. Catalogue of 45 reference Raman spectra of minerals concerning research in art history or archaeology, especially on corroded metals and coloured glass. Special Volume, Proceedings, Congress GEORAMAN2002, Prague, 2002. Spectrochimica Acta, Part A, 59, 2247-2266. BOUCHARD, M. & SMITH, D. C. 2005a. Database of 74 Raman spectra of standard minerals of relevance to metal corrosion, stained glass or

28

D.C. SMITH

Prehistoric rock art. h~: EDWARDS, H. G. M & CHALMERS, J. (eds) Raman Spectroscopy in Archaeology and Art Histotx. The Royal Society of Chemistry, London, 429-464. BOUCHARD, M. & SMITH, D. C. 2005b. Raman spectroscopic study of corroded metals, h~: RULL, F. SMITH, D. C. & EDWARDS, H. G. M. (eds) GEORAMAN and ARCHAEORAMAN. Universidad Valladolid, Valladolid (in press). BREWER, P., PASTERIS, J. D., MALBY, G., et al. 2002, Laser Raman spectroscopy used to study the ocean at 3600-m depth. LOS Transactions American Geophysical Union, 83, 469-470. BRODY, R. H., EDWARDS, H. G. M., FARWELL, D. W., O'CONNOR, S. & POLLARD, A. M. 1998. Real or take: FT-Raman study of ivories, h2: HEYNS, A. M. (ed.) ICORS, Capetown'98. Wiley, Chichester, 522-523. BUCHSENSCHUTZ, O., RALSTON, I. B. M. & SCHUBERT, F. 1998. Les fortifications. In: GRUEL, K. ~ VITALI, D. (eds) L'oppidum de Bibracte: un bilan de onze anndes de recherche (1984-1995): Dossier. Gallia, 55, 18-22. BURGIO, L. & CLARK, R. J. H. 2001. Library of FT-Raman spectra of pigments, minerals, pigment media and varnishes, and supplement to existing library of Raman spectra of pigments with visible excitation. Spectrochimica Acta, Part A, 57, 1491 - 1521. CALLIGARO, T., COLINART, S., POIROT, J.-P. & SUDRES, C. 2002. Nuclear instruments methods Physics Research B, 189, 320. COLOMBAN, P. & TREPPOZ, F. 2001. Identification and differentiation of ancient and modern European porcelains by Raman macro- and micro-spectroscopy. Journal of Raman Spectroscopy, 32, 93-102. COLOMBAN, P., SAGON, G. & FAUREL, X. 2001. Differentiation of antique ceramics fYom the Raman spectra of their coloured glazes and paintings, Journal of Raman Spectroscopy, 32, 351-603. COUPRY, C. & BRISSAUD, D. 1996. Applications in art, jewelry and forensic science. In: TURRELL, G. & CORSET, J. (eds) Raman Microscopy, Developments" and Applications. Academic Press, London, 421-453. COUpRY, C., SAGON,G. & Lautir, A. 1993. Contribution par spectrom&rie Raman ~ la connaissance des vitraux. Conservation commune d'un patrimoine commun, lkr Colloque du Programme framvallemand de Recherche pour la Conservation des Monuments historiques, Karlsruhe, 246-249. DI~LI~-DUBOIS, M. L., DHAMELINCOURT, P. & SCHUBNEL, H.-J. 1978. IXth General Meeting of the International Mineralogical Association, Novosibirsk, USSR, 4 - 9 September. Abstract volume. DgLE-DUBOIS, M. L., DHAMELINCOURT, P. & SCHUBNEL, H.-J. 1981a. Etude par spectrom&rie Raman d'inclusions dans ies diamants, saphirs et dmeraudes. Revue de Gemmologie, 63, 11 - 14. Dt~LI~-DUBOIS, M. L., DHAMELINCOURT, P. & SCHUBNEL, H.-J. 1981b. Etude par spectromdtrie

Raman d'inclusions dans les diamants, saphirs et ~meraudes. Revue de Gemmologie, 64, 13-14. DELE-DUBOIS, M. L., DHAMELINCOURT, P., POIROT, J.-P. & SCHUBNEL, H.-J. 1986a. Differentiation between natural gems and synthetic minerals by laser Raman micro-spectrometry. Journal of Molecular Structure, 143, 135-138. DELE-DUBOIS. M. L., POIROT, J.-P. & SCHUBNEL, H.-J. 1986b. Identification de micro-inclusions dans des rubis et ~meraudes de synthbse par spectroscopic Raman. Revue de Gemmologie, 88, 15-17. DELHAYE, M., GUINEAU, B., VEZIN, J. & COUPRY, C. 1985. La microsonde Raman au secours des oeuvres d'art. Mesures, 11, 119-124. DHAMELINCOURT, P. & BISSON, P. 1977. Principe et rralisation d'un microscope optique utilisant l'effet Raman. Microscopica Acta, 79(3), 267-275. D! LONARDO, G., MARTINI, C., OSPITALI, F., POLl, G., PRANDSTRALLER, O. & TULLINI, F. 2002. Raman and IR micro-spectroscopy for the study of corrosion products on archaeological bronze objects, hi: VAN GRIEKEN, R., JANSSENS, K., VAN'T DACK, L. & MEERSMAN, G. (eds) Proceedings of Art 2002, 7th hTternational Conference on 'Non-destructive Testing and Microanalysis for the Diagnostics and Conservation of the Cultural and Environmental Heritage', 2 - 6 June 2002, Antwerp, 1-7. EDWARDS, H. G. M. & TAIT, J. F. K. 1998. Fr-Raman spectroscopic study of decorated stained glass, Applied Spectroscopy, 52(5), 679. EDWARDS, H. G. M., BARRY, B. W., WILLIAMS, A. C. & RULL, F. 1996a. The Iceman: an FT-Raman spectroscopic study of 5200-year-old human skin. GEORAMAN-96. Terra Abstracts, Supplement 2. Terra Nova, 8, I. EDWARDS, H. G. M., DAFFNER, L. A., FALK, M. J., FARWELL, D. W., HERON, C. & QUYE, A. 1996b. FF-Raman spectroscopic analyses of ancient resins of archaeological significance, GEORAMAN-96. Terra Abstracts, Supplement 2. Terra Nova, 8, 1. EDWARDS, H. G. M., FALK, M. J., FARWELL, D. W. & JANAWAY 1996C. FF-Raman spectroscopic studies of linen samples from archaeological burial sites, GEORAMAN-96. Terra Abstracts, Supplement 2. Terra Nova, 8, 2. EDWARDS, H. G. M., DRUMMOND,L. & RUSS, J. 1998. Fourier-transform Raman spectroscopic study of pigments in native American Indian rock art: Seminole Canyon. Spectrochimica Acta, Part A, 54, i 849-1856. FROST, R. L., MARTENS, W., KLOPROGGE, T. & WILLIAMS, P.A. 2002a. Raman spectroscopy of the basic copper choride minerals atacamite and paratacamite: implications for the study of copper, brass and bronze objects of archaeological significance. Journal of Raman Spectroscopy, 33, 801. FROST, R. L., WILLIAMS, P. A., MARTENS, W. & KLOPROGGE, T. 2002b. Raman spectroscopy of the polyanionic copper (II) minerals buttgenbachite and connellite: implications for studies of ancient

NON-DESTRUCTIVE RAMAN SPECTROSCOPY copper objects and bronzes. Journal of Raman Spectroscopy, 33, 752. FRY, R. L., FROST, R. L., TURNER J. & HALLAM, D. 1998. Analysis of ceramics from the wreck of the 'Pandora' using Raman and FTIR spectroscopy. In: HEYNS, A. M. (ed.) ICORS, Capetown'98. Wiley, Chichester, 546-547. GENDRON, F. 1998. Les roches vertes en Mdsoamdrique. Archiologie du jade. Doctoral thesis, Ecole Hautes Etudes Sciences Sociales, Paris. GENDRON, F., SMITH, D. C. & GENDRON-BADOU, A. 2002. Discovery of jadeite-jade in Guatemala confirmed by non-destructive Raman microscopy. Journal of Archaeological Science, 29, 837-851. GREIFF, S. 1998. Naturwissenschaftliche Untersuchengen zur Frage de Rohsteinquellen ftir friihmittelalterlichen Almandingranatschmuck rheinfr~inkischer Provenienz. Jahrbuch des R6misch-Germanischen Zentralmuseums, Mainz, 45, 599-646. GRIFFITH, W. P. 1987. Advances in the Raman and infrared spectroscopy of minerals. In: CLARK, R. J. H. & HESTER, R. E. (eds) Spectroscopy of Inorganic-based Materials. Wiley, New York, 119-187. GUINEAU, B. 1987. L'6tude des pigments par les moyens de la microsp6ctrometrie Raman. In: DELAMARE, F., HACKENS, T. & HELLY, B. (eds) Datation-Caractdrisation des Peintures Paridtales et Murales (PACT-t7). European University Centre for Cultural Heritage, Ravello, 259-294. HARLOW, G. E. 1994. Jadeitites, albitites and related rocks from the Motagua Fault Zone, Guatemala. Journal of Metamorphic Geology, 12, 49-68. JAMBOR, J., DUTRIZAC, J. & ROBERTS, A. 1996. Clinoatacamite, a new polymorph of Cuz(OH)3CI, and its relation to paratacamite and 'anarakite'. Canadian Mineralogist, 34, 61-65. KIEFERT, L., H,~NNI, H. A. & OSTERTAG, T. 2001. Raman spectroscopic applications to gemmology. /17: LEWIS, I. & EDWARDS, H. G. M. (eds) A Handbook on Raman Spectroscopy. Marcel Dekker, New York, 469-489. KRESTEN, P., LARSSON, L. & HJARTHNER-HOLDAR,E. 1998. Thermometo' of ancient iron slags from Sweden, and of vitrified material from various hill-forts in western Europe. Geoarchaeology Laboratory, Uppsala, Research Report, R9601. LASNIER, B. 1989. GEORAMAN-89: CONTRIBUTIONS. Special Publication, Association Nationale de la Recherche Technique, Paris, 17. LIEM, N. Q., SAGON, G., QUANG, V. X., TAN, H. V. & COLOMBAN, P. 2000. Raman study of the microstructure, composition and processing of ancient Vietnamese (proto)porcelains and celadons (13-16th centuries), Journal of Raman Spectroscopy, 31, 933. LIEM, N. Q., THANH, N. T. & COLOMBAN, P., 2002. Reliability of Raman microspectroscopy in analyzing ancient ceramics: the case of ancient Vietnamese porcelain and celadon glazes. Journal of Raman Spectroscopy, 33, 287-294. MACQUET, C. 1994. Contribution h l'dtude des surfaces vitreuses anciennes; application b

29

l'amglioration des techniques de conservation. Th~se de doctorat, Universit6 de Poitiers. MAESTRATI, R. 1989. Contribution gt l'ddification du catalogue Raman des gemmes. DUG Diploma Memoir, Nantes University. MARTENS, W., FROST, R. L. & WILLIAMS, P. A. 2003. Raman and infrared spectroscopic study of the basic copper chloride minerals: implications for the study of the copper and brass corrosion and 'bronze disease'. Neues Jahrbuch far Mineralogie, Abhandlungen, 178(2), 197-215. MCBIRNEY, A., AOKI, K.-I. & BASS, M. N. 1967. Eclogites and jadeite from the Motagua Fault Zone, Guatemala. American Mineralogist, 52, 908-918. MCCANN, L. I., TRENTELMAN, K., POSSLEY, T. & GOLDING, B. 1999. Corrosion of ancient Chinese bronze money trees studied by Raman microscopy. Journal of Raman Spectroscopy, 30, 121-132. NASDALA, L., SMITH, D. C., KAINDL, R. & ZIEMANN, M. A. 2004. Raman spectroscopy: Analytical perspectives in mineralogical research. In: BERAN, A. & LIBOWlTZKY, E. (eds) EMU Notes in Mineralogy. Spectroscopic Methods in Mineralogy, 6, 281-343. OSr'ITALI, F. & SMITH, D. C. 2005. A Raman microscopic library of natural Mn oxides and oxyhydroxides. In: 3rd International Conference on the Application of Raman Spectroscopy in Art and Archaeology, Louvre, Paris, 31 August-3 September 2005, Abstract volume, 82. OSPITALI, F., SMITH, D. C. & LORBLANCHET,M. 2005. Raman microscopic analysis of prehistoric pigments in wall-painted caves: Roucadour, France. In: 3rd International Conference on the Application of Raman Spectroscopy in Art and Archaeology, Louvre, Paris, 31 August-3 September 2005, Abstract volume, 20. PINET, M. & SMITH, D. C. 1993. La micr0spectrom~trie Raman des grenats X3Y2Z3012: I. La s6rie calcique naturelle ouvarovite- grossulaireandradite. Schweizerische Mineralogische and Petrographische Mitteilungen, 73(1), 21-40. PINEY, M. & SMITH, D. C. 1994. La microspectrom6trie Raman des grenats X3Y2Z3012: II. La s6rie alumineuse naturelle pyrope-almandin-spessartite. Schweizerische Mineralogische and Petrographische Mitteilungen, 74(2), 161 - 179. PINEY, M., SMITH, D. C. & LASNmR, B. 1992. Utilit6 de la microsonde Raman pour l'identification non-destructive des gemmes, avec une s~lection representative de leurs spectres Raman. In: La Microsonde Raman en Gemmologie. Revue de Gemmologie, numdro spdcial hors sdrie, 11-61. QUAST, D. & SCHL13SSER, U. 2000. Mineralogische Untersuchungen zur Herkunft der Granate merowingerzeitlicher Cloisonn6arbeiten. Germania, 78, 75-96. RALSTON, I. B. M. 1983. Vitrified walls. Bulletin of Experimental Archaeology, 4, 10-11. RONDEAU, B. & SMITH, D. C. 2002. Non-destructive in situ identification of gemstones on two 'Navaratna' jewels by MRM (mobile Raman microscopy). In: 'ART 2002', 7th International Conference on Non-destructive Testing and Microanalysis for

30

D.C. SMITH

the Diagnostics and Conservation of the Cultural and Environmental Heritage, University of Antwerp, June 2002, Abstracts volume, 223. RULL-PEREZ, F. 2001. Applications of IR and Raman spectroscopy to the study of Medieval pigments. In: LEWIS, I. & EDWARDS, H. G. M. (eds) A Handbook on Raman Spectroscopy. Marcel Dekker, New York, 835-862. RULL-PEREZ, F., EDWARDS, H. G. M., RIVAS, A. & DRUMMOND, L. 1999. Fourier transform Raman spectroscopic characterization of pigments in the Mediaeval frescoes at Convento de la Peregrina, Sahagun, L6on, Spain, Part 1--Preliminary study. Journal of Raman Spectroscopy, 30, 301-305. SCHMETZER, K., KIEFERT, L., BERNHARDT, H.-J. & BEILI, Z. 1997. Characterisation of Chinese hydrothermal synthetic emerald. Gems & Gemmology, 33, 277-291. SCHREDER, B. & KIEFER, W. 2001. Raman spectroscopy on I I - V I semiconductor nanostructures. In: LEWIS, I. & EDWARDS, H. G. M. (eds) Handbook of Raman Spectroscopy. Marcel Dekker, New York, 491-547. SEAWARD, M. R. D. & EDWARDS, H. G. M. 1998. Biological origin of major chemical disturbances on ecclesiastical architecture studied by Fourier transform Raman spectroscopy. Journal of Raman Spectroscopy, 28(9), 691-696. SHARMA, S. K., ANGEL, S. M., GHOSH, M., HUBBLE, H. W. & LUCEY, P. G. 2002. Remote pulsed laser spectroscopy system for mineral analysis on planetary surfaces to 66 meters. Applied Spectroscopy, 56, 699-705. SHARMA, S. K., LUCEY, P. G., GHOSH, M., HUBBLE, H. W. & HORTON, K. 2003. Stand-off Raman spectroscopic detection of minerals on planetary surfaces. Special Volume, Proceedings, GEORAMAN-2002 Congress, Prague, 2002. Spectrochimica Acta, Part A, 59, 2391-2407. SMITH, D. C. 1987. The Raman spectroscopy of natural and synthetic minerals: a review. International Conference 'GEORAMAN-86'. Terra Cognita 7, 20-21. SMITH, D. C. 1988. A review of the peculiar mineralogy of the 'Norwegian coesite-eclogite province', with crystal-chemical, petrological, geochemical and geodynamical notes and an extensive bibliography. In: SMITH, D. C. (ed.) Eclogites and Eclogite-Facies Rock~. Developments in Petrology, 12, 1-206. SMITH, D. C. 1996. The importance of using a halfwave plate for the Raman spectroscopy of minerals which cannot be rotated under the laser beam, especially large objects of gemmological or archaeological interest. GEORAMAN-96, Terra Abstracts, Supplement 2. Terra Nova, 8, 23-24. SMITH, D. C. 1999. Letting loose a laser: MRM (mobile Raman microscopy) for archa~ometry and ethnomineralogy in the next millennium. Mineralogical Socie~ Bulletin, December, 3-8. SMITH, D. C. 2000. Pigments rouges et bleus sur cinq oeuvres d'Am6rique: analyse non-destructive par MRM (microscopie Raman mobile). Techne, 11, 69-83.

SMITH, D. C. 2001a. Recent MRM (mobile Raman microscope) analytical operations in situ in four French museums. Colloquium: 'Raman Spectroscopy in Archaeology and Art History', British Museum, bmdon, 20 November 2001. SMITH, D. C. 2001 b. Simulation of submarine archaeomerry by non-destructive physico-chemical analysis of gemstones by MRM (mobile Raman microscopy) in situ under impure water. Congrks "Archdomdtrie 2001", GMPCA, La Rochelle Universit3.', 108. SMITH, D. C. 2002a. ARCHAEORAMAN and mobile Raman microscopy (MRM): from pigments in aerial wall-paintings to gemstones in submarine archaeometry. Congress GEORAMAN-2002. Acta Universitatis Carolinae, Geologica, Praha, 46(1), 84-86. SMITH, D. C. 2002b. Semi-quantitative chemical analysis of garnet and jade in mounted jewels by non-destructive Raman microscopy: recent progress with the analytical method. Congress GEORAMAN-2002. Acta Universitatis Carolinae, Geologica, Praha, 46(1), 87-89. SMITH, D. C. 2002c. MRM (mobile Raman microscopy) in situ in four national museums. 'ART 2002', 7th International Conference on Nondestructive Testing and Microanalysis for the Diagnostics and Conselvation of the Cultural and Environmental Heritage, Universita' of Antwerp, Belgium, June 2002. Abstracts volume, 46. SMITH, D. C. 2003. In situ mobile subaquatic archaeometry evaluated by non-destructive Raman microscopy of gemstones lying under impure waters. Special Volume, Proceedings, GEORAMAN-2002 Congress, Prague, 2002. Spectrochimica Acta, Part A, 59, 2353-2369. SMITH, D. C. 2004a. An ancient Egyptian commemorative scarab carved in rock: non-destructive mineral identification by Raman microscopy, ln: FREDERICKS, P. M., FROST, R. L. & RINTOUL, L. (eds) XIXth International Conference on Raman Spectroscopy, ICORS, Gold Coast, Australia, 8-13 August 2004, Proceedings, (CD-ROM). SMITH, D. C. 2004b. Raman micromapping of physical and/or chemical transformations of minerals of interest in geology or archaeology. GEORAMAN2004, University of Hawaii, Honolulu, Hawaii, 6-11 June 2004. School of Ocean and Earth Science Technology Publications, 04-02, 71-72. SMITH, D. C. 2004c. Raman micro-mapping of chemical and/or physical mineral phase transformations involving jadeite, coesite, diamond or zircon in natural ultra-high pressure metamorphic environments (UHPM). ht: FREDERICKS, P. M., FROST, R. L. & RINTOUL, L. (eds) XIXth International Conference on Raman Spectroscopy, ICORS, Gold Coast, Australia, 8-13 August 2004, Proceedings (CD-ROM). SMITH, D. C. 2004d. Non-destructive semi-quantitative chemical analysis of a garnet gemstone by Raman microscopy compared with analysis by PIXE. hTternational Congress GEORAMAN-2004, Honolulu, UniversiO' of Hawaii, 6-11 June 2004.

NON-DESTRUCTIVE RAMAN SPECTROSCOPY School of Ocean and Earth Science Technology, Special Publications, 04-02, 73-74. SMITH, D. C. 2005a. Jewellery and precious stones. In: EDWARDS, H. G. M. & CHALMERS, J. (eds) Raman Spectroscopy in Archaeology and Art History. Royal Society of Chemistry, London, 335-378, SMITH, D. C. 2005b. The RAMANITA ~- method for non-destructive and in situ semi-quantitative chemical analysis of mineral solid-solutions by multidimensional calibration of Raman wavenumber shifts. Proceedings, 6th International GEORAMAN Congress, Hawaii, June 2004. Spectrochimica Acta, Part A 61, 2299-2314. SMITH, D. C. 2005c. Mesoamerican jade. In: EDWARDS, H. G. M. & CHALMERS, J. (eds) Raman Spectroscopy in Archaeology and Art History. Royal Society of Chemistry, London, 412-426, SMITH, D. C. 2005d. Raman mapping of the clinopyroxene/silica contacts in the quartz-jadeitite rock, possibly an 'Olmec Blue' jade, from Guatemala. 3rd International Conference on the Application of Raman Spectroscopy in Art and Archaeology, Louvre, Paris, 31 August-3 September 2005, Abstract volume, 63. SMITH, D. C. & BARBET, A. 1999. A preliminary Raman microscopic exploration of pigments in wall-paintings in the Roman tomb discovered at Kertch, Ukraine, in 1891. Journal of Raman Spectroscopy, 30, 319- 324. SMITH, D. C. & BOUCHARD, M. 2000a. Analyse de pigments des peintures pari&ales de Pergouset par microscopie Raman, In: LORBLANCHET, M. (ed.) La Grotte de Pergouset. DAF, Paris, 174. SMITH, D. C. & BOUCHARD, M. 2000b. PETRORAMAN. Archdom~trie, Dossiers de l'Archdologie, 253, 54. SMITH, D. C. & BOUCHARD, M. 2002. Bronze disease: non-destructive distinction of three polymorphs of Cu2CI(OH)3 by Raman microscopy (RM). 'ART 2002', 7th International Conference on Nondestructive Testing and Microanalysis for the Diagnostics and Conservation of the Cultural and Environmental Heritage, University of Antwerp, June 2002, Abstracts Volume, 227. SMITH, D. C. & CARABATOS-NEDI~LEC, C. 2001. Raman spectroscopy applied to crystals: phenomena and principles, concepts and conventions. In: LEWIS, I. & EDWARDS, H. G. M. (eds) A Handbook on Raman Spectroscopy. Marcel Dekker, New York, 349-422. SMITH, D. C. & EDWARDS, H. G. M. 1998. A wavenumber-searchable tabular indexed catalogue for 'ARCHAEORAMAN~':~': Raman spectra of geomaterials and biomaterials of interest in archaeology (sensu lato). In: HEYNS, A. M. (ed.) ICORS Capetown'98. Wiley, Chichester, 510-511. SMITH, D. C. & GENDRON, F. 1997a. Archaeometric application of the Raman microprobe to the nondestructive identification of two Pre-Columbian ceremonial polished 'greenstone' axe heads from Mesoamerica. Journal of Raman Spectroscopy 28, 731-738. SMITH, D. C. & GENDRON, F. 1997b. New locality and a new kind of jadeite jade from Guatemala: rutile-

31

quartz-jadeitite. Fifth International Eclogite Conference. Terra Nova, 9, Abstract Supplement 1, 35. SMITH, D. C. & OSPITALI, F. 2005. First use of a handheld self-contained ultra-mobile Raman microscope for non-destructive in situ analysis of gemstones inlaid in XVIIth century Florentine stone marquetry tables. 3rd International Conference on the Application of Raman Spectroscopy in Art and Archaeology, Louvre, Paris, 31 August-3 September 2005, Abstract volume, 66. SMITH, D. C. & PER1N, P. 2003. Non-destructive in situ MRM (mobile Raman microscope) semiquantitative chemical analysis of garnets in Merovingian cloisonn~ style jewellery from Vicq, France. International Congress 'Application of Raman Spectroscopy in Art and Archaeology', Ghent, September 2002, 62. SMITH, D. C. & PINET, M. 1989. A method for single variable plotting of multi-dimensional chemical data for comparing Raman wavenumbers with chemical variations in solid-solutions. GEORAMAN-89: Contributions. Special Publication, Association Nationale de la Recherche Technique, Paris, 24. SMITH, D. C. & ROBIN, S. 1997. Early-Roman Empire intaglios from 'rescue excavations' ign Paris: an application of the Raman microprobe to the non-destructive characterisation of archaeological objets. Journal of Raman Spectroscopy, 28(2-3), 189-193. SMITH, D. C. & RONDEAU, B. 2001. Non-destructive mineralogical analysis of precious stones in situ under thick glass by MRM (mobile Raman microscopy). Congress 'Archdomdtrie 2001', GMPCA, La Rochelle Universi~. , La Rochelle, 101. SMITH, D. C. & VERNIOLES,J. D. 1997. The temperature of fusion of a Celtic vitrified fort: a feasibility study of the application of the Raman microprobe to the non-destructive characterisation of unprepared archaeological objects. Journal of Raman Spectroscopy, 28(2- 3), 194-197. SMITH, D. C., BOUCHARD, M. A. & LORBLANCHET, M. 1999a. An initial Raman microscopic investigation of prehistoric rock art in caves of the Quercy district, S. W. France. Journal of Raman Spectroscopy, 30, 347-354. SMITH, D. C. EDWARDS, H. G. M. and Russ, J. 1999b. Congress GEORAMAN'99, Abstracts. Valladolid University Press, Valladolid, 61-62. SMITH, D. C., CARABATOS-NEDI~LEC, C. BOUCHARD, M. 1999c. VITRORAMAN: establishing a database on the Raman spectra of pigments on and in stained glass. GEORAMAN'99, Abstracts. Valladolid University Press, Valladolid, 36-37. SMITH, D. C., EDWARDS, H. G. M., BOUCHARD, M., BRODY, R., RULL-PEREZ, F., WITHNALL, R. & COUPRY, C. 2000. MRM (mobile Raman microscopy): a powerful non-destructive polyvalent in situ arch~eometric tool for microspectrometrical analysis of cultural heritage in the next millennium (ARCHAEORAMAN): geomaterials, biomaterials and pigments. In: GUARINO, A. (ed.) 2nd International Congress on 'Science and Technology for the Safeguard of Cultural Heritage in the Mediterranean Basin, Nanterre, Paris, 5 - 9

32

D.C. SMITH

July 1999, Proceedings, Vohone 2. Elsevier, Amsterdam, 1373-1375. SMITH, D. C., GONTHIER, E. & REINHARDT,A. 2003a. Raman Microscopy of Chinese nephrite jades, both ancient cultural artefacts and geological sources in China. International Congress 'Application of Raman Spectroscopy in Art and Archaeology', Ghent, September 2002, 85. SMITH, D. C., PI~RIN, P., KAZANSKI,M. & GABUEV, T. 2003b. Andradite-rich pentary garnet compositions deduced from non-destructive in situ MRM (Mobile Raman Microscope) semi-quantitative chemical analysis of Vth century cloisonn~ garnets from an Alan warrior burial mound at Brut, Northern Ossetia, Russian Federation. International Congress "Application of Raman Spectroscopy in Art and Archaeology', Ghent, September 2002, 39. VANDENABEELE, P., MOENS, L., EDWARDS, H. G. M. & DAMS, R. 2000. Raman spectroscopic database of azo-pigments and application to modern art

studies. Journal of Raman Spectroscopy, 31(6), 509-517. WANG, A., JOLLIFF, B. L. & HASKIN, L. A. 1996. Raman system for robotic exploration on planets. GEORAMAN-96, Terra Abstracts, Supplement N 2. Terra Nova, 8, 25. WANG, C., LU, B., ZUO, J., SUZUKI, M. & CHASE, W. T. 1995. Structural and elemental analysis of the nanocrystal SnO2 in the surface of ancient Chinese black mirrors. Nanostructured Materials, 4, 489-496. WHITE, W. B. 1975. Structural interpretation of lunar and terrestrial minerals by Raman spectroscopy. In: KERR, C. (ed.) hlfrared and Raman Spectroscopy of Lunar and Terrestrial Materials. Academic Press, New York, 325-357. W1THNALL, R. 1999. h7 situ identification of pigments, manuscripts and prints using Raman microscopy. GEORAMAN'99. Valladolid University Press, Vailadolid, 15- ! 6.

The Neolithic pottery of Abri Pendimoun (Castellar, France): a petro-archaeometric study E. B A S S O 1, D. B I N D E R 2, B. M E S S I G A 1'3 & M. P. R I C C A R D I 1'3

1Dipartimento di Scienze della Terra, Universit~ degli Studi di Pavia, via Ferrata n. 1, 27100 Pavia, Italy 2CEPAM-CNRS, rue Albert Einstein n. 250, Valbonne-Sophia Antipolis 06560 France, 3CISRiC-Beni Culturali, Universith degli Studi di Pavia, via Ferrata n. 1, 27100 Pavia, Italy (e-mail: [email protected]) Abstract: Middle and Late Neolithic ceramics from Abri Pendimoun (Castellar, France)

and their geological raw materials have been investigated to characterize the ceramic bodies and to determine the possible provenance of raw materials. Petrographic, mineralogical and energy-dispersive spectrometry analyses were undertaken to define the compositional parameters of sherds and to clarify the relationship between Square Mouthed Pottery-phase I (VBQ I) and Chassey Culture, The ceramic bodies were generally made from glauconite-rich layers and terra rossa, unprocessed or mixed in variable proportions. Different kinds of temper, such as carbonates and/or aplite fragments, were added to the mixtures. Although most of the analysed ceramics were produced locally, a few mixtures show the addition of exogenous rocks. Although these ceramics could be interpreted as imported, we demonstrate that local clayey materials were used at Abri Pendimoun. The hypothesis that pottery was imported can therefore be ruled out. A small amount of crushed calcite (5%) was added to some glauconitic pellet mixtures. Pots made with this mixture are normally referred to the VBQ I. This combination of mixture and shape indicates that there was an important link between the VBQ I and Chassey Cultures.

The Square Mouthed Pottery-phase I Culture (VBQ I) played an important role in the development of Chassey Culture, as seen at the Villa Giribaldi site in Nice (Binder 1990; Binder et al. 1994) and in the diffusion of Chassey Culture in northern Italy (Crepaldi 2001). The VBQ I appears in northern Italy from the sixth millennium BP: Chassey Culture arose in Provence during the first half of the same millennium. The expansion of Chassey Culture in Italy (Fig. 1) is very complex and it is today acknowledged that this cultural change was accompanied by ethnic transfer from southern France to the east, contributing to the spread of the Western Neolithic in Italy (Bagolini & Pedrotti 1998). The occurrence of Bedoulian chert from Provence in Chassey Culture layers at the Arene Candide cave (Western Liguria) (Starnini & Voytek 1997; Binder 1998) demonstrates raw material circulation as a factor in spreading the pot shapes, styles and technique of Chassey Culture in Italy (Binder 1998). The VBQ I influences in France are less clear. Figure 1 shows the VBQ I sites recorded in Provence.

The site of Abri Pendimoun is in a key position along the communication route between Liguria (Italy) and eastern Provence (France). The clear evidence of contacts with the VBQ I, seen in the geometric 'graffita' decorations on potsherds from the more ancient horizons (Binder 2000), makes Pendimoun an important site in the study of interactions between the VBQ I and Chassey Cultures in the transalpine area, which occurred during the Middle and Late Neolithic age. This pottery is therefore particularly important, not only because it enriches the collection of artefacts from the Middle and Late Neolithic, but also because the technological and cultural evidence may be extended to coeval archaeological contexts. These topics necessarily touch on a broader issue in the archaeology of ceramic production, i.e. the concept of 'local pottery' or 'trade pottery', which in turn has important implications for the reconstruction of ancient economies. Prehistoric ceramic products are generally assumed to be local, although the temper may well have been obtained elsewhere (Martineau

From: MAGGETTI,M. & MESSIGA, B. (eds) 2006. Geomaterials in Cultural Heritage. Geological Society, London, Special Publications, 257, 33-48. 0305-8719/06/$15.00 ~ The Geological Society of London 2006.

34 J

E. BASSO E T AL. s u ~ s s E

z_.C

T~no

:++i

+,.m

o

Pia6e.za

:

~

-..

~

"

'

"

6

"+,+/

~

,"~v..,.

k --

'~ ./

II

',...,

"

Fig. 1. Simplified index map of Chassey sites in Italy and of VBQ I sites in Provence (modified from Crepaldi 2002).

2000; Di Pierro 2003). The choice of temper may indeed have been 'cultural', sometimes determining the selection of source areas that were far from the site (Constantin & Courtois 1985). The definition of 'local pottery' has been discussed in an ethno-archaeological study by Arnold (1985), who redefined the term in 2001 (Arnold 2001). Arnold considered that the clay and temper used to manufacture pottery should be sought within a 1 - 2 k m radius of the production site, whereas a vessel produced with material from more than 10 km from the archaeological site was brought to the site for reasons not strictly related to its manufacture. The concepts of local or imported are still unclear, as the provenance of raw materials seems to vary in relation to different parameters, including the geological and/or geographical context, the cultural traditions of certain groups, trade among groups, and social-economic factors influencing the selection of source areas (Martineau 2000). Actually, considering how demanding prehistoric populations were in the selection of raw materials for lithic production, Arnold's affirmation is not necessarily true (Bir6 1998). Indeed, there is clear evidence of differential management of lithic resources, and it is likely that obsidian (Crisci et al. 1994) and Bedoulian chert were transported over long distances (Binder & Perlrs 1990) to be traded for 'greenstone' (i.e. eclogites) (Odetti 1991; D'Amico 2005; D'Amico & Starnini 2005). Even assuming the local origin of many ceramics, there may be large technological and

decorative variations as a result of, for instance, a widespread exchange of ideas and traditions. Archaeological sites, especially prehistoric ones, rarely coincide with the source area of raw materials. At Abri Pendimoun, archaeological evidence, such as the discovery of potter's tools and raw materials (Binder 2002), suggests that pottery was produced locally. In this study the term 'local pottery' is applied to pottery produced from natural materials available over a sometimes vast geographical area containing the archaeological site. The local boundaries may be natural or man-made barriers that limited access to other possible source areas. The investigated ceramic material, abundant in all stratigraphic levels, provides a detailed record of the evolution of shapes, decoration styles, surface treatments and of ceramic mixtures at the site between Ancient and Late Neolithic time. This work presents a petro-archaeometric study of Middle and Late Neolithic ceramics (1245 artefacts) from the Abri Pendimoun excavation. Geomaterials possibly used in the ceramic production will be also studied. The aim of the research is to identify raw materials (geomaterials) and production technology used to make the clay body mixture, and to give an insight into relationships between VBQ I and Chassey Culture in Provence.

Archaeological framework First excavated in 1955-1956 by Louis Barral (curator of the Museum of Prehistoric Anthropology in Monaco) and his staff, the Abri Pendimoun site (Castellar, Alpes Maritimes, France) was then known for the numerous finds from the Ancient Neolithic and for an important sepulchre ascribed to the Cardial period of the Ancient Neolithic in Provence (Barral 1958). These elements made it an important point of reference between Fontbr~goua to the west (Courtin 1976) and the Arene Candide cave to the east (Bernab6 Brea 1946, 1956) for analysing the spread of Neolithic culture in the northwestern Mediterranean. The Abri Pendimoun ('abri', rock shelter) documents a stratigraphic sequence from the upper Epipalaeolithic (about 10th millennium BP) to the final Neolithic (about third millennium BP) (Binder et al. 1993). The Pendimoun excavations reveal one of the most peculiar and detailed stratigraphic sequences (including the seventh and sixth millennia BP) studied so far in the western Mediterranean area. The sequence shows great research potential, and its peculiar characteristics may help to solve some issues

NEOLITHIC POTTERY OF ABRI PENDIMOUN regarding the Mediterranean Neolithic, such as intercultural relations in the area between the domains of Provence and Italy. The interpretations, already discussed by Binder et al. (1993), make Abri Pendimoun one of the most important sites providing insight into the spread of Neolithic culture in western Europe.

Geological setting The rock shelter is a natural cavity formed at the contact of limestones (KimmeridgianPortlandian) with grey marls (Neocomian), which are in turn interbedded with glauconite-rich

D [~G: [~

E: debris chaotic deposits Annot sandstone and Otigocene flysch E~_~:Melobesiae limestones (Upper Eocene)

l

layers. The grey marls are in contact with yellow marls resulting from the alteration of the limestones. This contact enhances water circulation, as evidenced by the numerous springs at the foot of the Ormea cliff. The principal geological formations in this area (Fig. 2) are Upper Cretaceous marly-calcareous undifferentiated rocks (C3_7), with massive limestones, and Neocomian units, characterized by marly limestones, calcareous marls and layered marls interbedded with glauconite-rich layers (N|_4). The Portlandian limestones (J9) a r e massive, and their colour varies from grey (northwestern area) to white (southern area). The Upper Jurassic is

N,~: Marly limestones, calcareous marls, marls interbedded with gtauconitic layers (Early Cretaceous)

1"[]]]] Jg-J~:Limestones and marls (Jurassic) ]T~:

marls and gypsum (Late Triassic)

~ z ] Es: Nummulitic limestone (Middle Eocene)

] M e d i t e r r a n e a n Sea

[~

" ' : France-Italy boundary

C3_~:marls and limestone (Cretaceous)

35

Fig. 2. Location of the archaeological site (A) and sampling points (.) of raw materials, on a simplifiedgeological map (after Geze 1968). The Ormea cliff is located near the Abri Pendimoun site. The numbers correspond to the samples listed in Table 2.

36

E. BASSO ETAL.

represented, in the south, by a pseudo-oolitic massive limestone.

Table 1. Ceramic finds analysed in thin section Thin section

Inventory no.

AP37 AP4 AP5 AP6

AP 23075 AP 105 LI9 AP 11003 AP 112 LI9 12-14 R3 AP 11723 AP 12 M19 AP 122 LI9 12-14 AP 12416 AP 190 M21 n60-82 AP 20606 AP 20645 + 20763 AP 22708 AP 26 M20 AP 3 MI9 7 AP 77 LI9 AP M21 CPE diag str 11 AP 11223-4 AP 68 L19 8-10 RI

Subgroup

Sampling Ceramic finds The study of ceramic finds entailed a preliminary phase, in which all potsherds (1245 finds) ascribed to the Middle and Late Neolithic were analysed under the optical stereomicroscope. A selection of 46 ceramic fragments was used for archaeometric investigation (Table 1). These were chosen from among those artefacts belonging to the most representative groups of mixtures; also 'anomalous' finds, i.e. with unusual textural and/or compositional characteristics with respect to those of the main ceramic groups, were taken into account. R a w materials To identify the source area of raw materials, samples of geological materials (15 specimens) suitable for the production of ceramics were collected within an approximately 5 km radius of the archaeological site (Fig. 2). The sampling was determined by: (1) geological knowledge of the area; (2) petrographic data obtained from ceramic artefacts; (3) excavated raw materials. The geological formations involved were (1) Neocomian marly carbonate clays and the intercalated glauconite-rich layers (Nj_4); (2) Portlandian (J9) and Upper Cretaceous (C3_7) marly clays; (3) terra rossa (Fig. 2; Table 2).

Analytical strategy The sampling strategy strictly follows the multidisciplinary investigative procedure defined by Basso (2004). This procedure involves three stages ( M A C R O 1 - M I C R O - M A C R O 2 ) and entails the study of finds at different scales of observation, as well as a complete integration of archaeometric and archaeological data. This approach aims to apply both the first (MACRO1) and last (MACRO2) stages to all the excavated ceramic materials using an optical stereomicroscope (a non-invasive and low-cost procedure). The archaeometric stage sensu stricto, completed at a detailed observation scale (MICRO), was applied to samples containing variations in textural and compositional characteristics identified during the MACRO1 stage. The extremely detailed observational characters, obtained at the MICRO scale, were then applied to all samples during the following MACRO2 stage.

Wall thickness (mm)

1G IG 1G 1G

8.7 6.8 5.6 9.8

1G 1G 1G IG 1G 1G IG 1G IG IG IG 1G

4.4 6.0 7.8 4.2 5.0 5.0 4.0 5.4 7.0 7.2 7.3 6.5

1G IG

4.2 6.0

12045 13586 74 L19 8-10 R3 HS 2002

IGS IGS 1GS I GS

7.8 5.0 6.4 9.0

AP30 AP31

AP 12454+ 12436 AP 12346

IGCc 1GCc

5.2 5.5

API9 AP7 API2 AP23 AP24 AP26 AP28 AP 16 API8 AP27

AP AP AP AP AP AP AP AP AP AP

1GC IGC 1GC 1GC IGC 1GC 1GC 1GC 1GC 1GC

7.3 4.2 6.3 10.6 9.0 5.3 7.6 4.8 6.5 6.4

AP2 AP3 AP33

AP 10911 AP 8955

IGSCc 1GSCc

6.8 7.0

2S

7.3

2S 2S 2S 2S

9.3 4.3 5.5 5.9

2S 2S

5.5 5.5

AP8 AP9 AP10 AP11 AP13 AP14 AP15 AP17 AP20 AP21 AP22 AP25 AP29 AP32 AP34 AP36 AP38 AP39

AP AP AP AP

23720 11272 13541 83 K19...RI 97 LI9 12-14 RI 13639 23072 21 M20 7 23074 22699

AP35 AP40 AP41 AP42 AP44 AP45

AP 103 L 19 12-14 R2 AP 12539 AP 9781 AP 10313 AP 8972 + 8973 + 8977 + 9038 AP 8996 AP 9361

API

AP HS

2C

6.0

AP43 AP46

AP 20225* AP 5243

3F 3F

8.0 9.0

*VBQ ! fragment.

NEOLITHIC POTTERY OF ABRI PENDIMOUN Table 2. Geomaterials sampled near the archaeological site Sample no.

Geomaterial

Geological formation

Location no.*

AP.S2 AP.S3 AP.S4 AP.S5 AP.S9 AP.S10 AP.S 11 AP.S14 AP.S15 AP.Sl6 AP.S17 AP.S 18 AP.SI9 AP.S20 AP.S24

Glauconitic layer Carbonatic clay Marly clay Gauconitic layer Gauconitic layer Gauconitic layer Gauconitic layer Terra rossa Marly clay Gauconitic layer Gauconitic layer Gauconitic layer Marly clay Terra rossa Giauconitic layer

Nl_ 4 J9

2 3 4 5 9 10 11 14 15 16 17 18 19 20 24

C3_7 N1_4 Nl_ 4 NI_ 4

N1-4 E C3_7 NI_ 4 NI_ 4

N 1-4 C3_ 7

E N1_4

*Numberscorrespond to sampling points in Figure 2. In this way, the dataset acquires statistical validity, as all typologies of ceramic mixtures from the excavation area are included and it can be integrated with archaeological data. In the MACRO 1 stage, the most representative artefacts of each group and subgroup of ceramic mixtures were selected, along with some specimens showing 'anomalous' characteristics. The MICRO stage was applied to 46 highly fragmented ceramic finds representing common wares (plates, bowls, tall pots, jars, basins, cups, etc.), whose shape cannot always be reconstructed. The MACRO1 stage

The diagnostic character for the pottery of Abri Pendimoun, identified during the first stage of investigation (MACRO1), is the presence or

37

lack of spheroidal glauconitic pellets. The most statistically diffuse texture is the 'hiatal' one, as defined by Maggetti (1994), which was identified with certainty under the stereomicroscope by the evident discontinuity between the grain size of non-plastic inclusions (temper and framework) and that of the matrix. All 1245 ceramic finds have been grouped into three petrographic groups (Table 3; Fig. 3a). Only a few finds lack non-plastic inclusions and have thus been placed in a specific petrographic group. Subgroups were defined according to the presence of glauconitic pellets and their association with other non-clayey materials (Table 3; Fig. 3b), whose mineralogy was determined by evaluating the hardness of single fragments, their shape and colour. The carbonate fragments are softer and may be scratched with a metal point, whereas silicate fragments are harder and cannot be scratched. Fragments of non-uniform colour were classified as lithic clasts. Most of samples belong to Chassey Culture; the 41 remaining samples are related to the VBQ I (Basso 2004). These latter in turn have only one sample belonging to group 1. The MICRO stage

The multi-analytical strategy (MICRO) applied to the study of ceramic artefacts and raw materials combines textural (using polished, C-coated thin sections), microchemical and bulk analyses. The textural analysis of thin sections is the most straightforward approach to investigate ancient ceramics (Courtois 1976; Maggetti 1994; Dickinson & Shutler 2000). Petrographic studies under the optical microscope provide data on the mineral composition and proportion (modal amount) of temper, non-plastic inclusions and clay matrix. Furthermore, the

Table 3. Groups and subgroups recognized under optical stereomicroscope (MACROI column) and under polarized-light microscope, in thin section (MICRO column) Group

1

Fabric

Hiatal

Glauconitic pellets

Yes

Subgroup MACRO 1

MICRO

1G 1GS

1G IGS 1GCc 1GC

1GC 1GSC

2 3

Hiatat Seriate

No No

Temper

{ 2Cc 2s 3F

IGSCc 2S 2Cc 3F

Untempered Aplitic rocks Calcite (a) Glauconitic limestone (b) Oolitic limestone (c) Micritic limestone Aplitic rocks + calcite Aplitic rocks Calcite Untempered

38

E. BASSO ETAL.

(a)

2S (3%)~2Cc (1%)

(b)

i'fi'iil

(1%)

1GS (13%)

1245 SAMPLES Group 1 (95%)

1G (69%)

I

Fig. 3, (a) Finds distribution among the three groups (MACRO phase). (b) Finds distribution among subgroups (MACRO phase). G, glauconitic pellets: S, silicate fragments:C, carbonate fragments:Cc, sparry calcite; F, fine mixture. scanning electron microscope (SEM) is used for the microstructural and microchemical investigation of the intergranular clay matrix. The analytical procedure combines microanalytical data on the clay matrix with mineralogical (qualitative composition) and petrographic data on the ceramic bulk sample. Micro-analytical data were determined at the University of Siena (Earth Science Department) using a Philips XL 30 SEM equipped with a Philips EDAX-DX4 energy dispersive spectrometer. The expected error for major and minor element measurements is 5% and 15%, respectively. The qualitative mineralogical composition was obtained by X-ray powder diffraction (XRPD) and polarized-light microscope examination of thin sections. X-ray powder samples were prepared by crushing in an agate mortar followed by addition of 10% diluted HC1. This last procedure was adopted to avoid the interference due to the presence of abundant calcite, which could mask the diagnostic reflections of firing phases such as gehlenite, pyroxene and anorthite. This procedure was not applied to raw materials. Bulk chemical data was determined at Activation Laboratories Ltd (Ontario, Canada) using the fusion inductively coupled plasma-optical emission spectrometry (ICP-

OES) technique. Samples were prepared and analysed in batches. Each batch contained a method reagent blank, certified reference material, and 17% replicates. Samples were mixed with a flux of lithium metaborate and lithium tetraborate and fused in an induction furnace. The molten melt was immediately poured into a solution of 5% nitric acid containing an internal standard, and mixed continuously until completely dissolved (c. 30min). The samples were then run for major oxides on a combination simultaneous-sequential Thermo Jarrell-Ash ENVIRO II ICP. The sample solution was also spiked with internal standards and was further diluted and introduced into a Perkin Elmer SCIEX ELAN 6000 inductively coupled plasma mass spectrometer using a proprietary sample introduction technique. Calibration was performed using USGS and CANMET certified reference materials (Table 4).

Results The results of the petro-archaeometric and chemical analyses of ceramic finds and raw materials are discussed for each mixture group and for each typology of raw materials; only the clay matrix petrography is described separately.

Table 4. Typical ICP analysis (wt%) of some USGS and Canmet Standards (%) (detection limits are given in parentheses)

Standard

SiO2 (0.01)

A 1 2 0 3 Fe203 (0.01) (0.01)

SY3 MRG1 DNC1 BIR1 W2 G2 STM1

59.51 39.43 46.91 47.78 52.58 68.72 59.64

11.62 8.59 18.46 15.43 15.35 14.95 18.07

6,47 17.93 9.76 11.52 10.72 2.65 5.24

MnO (0.001)

MgO (0.01)

CaO (0.01)

Na20 (0.01)

K.O (0.01)

TiO2 (0.001)

P205 (0.01)

0.32 0.17 0.15 0.17 0.16 0.03 0.22

2.54 13.74 10.05 9.70 6.37 0.71 0.07

8.25 14.77 11.27 13.75 10.98 1.87 1.09

4.17 0.73 1.99 1.86 2.31 4.08 8.87

4.23 0.18 0.24 0.02 0.64 4.48 4.24

0.14 3.78 0.47 0.95 1.05 0.48 0.13

0.52 0.07 0.07 0.02 0.12 0.13 0.1

NEOLITHIC POTTERY OF ABRI PENDIMOUN The textural and compositional parameters (i.e. grain size, shape, mineralogy of nonplastic inclusions and non-plastic inclusions/ clay matrix ratio) allowed the detailed definition of the different types of ceramic mixtures. The presence or lack of glauconitic fragments and of other kinds of non-plastic inclusion defines the petrographic groups of ceramic mixture (Fig. 3a). Within these groups, the association of different types of temper defines seven subgroups (Fig. 3b; Table 3) and 10 types of mixture (Table 3). Artefacts with a seriate texture form a separate group (Table 3). Table 5 reports the mineralogical association of the non-plastic inclusions of the ceramic finds according to group and subgroup mixtures. Two types of clay matrix, dark brown or light brown-reddish, may be identified in thin section under the polarized microscope. Often, in the dark brown matrix, the framework is absent or undetectable. The framework essentially consists of small quartz fragments sometimes associated with plagioclase, K-feldspar, rare muscovite lamellae and small rock fragments (microcrystalline siliceous and sedimentary rocks). In contrast, the light brown-reddish matrix has a rather abundant framework of <200 txm grain size consisting of quartz, K-feldspar, plagioclase, muscovite and/or biotite as major phases, associated with rare fragments of microcline and zircon. Occasionally, fragments of calcite, microcrystalline silicate rocks and, occasionally,

small fossils observed.

39

with

carbonate

shells

were

Ceramic finds Group 1. This group always shows the presence of glauconitic pellets, sometimes associated with rock fragments (silicate or carbonate) and/or sparry calcite. These are rounded, 25 lxm to a few millimetres in size (Fig. 4a) and, in contrast to the bright green colour of unfired materials (Fig. 5a), they are yellowishbrown, brown, red, or black. The glauconitic pellets are cracked during firing, and the cracks often are filled with secondary calcite aggregates of a few tens of microns in size. Silicate rock temper is commonly of millimetre scale (5001xm to 2mm), with an irregular, angular and subordinately subrounded morphology (Fig. 4b), and occurs in amounts ranging from 10 to 15 vol% in subgroup 1GS. In the fragments, the mineralogical associations are quartz + plagioclase, quartz + plagioclase + K-feldspar and quartz + muscovite + plagioclase (feldspars are well preserved). Textural relations among minerals suggest that fragments derive from igneous rocks such as aplites. This kind of temper is associated with glauconitic pellets in subgroup 1GS (Fig. 4b) and with glauconitic pellets and sparry calcite in subgroup 1GSCc. In this latter subgroup calcite and aplite fragments do not exceed 10 vol%. The presence

T a b l e 5. Petrographic composition of sherds (rough average for every subgroup)

Group

1

2

3

L

Subgroup T/M

G

GS

GCc

GC.a

GC.b

G C . c GSCc

>0.4

>0.4

>0.4

>0.4

>0.4

>0.4

Glauconitic pellets Quartz Micas K-feldspar Plagioclase Sparry calcite Calcite Aplite Sandstone Glauconitic limestone Oolitic limestone Micritic limestone Jasper Fe-oxides

: ! ..... :. : ,

S

>0.4 |

Cc

>0.4 ]

0.1
>0.4 [

m

....... ......... I

J '

- ' ,.

Bt ~

,

.

.

.

. ~i

..

F

......... ~"

,.I :~ii~:.

....

i I

,,,,

: . .". .

•

The estimated value of 0.4 is the minimum T/M ratio calculated for each subgroup. G, glauconitic pellets; GS, glauconitic pellets + silicate grains; GCc, glauconitic pellets + carbonate grains; GC, glauconitic pellets + carbonate grains; GSCc, glauconitic pellets + silicate grains + calcite; S, silicate grains; Cc, calcite; F, fine; T/M, non-plastic inclusions/clay matrix ratio.

40

E. BASSO ET AL.

of carbonate rock fragments characterizes subgroup 1GC (Table 5). Fragments have an irregular, subangular shape and vary in size from a few hundred microns to 1 mm. The abundance of carbonate rock fragments ranges from 5 to 20 vol%. Microscopic analysis allowed the identification of three lithologies: (I) glauconitic limestones; (2) oolitic and/or fossiliferous limestones; (3) micritic limestones. These types of temper are always associated with glauconitic pellets and define three types of mixture within subgroup 1GC (Table 5): glauconitic pellets and glauconitic limestone (1GC.a, Fig. 4c), glauconitic pellets and oolitic and/or fossiliferous limestone (1GC.b), and glauconitic pellets and micritic limestone (1GC.c). The sparry calcite as temper is rather sporadic and often associated with other types of tempers (subgroups 1GC c and 1GSCc). In subgroup I GCc (Fig. 4d), a minimum percentage of sparry calcite (5%) is associated with the glauconitic pellets.

. t

j,~"

..It

This group is characterized by the constant presence of a clay mineral peak at d = 10.1 A, which in this case may correspond to glauconite. Quartz is the most abundant mineral detected by XRPD (Table 6). The intensity of quartz reflections in group I GS is the highest (Table 6). The major amount of quartz and the constant presence of feldspar reflections (plagioclase and K-feldspar) are attributed to the rock fragments of the non-plastic fraction. Neoformed minerals such as pyroxene and/or hematite were identified in rare cases; in they occur particular in groups I GS, 1GCc and 1GC (Table 6). Nevertheless, the presence of firing phases cannot be excluded for two reasons: either the dimensions of crystals are too small to yield a significant XRD response, or neoformed phases may have been weathered during burial, as emphasized elsewhere for gehlenite (Courtois 1973; Heimann & Maggetti 1981 ; Maggetti 1994). In this group the clay matrix corresponds to the dark brown typology. In the matrix, S i O 2 and

.

Fig. 4. Micrographs of ceramic mixtures. (a) Hiatal texture, subgroup 1G (plane-polarized light); (b) hiatal texture, subgroup 1GS (crossed Nicols); (c) hiatal texture, subgroup IGC.a (plane-polarized light); (d) hiatal texture, subgroup 1GCc (crossed Nicols);

NEOLITHIC POTTERY OF ABRI PENDIMOUN A1203 contents very widely in group 1, from 57 to 63.1 wt% and from 8.2 to 21.8 wt% on average, respectively (Table 7). MgO and FeO show a positive correlation when the clay matrices of group 1 and the glauconite-rich layers are considered. Only a group of finds and of glauconite-rich layers varies from this general trend in MgO content. Group 1 contains up to 95% of finds (Fig. 3a). In this group, the VBQ I pottery corresponds to the subgroups 1GCc (56%), 1G (34%) and 1GS (10%) (Basso 2004).

Group 2. Ceramics ascribed to this group contain aplite fragments in subgroup 2S (Fig. 4e; Table 5) and sparry calcite in subgroup 2Cc (Fig. 4f; Table 5). In subgroup 2S, the abundance of igneous rock fragments (25-30 vol%) is greater than in subgroups 1GS and 1GSCc. Rarely, only sparry calcite is used as a temper (subgroup 2Cc, Fig. 4f). In this latter group, consisting of a single specimen, rhombohedral calcite is 100 ~m to 1 mm in size. Calcite

41

never exceeds 20 vol% and the fragments never show any evidence of decarbonation. Group 2 is characterized by the almost total absence of mica/illite-glauconite reflections. Quartz is the most abundant phase, followed by K-feldspar and plagioclase (Table 6). Neoformed pyroxene, probably diopside, is scarce or rare. This mineral was detected only by XRPD. The clay matrix corresponds to the reddish typology. Group 2 contains up to 4% of finds (Fig. 3a). In this group, the VBQ I pottery corresponds to subgroup 2S (20%) (Basso 2004).

Group 3. Group 3 only contains subgroup 3F, it lacks temper fragments and is characterized by a fine-grained mixture (Fig. 4g). The mineralogical composition is similar to that of Group 2, except for a minor content of quartz and feldspars (Table 6). In this group the clay matrix corresponds to the light brown or reddish typology.

Fig. 4. Continued(e) hiatal texture, subgroup 2S (crossed Nicols); (f) hiatal texture, subgroup 2Cc (crossed Nicols); (g) seriate texture, subgroup 3F (plane-polarizedlight). G.p., glauconiticpellets; A.r., aplite; G.I., glauconitic limestone; Cc, sparry calcite.

42

E. BASSO ETAL.

-

•

,.-

-

? ~.~"

~

,

"

~

~

,

~.-~-

. • '"

q ."

"

-,j

,'~ ,~

-

.

~.,

:

&~ ,.

,

,

~ .

•

-

.

~

~ .

. .

.

,"

~

~"

;6.

,,

'

Jl~

o,

~

f

" ~ .

-

lu,r

,~..

/

-,

• .. .

-" i

.

~,

•

*

I~.

,

&

i t

"

.," ~.

.-

•

,

'adD~ill

.

.

b

;.

a

."

'

. .

~

"~i[,

"

Fig. 5. Micrographs of raw materials. (a) Carbonate clay (plane-polarized light); (b): glauconitic layer (plane-polarized light); (c) terra rossa (plane-polarized light); (d) clay pellets (subgroup IGC; crossed Nicols). C.p., clay pellet.

Group 3 contains up to 1% of finds (Fig. 3a). In this group, the VBQ I pottery comprises 10% of finds (Basso 2004).

Raw materials Carbonate and marly clays. These are very fine, fat clays (Fig. 5b). The variable carbonate component is easily detected by XRD (Table 6). The framework consists of small angular quartz fragments, some muscovite lamellae and rare sedimentary rock fragments. Small glauconitic pellets (< 150 p.m) are scanty. Small gastropod fossils in the marlier portions may also occur. Clay minerals are chiefly i l l i t e + k a o l i n i t e _ chlorite (Table 6). The marly layers contain a moderate amount of kaolinite and high contents of calcite (Table 6); they show low SiO2 and A1203 contents and very high values of CaO (Table 7). The carbonate layers are the richest in calcite, with moderate K-feldspar and low quartz contents (Table 6). In Table 7, they show A1203 and CaO contents of 13 wt% and 15 wt%, respectively, associated with low values of MgO and Fe203.

Glauconite-rich layers. These layers consist of abundant glauconitic pellets held together by a very fine matrix (Fig. 5a). The pellets are rounded, with diameters ranging from fine (50-250 p.m) to coarse (250-700 txm), depending on the position within the sampled layer: the coarser pellets are at the bottom of the layer. Although the pellets are bright green under the optical microscope, they sometimes show yellow to green colour. The matrix, which is dark green under the optical microscope, binds the pellets and is very fine-grained. Commonly it contains a framework consisting of small, angular quartz fragments, calcite fragments and sporadic muscovite lamellae. The glauconiterich layers are inhomogeneous in both clay matrix/pellets ratio and mineralogical composition of the matrix; they show also a wide chemical variability (Table 7). Terra rossa. This is a clayey fine-grained material, bright red in colour, with quartz, muscovite lamellae and rare calcite fragments (Fig. 5c). XRPD analysis reveals an association of clay minerals consisting of chlorite + illite + kaolinite (Table 6). Hematite is also present•

43

NEOLITHIC POTTERY OF ABRI PENDIMOUN Table 6. Semi-quantitative contents of minerals in sherds and raw materials, based on X R P D data

Sample

Chl

M/I-G1

Kaol

Qz

P1

K-f

Cc*

Dol

Px

Hem

CaO

+

+

-

+

Subgroup 1G

AP6 AP1 AP8 AP9 AP15 AP21 AP22 AP25 AP29

++ ++ ++ + ++ ++ ++ ++ +

+ ++ ++ + + + + ++ +

++

XX X

++

++ +

++ + + + +

++ ++ ++ ++ X

+ -

X XX XX

++

++

++ + ++

++

-

+

X

++ X

-

X -

XXX XX

-

X

+

+

+

Subgroup 1GS

AP34 AP38

+ +

+ +

-

+

+ +

Subgroup 1 GCc

AP3 AP31 Subgroup 1GC

AP19 AP23 AP24 AP27 AP28

_

D

m

m

_

+

-

+

Subgroup 2S

AP33 AP42 AP45

++

+

+ +

Subgroup 3F

AP43 R a w materials

Carbonatic clay Marly clay Terra rossa

-

+ + -

Chl, Chlorite;M/I-G1,mica/illite-glauconite;Kaol,kaolinite;Qz, quartz; PI, plagioclase;K-f, K-feldspar;Dol,dolomite;Px, pyroxene; Hem, hematite;CaO, calciumoxide. Abundance:XXX, very high (>40%); XX, high (30-40%); X, moderate(15-30%); ++, low (5-15%); +, scarce (3-5%); -, rare (<3%). *Cc not detectedin sherdsbecauseof the HCI treatmentof samples.

The chemical composition of terra rossa has high A1203 and Fe203 contents and low values of MgO and CaO (Table 7). Discussion

and conclusions

The ceramics studied here belong to three mixture groups corresponding to different 'recipes'. The ternary diagram in Figure 6a shows a good match between the compositional field of clay matrix in subgroup 1G and glauconite-rich layers sampled in the vicinity of the rock shelter. This suggests that an unprocessed mixture was used. The hiatal texture of finds in this subgroup has led to an important technological consideration, because the hiatal texture does not always indicate the addition of

temper. In fact, in the outcrop the glauconiterich layers are characterized by the hiatal texture, with a gap between the dimension of pellets and the fine matrix (Fig. 5a). In the outcrop the inhomogeneity of the geological material explains the wide variation of CaO and A1203 contents in some glauconite-rich layers. The composition of the clay matrix of these subgroups is compatible with a mixing of glauconite-rich layers and terra rossa (Fig. 6 b e). The petrographic evidence (Figs 4 and 5) and the compositional similarities between the clay matrix of the artefacts and the raw materials (Fig. 6), available close to the site (marls and glauconite-rich layers) or nearby (terra rossa) suggest that the pottery was produced locally.

0.91±0.5) 0.51±0.1) 0.61±0.1) 0.4 (±0.3) 0.3 (±0,51

1.1 +11.3) 1.5 ±1.1) 1.3 ±0.3) 0.6 +_0.9) 7.7 _+5.8) 14.8 4- 1.9) 1.1 ±0.3) 3.1 __+0.1) 4.7 ±2.5) 1.0 +0.2) 37.8 ±7.21

3.9 (+0.7) 8.7 ( +_0.2) 2.8 ( + 0.6)

0.0 0.2 0.0

12.8 _+11.8) 19.9 +0.5) 8.2 ( + 2.0)

O.7 1.0 O.4 (4-11.11

64.3 ( ± 1,2) 59.0 ( 4- 1.9) 48.1 (+_5.4)

3 3 5

2.1 ±0.7)

(1.3 ±0.2)

29.3 (+5.3)

0.2 (+0.1)

4.0 ± 1.51

(1.2 (_+0.1)

54.7 ( + 2.8)

15

4.4 ±0.8)

0.3 +0.4)

24.3 (+3.11

0.2

8.6 ± 1.4)

0.2 ( + 0.4)

56.2 ( _ 2.7)

40

4.9 _+11.41

0.2 ±0. I)

26.3 ( ± 3.0)

0.2 ( + 0 . I)

8.3 + l.l)

0.2 ( ± 0 , 1 )

54.9 ( ± 3,5)

10

4.6 ( ± 0 . 5 )

(1.2 ±0.11

25.4 ( ± 3.0)

0.2 ( ___0. I )

8.5 + 1.2)

0.2 (_+0.1)

56.5 _+3.11

511

0. I

4.7 ( ± 0 . 3 )

±0.l)

0.I

8,6 +_ 1.0)

0.1 (+_0.11

58.3 + 1.6)

27

-I-0.11

24.0 ( + 1.5)

±0.1)

0.1

+0.1) _+0.11

7.8 ( + 11.8)

(4-4.4) + 1.1) + 5.9) +0.7) ±5.5)

0.1 0.4 0.1

0.8 ( ± 0 . 4 )

1.0 (+_0.3)

3.7 ( + 0 . 2 )

0. I

27.5 ( ± 1.6)

0.6 (+0.2)

1.3 (_+11.3)

29.2 _+4,0)

0.2 4-0.1)

23.8 10.2 15.0 7.0 20.7 9.2 16.1 7.5 9.4 20.4

0. I (+0.1)

( + 1.9) (+2.11 ( + 2.3) ( + 1.7) 4.6 (_+0.5)

0.2 +0. I)

-t- 1.6) -t-0.9) ± 2.2) _+4.7)

±0.1) +0.1) _+0,11 +0.1) +_0.1)

59.9 ± 1.7)

( + 11.2) ( + 0.2) (+2.11 ( + 0.5)

0.7 (+0.3) 1.0 (+0.7) 0.81+0.31 1.21_0.3) 0.6 (+0.11 0.3 0.7 14-11.1) 0.81_0.5) 0.8 (±0.4) 0.41±0.41

3.8 (+2.0) 2.3 (_+0.8) 4.9 (+2.2) 2.1 (±0.1) 4.01±0.1) 1.7 2.9 (+0.5) 2.61±11.9) I.I (_+0.3) 4,71_+3.91

4.0(+0.5) 1.5 (_+0.4) 3.8(_____1.0) 2.4(__+0.2) 4.1 ( + I . I ) 3.5 4.1 (±0.6) 1.8 ( 4- 0.6) 2.4(+0.4) 3.9(+__1.41

0.1 (+0.1) 0.1 (±0.1) 0.1 (__+0.1) 0. I ( + 0 . 2 ) 0.21+___0.11 0.1 0.31±0.1) 0.2 ( ± 0. I ) 0.1 (±0.1) 0.2(__+0.3)

0.2 0.0 0.2 0.1 0,2 0. I 0.3 0.2 0. I 0.0

10

( + 2.8) ( + 1.3) (+3.9) (4- I.I) (+2.4)

6.6 (4- 1.9)

8.2 20.1 13.1 21.8 10.4 15.2 14.5 21.7 24.7 9.1

0.2 ( _ 0 . 1 )

(_+0.1) (_+ 1.41 ( ± 0.2) ( + 0.2) (+_0. I)

Na20 (wt%)

CaO (wt%)

MgO (wt%)

Fe203t (wt%)

MnO (wt%)

Cr203 (wt%)

55,2 + 5.2)

0.2 1.0 0.4 0.5 0.3 1.7 0,5 0.6 1.2 0.4

A1203 (wt%)

33

_+ 1.2) +_3.71 ± 3.0) _+6.4)

± 4.6) 4- 3.2) +2.5) +2.0) ± 4.9)

TiO2 (wt%)

57.0 63.1 58.4 61.5 58.2 67.2 58,1 62.1 58.7 57.5

SiO2 (wt%)

30 10 22 4 50 1 10 30 12 25

Number of analyses*

*Number of analyses per group. +Calculated. SXRF data.

l G matrix IGS matrix 1GCc matrix IGCc clay pellets 1GC matrix 1GC clay pellets 1GSCc matrix 2S matrix 3F matrix Glauconitic layers matrix I G glauconitic pellets 1GS glauconitic pellets 1GCc glauconitic pellets 1GC glauconitic pellets 1GSCc glauconitic pellets Glauconitic layers pellets Glauconitic pellets of finds Carbonatic clay:I: Terra rossa~Marly clay:l:

Average compositions

0.5 ±O.4) (1.6 (+0.8) 11.11 0.2 ( ± 0. I ) 11.2 0. I

9.3 _+ 1.0) 6.4 + 1.01 2.2 +0.1) 3,0 _+0.1) 1.5 _+0.3)

+11.21

7.6 ± 1.8)

0.1

0,4 +11.3)

_+11.6) 6.6 ___I.I)

6.1

0.0

+O.3) ±0.4) ±0.1) ±_0.8)

±0,31 i 0.1 ) +0.4)

+0.5)

4.4 ( + 0 . 6 )

0.5 0.0 0.3 0.0 0.5 0.3 0.7 0.4 0. I 0.8

0.2 +0.2)

(±0.31 (___0.7) (___ 1.0) 14-2.1)

( + 2.2) ( _+0.5) ( + 0.6) ( ± 0.3) (_+ 1.2)

P20.~ (wt%)

7.4 (_+2.5)

5.0 2.7 4.7 4.5 4.5 2.5 5.0 3,2 3.3 6.6

K20 (wt%)

T a b l e 7. Average chemical contents determined by SEM-EDS fi)r each group of mixtures, and standard deviations for clay matrix of finds and raw materials'; bulk chemical data for raw materials are also reported

NEOLITHIC POTTERY OF ABRI PENDIMOUN (a)

Fe203 (100%)

1

CaO(100%) (c) 1G

Fe~O~(100%)

(b)

1G

AI203(100%) CaO (100%)

Fe203(100%) / ~

(d)

AI20~(100%) CaO (100%)

AI203(100%)

Fe203(100%)

Fe203(100%)

/ ~

2

CaO(100%) 1

AI203(100%) Fe203(100%)

CaO(100%) (e)IG

45

AI203(100%)

AI2Os(100%) CaO (100%)

2 %...,,s

3('

80

90

,/

4" "'" ",

5 l' "

10¢ _

110

12©

Fig. 6. FeO-CaO-A1203 diagram, showing not only the correlation between the clay matrix in finds and the local raw material compositions, but also the mixture between several clay materials. Differences in chemical composition of glauconite-rich layers are due to a natural chemical inhomogeneity. Clay matrix chemical compositions: 1, subgroup 1G; 2, subgroup 1GC; 3, subgroup 1GCc; 4, subgroup 1GS; 5, subgroup 1GSCc; 6, group l; 7, group 2; 8, group 3. Raw materials: 9, terra rossa; 10, carbonate clays; 11, marly clays; 12, glauconite-rich layers; 13, clay pellets.

These mixtures were sometimes tempered with carbonate rocks (subgroup 1GC), sparry calcite (subgroup IGCc) and/or aplite (subgroups 1GSCc and 1GS). Table 8 summarizes all the recipes used in the ceramic production at Abri Pendimoun during the Middle and Late Neolithic.

The association of glauconitic pellets and carbonate rocks in the pottery may be ascribed to a natural mixture when the carbonate temper is less than 20% (Table 8), because in the literature this value is considered a minimum technological value suitable to produce a good ceramic

46

E. B A S S O

ET AL.

Table 8. Synoptic table of the recipes used in the ceramic production of Abri Pendimoun with the percentage of temper Subgroup Glauconitic layers

Raw materiaL~ Limestones

Terra rossa

Calcite

Aplites

1G

869

5~20% • ,. . . . . . . . 1G C c

Number cd Number o f linds VBQ / f i n d s

'

-+'

.... +

|~.

!

. . . .

37

.%2tWo ,

"

14

<5%

114

23

,>20%

147

1


14

25-30*/,

36

.~.

1GS 1GSCc

'

. . . . . . . . . . . . . . . ~+.....

. . . .

•

~
.

2S 2Cc 3F The last two columns various subgroups.

,

-

,

19

. ,r~ report the numbers

o f all f i n d s c o n s i d e r e d

(Hoard et al. 1995). In subgroup IGC, the low content of carbonate rock fragments (around 5%) can be interpreted as natural mixture; in contrast, in a mixture with around 20 vol% of carbonate fragments, these fragments were intentionally added. The use of limestone and monocrystalline calcite as temper in ceramics has long been established (Arnold 1985; Porat 1989): pure crystalline calcite was used for cooking ware, whereas limestones were employed only in vessels not used for cooking (Arnold 1985). Studies on limestone and/or calcite ceramic mixture (Heller-Kallai et al. 1987; Shoval et al. 1993) indicate that monocrystalline calcite, stable at higher temperatures, is more suited to the production of cooking ware, whereas fragments of limestone were apparently employed in the production of vessels not used for cooking. Indeed, monocrystalline calcite produces pottery with lower mechanical resistance, which degrades rapidly when the firing temperature exceeds 700 °C in an oxidizing atmosphere. A semi-reducing atmosphere, however, increases the stability of calcite and of mixtures to 800 ~C (Peters & Iberg 1978; Letsch & Noll 1983; Fabbri et al. 1997). At Abri Pendimoun it is impossible to define the functionality of carbonate tempers because the artefacts are too fragmented to reconstruct vessel shapes and their usage. The presence of crushed sparry calcite, always less than 5% (subgroup 1GCc, Table 8), which at Abri Pendimoun seems to be strictly linked to the 'graffita' pottery of the Square Mouthed Pottery phase I Culture, may be mainly dictated by cultural rather than technological factors. This combination of mixture and shape indicates that there was an important link between the VBQ I and Chassey Cultures. A small percentage of ceramic finds (subgroups 1GS, 1GSCc and 2S) were produced by

2

9

,

in t h i s w o r k a n d the n u m b e r s

of VBQ

1 finds belonging

1 to the

adding crushed aplite to the local clay materials, although aplite was not available in the vicinity of the site. The presence of these fragments is difficult to explain, as their association with glauconitic pellets has not been found in raw materials. In Provence, igneous and metamorphic rocks crop out in the ancient Maures, Est&el and Tanneron massifs and in the Southern Alps (Ricq-de Bouard 1996). The Alpine Argentera massif essentially consists of granites, gneiss and silicate rocks such as aplite. However, the main sources of crystalline rocks in Provence are Tertiary and Quaternary sediments, which often contain pebbles and cobbles of igneous and metamorphic rocks. In eastern Provence, the sediments in the Nice region contain Argentera rock fragments of Pliocene conglomerates transported by the Var, Tin6e and Vesubie rivers (Ricq-de Bouard 1996). The temper undoubtedly consists of crushed aplites rather than natural sand, because of the angular shape of rock and mineral fragments (Fig. 4b and e). Also the good preservation of alterable minerals (feldspars), which disappear in detrital formations far from source rocks, corroborates this hypothesis. In the mixture 1GS and 1GSCc, the low quantity of the temper does not represent a functional tempering (Tire et al. 2001). Moreover, we could infer that the addition must have been intentional because of the non-local origin of the material. At Abri Pendimoun, the addition of crushed aplite is < 2 0 vol% for glauconitic pellet-bearing mixtures, whereas it is >20 vol% for the terra rossa mixtures (Table 8).The scanty knowledge of the function of the pots prevents speculation about a technological or cultural reason for the tempering. Probably the pots' shape and their use constrain both mineralogical composition and temper/matrix ratio of the mixtures.

NEOLITHIC POTTERY OF ABRI PENDIMOUN The use of glauconitic material in 95% of the pottery (Fig. 3a) indicates that raw materials and the production technology remained constant through the Middle and Late Neolithic. Current literature data (Nung/isser & Maggetti 1978; Nung/isser et al. 1985, 1992; Maggetti 1994; Martineau et al. 2000; Di Pierro 2003) highlight that the temper was carefully selected from outcrops even quite distant from the archaeological site. In particular, in Prehistoric times it was common practice to add granitic rocks to ceramic mixtures. These lithologies are easily crushed when heated and rapidly quenched in water (Nung/isser et al. 1992), as thermal shock causes 'self-crushing'. Another aspect regarding the functionality of mixtures is linked to the use of temper mixtures with white rocks, either silicates or carbonates. The technological aspects linked to the use of carbonate rocks are well documented in literature, whereas issues linked to the addition of silicate rocks have been less tackled. Experimental data presented by Kilikoglou et al. (1995, 1998) demonstrate that concentrations of quartz temper > 20 vol% result in a decreasing fracture strength. Did the addition of granites or aplites actually fulfil a technological requirement (for instance, for quartz) or was it simply dictated by tradition? This remains an open issue. We would like to thank the Earth Science Department (University of Siena) for the SEM analyses. We also wish to thank J.-C. l~challier, who has provided his geological survey data. A special acknowledgement is due to F. Crepaldi, who deals with archaeological aspects of VBQ I-Chassey interactions. The authors would like to express their gratitude to the anonymous referees, who improved the first draft of this paper.

References ARNOLD, D. 1985. Ceramic Theory and Cultural Process. Cambridge University Press, Cambridge. ARNOLD, D. 2001. Linking society with the compositional analyses of pottery: a model from comparative ethnography. In: LIVINGSTONE SMITH, A., BOSQUET,D. & MART1NEAU,R. (eds) Pottery Manufacturing Process: Reconstitution and Interpretation. British Archaeological Reports International Series. Archaeopress, Oxford, 1349, 1-12. BAGOLINI, B. & PEDROTTI,A. 1998. L'Italie septentrionale. Atlas du Ndolithique europgen. L'Europe occidentale, 2A, Etudes et Recherches Archdologiques de l'Universit~ de Li6ge, 46, 233-341. BARRAL, L. 1958. L'homme cardial de Castellar. Bulletin du Musde d'Anthropologie Prdhistorique de Monaco, 5, 135-164. BASSO, E. 2004. lndagini mineralogico-petrografiche applicate allo studio di reperti archeologici di eta neolitica del piacentino e del sito archeologico di Abri Pendimoun (Alpi Marittime--Francia). PhD thesis, Universit~i degli Studi di Pavia.

47

BERNAB() BREA, L. 1946. Gli scavi nella Caverna delle Arene Candide, 1. The Istituto di Studi Liguri, Bordighera. BERNABO BREA, L. 1956. Gli scavi nella Caverna delle Arene Candide, 2. Istituto di Studi Liguri, Bordighera. BINDER, D. 1990. N~olithique moyen et sup~rieur dans l'aire liguro-provenqale: le cas de Giribaldi (Nice, Alpes-Maritimes, France). In: GUILAINE,J. & GUTHERZ, X. (eds) Premibres communautds paysannes. Autour de Jean Areal, Montpellier, 147-161. BINDER, D. 1998. Silex blond et complexit6 des assemblages lithiques dans le N6olithique liguroprovenqale. Rencontres mdridionales de Prdhistoire rdcente, Juan-les-Pins. Association pour la Promotion et la Diffusion des Connaissances Archaeologiques, Juan-les-Pins. Aries, 111-128. BINDER, D. 2000. Mesolithic and Neolithic interaction in southern France and northern Italy: new data and current hypotheses. In: PRICE, T. D. (ed.) Europe's First Farmers. Cambridge University Press, Cambridge, 117-143. BINDER, D. 2002. Castellar--Abri Pendimoun (Alpes Maritimes). Deuxibme rapport interm6diaire, Campagne de 2001. Centre d'~tudes Pr~histoire, Antiquitd Moyen Age (CEPAM-CNRS). BINDER, D. & PERLI~S,C. 1990. Strat6gies de gestion des outillages lithiques au N6olithique. Paldo, 2, 257-283. BINDER, D., BROCHIER,J.-E., DUDAY,H., HELMER,D., MARINVAL,P., THII~BAULT,S. & WATTEZ,J. 1993. L'Abri Pendimoun ~ Castellar: nouvelles donn~es sur le complexe culturel de la c6ramique imprim6e m6diterran6enne dans son contexte stratigraphique. Gallia Prdhistoire, 35, 177- 251. BINDER, D., GASSIN, B. & St~NI~PART, I. 1994. l~16ments pour la caract~risation des productions c~ramiques n6olithiques dans le Sud de la France. L'exemple de Giribaldi. Terre Cuite et Socidtd. La Ciramique, Document Technique, Economique, Culturel. Association pour la Promotion et la Diffusion des Connaissances Archaeologiques, Juan-les-Pins. Aries, 255-267. BIRIS, K. T. 1998. Stones, numbers---history? The utilization of lithic raw materials in the Middle and Late Neolithic. Journal of Anthropological Archaeology, 17(1), 1-18. CONSTANTIN, C. & COURTOIS, L. 1985. Le mat6riau c6ramique comme caract~ristique culturelle. L'exemple du d6graissant pendant le N~olithique dans le Bassin Parisien. Etudes des Cdramiques en Archdologie. Documents et travaux Institut G~ologique Albert-de-Lapparont, 9, 19- 26. COURTIN, J. 1976. La baume Fontbr~goua (Salernes, Var). 9e Congrds de l'Union lnternationale des Sciences Prdhistoriques et Protohistoriques, 21-29. COURTOIS, L. 1973. Ph6nom~nes de r6g6n~ration aprbs cuisson de certaines cdramiques anciennes. Comptes Rendus de e'Academic des Sciences, Serie D, 276, 2931-2933. COURTOIS, L. 1976. Examen au microscope pgtrographique des cdramiques archdologiques. Centre de recherches arch6ologiques, Valbonne.

48

E. BASSO ET AL.

CREPALDI, F. 2001. Le Chassfen en Ligurie. Bulletin de la Socidtg Pr~historique Franqaise, 98(3), 485 -494. CREPALDI, F. 2002. Tecnologia e tipologia degli aspetti di tradizione chasseana in Italia settentrionale. In: FERRARI, A. & VISENTINI, P. (eds) II declino del mondo neolitico. Ricerche in Italia centrosettentrionale fra aspetti peninsulari, occidentali e nord-alpini. Quaderni del Museo Archeologico del Friuli Occidentale, Pordenone, 4, 157-166. CRISCI, G. M., RICQ-DEBOUARD,M., LANZAFRAME,U. & DE FRANCESCO, A. M. 1994. Les obsidiennes du Midi de la France. Nouvelle m&hode d'analyse et provenance de l'ensemble des obsidiennes n6olithiques du Midi de la France. Gallia Prghistoire, XXXVI, 299-309. D'AMICO, C. 2005. Neolithic 'greenstone' axe blades from North-western Italy across Europe: a first petrographic comparison. A rchaeometo', 47(2), 235- 252. D'AMIcO, C. & STARNINI, E. 2005. Prehistoric polished stone artefacts in Italy: a petrographic and archaeological assessment. In: MAGGETTI, M. & MESSIGA, B. (eds) Geomaterials in Cultural Heritage. Geological Society, London, Special Publications, 257, 257-272. DICKINSON, W. R. & SHUTLER, R. J. 2000. Implications of petrographic temper analysis for Oceanian prehistory. Journal of World Prehistoo', 14(3), 203-266. DI PIERRO, S. 2003. Ceramic production technology and provenance during the Final Neolithic: the Portalban settlement, Neuchatel lake, Switzerland. Revue d'Archdomdtrie, 27, 75-93. FABBRI, B., GUALTIERI, S. & SANTORO, S. 1997. L'alternativa chamotte/calcite nella ceramica grezza: prove tecniche, la Giornata di Archeometria della Ceramica, Bologna. University Press, Bologna, Imola, 183-190. GEZE, B. 1968. Carte g~ologiques de France, feuille Menton-Nice, 973. Bureau de Recherches G6ologiques et Mini~res, Orleans. HEIMANN, R. B. & MAGGETTI, M. 1981. Experiments on Simulated Burial of Calcareous Terra Sigillata (Mineralogical Change). Preliminao, Results. British Museum Occasional Papers, 19, 163-177. HELLER-KALLAI, L., MILOSLAWSK1, I. & AIZENSHTAT, Z. 1987. Volatile products of clay mineral pyrolysis revealed by their effect on calcite. Clay Minerals, 22, 339-348. HOARD, R. J., O'BRIEN, M. J., GHAZAVYKHORASGANY, M. & GOPALARATNAM, V. S. 1995. A materialsscience approach to understanding limestone-tempered pottery from the Midwestem United States. Journal of Archaeological Science, 22, 823-832. KILIKOGLOU, V., VEKINIS, G. & MANIATIS, Y. 1995. Toughening of ceramic earthenwares by quartz inclusions: an ancient art revisited. Acta Metallurgica et Materialia, 43, 2959-2965. KILIKOGLOU, V., VEKINIS, G., MANIATIS, Y. & DAY, P. M. 1998. Mechanical performance of quartz-tempered ceramics. Part I: strength and toughness. Archaeomet©', 40, 261-279. LETSCH, J. & NOLL, W. 1983. Phase formation in several ceramic subsystems at 600 ':C-1000 °C as a function of oxygen fugacity. Ceramic Forum-

International Berichte der Deutschen Keramischen Gesellschaft, 60(7), 259- 267. MAGGETTI, M. 1994. Mineralogical and petrographic methods for the study of ancient pottery. In: 1st European Workshop on Archaeological Ceramics, Universith degli Studi di Roma La Sapienza, Rome, 23-35. MARTINEAU, R. 2000. Poterie, techniques et socidtds. Etudes analytiques et expdrimentales b Chalain et b Clain,aux (Jura), entre 3200 et 2900 av. J.-C. PhD thesis, Universit~ de Franche-Comt6, UFR des Sciences de 1' Homme, du Langage et de la Soci&& MARTINEAU, R., CONVERTINI, F. & BOULLIER, A. 2000. Provenances et exploitations des terres poterie des sites de chalain (Jura), aux 31e et 30e si~cles avant J.-C. Bulletin de la Socidtg Pr~historique Franqaise, 97(1), 57-71. NUNGASSER, W. & MAGGETTI, M. 1978. Mineralogisch-petrographische Untersuchung der neolithischen T6pferware von Burg~ischisee. Bulletin de la Socidtd Fribourgeoise des Sciences naturelles, 67(2), 152-173. NUNG,~.SSER, W., MAGGETTI, M. & STOCKLI, W. E. 1985. Neolitische Keramik von Twann--Mineralogische und Petrographische Untersuchungen. Jahrbuch der Schweizerischen Gesellschaft fiir Urund und Friihgeschichte, 68, 7-39. NL'NGASSER, W., MAGGETTI, M. & GALETTI, G. 1992. Analyse der Scherbensubstanz mit Mikroskop und R6ntgenlicht. /n: BILL, J., NUNGASSER, W. & GALETTI, G. (eds) Liechtensteinische Keramikfunde der Eisenzeit. Jahrbuch der Historischen Vereins ftir das Ffirstentum Liechtenstein, 91,

119-165. ODETTI, G. 1991. I1 Neolitico medio ligure e le influenze chasseane, ldentitg du Chassden, Actes du Colloque btternational de Nemours 1989, M6moires du Musge de Pr6histoire d'Ile-de-France, 4, 59-68. PETERS, T. J. & IBERG, R. 1978. Mineralogical changes during firing of Ca-rich brick clays. American Ceramic Society Bulletin, 57, 503-506. PORAT, N. 1989. Petrography of pottery from Southern Israel and Sinai. In: MIROSCHEDJI (ed.) L'urbanisation de la Palestine b l'age du Bronze Ancien. British Archaeological Reports. International series, Oxford, 527, 169-188. RICQ-DE BOUARD, M. 1996. Pdtrographie et socidtds ndolithiques en France mdditerrandenne. Monographie du CRA, 16, CNRS. SHOVAL, S., GAFT, M., BECK, P. & KIRSH, Y. 1993. Thermal behaviour of limestone and monocrystalline calcite tempers during firing and their use in ancient vessels. Journal of Thermal Analysis, 40, 263-273. STARNINI, E. & VOYTEK, B. 1997. The Neolithic chipped stone artefacts from the Bernabb BreaCardini excavations. Arene Candide: a functional and environmental assessment of the Holocene sequence, Memorie dell'lstituto Italiano di Paletnologia Italiana, 5, 349-426. TITE, M. S., KILIKOGLOU, V. & VEKINIS, G. 2001. Strength, toughness and thermal shock resistance of ancient ceramics, and their influence on technological choice. Archaeometo', 43(3), 301-324.

Provenance and technology of Apulian Neolithic pottery R O C C O L A V I A N O I & I T A L O M. M U N T O N I 2'3

1Dipartimento Geomineralogico, University of Bari, Via E. Orabona 4, 70125 Bari, Italy (e-mail: rocco.laviano @geomin, uniba, it) 2Museo delle Origini, Universith degli Studi di Roma 'La Sapienza' University, Piazzale Aldo Moro 5, 00185, Rome, Italy 3Department of Archaeometry, Science Faculty, University of Bari, Via E. Orabona 4, 70125 Bari, Italy Abstract: Apulia is the best-represented region in Italy as far as archaeometric analyses of Neolithic pottery are concerned. Cross-checked use of petrological (optical microscopy), mineralogical (X-ray powder diffraction) and chemical analyses (X-ray fluorescence) have been performed, in the Dipartimento Geomineralogico of Bari University, on 375 Early to Late Neolithic (from the seventh to the fourth millennium Bc) pottery samples from the Tavoliere and Murge areas. A correlated analysis of 134 samples of the main clayey deposits of the two areas was also conducted. Generally local clays were used and, in some cases, the exploitation of a range of different local fabrics has been verified. In Middle Neolithic sites, the use of non-local clay, probably imported, has been also determined. Few finished pots were actually exchanged at an inter-site scale during the Neolithic. Preparation of raw materials has shown different choices followed by ancient potters. Clays are usually more or less refined and the use of mineral temper such as sand, quartz, calcite and grog has been found. The maximum temperature reached during firing is usually between 600-700 and 850 °C. For some Middle Neolithic fine painted pottery higher temperatures have been suggested (between 850 and 1050 °C), revealing a better firing control and the use of kilns.

Archaeometric analysis of Italian Neolithic pottery, developed only in the last 20 years, has assumed a particular importance in relation to innovations in pottery production and the emergence of productive economies and structured settlements from the end of the seventh millennium Bc. On a regional scale Apulia is the best represented, in terms of both sampled sites (about 45 settlements) and analysed fragments (more than 600). The chronological range and the archaeological facies of the Neolithic period, from Early to Late, have been entirely covered. Many Italian (from Milan, Florence, Genoa and Bari University) and foreign teams (from the UK and Canada) have analysed Apulian Neolithic pottery (Muntoni 2002a). Within the framework of Matson's concept of Ceramic Ecology (Matson 1965; Kolb 1989), the emphasis here will be on research strategies, the relation between goals and methods, and on the sampling techniques of such studies. The ideal development should be the shift from pottery as a simply finished product, whose origin is to be understood, towards an ecological

and sociocultural frame of reference in which physical and chemical data are placed. In light of awareness of both the potentialities and limits of analytical techniques, the status of technological and archaeometric studies of Apulian Neolithic pottery will be briefly reviewed (Muntoni 2002b). The number of analysed samples per site is normally small: the number of fragments varies between one and 49 sherds per site, sampled from surveyed or excavated materials. Of the various archaeometric analytical methods, the most widespread technique is petrological examination on thin-sections optical microscopy (OM), often employed alone or in conjunction with other mineralogical methods, such as X-ray powder diffraction (PXRD) and/or thermal analyses (thermogravimetric analysis (TGA); differential thermal analysis (DTA)). Chemical instrumental neutron activation analysis (INAA), inductively coupled plasma-atomic emission spectrometry (ICP-AES) or morpho-chemical (scanning electron microscopy plus energydispersive spectrometry (SEM-EDS)) analyses

From: MAGGETTI,M. & MESSIGA,B. (eds) 2006. Geomaterialsin CulturalHeritage. Geological Society, London, Special Publications, 257, 49~52. 0305-8719/06/$15.00 © The Geological Society of London 2006.

50

R. LAVIANO & I. M. MUNTONI

are even now rarely used in Apulian Neolithic pottery studies,

arrangements involved in the production, use and distribution of ceramics.

Our approach to archaeometric analyses

Analytical methods

Since 1990 a new approach to the archaeometric analysis of Apulian Neolithic pottery has been applied in the Dipartimento Geomineralogico of Bail University, in collaboration with different archaeological teams from the 'La Sapienza' University, Rome, and the Soprintendenza per i Beni Archeologici della Puglia and, more recently, from the Dipartimento di Beni Culturali e Scienze del Linguaggio of Bari University. A crucial problem in the study of prehistoric ceramics is the fabric variability. The use of appropriate sampling strategies, number of samples and archaeometric techniques is essential if meaningful technological studies are to be done. To verify adequately the variability and heterogeneity within and between pastes, archaeologists had to construct rigorous and explicit sampling schemes that could encompass the full range of variability. Cross-checked use of mineralogical and chemical analyses was then employed: complete combinations of petrological (OM), mineralogical (PXRD) and chemical (X-ray fluorescence (XRF)) analyses have been performed on 375 archaeological samples, all analysed in the Dipartimento Geomineralogico of Bail University (Table 1). Twenty-three settlements, located in the Tavoliere plain and in the Murge plateau along the Adriatic coast, were sampled (Fig. lb). A correlated analysis of 134 pelitic and clayey samples was also conducted (Laviano & Muntoni 2004). Only the analysis of clay sources can give more information on which clay source was used, and why, and what kind of problems were encountered by the ancient potters in the modification of the materials. 'Any investigation of ceramic technology at an archaeological site or region is most fruitfully conducted within the perspective of ceramic ecology: it is essential to search out deposits and sample them so as to compare the pottery of interest with the properties of local clays' (Bishop et al. 1982, p. 319). As all the archaeometrical data are separately published or are still in press, in this paper we aim for the first time at a geographical and chronological synthesis of our results. The reconstruction of the working sequence used in pottery manufacturing (from raw material provenance, to preparation of bodies and firing techniques) should help to obtain insights into the potter's role as an active and controlling agent in the procedure of a specific pottery manufacture, and into the pottery economics or the socio-economic

Mineralogical studies were carried out by PXRD using a Philips diffractometer (PW 1710) with Ni-filtered Cu K,~ radiation and employing NaF as internal standard. Petrological observation was made on thin sections, with a polarized light microscope (OM). Modal analysis was carried out using a Swift & S. Point Counter on 2500-4500 points for each sample (according to their wall thickness), with a line distance of 0.05 mm and a lateral step of 0.2 mm. Major and trace element determination was performed by XRF, using a Philips PW 1480/10 spectrometer (Cr anticathode for major and minor elements, Rh anticathode for Rb, Sr, Y, Zr, Nb and W anticathode for Ce, La, Ba, Ni, Cr, V), following the analytical techniques outlined by Franzini et al. (1972, 1975) and Leoni & Saitta (1976). About 5 g of representative powder for each sample was subjected to XRF. Two reference standards (AGV-1 of the USGS, USA and NIM-G of NIM, South Africa) were used to check the accuracy of the analytical data. Loss on ignition was determined by heating the samples at 1000 °C for 12 h; then PXRD patterns of the same previously heated samples, for the identification of mineralogical changes, were recorded at room temperature.

The geological context The Tavoliere plain

The Tavoliere plain (Fig. 1a), the most extensive one in Southern Italy, is a Mesozoic-Palaeogene limestone depression filled with marine deposits of Plio-Pleistocene silty clay (Bradanic cycle), often overlain by post-Calabrian marine sands (terraced marine deposits), Upper Pleistocene (terraced alluvial deposits) and Holocene alluvial and lacustrine deposits of continental origin. Marine Plio-Pleistocene clays of the Bradanic cycle, also named the Argille Subappennine, crop out along the western margin of the Tavoliere plain. The depth of the outcrops may vary from a few metres to 350 m. The clays consist of silty clay or clayey silt, with little sand, and have (Balenzano et al. 1977; Dondi et al. 1992) a very similar mineralogical composition (clay minerals, carbonates, quartz and feldspars). The clay minerals are a mixture of 2 M illite, Mgbearing smectite, Fe-bearing chlorite, kaolinite and randomly interstratified illite-smectite with 30-70% montmorillonite-like layers. Natural

51

APULIAN POTTERY

Table 1. Sampled Neolithic settlements of Tavoliere and Murge areas Code

Samples Excavation

Tavoliere Monte Aquilone Masseria Valente Coppa Nevigata Masseria Candelaro Masseria Santa Tecchia Podere 96 Masseria Cascavilla Capo di Lupo Masseria Centonze Masseria Mischitelli Casello Amendola Total

MA MV CN MC ST P96 MCS CL MCZ MM CA

15 13 8 61 11 7 6 5 3 3 2 134

X X X X X

Murge Balsignano Pulo di Molfetta Ciccotto Madonna delle Grazie Torre delle Monache Santa Barbara Setteponti Cala Colombo Grotta della Tartaruga Grotta Scanzano Masseria Chiancudda Cala Scizzo Total

BALS PU CC MG TM SB SP CCL GT GS CH CS

30 59 33 6 6 43 15 10 10 9 1 19 241

X X X

non-plastic material consists of carbonates (calcite, as bioclastic or detrital granules, and dolomite), quartz and feldspars (orthoclase, microcline and Na-plagioclase). Very different Holocene alluvial clays can be found on the coastal plain, deposited by the numerous ancient rivers and streams. The clay composition is very variable depending on the erosion of different clayey and arenaceousmarly deposits (Cassano et al. 1995b; Eramo et al. 2004). Alluvial clays have volcanic minerals and rock fragments as a distinctive feature. Heavy minerals are represented by dominant diopside-augite pyroxene, magnetite, biotite and garnet, together with debris of volcanic glass. Marine and alluvial clays are characterized by the relative abundance of SiO2, A1203, CaO, Fe203, K20 and MgO. Because of their mainly calcareous composition (up to 17 wt% CaO), both groups can be classified as marly clays. In general, clay fractions ( < 2 Ixm) have a lower CaO content than the whole specimens, whereas the A1203 and Fe203 concentration is higher.

Survey

Early Neolithic

Middle Neolithic

X X X X X X X X X X

X X X X X X

X X

X X X X X

Late Neolithic

X X

X X X X X X X

The Murge plateau

The large geologically homogeneous Murge plateau (Fig. 1a) is formed by the limestone formations of Calcare di Bari and Calcare di Altamura, with terra rossa deposits present in the sequence. Terra rossa are silty-clayey continental sedimentary deposits, very poor in carbonate (Dell'Anna 1967; Dell'Anna & Garavelli 1968; Dell'Anna et al. 1973) composed of dominant clay minerals (illite and kaolinite) and Fe-oxides or hydroxides, with subordinate quantities of quartz, feldspars, micas and pyroxenes. SiO2, A1203 and Fe203 are the main oxides, both in the clay fraction and in the whole specimen. Marine Plio-Pleistocene silty clays of the Bradanic cycle (Argille Subappennine) crop out extensively along the western margin of the Murge plateau (Dell'Anna & Laviano 1991) and locally in the Rutigliano area (Dell'Anna 1969; Moresi 1990).

Raw material provenance Our studies indicate that generally local clays were used for Neolithic pottery production, in

52

R. LAVIANO & I. M. MUNTONI (a)

o. . . . . . . . . .

50 km i~,,. N

-%

:----" ( :i4~-~.~ -.

Adriatic

"\

[~

Alluvial and lacustrine deposits (Holocene) Terraced alluvial deposits (Upper Pleistocene)

[]

Terraced marine deposits (Upper-Middle Pleistocene)

[]

Bradano Units (Pliocene-Lower Pleistocene) Calcarenites (Miocene)

I~ [~

Calcarenites (Eocene) Limestone (Mesozoic)

-

~,

.

[]

[]

sea

,

.

Apenninic Chain Units (from Cretaceous to Pliocene)

~;:-: :

Ionian sea

(b)

50 km v o l i e r e

\--"-'--~r-,~. h! D , 3CCL

Fig. 1. (a) Geological and geomorphological map of Apulia (from Caldara et al. 1990, fig. 5). (b) Location of sampled Neolithic settlements (*) of Tavoliere and Murge areas.

APULIAN POTTERY

53

both the Tavoliere and Murge areas, with some differences between the two areas.

Early Neolithic (6200-5500 BC) of the Tavotiere plain Plio-Pleistocene silty clay and Holocene alluvial deposits of continental origin were exploited in the Tavoliere plain (Cassano et al. 2004). Archaeometric data from eight neighbouring Early Neolithic villages, located in a small area (150 km 2) of the plain near the Adriatic coast, show that ceramic production seems fairly homogeneous (raw material supply, grain-size variability and firing techniques) in these villages, although some degree of variation in grain-size composition of natural non-plastic material is also present between different wares (Cassano et al. 1995b; Muntoni & Laviano 2005). As far as the ecological and technological aspects are concerned, the same alluvial deposits, as shown by the well-preserved mineralogical component, from three river valleys were exploited. This hypothesis is also sustained by the presence in the pottery of clasts of pyroxenes and volcanic glass typical of alluvial clays. Behavioural and technical similarities in Early Neolithic pottery technology are thus confirmed. The concentration of chemical elements is very useful to identify a strong affinity between Early Neolithic samples that can be all defined as 'Ca-rich'. A very few finished pots were considered as outliers (Fig. 2a), probably exchanged at an inter-site scale during Early Neolithic. The ternary diagram (Fig. 3a) also shows a good chemical correspondence between pottery samples and alluvial clay deposits from the Tavoliere plain. The difference of CaO content between some pottery samples and alluvial clays (Fig. 3b) is due to the compositional variability of alluvial deposits and to the addition of calcareous sand in some samples of coarse ware, mainly from Monte Aquilone village. Three principal criteria have been used to identify temper: (1) the hiatal distribution of the inclusions; (2) a proportion of inclusions >30%; (3) the relative abundance of micritic bioclasts (mainly bivalves), limestone clasts and fine-grained calcite. XRF concentrations have also shown very fine local distinctions (Fig. 2b) between sites located in different alluvial basins (Muntoni & Laviano 2005). Therefore, villages collected their clays in areas around the settlement itself in the nearest alluvial basin. Thus groups that had the same taste and behavioural choices

300

IOO08

~ Sr

300

400 2 0 0 ~ 0 0 ~ / / " lO0

Zr

(b)

1000

4

o

o

F

~

,oo

1000~ , ~ J

.

~ .

.

.

~

~

..... .I

~

300

Fig. 2. (a) Crv. Sr v. Zr plot (ppm) for all Early Neolithic pottery samples (+, n ----73) from the Tavoliere area (note the cluster and the few samples (labelled) that could be considered as outliers). (b), Ba v. Sr v. Zr plot (ppm) for all Early Neolithic pottery samples from the Tavoliere area and the samples (marked) from sites located in two alluvial basins (D, samples from CN and MV villages; V, samples from CL, MM and MCS villages).

were completely autonomous as far as raw material supply and pottery manufacture are concerned.

Early Neolithic (6200-5500 BC) of the Murge plateau The exploitation of different local deposits in the same Early Neolithic site has been verified with an extended range of analysed samples, as in the large karst doline of Molfetta (locally known as 'pulo') or in the village of Balsignano, both located near the Adriatic coast of the Murge plateau (Muntoni 2003).

R. LAVIANO & I. M. MUNTONI

54

(a)

Si02 ._ ;,,,

, ,,

kj

/'

\

. -

-~/ C a - r i c h "

./

\

~.~

-.// "-\ /.' , ~" /.

CaO+MgO!

~"'Gh

\

.

\,

,./.,~ C a - p o o r ,,,.

'\

" ,.

!, An

.

, ]",

',

~',.

..........

[ ~ ~-

,

,,, •

,

'

'

0

(b)

~'~

',

25

5O

75

AI203 100

60.00 + +

5000

4-

t,++t -,*+ ~

+

40.0O

+

.~

30.00

+

,4+

2000

10.00

0.00 0.00

500

10.00

15.00

20.00

25 00

30.00

CaO

Fig. 3. (a) (CaO + MgO)-A1203-SiO2 wt% diagram (Di, diopside; Gh, gehlenite; An, anorthite); (b) S i O 2 v . CaO plot (wt%). Early Neolithic pottery samples (+, n = 73) from the Tavoliere area and the fields (A, whole sample; B, clay fraction) of alluvial deposits (n = 7).

Three paste groups with different dominant mineralogical constituents (predominant quartz and calcite, with little K-feldspar and plagioclase) were found in the Early Neolithic strata of Pulo di Molfetta (Laviano & Muntoni 2003). Raw material variability is evident mainly on a synchronous level: three pastes (Cal fabric, Qtz + Cal fabric and Qtz fabric) were used in the same village and in the same archaeological horizon. The dominant clastic constituents are all compatible with the geological deposits that crop out in the site hinterland. The differences in mineralogy and grain-size distribution (detected by PXRD and OM) between Qtz + Cal and Qtz fabrics could be explained if we hypothesize that Neolithic potters collected different clays in more areas around the settlement itself. The large amount of carbonate

fossil fragments (mainly molluscs and rare benthonic Foraminifera), which characterize the Cal-rich fabric, is due to an intentional addition of a calcareous temper. In this case two principal criteria have been used to identify temper: (1) a proportion of inclusions >_30%; (2) the exclusive presence of carbonate fossil fragments (from 2 0 0 - 4 0 0 p,m to 1.5-2 mm). XRF analyses are consistent with mineralogical data (obtained by PXRD and OM). SiO2, CaO and AI203 are the dominant oxides, with some variations in CaO percentage. The ternary diagram show a clear differentiation of the Cal-rich samples, the Qtz + Cal-rich samples being characterized by relatively more balanced amounts of the three main oxides, whereas the Qtz-rich samples are characterized by higher quantities of Qtz (Fig. 4a). This last group shows a good overlap

APULIAN POTI'ERY

55

(a)

\

Di

\

x,\\\

,/ ×\

~

"

"7

\

/

./

,.~ "-/ ~'~ .... /

'

'

0

(b)

/ /

. /

,

......... ~ \

\

'"'

,

"X

~

............ ::~ \

\

/'"" \

/

\ ,

'

\

•

G

\

\\\X \ \

-./

CaO+MgO(

V

//

!

\

,

\ ,

1 ' ' ' '

r

25

50

'

X" \

// /'

'\,

XX

,

xe-

\ ,

,

,,

\

I

'

/

o

''

75

/AI203 100

50 A

~ ........ t

-

t

roe

30 Nb

25

20

T

!

T-

15

!

4 I

J

i

;

q

5

o

[a r'l

'

5

10

15

', -

+-

~

"

' " ai -

IArgitle diRutigliano ~............. . . . . . . . . . . . .

L 0

a

I Z3.-=,.-.-~-,~_ --

20

i

'~

25

30

I

t

!

, 35

40

45

50

Y

Fig. 4. (a) (CaO + MgO)-A1203-SiO2 wt% diagram (Di, diopside; Gh, gehlenite; An, anorthite); (b) Nb v. Y plot (ppm). Early Neolithic Cal-rich (×), Qtz + Cal-rich (El) and Qtz-rich (A) pottery samples (n = 47) from Pulo di Molfetta and the fields (whole sample) of Argille di Rutigliano (n = 5) and terra rossa (n = 25).

with the chemical composition of the terra rossa and the Qtz + Ca1 group with the Argille di Rutigliano of the Murge plateau. Also, some trace elements, such as niobium (Nb) and yttrium (Y), are important for distinguishing the three groups of pottery (Fig. 4b). In a few cases, different paste types can be actually related to typological groups of vessel form or archaeological classes, as in the Early Neolithic village of Balsignano (Modugno). At this site (Muntoni 2003) all Neolithic fragments were macroscopically analysed and attributed to the paste groups. Only four basic vessel forms were distinguished in the analysed materials: dishesplates, bowls, large jars and collared jars. The occurrence of bowls and large jars in pastes is different (Muntoni 2003, table 60). Bowls are more concentrated (77%) in Q t z + Cal-rich fabrics, with few carbonate clasts: the high

quantity of quartz inclusions was probably reserved for pots generally used for serving or display. Large jars are more concentrated (60%) in Cal-rich fabrics, with carbonate rock and fossils: the higher proportion of carbonate inclusions, mainly in coarse wares, could suggest their use as cooking and/or storage vessels. In low-fired cooking vessels, which are heated and cooled during use, because the thermal expansion of calcite is similar to that of average fired clays (Bronitsky & Hamer 1986; Fabbri et al. 1997), stresses owing to differential expansion of the clay matrix and temper are usually minimal.

Middle Neolithic (5500-4400 or 4200 Bc) of Tavoliere and Murge Middle Neolithic pottery shows a substantial shift in the whole production sequence; in

56

R. LAVIANO & I. M. MUNTONI

particular, the systematic exploitation of Marine Plio-Pleistocene silty clays, even in sites where they could be considered non-local materials, has been determined in both the Tavoliere (Cassano et al. 1995b; Muntoni 1999) and the Murge areas (Muntoni 2003; Muntoni et al. 2006). All red and/or brown painted pots (figulina), which are typical of the Middle Neolithic of Southern Italy, show a very fine paste texture, with a fairly fat clay matrix: of the sheet silicates only mica crystals are recognizable. Non-plastic inclusions are homogeneous fine-grained (< 150-200 ~m) quartz and scarce carbonate fossils (mainly benthonic Foraminifera). Clasts of feldspars and iron oxides or hydroxides are also present. XRF analyses showed that all samples are Ca-rich. The ceramic diagram (Fig. 5) shows an almost complete overlap of the samples in the central part, corresponding to the chemical composition of Plio-Pleistocene Apulian silty clays. A few samples from the village of Masseria Candelaro in the Tavoliere area are characterized by very high quantities of calcite, whereas others from the northern Murge sites of Pulo di Molfetta and Grotta Scanzano have lower quantities. The former shows a more Cal-rich fabric with carbonate fossil fragments (mainly molluscs, such as bivalves and gastropods), whereas the latter has a Qtz-rich fabric with very few carbonate inclusions. Also, trace

element concentrations confirm the homogeneity between samples. This finding is in agreement with the geochemical homogeneity of the PlioPleistocene Apulian clays. Nevertheless, some trace elements, such as Ba, Sr and Zr (Fig. 6), are important for distinguishing sub-groups of pottery related to their geographical setting. Mineralogical and chemical data clearly show in the two areas the exploitation of the PlioPleistocene silty clay, which in some cases crops out more than 30 km from the sites. The use of specific clay-beds shows a more complex clay supply activity, involving perhaps the whole group of people. Such an activity might be distinct from individual and domestic tasks, and may suggest that local production was no longer domestic. In Middle Neolithic societies, pottery production, mainly of fine painted ware, probably evolved from a domestic mode of production to an incipientspecialization stage (Rice 1981; Van der Leeuw 1984). This stage would include an increasing standardization of paste composition, reflecting greater exploitation of particular kinds of clays. In addition, a greater skill is more evident in manufacturing and firing (up to 1000 C ) technology. Middle Neolithic black household pots, analysed in two Murge settlements (Setteponti and Santa Barbara), show a silicate matrix with angular to sub-angular coarse-grained alabastrine

SiO2&, ./ "v~.. i

)...-"

"-/ !

', Ca-rich ,,f.

,>,'

~

.// ,"

* \ ,

•- .. ............

0

'

""> ),~ i

, '\

~

.

25

,,

\

~/"'~' •/ ,

r

\ ; An

,

,I

Ca-poor x

',i f

\

~,,

C a O + M a O

./

.

.

.

'.f-

:~ ...........

"'/Gh

~ .....

50

O

~__. !

L.f"

75

AI20 3 t00

"

Fig. 5. (CaO + MgO)-A1203-SiO 2 wt% diagram (Di, diopside; Gh, gehlenite; An, anorthite). Middle Neolithic fine painted pottery samples (n = 127) from the Tavoliere plain (•), the Bradanic trough (grey dots), the northwestern (x) and southeastern (D) areas of the Murge plateau, and the fields (whole sample) of Plio-Pleistocene silty clays (n = 89).

APULIAN POTTERY

(a)7°°] Ba

Non-calcareous clay (terra rossa) could be used as a raw material, tempered with crushed alabastrine limestone clasts.

/

! 600~

o

i

Late Neolithic (4400 or 4 2 0 0 - 4 0 0 0 Bc)

o

of the Murge plateau

oo

6oo! o

o o**

•

P* " 8 o

*

o~

.

400.

" ~.+.

o

o

oe.~..-

, c~°.

300 o

o

o

•

o @@

200 200

(b) 700-

400

600

800

10'00 12'00 14'00Sr

Ba

600-

500.

•

o

°

o

i

*

*

• o~

° •

4004

j.: •'.., • i

•

}

° O

]

• '4~'0

0,

111¢* O0 * °°

. *. +

GO

300~

•@~.

"z~" •

°

,,

i "; 200

57

"

•

6o

8'0

16o 1to 1~,o 16o 180 260 220 Zr

Fig. 6. (a), Bav. Sr plot (ppm) for Middle Neolithic fine painted pottery samples (small dots) and the samples from the Bradanic trough (grey dots) and the southeastern area of the Murge plateau (R); (b) Ba v. Zr plot (ppm) for the Middle Neolithic fine painted pottery samples (small crosses) and the samples from the Tavoliere plain (e) and the Bradanic trough (grey dots).

limestone clasts; quartz and Fe-oxides or hydroxides as aggregates and pisoliths were also observed as natural non-plastic inclusions. PXRD analyses confirm the presence of predominant calcite, accompanied by variable amounts of quartz and feldspar. XRF analyses are consistent with the mineralogical data: SiO2, C a • and A1203 are the dominant oxides (Geniola et al. 2005; Muntoni et al. 2006).

Late Neolithic pottery (black and plain household wares) was sampled only from the Cala Scizzo cave (Bail province), located on the Adriatic coast of the Murge plateau (Geniola et al. 2005). As far as the ecological and technological aspects are concerned, archaeometric data suggest the use of two different (noncalcareous and calcareous) clays for production for the two archaeological wares. Raw material variability is again evident (as in the Early Neolithic pottery of the same area) mainly on a synchronous level. Petrological and mineralogical data, and chemical concentrations of SiO2, A1203 and C a • allow the identification of two material groups, with different dominant clastic constituents. In the black household pottery, the silicaterich matrix is dominant and non-plastic inclusions are coarse-grained quartz and feldspar clasts, with no carbonate rocks. These samples are also distinguished by the highest AlzO3 and SiO2 values (Fig. 7a) related to clay matrix abundance. Terra rossa, which is very poor in carbonate (Dell'Anna 1967; Dell'Anna & Garavelli 1968; Dell'Anna et al. 1973), could be used as a raw material for this ware. Plain household pottery shows an abundant sheet silicate matrix, in which only the micas are recognizable; nonplastic inclusions are homogeneous fine-grained minerals such as quartz and carbonate fossils (planktonic Foraminifera). This ware is also distinguished by similar amounts of the three main oxides. The Argille di Rutigliano, marly clays cropping out not far from the considered site (Moresi 1990), are consistent with the sheet silicate clay matrix of plain household pottery (Fig. 7a). As regards trace elements, rare earth elements (REE), such as lanthanum (La) and cerium (Ce), clearly show only two clusters (Fig. 7b): the former contains plain household pottery and the latter black household samples.

Preparation of bodies Preparation of raw materials has shown the different choices followed by ancient potters in the preparation of bodies. Clays are usually more or less refined, and in some cases the use of mineral temper (such as fossiliferous sand, alabastrine limestone and calcite), grog or

58

R. LAVIANO & I. M. MUNTONI (a)

c>Si02~

J

'

~/'~

~'Argille di Rutigt!ano

'

,/

\\\

/

-,/ / AI~

/ ,-

\ \

"/

CaO+MgOI

L

,

~

0

(b)

V Gh \

,

i \

~

\\

/ \\

\

\

, I 25

'

'

"

; '\'\

;

/

\

, \

~

~'~-

\

, J 50

'

~

'

/."~,. \

, \

'

/

\

, 1 75

'v" '

'

~

'

IAI203 100

2oo

180 160 140 120 Ce

too

8o

_ _~

.......... " ~ - - -

~ _ . _ ~

6o ~.~Argille

di Rutigliano

40 2o 0

J

10

1

20

30

40

50

80

70

80

90

100

La

Fig. 7, (a) (CaO + MgO)-A12Os-SiO2 wt% diagram (Di, diopside; Gh, gehlenite; An, anorthite; (b) Ce v. La plot (ppm). Late Neolithic plain household (x) and black household (O) pottery samples (n = 19) from Cala Scizzo, and the fields (whole sample) of Argille di Rutigliano (n = 5) and terra rossa (n = 25).

vegetable material has been found, with variations in its incidence from sample to sample, from site to site and from area to area. The Tavoliere samples (Cassano et al. 2004) showed that clays were generally used as they naturally occurred, as has been proven by the mineralogical component, which compares well with that of local clays, although some degree of variation in grain size and percentage occurs. Microscopic observation also revealed that fossiliferous sand or crushed sparry calcite was occasionally (n = 18) employed as temper in coarse ware in only three Early and Middle Neolithic settlements (Monte Aquilone, Santa Tecchia and Masseria Candelaro). In the Murge area ancient potters followed different choices in the preparation of pastes.

Analytical data have shown that the main degree of variation was in paste preparation and grain-size (from 5 0 - 2 5 0 to 5 0 - 6 0 0 p~m) composition. Some Early Neolithic Qtz-rich samples from Pulo di Molfetta and Ciccotto (Muntoni 2003) are characterized by the presence of grog fragments (n = 9, mainly from Pulo) or by the presence of curvilinear and long pores probably derived from vegetable material burnt during firing (n = 5, from Pulo). The large amount of fossils, which characterized the Cal-rich Early Neolithic coarse wares of Pulo di Molfetta and Balsignano, is due to an intentional addition of a calcareous temper to terra rossa. The presence of this particular ware in two different Neolithic villages could also indicate the same appreciable intra-group

APULIAN POTTERY choice as a response to functional or/and social constraints. Only Middle Neolithic black household pots from the Murge settlements of Setteponti and Santa Barbara (about 90 km apart) are systematically tempered by angular to subangular coarse-grained alabastrine limestone clasts (Geniola et al. 2005; Muntoni et al. 2006). In this case three principal criteria have been used to identify temper: (1) the hiatal distribution of the inclusions; (2) a proportion of inclusions _>30%; (3) the angular outlines of the grains. Such data could give some positive insight into the inter- and intra-group organization of the many Middle Neolithic communities of the Murge plateau, who shared so many other common behavioural features (Cassano 1993).

Firing techniques Mineralogical and petrological data gave some insight concerning pottery firing temperatures: the maximum temperatures reached during firing have been inferred to be usually between 600-700 and 850 °C in both the considered areas. As kaolinite, which is a common component of the Apulian clays, has disappeared in every sample, one could suggest that temperatures exceeding 600 °C were always reached. Early Neolithic pots are usually light buff coloured; pottery was probably fired in pits and several factors determined the success or outcome of the firing process (for some experimental tests, see Cassano et al. 1995a). For Early Neolithic firing technology, mineralogical data show that maximum temperatures did not exceed 800 °C. However, on the basis of the presence in X-ray patterns (Table 2) of very weak peaks of some clay minerals (illite plus muscovite and minor quantities of smectite), one could argue that some Early Neolithic samples were fired at temperatures that could have reached 600-700 °C. The high peaks of primary calcite in Early Neolithic Cal-rich samples (Table 2) show that these coarse pots also were fired at a temperature not exceeding 700 °C. Other samples, as the lower amount of clay minerals, together with the presence of calcite, shows (Table 2), could have been fired at higher temperatures, but still not over 800 °C. In Middle and Late Neolithic villages, black burnished and plain red and/or brown painted pottery are associated in archaeological contexts. For these pots, involving different firing structures and techniques, greater efforts were made by ancient potters to control the amount of oxygen that entered the firing structures, to produce an oxidizing or reducing atmosphere.

59

Mineralogical and chemical data show that Middle-Late Neolithic black household and Late Neolithic plain household pots were fired at a temperature not exceeding 600-800 °C. For Middle Neolithic fine painted pottery (the so-called f i g u l i n a ) higher temperatures have been suggested, revealing a better firing control (temperature, rate of heating and oxidizing atmosphere) and the use of kilns. The absence of a dark core and the low birefringence of the matrix confirm a high degree of sintering. Such samples show gehlenite and pyroxene neoformations, only apparent by PXRD analysis, whereas clay minerals are absent (Table 2). In the same group one can see decreasing amounts of calcite, which, as microscopic observation on thin section shows, was recrystallized in pottery pores. On the basis of such data one can suggest that temperatures between 850 and 1050 °C were obtained. In some settlements (Masseria Candelaro in the Tavoliere plain and Pulo di Molfetta and Ciccotto in the Murge plateau) a clear differentiation between Early and Middle Neolithic ceramics (the latter fired at higher temperatures) has been found (Cassano et al. 2004; Muntoni 2003). Unfortunately, in the Tavoliere and Murge areas very few fire structures have been identified, probably because of a lack of extensive archaeological excavations, and no direct connection with ceramic firing (rather then with baking or roasting) can yet be established. PXRD analyses of representative samples heat-treated at 1000 °C (Table 2) confirm that pyroxene and gehlenite synthesis is dependent on calcite and clay mineral abundance (Maggetti 1982). The presence of a high quantity of neoformed pyroxene and gehlenite was detected in Qtz + Cal-rich samples; their concentration increases in Middle Neolithic fine painted pots and in Early Neolithic Qtz ÷ Cal-rich samples whereas they are initially absent. In Qtz-rich samples, found only in Early (Pulo di Molfetta and Ciccotto) and Late Neolithic (Cala Scizzo) sites of the Murge region, characterized by very low quantities of calcite, only hematite is the secondary product of firing. Only in Cal-rich heat-treated samples, characterized by coarsegrained carbonate fossils (Early Neolithic Pulo di Molfetta and Balsignano coarse ware) and alabastrine limestone clasts (Middle-Late Neolithic black household ware), was the co-occurrence of diopside, gehlenite and hematite observed. CaO and a neoformed calcium silicate very similar to larnite were also detected, probably formed as a result of the high quantities of CaCO3 in the paste.

FP FP FP FP FP FP FP FP FP FP BH BH BH BH BH

BH BH BH PH PH PH

Middle Neolithic PU30 PU53 MC23 MC24 MC30 SP04 GT02 GS09 CC08 CC 18 SPI4 SPI5 SP18 SB01 SB 11

Late Neolithic CSI0 CS02 CS01 CSll CSI2 CSI7

X XX tr X X

X X X X XX

X

X X tr tr X tr

XX XX XX XX XX XX XX

Sm

tr

tr X X tr

tr X XXX XX XX X XX

tr

tr tr

X X X XX XX XX

XX X XX X XX XX XXX

I11 + Ms

XXXXX XXXXX XXXXX XXXXX XXXXX XXXXX

XXXXX XXXXX XXXXX XXXXX XXXXX XXXXX XXXXX XXXXX XXXXX XXXXX XXXX XXXX XXXX XXXXX XXXX

XXXX XXXX XXXXX XXXXX XXXX XXXXX XXXXX XX XX XX XXX XXXXX XXXXX XXX XXXXX XXXXX XXXXX

Qtz

XXX X X XX X XX

XXXX XXX XXX XX XX XXX XX XXX XXXX XXXX X X X X XXX

X X XX X X XXX XXX tr tr tr tr XX XX XXXX XXXX XXX XXXX

Feld

tr XXXX XXXXX XXXX

X X X X tr tr XX XX XXXXX XXXXX XXXXX XXXXX XXXXX

tr

XXXX XXXXX XX XXXXX XXXXX XXXXX XXXXX XXXXX XXXXX XXXXX XXXXX XXXXX XXXXX XXXXX

Cal

X tr

XX XX XX X XX XX XX XX XX XX

Px

X XXX XX XX XX XX XX XX XX XX

Gh

X tr tr

tr X tr X tr X

Hem

XXXXX XXXXX XXXXX XXXXX XXXXX XXXXX

XXXXX XXXXX XXXXX XXX XXXXX XXXXX XXXXX XXXXX XXXXX XXXXX XXXXX XXXXX XXXXX XXXXX XXXXX

XXXX XXXX XXXXX XXX XXXX XXXXX XXXXX XX XX XX XXX XXX XXXX XXX XXXXX XXXXX XXXXX

Qtz

XXXXX XX XX XXX XXX XXX

XXXXX XXXX X X XXX XX XX XXX XXXX XXX XX XXX XX XXX XX

X XX XXX XXX XX XXX XXXX tr tr tr tr XX XXX XX XXXX XXX XXX

Feld

X X XXX XXX XXX

XX XX XXX XX XXX XX XXX XX XXX XXX XX XX XX X XX

XX XXX XX XXX XXX XXX XXX X X X X XX XXX XX XX X

Px

Gh

XX XX XX tr

XXXX XX XXX

XX X tr tr tr X tr X X tr X X tr XX XX

XX XX XX

Hem

tr

X XX XXXX XXXXX XX XX XX XX X X XX XX XXX X X

X XX X XX XX XX XX XXXX XXXX XXX XXX XXX XX XXXX

heat-treated at 1000 C

Sm, smectite; I11, illite; Ms, muscovite; Qtz, quartz; Feld, k-feldspar and plagioclase; Cal, calcite; Px, pyroxene; Gh, gehlenite; Hem, hematite (symbols as in Kretz 1983); number of X is in relationship with mineralogical phase abundance; tr, traces. Archaeological wares: C, coarse; B, burnished; P, painted; FP, fine painted; BH, black household; PH, plain household.

C B P P P C C C C C C C P C B C B

Ware

as-received

Mineralogical composition (by PXRD) of representative pottery, samples

EarlyNeolithic CN03 MCI2 MM1 MA8 MV8 PU02 PU26 PUI6 PUI9 BALS24 BALS30 MCSI MA4 MAI0 PU01 PU05 CC27

T a b l e 2.

¢~ o

APULIAN POTTERY

61

Concluding remarks

References

The review of Apulian Neolithic settlements providing archaeometric pottery data has shown that ceramic studies employing laboratorybased techniques remain one of the most active areas of research in Italian archaeometry. A complete combination of petrological (OM), mineralogical (PXRD) and chemical (XRF) analyses is more informative and capable of greater discriminatory power. The correlated analysis of clay sources is most fruitfully conducted within the perspective of gathering information about which clay source was used, and why, and what kind of problems were encountered by the ancient potters in the modification of the materials. The main patterns of variation of Neolithic pottery production, from the seventh to the fourth millennium Bc, may be summarized in relation to provenance of raw materials and/or finished pottery artefacts, preparation of raw materials and firing techniques. Our studies indicate that generally local clays, the PlioPleistocene silty clay of the Bradanic cycle (Argille Subappennine and Argille di Rutigliano) and silty-clayey continental sedimentary deposits (terra rossa), were used for Neolithic pottery production. In some cases the exploitation of a range of different local fabrics has been verified. In Middle Neolithic sites the systematic use of Plio-Pleistocene silty clay has been also determined, even in some sites where such clay cannot be strictly considered as a local raw material (sometimes cropping out more than 30 km from the sites). Few finished pots were actually exchanged at an inter-site scale during the Neolithic. In the preparation of raw materials, different choices were followed by ancient potters in the preparation of pastes. Clays are usually more or less refined and, in some cases, the use of mineral temper, grog or vegetable material has been found, with variations in their incidence from sample to sample and from site to site. Firing techniques have been also considered; the maximum temperatures reached during firing are usually between 6 0 0 - 7 0 0 and 850 °C. For Middle Neolithic fine painted pottery higher temperatures have been suggested (between 850 and 1050 °C), revealing a better firing control (temperature and atmosphere) and the use of kilns. The different sources of ceramic variation, their relative frequency and the potter's role as a controlling agent in pottery manufacture will be further explored in the next stage of our research.

BALENZANO, F., DELL'ANNA, L. & DI P1ERRO, M. 1977. Ricerche mineralogiche, chimiche e granulometriche su argille subappennine della Daunia. Geologia Applicata e Idrogeologia, XII(II), 33 -55. BISHOP, R. L., RANDS, R. L. & HOLLEY, G. R. 1982. Ceramic compositional analysis in archaeological perspective. In: SCmFFER, M. B. (ed.) Advances in Archaeological Method and Theory, 5. Academic Press, New York, 275-330. BRONITSKY, G. & HAMER, R. 1986. Experiments in ceramic technology: the effect of various tempering materials on impact and thermal-shock resistance. American Antiquity, 51, 89-101. CALDARA, M., FATIGUSO, R., GARGANESE, V. & PENNETTA, L. 1990. Bibliografia geologica della Puglia. SAFRA, Bari. CASSANO, S. M. 1993. La facies Serra d'Alto: intensificazione delle attivit~ produttive e aspetti del rituale. Origini, XVII, 221-253. CASSANO,S. M., EYGUN,E., GARIDEL,Y. & MUNTON1, I. M. 1995a. Pottery making in southern Italy Neolithic: an experimental study. In: VENDRELL-SAZ, M., PRADELL,T., MOLERA,J. & GARCIA,M. (eds) Estudis sobre cerhrnica antiga. Actes del simposi sobre ceramica antiga. Universitat de Barcelona, Barcelona, 11- 16. CASSANO, S. M., LAVIANO, R. & MUNTONI, I. M. 1995b. Pottery technology of early Neolithic communities of Coppa Nevigata and Masseria Candelaro (Foggia, Southern Italy). In: FABBRI, B. (ed.) The Cultural Ceramic Heritage European Ceramic Society Fourth Conference. Gruppo Editoriale Faenza, Faenza, 14, 137-148. CASSANO, S. M., ERAMO, G., LAVIANO, R. & MUNTONI, I. M. 2004. Analisi archeometriche delle ceramiche. In: CASSANO, S. M. & MANFREDINI, A. (eds) Masseria Candelaro. Vita quotidiana e mondo ideologico in una communitgt neolitica del Tavoliere. Claudio Grenzi, Foggia, 221-249. DELL'ANNA, L. 1967. Ricerche su alcune terre rosse della Regione Pugliese. Periodico di Mineralogia, XXXVI(2), 539-592. DELL'ANNA, L. 1969. Ricerche mineralogiche e chimiche sulle 'Argille di Rutigliano'. Periodico di Mineralogia, XXXVIII(3), 515-577. DELL'ANNA, L. & GARAVELLI,C. L. 1968. Su alcune 'terre rosse' della Puglia settentrionale. Grafiche Rossi, Bari. DELL'ANNA, L. & LAV1ANO,R. 1991. Mineralogical and chemical classification of Pleistocene clays from the Lucanian Basin (Southern Italy) for the use in the Italian tile industry. Applied Clay Science, 6, 233-243. DELL'ANNA, L., DI PIERRO, M. & QUAGLIARELLAASCIANO, F. 1973. Le 'terre rosse' delle Grotte di Castellana (Bari). Periodico di Mineralogia, XLII(1-2), 23-67. DONDI, M., FABBRi, B. & LAVIANO,R. 1992. Characteristic of the clays utilized in the brick industry in

62

R. LAVIANO & I. M. MUNTONI

Apulia and Basilicata (Southern Italy). Mineralogica Petrographica Acta, XXXV(A), 181 - 191. ERAMO, G., LAVIANO,R., MUNTONI, I. M. & VOLPE, G. 2004. Late Roman cooking pottery from the Tavoliere area (Southern Italy): raw materials and technological aspects. Journal of Cultural Heritage, 5, 157-165. FABBRI, B., GUALTIERI, S. & SANTORO, S. 1997. L'alternativa chamotte/calcite nella ceramica grezza: prove tecniche. In: Santoro BIANCHI, S. & FABBRI, B. (eds) II contributo delle analisi archeometriche allo studio delle ceramiche grezze e comuni. II rapporto forma/funzione/impasto. University Press Bologna, Imola, 183-190. FRANZINI, M., LEONI, L. & SAITTA, M. 1972. A simple method to evaluate the matrix effects in X-ray fluorescence analysis. X-Ray Spectrometr3', 1, 151-154. FRANZINI, M., LEONI, L. & SA1TTA, M. 1975. Revisione di una metodologia analitica per fluorescenza X, basata sulla correzione completa degli effetti di matrice. Rendiconti della Societh ltaliana di Mineralogia e Petrologia, 31,356-378. GENIOLA, A., LAV1ANO, R. & MUNTONI, I. M. 2005. Pottery production in Late Neolithic cult sites of Santa Barbara and Cala Scizzo (Apulia, Southeast Italy). In: PRUD~.NCIO, M. I., D1AS, M. I. & WAERENBORGH, J. C. (eds) Understanding people through their potter3'. Proceedings of the 7th European meeting on Ancient Ceramics, Lisbon, 27-31 Ottobre 2003, Instituto Portugu6s de Arqueologia, Lisbon, Trabaihos de Arqueologia, 42, 89-101. KOLB, C. C. 1989. Ceramic Ecology, 1988. Current Research on Ceramic Material. British Archaeological Reports, Oxford, 513. KRETZ, R. 1983. Symbols for rock-forming minerals. American Mineralogist, 68, 277-279. LAVIANO, R. & MUNTON1, I. M. 2003. Early and Middle Neolithic pottery production at 'Pulo di Molfetta' (Apulia, Italy): social, chronological and functional implications of raw materials variability, hT: Dt PIERRO, S., SERNEELS, V. & MAGGETTI, M. (eds) Ceramic in the Society. Proceedings of the 6th European Meeting on Ancient Ceramics. University of Fribourg, Fribourg, 163-173. LAVIANO, R. & MUNTONI, I. M. 2004. Le argiile e l'archeometria delle ceramiche. Scelte tecnologiche delle comunita neolitiche in Puglia ricostruite con modeme metodiche analitiche. In: INGRAVALLO, E. (ed.) 11 fare e il suo senso. Dai cacciatori paleo-mesolitici agli agricoltori neolitici. Congedo, Galatina, 113-164. LEONI, L. & SMTTA, M. 1976. Determination of yttrium and niobium on standard silicate rocks by

X-ray fluorescence analysis. X-Ray Spectrometr3", 5, 29-3O. MAGGETTI, M. 1982. Phase analysis and its significance for technology and origin, hi: OLIN, J. S. & FRANKLIN, A. D. (eds) Archaeological Ceramics. Smithsonian Institution Press, Washington, DC, 121-133. MATSON, F. R. (ed.) 1965 Ceramics and Man. Aldine, Chicago, IL. MORESl, M. 1990. Genesi ed evoluzione di depositi argillosi pleistocenici in Puglia. Mineralogica Petrographica Acta, XXXIII, 283-295. MUNTONI, I. M. 1999. From ceramic production to vessel use: a multi-level approach to the Neolithic communities of the Tavoliere (Southern Italy). In: OWEN, L. R. & PORR, M. (eds) Ethno-Analogy and the Reconstruction of Prehistoric Artefact Use and Production. Mo Vince, Tfibingen, 237-254. ML'NTONi, I. M. 2002a. The application of archaeometric analyses to the study of Italian Neolithic pottery: some methodological considerations. In: Archaeometrv in Europe in the Third Millennium. Atti del Com'egno h~ternazionale, Roma, 29-30 Marzo 2001. Accademia Nazionale dei Lincei, Roma. 203-213. ML'NTONI, I. M. 2002b. Le analisi archeometriche di ceramiche neolitiche in Italia: storia degli studi, strategie di campionamento, tecniche analitiche e obiettivi delle ricerche. Origini, XXIV, 165-234. ML'NTONI, I. M. 2003. Modellare l'argilla. Vasai del Neolitico antico e medio nelle Murge pugliesi. Istituto Italiano di Preistoria e Protostoria, Firenze. MUNTONI, I. M. & LAVIANO, R. 2005. La produzione ceramica nel Neolitico antico del Tavoliere (FG): verso un modello di interazione tra le diverse comunit~ di viilaggio, hi: FABBRI,B., GUALTIERI,S. & VOLPE, G. (eds) Tecnologia di lavorazione e impieghi dei manufatti. Atti della 7" Giornata di Archeometria della Ceramica, Lucera, 10-11 Aprile 2003. Edipuglia, Bari, 61-69. MUNTONL I. M., LAVlANO, R. & RADINA, F. 2006. Materie prime e tecnologia di produzione della ceramica 'Serra d'Alto' helle Murge pugliesi. In: FABBRI, B., GUALTIERI, S. & ROMITO, M (eds) l_xt ceramica in Italia quando I'Italia non c'era. Atti della 8" Giornata di Archeometria della Ceramica, Vietri sul Mare, 27-28 Aprile 2004. Edipuglia, Bari, (in press). RICE, P. M. 1981. Evolution of specialized pottery production: a trial model. Current Anthropology, 22(3), 219-240. VAN DER LEEUW, S. E. 1984. Dust to dust: a transformational view of the ceramic cycle. In: VAN DER LEEUW, S. E. & PRITCHARD, A. C. (eds) The Matt3' Dimensions of Potter3,. University of Amsterdam Press, Amsterdam, 707-774.

Late La T~ne pottery from western Switzerland: one regional or several local workshops? MARINO MAGGETTI & GIULIO GALETTI

University of Fribourg, Department of Geosciences, Mineralogy and Petrography, Chemin du Musde 6, CH-1700 Fribourg, Switzerland (e-mail: [email protected])

Abstract: A total of 203 pieces of fine ceramic and four clays from seven sites of western Switzerland (Bern, Gen6ve, Grotte du Four, La T~ne, Matin, Saint-Triphon-Massongex and Yverdon) were studied chemically and mineralogically to determine if there was local production at each site and if trade links existed between the sites. Firing wasters from Bern and the region of Gen~ve indicate local ceramic production. The sherds are often contaminated with secondary phosphorus and, in the case of Bern, copper. Most of the fine ceramic is CaO-poor, contrasting with the CaO-rich clays. Based on the chromium and nickel concentrations, it can be subdivided into two distinct groups. The majority of the sherd populations from Gen~ve, Saint-Triphon and Massongex, as well as a few specimens from Bern, La T~ne and Yverdon, have high Cr and Ni values. The remaining sherds have low Cr and Ni concentrations. The analyses show that: (1) the fine ceramic from each of the seven sites forms an often inhomogeneous and widely dispersed group, distinct from the others; consequently, it is most probably a local or regional product; (2) ceramic import is probable for one piece from Grotte du Four (provenance Yverdon); (3) the Late La T~ne fine ceramic was manufactured mainly from silicate or silicate-carbonate, fat to lean clays.

Late La T6ne (LT D, c. 150-30 BC) ceramics from the NW of Switzerland (locations: BaselGasfabrik, Basel-Mtinsterhtigel, Sissach-Brtihl, Waldenburg-Gerstelfluh) were analysed by Maggetti & Galetti (1981) and Maggetti et al. (1988). They showed that the ceramic of Basel-Gasfabrik and of the pottery centre of Sissach-Brtihl is a CaO-poor fine ware, which forms two well-defined, homogeneous chemical reference groups. It must therefore be assumed that pottery was produced not only in SissachBriihl, but also in Basel-Gasfabtik. The similar chemical composition of some objects from Basel-Mtinsterhtigel and Waldenburg-Gerstelfluh to these reference groups suggests local ceramic trade. This study focuses on 203 La T~ne fine ceramic fragments from eight other sites in western and southwestern Switzerland, i.e. Bern, Gen~ve, Grotte du Four, La T6ne, Matin, Saint-Triphon, Massongex and Yverdon (Fig. 1). For further comparison, the four clays excavated at the production sites were included in the study. Based on firing wasters, local production was indicated for the sites of Bern and Gen~ve only. As a result, it was necessary to determine if the remaining samples originated from these two production sites, or if they formed different chemically, mineralogically and petrographically distinguishable groups. A

significant enough difference among the seven provenances would point to local ceramic production, i.e. manufacture at different places simultaneously, rather than large isolated workshops in the area of present-day western Switzerland, with regional and national trade.

Samples A total of 203 grey to light grey fine ceramic samples were provided by the archaeologists involved in the project. They belong typologically mainly to bottles and bowls and were made with a wheel. Dating and archaeological literature are as follows.

Bern BE. Middle to Late La T~ne, LT C 1 - L T D2 (c. 250-c. 30 Bc): Mtiller-Beck (19631964), St/ihli (1977), Bacher (1989), Mtiller (1990, 1996), Kohler (1991). Genbve GE. Middle to Late La T~ne, LT C 2 LT D2 (c. 200-c. 30 ~c); Paunier (1981), Kaenel (1990), Bonnet (1997). Grotte du Four (Boudry NE). Middle to Late La T~ne, LT C 2 - L T D2 (c. 2 0 0 - c . 30 BC); Kaenel ( 1991). La Tkne (Marin-Epagnier NE). Middle La T~ne, LT C (c. 2 5 0 - c . 150 BC); Schwab (1989), MUller (1990), Egloff (1991).

From: MAGGETTI,M. & MESSlGA,B. (eds) 2006. Geomaterials in CulturalHeritage. Geological Society, London, Special Publications, 257, 63-80. 0305-8719/06/$15.00 © The Geological Society of London 2006.

64

M. MAGGETrI & G. GALETTI structure types was made, based on the matrix (fat or lean). Clay pellets were not considered in the determination of grain sizes, because they dissolve during soaking.

X-ray diffraction (XRD).

A total of 199 fine ceramic samples were X-rayed under standard conditions (Cu K~, 3-65°20, operating conditions 30 kV and 40 nA) on a Siemens Kristalloflex D500 generator.

!

X-ray fluorescence analysis.

B~lrn Yverdon

Triphon

Fig. 1. Location of the studied sites. Gen~ve = Annecy (France), Dardagny-Brive, Gen~ve, Meinier, Momex and Vandoeuvre-Pressy.

Marin (Marin-Epagnier NE). Late La T~ne, LT DI (c. 150-c. 80 BC); Arnold (1992). Massongex VS. Late La T6ne, LT D (c. 150c. 30 Bc): Haldimann et al. (1991), Curdy et al. (1997). Saint-Triphon (Ollon VD). Late La T6ne, LT D (c. 150-c. 30 BC); Kaenel (1990). Yverdon (Yverdon-les-Bains VD). Middle to Late La T~ne, LT C 2 - L T D2 (c. 200-c. 30 BC); Curdy et al. (1992, 1995), Brunetti (2005). Methods

Powder preparation. For chemical analysis and X-ray diffraction (XRD), 4 - 5 g per sample were ground to a fine powder in a tungsten carbide mill after abrasion of possibly contaminated surface areas. Optical microscopy. Thin sections were prepared when sufficient original material was present. After microscopic analysis, an approximate classification of the 97 samples into

Determination of major elements (Si, Ti, AI, Fe, Mn, Mg, Ca, Na, K, P) and trace elements (Ba, Cr, Cu, Ga, Nb, Ni, Pb, Rb, St, Th, V, Y, Zn, Zr) was performed on all samples. Circular tablets (40 mm diameter) of glassy material (calcined powder) were used for the major elements. Circular tablets of pressed powders (non-calcined) were used for the trace elements. The preparation of the glass tablets was carried out as follows: after calcination of the powdered sample for l h at 1000 ~C, 1.2 g of this calcined material was mixed with 5.7 g of lithium tetraborate and 0.3 g of lithium fluoride. This mixture was then placed in a P t - A u crucible and melted at 1150 °C before being poured into a preheated mould and cooled with compressed air. The preparation of the pressed tablets was carried out as follows: 2.5-3.0 g of initial powder were mixed with < 0 . 3 m l of a moviol-saturated aqueous solution. This was then added to a 32 mm mould and subjected to a pressure of 6 tons for at least 1 rain. The tablet was carefully removed and placed on a bed of 7 g of boric acid in a 40 mm mould. This was again subjected to a pressure 6 tons for at least 1 min. The resulting tablet was dried in a vacuum for 24 h at room temperature. Analytical measurements were performed using a Philips PW 1400 X-ray spectrometer with Cr anticathode. The conversion of the measured values to weight percentage concentrations utilized standardization curves established on reference samples (e.g. USGS, NIM, ANRT). The results of the measurements of the major elements of the matrix were corrected with Philips alpha coefficients.

FeO. Determination of FeO was by the 2.2 dipyridilic method (Lange & Vejdelek 1980), using a Philips Pye-Unicam PU 8650 at 528 nm. Statistics.

Multivariate analyses were performed using SPSS 11.0, neglecting P205 and Cu (contamination effects) as well as Pb and Th (too many blank values) and FeO. For cluster analysis, the raw data (20 elements), average linkage and Ward linkage, and z-scores were

LATE LA TENE POTTERY used. For the factor and the discriminant analyses log-transformed data (20 elements) were used.

Chemical contamination In contrast to the clays, which can contain a maximum of c. 0.2 wt% P205 (Koritnig 1978), ancient ceramics are often characterized by distinctively higher concentrations. In most cases, this can be interpreted as a phenomenon relating to the contamination through migrating P-rich solutions during burial (see Collomb & Maggetti 1996). However, with a concentration of up to 11 wt%, most ceramic pieces from Bern greatly exceed this value. To minimize this secondary phosphorus contamination, all analyses shown in Table 1 were recalculated to 0 wt% P205, and standardized to a total of 100 wt%. These values were used in the following paragraphs. Some Bernese samples show an increased copper content up to a maximum of 2540 ppm. This is due to the use of a Cu-bearing marker for the annotation of these objects (i.e. BE 20, 27, 58, 66, 74, 85 and 92). Copper was therefore not taken into account for the multivariate statistics. In addition to phosphorus, other elements may have been mobile, a possibility that is virtually impossible to confirm. The discussion is therefore based on the assumption that no elements other than P and Cu were affected by secondary processes.

Pottery from the production sites of Bern and Gen~ve Bern A new reference group. A total of 87 samples were analysed (Tables 1 and 2), comprising 85 fine ceramic sherds from two neighbouring excavation sites, Tiefenau/Heiligkreuz and Engemeistergut, as well as two clays, BE 40 from the first excavation site, mixed with sherds, and BE 52 from a trench close to the first excavation site and corresponding to the clayey substratum of the Celtic settlement. The identification of four kiln wasters (BE 38, 44, 49, 73) indicates local ceramic production. The products show a wide scatter in their CaO content, but the bulk of the Bernese fine ceramic is relatively CaOpoor, with a blurred transition to CaO-richer specimens (Fig. 2a). Both clays are very CaOrich and belong therefore to marls rather than to clays (Table 1). Some pottery samples stand out markedly from the main body with regard to one or several chemical parameters (Fig. 2a-d). These are BE 28 (low Zr and Y contents), BE 68 (highest CaO with lowest A1203

65

and Zr values), BE 53 (lowest Na20, MgO and MnO, and highest TiO2, Y and Zr values), and BE 26, 48, 71 (increased Cr and Ni contents). There may be several possible explanations for the lack of compatibility between outliers and the fine ceramic group. (1) The outlier may be of local origin but the amount of sampled material per sherd is insufficient and not representative of a single object; or the outlier may have been affected by secondary contamination during use or burial; or there may be coincidental fluctuation of the chemical composition of the otherwise homogeneous raw materials, i.e. fluctuations that were not eliminated during clay preparation. (2) The outlier is not of local production. For fine ceramic, the population investigated here, the amount of material selected for a single analysis, i.e. 4 - 5 g, is sufficient (Schneider 1989). Also, the contamination hypothesis can be ruled out because elements such as Zr,Y, Mg, etc. are normally not affected by secondary processes. What about the last two possibilities? The use of an illitic-chloritic clay as raw material can be inferred from the negative A1203-SIO2 (Fig. 3a) and the positive A1203TiO2 (Fig. 3b) and AlzO3-FezO3totcorrelations (Fig. 3c). The amounts of calcite vary (Fig. 2a). This substantiates results obtained by XRD phase analysis, revealing relict primary clay phases such as quartz, illite, chlorite, plagioclase and potash feldspar (+_calcite) in the samples with the lowest firing temperatures of c. 500650 °C. Microscopic examination allows for allocation of the samples to three fabrics, as follows. Fabric 1. Very fat, silicate to silicatecarbonate matrix. Small amounts of mostly silicate non-plastic fragments with a maximum grain diameter of 2.52 mm (BE 7, 8, 10, 14, 15, 19-22, 24, 27-30, 34, 35, 39, 41-43, 46-48). Fabric 2. Lean, silicate matrix. Relatively large amounts of silicate non-plastic fragments (sandstone) with a maximum grain diameter of 1.08 mm and a serial granulometry (BE 13, 17, 26, 45). Fabric 3. Carbonate-silicate matrix. Nonplastic constituents with a maximum diameter of 2.23 mm (clays BE 40, 52). CaO-poor as well as CaO-rich (e.g. BE 41 with 13.6 wt% CaO) fine ceramic samples have been allocated to fabric 1, despite distinct variations in CaO, because the matrix does not differ significantly between the specimens. It can be concluded that geologically similar material, with a considerable fluctuation in microcrystalline calcites, has been used in the production of ceramics belonging to fabric 1. The assumption that these raw materials share a similar geological history

66

M. MAGGETTI & G. GALETTI

Table 1. Chemical analyses (wt% ) No.

SiO2 TiO 2 AI203 Fe203* MnO MgO

CaO

Na,O K20 P205

Total

FeO

LOI

H20-

63.77 50.58 52.22 59.05 56.37 60.01 63.18 56.30 51.84 51.41 57.15 52.99 64.51 55.50 62.11 51.78 60.60 47.62 56.33 57.70 59.77 57.20 62.21 45.95 51.24 54.40 51.42 53.12 53.54 49.88 51.53 53.35 47.53 56.70 56.35 44.70 50.77 47.79 62.31 57.78 56.28 60.31 58.34 55.39 64.80 49.12 59.09 63.14 60.54 58.47 47.07 53.67 53.20 55.31 50.79 45.57 54.81 61.59 49.38 54.40 61.18 51.20 52.45 57.51

3.45 12.6 7.31 0.90 7.29 2.93 1.43 2.14 2.01 8.25 3.67 5.08 1.49 2.05 1.18 4.97 2.16 3.58 2.16 1.94 0.96 1.87 1.80 4.90 7.09 !.88 11.50 3.52 1.85 3.01 3.20 2.70 3.96 5.05 16.50 12.60 3.85 6.59 2.82 1.42 2.79 1.67 2.31 2.20 1.40 23.70 2.41 1.36 5.41 9.10 4.46 3.15 1.82 2.13 3.50 9.43 3.87 1.74 5.57 3.48 1.41 18.40 8.53 2.40

1.18 0.96 0.85 0.92 1.02 1.18 1.09 0.99 0.89 0.93 0.92 0.81 1.03 0.93 1.06 0.76 1.13 0.62 1.00 1.25 0.91 0.98 1.30 0.52 0.75 1.02 0.91 0.77 0.78 0.70 0.76 0.80 0.86 0.88 0.96 0.75 0.78 0.56 1.13 0.95 1.22 1.22 1.09 0.99 !.27 1.03 0.20 1.08 1.01 1.00 0.63 0.86 1.01 !.17 0.81 0.54 0.83 1.07 0.79 0.98 1.05 0.93 0.98 1.03

99.93 99.99 99.81 100.00 99.66 100.06 100.89 100.54 100.39 99.81 100.16 99.97 100.14 100.13 100.21 100.56 100.23 99.56 100.23 100.22 99.93 100.27 100.20 99.87 100.50 100.02 99.47 99.85 99.83 99.67 99.49 100.07 99.72 99.81 99.63 99.73 99.94 99.76 99.67 99.73 99.66 100.08 99.87 99.60 100.02 99.13 99.98 99.62 100.42 100.51 99.59 100.11 100.10 100.13 99.95 99.57 100.06 100.40 99.78 99.97 100.51 98.93 99.37 100.11

4.45 1.42 1.89 2.98 1.68 1.37 0.58 3.14 2.81 1.66 4.00 3.30 t.03 1.64 1.13 2.55 1.03 1.89 1.59 1.49 5.37 1.42 0.60 1.59 1.89 1.79 3.24 3.17 2.15 2.62 2.78 2.77 1.35 1.44 1.09 0.99 3.02 1.98 4.45 2.85 0.70 0.55 1.74 0.31 0.19 1.96 0.34 0.72 3.00 1.56 0.51 2.86 0.28 1.75 1.35 2.67 2.40 0.77 1.68 2.93 1.62 1.03 1.92 1.38

1.00 11.00 5.32 2.84 6.38 4.57 2.93 2.88 3.89 4.77 2.77 4.22 4.55 5.11 4.16 4.63 4.70 5.61 4.71 3.49 2.20 5.12 4.87 5.90 4.68 5.34 2.35 5.04 4.54 5.05 4.71 5.79 7.95 5.77 12.5 4.43 6.09 6.94 3.82 2.97 4.43 4.09 4.56 4.72 3.23 18.70 6.96 4.94 1.08 3.14 6.77 5.00 4.06 4.05 6.02 6.51 5.08 4.71 7.05 5.18 4.62 14.70 9.19 5.48

1.40 5.25 6.30 3.84 5.61 5.43 3.52 3.78 5.10 4.39 4.41 5.71 4.22 6.72 3.69 7.42 4.94 8.59 6.14 4.15 3.61 5.62 4.32 9.32 7.07 7.66 3.58 7.16 6.90 7.94 7.41 6.78 10.30 7.79 4.56 5.87 7.69 8.49 3.85 3.23 3.78 4.75 5.12 5.53 3.36 1.46 8.37 3.92 0.79 1.82 6.55 4.84 3.56 3.66 6.55 6.90 5.42 4.15 7.59 5.77 3.89 2.82 5.00 4.86

Bern BE 6 BE7 BE 8 BE9 BE 10 BEll BE 12 BE13 BE14 BE15 BE16 BE 17 BE 18 BE 19 BE20 BE21 BE22 BE23 BE24 BE25 BE 26 BE27 BE 28 BE 29 BE 30 BE31 BE 32 BE33 BE 34 BE35 BE36 BE37 BE38 BE 39 BE40 BE41 BE42 BE43 BE 44 BE 45 BE 46 BE 47 BE48 BE49 BE50 BE52 BE53 BE 54 BE55 BE56 BE57 BE58 BE59 BE60 BE61 BE62 BE63 BE 64 BE65 BE 66 BE67 BE68 BE69 BE70

0.77 0.82 0.88 0.91 0.83 0.77 0.89 0.93 1.00 0.83 0.93 0.90 0.80 0.93 0.84 0.95 0.86 1.02 0.90 0.90 0.92 0.93 0.86 0.99 0.94 0.96 0.84 0.97 0.97 1.00 0.98 0.96 1.04 0.92 0.70 0.91 0.97 0.97 0.83 0.91 0.88 0.87 0.84 0.93 0.78 0.64 1.15 0.82 0.82 0.81 0.99 0.97 1.00 0.93 1.00 0.93 0.93 0.88 0.97 0.91 0.87 0.71 0.86 0.91

17.44 17.66 19.32 20.81 17.58 17.48 18.80 20.66 22.43 19.60 19.23 20.99 18.68 20.77 19.47 21.03 18.56 22.41 20.10 19.79 19.22 20.57 18.42 22.48 20.64 21.18 18.57 21.39 22.17 21.88 21.49 21.43 22.56 20.29 14.23 20.23 21.36 21.17 18.54 20.89 19.75 18.93 18.69 20.78 17.47 12.98 19.15 18.80 18.55 17.54 22.10 21.32 22.18 20.54 21.96 20.87 20.93 19.22 21.00 20.07 19.20 15.20 18.57 19.87

6.47 7.00 7.73 7.98 7.00 7.21 7.39 8.37 9.48 8.50 7.65 7.69 6.07 8.25 7.88 8.17 7.49 8.84 7.97 8.01 8.98 8.26 7.46 8.81 8.11 8.20 7.29 8.22 8.68 8.75 8.30 8.49 8.97 8.18 5.61 7.95 8.59 8.43 7.41 8.32 8.04 7.32 8.03 8.16 6.75 5.12 7.56 7.52 7.37 7.06 8.48 8.22 9.12 8.40 8.67 8.03 8.01 7.61 8.16 7.81 7.65 5.98 7.26 7.90

0.06 0.13 0.15 0.10 0.18 0.13 0.17 0.14 0.14 0.17 0.17 0.13 0.t0 0.19 0.12 0.17 0.18 0.15 0.14 0.13 0.16 0.17 0.09 0.20 0.14 0.15 0.13 0.17 0.15 0.16 0.17 0.15 0.16 0.15 0.12 0.15 0.18 0.16 0.15 0.16 0.14 0.16 0.21 0.22 0.13 0.11 0.06 0.13 0.14 0.13 0.19 0.17 0.16 0.12 0.17 0.17 0.15 0.14 0.19 0.11 0.12 0.11 0.14 0.15

2.44 2.10 2.41 3.18 2.12 1.65 2.41 3.45 3.35 3.85 2.66 3.21 2.00 1.99 2.45 2.67 2.14 2.67 1.97 2.51 3.57 2.24 1.98 2.70 2.59 2.07 2.82 2.69 2.13 2.67 2.80 2.47 2.06 2.64 2.02 3.04 2.62 2.51 2.77 3.18 2.72 1.75 2.69 2.01 1.91 3.67 0.75 2.34 2.73 2.51 2.85 2.81 3.54 2.97 2.51 2.57 2.72 2.57 2.51 2.30 2.31 2.17 2.06 2.36

3.13 2.94 2.58 3.14 2.82 2.41 2.48 2.55 3.10 2.82 2.15 2.56 2.70 3.27 3.30 2.26 3.09 2.25 3.15 3.32 1.71 3.33 2.89 2.31 2.35 3.16 2.48 2.36 2.79 2.16 2.00 2.31 2.77 2.79 2.45 2.17 2.34 2.10 3.53 2.62 3.45 2.62 2.84 3.08 2.98 2.59 2.41 3.21 3.51 3.27 2.15 2.20 2.64 3.08 2.44 1.89 2.22 3.34 2.56 2.23 3.23 2.63 3.00 3.34

1.22 5.21 6.36 3.01 4.45 6.29 3.05 5.01 6.15 3.45 5.63 5.71 2.76 6.25 1.80 7.80 4.02 10.40 6.51 4.67 3.73 4.72 3.19 ll.00 6.65 7.00 3.54 6.64 6.77 9.46 8.26 7.41 9.81 2.21 0.66 7.25 8.48 9.48 0.18 3.50 4.39 5.23 4.83 5.84 2.53 0.14 7.19 1.21 0.36 0.62 10.70 6.73 5.44 5.48 8.11 9.57 5.59 2.25 8.64 7.67 3.50 1.63 5.52 4.64

(Continued)

LATE LA TENE POTTERY

67

Table 1. Continued No.

SiO2 TiO2 A1203 Fe203* MnO MgO

BE71 BE72 BE73 BE74 BE75 BE76 BE77 BE 78 BE79 BE 80 BE81 BE82 BE83 BE84 BE85 BE86 BE87 BE88 BE89 BE90 BE91 BE92 BE93

62.84 50.35 60.97 51.35 54.35 53.55 55.99 51.89 62.24 49.25 48.80 56.27 55.34 50.19 59.06 49.81 56.02 53.47 58.64 50.65 51.25 53.08 55.08

0.75 0.96 0.82 0.97 0.95 0.94 0.94 0.88 0.84 0.99 0.98 0.96 0.95 0.98 0.87 0.93 0.95 0.98 0.93 0.99 0.98 0.97 0.95

18.09 21.10 18.41 21.48 20.78 20.43 21.21 19.45 18.57 22.30 21.97 21.55 21.42 21.77 19.71 20.88 21.39 21.64 20.25 21.74 21.69 21.39 21.38

7.60 7.89 7.35 8.42 8.21 8.10 8.48 7.64 7.14 8.42 8.70 8.48 8.34 8.40 7.61 7.99 8.27 8.55 8.20 8.60 8.56 8.22 8.22

0.16 0.19 0.15 0.17 0.18 0.16 0.23 0.15 0.10 0.18 0.21 0.17 0.15 0.17 0.15 0.14 0.14 0.18 0.14 0.19 0.21 0.19 0.15

2.59 2.44 2.76 2.22 2.23 2.27 2.49 2.71 1.68 2.46 2.42 2.96 2.82 2.92 2.25 2.79 2.04 2.56 2.46 2.82 2.81 2.81 2.01

Gen~ve GEl GE2 GE 3 GE4 GE 5 GE6 GE 7 GE 8 GE 9 GE 10 GE 11 GE 12 GE 13 GE 14 GE 16 GE 17 GE 18 GE 19 GE20 GE21 GE22 GE23 GE24 GE 25 GE26 GE27 GE28 GE29 GE 30 GE31 GE 32 GE 33 GE34 GE39 GE40 GE41 GE42 GE43 GE44 GE45 GE46

64.92 66.12 65.81 67.06 64.38 64.63 64.15 64.96 64.54 65.66 64.80 65.15 64.84 64.79 65.07 65.21 64.76 65.65 65.22 64.84 59.17 67.44 66.68 64.60 66.76 67.45 68.34 67.47 65.51 65.29 65.44 67.96 67.06 65.39 65.01 64.39 64.29 60.45 61.76 64.57 64.93

0.79 0.78 0.78 0.72 0.82 0.80 0.81 0.80 0.79 0.80 0.78 0.79 0.79 0.79 0.79 0.77 0.81 0.78 0.80 0.80 0.75 0.74 0.73 0.78 0.77 0.74 0.70 0.74 0.76 0.78 0.76 0.70 0.76 0.80 0.80 0.81 0.80 0.45 0.42 0.83 0.80

17.88 17.35 17.24 16.42 18.31 18.13 18.18 17.91 17.96 17.86 17.71 17.62 18.01 17.90 17.77 17.26 18.18 17.32 17.98 18.03 17.07 16.71 16.50 17.20 16.94 16.77 16.37 16.65 17.04 17.43 17.05 16.30 16.71 17.93 18.14 18.25 18.08 8.90 8.27 18.72 17.41

7.51 7.23 7.33 7.19 7.69 7.61 7.62 7.51 7.46 7.45 7.41 7.40 7.54 7.52 7.47 7.30 7.62 7.30 7.55 7.50 6.75 7.13 7.26 7.21 7.23 7.18 6.95 7.02 7.29 7.38 7.25 6.91 7.15 7.53 7.59 7.54 7.46 3.33 2.99 7.67 7.57

0.19 0.19 0.19 0.18 0.19 0.18 0.18 0.18 0.18 0.18 0.18 0.17 0.18 0.18 0.18 0.17 0.18 0.17 0.18 0.17 0.13 0.17 0.17 0.14 0.18 0.17 0.16 0.17 0.18 0.17 0.20 0.17 0.19 0.18 0.18 0.18 0.17 0.09 0.09 0.16 0.18

CaO

H20-

Na20 K20 P20~

Total

FeO

LOI

1.18 5.88 5.07 3.38 3.00 3.46 2.93 8.53 1.59 4.48 4.20 2.50 3.67 5.25 1.70 6.16 1.64 3.21 1.62 3.14 3.16 3.32 1.92

1.21 0.65 1.09 0.86 0.99 0.99 0.82 0.87 1.16 0.77 0.75 0.88 0.85 0.69 1.09 0.78 0.88 0.85 1.05 0.71 0.72 0.81 0.95

2.02 2.24 3.43 2.66 3.35 3.13 2.98 2.51 2.66 2.70 2.55 2.44 2.59 2.21 3.25 2.41 2.95 2.59 3.54 2.01 2.02 2.09 3.02

3.77 8.21 0.29 8.20 6.19 6.84 3.96 5.17 4.38 8.40 9.43 3.89 4.33 6.92 4.53 7.66 5.80 6.11 3.58 8.97 8.74 6.73 6.85

100.21 99.89 100.35 99.73 100.23 99.86 100.02 99.80 100.37 99.96 100.02 100.10 100.45 99.49 100.22 99.55 100.08 100.14 100.39 99.83 100.12 99.60 100.52

2.76 1.97 5.49 1.73 1.34 1.83 2.10 2.63 1.45 1.78 2.11 3.99 2.98 2.23 1.41 2.16 1.99 2.36 0.95 2.90 2.84 2.92 1.93

3.66 6.51 0.22 6.12 6.05 6.06 4.61 3.24 3.73 7.88 7.57 3.74 4.93 6.27 5.45 5.15 5.70 6.67 5.29 5.93 5.78 5.55 5.69

3.92 7.67 0.14 7.06 6.53 6.12 5.44 4.18 4.03 7.53 8.33 3.66 4.84 5.77 4.20 5.19 5.22 6.49 4.12 6.22 6.09 5.42 6.08

3.34 0.85 3.35 1.01 3.34 0.72 3.18 0.93 3.72 0.87 3.46 0.79 3.60 0.80 3.42 0.79 3.50 0.87 3.50 0.85 3.44 0.85 3.33 1.01 3.53 0.79 3.46 0.78 3 . 4 1 0.83 3.44 0.81 3.39 0.75 3.17 0.92 3.35 0.96 3.30 0.92 2.71 8.87 3.03 0.74 3.01 1.65 3 . 3 1 2.08 3.33 0.81 3.04 0.78 2.79 0.88 2.84 0.82 3.47 0.71 3.26 1.19 3.62 1.12 2.76 0.88 3.21 0.91 3.42 0.82 3.48 0.85 3.39 0.85 3.33 1.08 2.05 21.6 2.18 21.5 3 . 7 1 0.97 3.71 1.68

1.14 1.19 1.25 1.38 1.13 1.14 1.22 1.08 1.17 1.10 1.19 1.10 1.11 1.10 1.22 1.20 1.01 1.17 1.09 1.21 1.01 1.41 1.31 1.27 1.32 1.21 1.23 1.32 1.19 1.17 1.20 1.28 1.29 1.09 1.08 1.21 1.10 1.10 1.35 1.05 1.29

3.24 2.97 3.35 2.85 3.15 3.22 3.27 3.12 3.29 3.18 3.32 3.31 3.17 3.17 3.24 3.12 3.18 3.09 3.21 3.16 3.21 3.09 2.87 3.37 3.05 2.96 2.89 3.00 3.25 3.30 3.07 2.87 3.00 3.12 3.18 3.32 3.30 1.94 1.54 3.19 2.94

0.18 0.18 0.15 0.14 0.20 0.12 0.14 0.11 0.13 0.15 0.14 0.22 0.10 0.10 0.12 0.18 0.10 0.21 0.18 0.14 0.38 0.13 0.30 0.41 0.11 0.13 0.18 0.16 0.15 0.45 0.21 0.17 0.16 0.12 0.11 0.14 0.27 0.46 0.29 0.11 0.31

100.05 100.37 100.16 100.04 100.46 100.07 99.97 99.88 99.88 100.73 99.84 i00.11 100.07 99.80 99.63 99.47 99.98 99.79 100.51 100.07 100.05 100.59 100.49 100.36 100.51 100.43 100.51 100.19 99.54 100.42 99.92 99.99 100.44 100.40 100.42 100.06 99.88 100.39 100.37 100.07 100.82

5.46 5.75 5.62 5.05 4.66 5.87 5.54 4.24 2.58 5.38 4.63 1.74 3.14 3.48 3.95 4.37 4.39 1.75 1.32 3.97 1.39 0.04 0.05 0.10 1.06 3.75 3.93 3.76 3.24 3.06 2.94 3.80 1.73 4.54 3.07 4.87 3.20 0.49 0.29 4.92 4.82

0.02 0.04 0.00 0.21 0.34 0.00 0.00 0.39 0.85 0.35 0.40 1.98 0.51 0.45 0.68 0.18 0.44 1.92 2.07 1.13 2.29 1.00 1.45 2.88 0.45 0.47 1.36 0.99 0.46 2.55 0.77 1.14 0.87 0.50 0.52 0.60 2.49 14.7 14.8 0.34 1.27

0.11 0.10 0.05 0.15 0.19 0.12 0.09 0.20 0.18 0.18 0.17 0.75 0.20 0.19 0.18 0.11 0.19 0.68 0.87 0.16 0.55 0.24 0.27 1.05 0.10 0.20 0.33 0.27 0.20 1.05 0.26 0.32 0.38 0.22 0.17 0.23 0.76 2.16 1.25 0.19 1.27

(Continued)

68

M. MAGGETTI & G. GALE'Iq'I

Table 1. Continued No.

SiO2 TiO2 A1203 Fe203* MnO MgO

CaO

Na20 K20 P_,Os Total

FeO

LOI

GE47 GE 48 GE49 GE 50 GE51 GE52 GE53 GE 54 GE 55 GE56 GE57 GE58 GE59 GE60 GE61 GE62 GE63 GE 64 GE65 GE 66 GE 67

65.09 64.26 65.47 64.79 64.20 66.21 65.39 67.35 64.43 69.27 71.64 66.40 64.09 66.98 66.44 66.12 64.98 66.92 66.78 65.60 63.65

0.88 0.86 0.71 1.02 0.79 0.84 1.02 0.76 1.76 4.65 3.53 0.92 1.28 0.69 0.76 0.94 1.32 1.68 0.98 0.94 1.46

1.18 1.18 1.25 1.04 1.06 1.25 1.14 1.29 1.22 0.94 i.00 1.24 1.06 1.23 1.18 !.08 1.19 1.37 1.28 1.14 0.96

5.23 2.86 5.74 4.88 6.9l 4.25 5.49 3.69 0.20 3.04 3.03 3.69 4.17 5.08 3.93 3.02 3.35 3.88 4.16 5.29 5.65

0.13 0.78 0.00 0.19 0.00 0.28 0.10 0.57 1.14 1.04 1.38 0.51 1.00 0.63 0.33 4.10 0.50 0.51 0.59 0.28 0.36

18.06 18.56 17.51 18.11 18.67 17.35 17.76 16.98 17.49 14.40 13.46 17.13 18.22 17.01 17.19 17.89 17.28 15.35 16.56 17.46 18.25

7.63 7.77 7.39 7.58 7.81 7.36 7.56 7.13 7.37 5.92 5.36 7.41 7.53 7.48 7.50 8.05 7.40 6.65 7.15 7.46 7.52

0.18 0.21 0.19 0.19 0.18 0.17 0.18 0.20 0.18 0.09 0.14 0.18 0.19 0.17 0.17 0.19 0.20 0.17 0.18 0.18 0.17

Grotte du Four GF 1 62.6 1 . 1 5 21.59 GF2 62.98 0.82 19.01 GF 3 69.97 0.72 15.71

7.50 7.54 6.17

0.09 1.67 0.19 3.37 0 . 1 1 2.18

1.86 0.56 3.38 1.07 1 . 3 8 4.29 1.51 1 . 3 5 2.68

0.27 100.70 2 . 8 1 0.26 100.91 3.79 0.17 100.57 4.06

1 . 7 6 0.48 i.58 0.65 0.59 0.21

La T~ne LT 1 68.60 0.78 LT2 69.33 0.81 LT 3 63.74 0.83 LT4 61.07 0.76 LT5 62.58 0.98 LT6 64.68 0.85 LT7 65.06 0.82 LT8 67.15 0.79

15.72 17.58 18.10 16.77 20.99 18.60 18.34 16.78

4.85 4.73 7.79 6.21 5.60 5.32 6.94 6.89

0.03 0.03 0.25 0.09 0.06 0.06 0.10 0.16

1.99 2.31 3.41 2.32 2.73 2.59 2.90 2.60

4.72 1.71 1.80 8.50 3.51 3.53 1.45 1.52

0.99 1.18 0.83 1.12 0.90 1.09 1.45 1.30

2.84 2.5 3.22 3.I2 2.41 3.27 3.23 3.13

0.17 0.13 0.17 0. t9 0.75 0.15 0.17 0.19

100.69 100.31 100.14 100.15 100.51 100.14 100.46 100.51

2.52 3.04 4.45 2.85 2.98 2.61 3.77 3.29

2.85 3.85 0.91 5.45 4.07 1.76 0.84 0.67

0.67 1.03 0.38 1.27 3.42 0.68 0.25 0.31

Marin ME 1 ME2 ME3 ME4 ME5 ME6 ME 7 ME8 ME9 ME 10 ME l l

25.08 21.94 19.60 15.69 15.85 19.52 23.96 23.94 21.66 15.94 21.85

9.77 8.47 7.89 5.53 7.53 8.65 9.67 10.00 9.53 7.45 8.04

0.12 0.20 0.52 0.54 0.52 0.26 0.15 0.23 0.34 0.49 0.15

3.42 2.56 3.08 1.96 2.16 3.14 3.66 2.69 2.97 1.99 2.85

1.16 1.53 5.66 1.39 1.44 5.75 1.28 1.34 1.81 1.32 2.00

0.86 1.35 0.87 i.16 1.13 0.84 1.25 1.26 1.68 1.12 0.94

2.93 3.35 3.50 2.39 2.51 3.60 2.76 3.65 3.87 2.45 2.00

0.90 1.78 0.61 0.52 1.08 0.94 0.58 1.32 2.13 0.52 1.31

100.39 100.50 100.28 100.75 100.92 100.45 100.45 100.87 100.41 100.63 100.64

1.75 1.68 0.42 0.37 0.69 1.02 4.61 1.19 1.77 0.35 3.43

4.33 4.55 7.32 4.57 3.84 7.25 3.67 5.52 4.40 3.96 4.21

5.60 6.68 3.72 3.75 3.47 3.51 6.45 6.61 5.74 3.79 7.64

16.74 19.14 17.07 17.61 18.79 17.99 18.12 17.39 18.78 16.32 18.32 16.01 17.74 15.94 17.63

7.24 7.85 7.38 7.69 7.46 7.60 7.76 7.64 7.28 6.82 6.97 5.97 6.44 4.79 7.10

0.19 0.17 0.22 0.18 0.14 0.17 0.19 0.17 0.14 0.14 0.11 0.05 0.07 0.04 0.17

3.64 3.08 3.72 4.01 3.13 3.23 3.65 3.66 3.50 2.35 3.43 2.14 2.76 !.66 3.56

1.40 2.18 0.99 1.74 7.39 1.65 1.22 1.84 8.75 8.08 9.82 1.82 6.04 5.42 0.93

1.25 0.91 1.11 1.21 0.90 1.07 1.05 1.21 0.86 1.32 0.99 1.40 1.11 0.73 1.16

3.09 3.20 3.03 3.11 3.45 3.02 3.21 2.99 3.49 2.95 3.46 2.75 3.19 2.31 3.22

0.35 1.78 0.17 0.21 1.61 1.33 0.43 0.42 1.37 1.64 0.29 0.18 0.34 1.27 0.18

99.83 100.13 100.51 100.39 100.02 100.01 100.16 100.44 100.21 99.90 99.70 99.78 100.02 99.82 100.41

4.52 3.37 4.20 3.39 2.86 3.93 2.98 3.17 4.09 3.09 1.19 0.21 0.43 0.24 5.09

0.61 3.69 0.33 0.63 2.53 2.23 1.27 1.20 1.13 3.13 0.54 0.62 0.75 4.38 0.47

0.22 1.76 0.09 0.21 1.33 1.18 0.41 0.39 0.67 1.65 0.43 0.39 0.47 1.34 0.39

55.05 58.39 57.71 70.75 67.87 56.92 56.09 55.41 55.50 68.52 60.56

0.82 0.87 0.78 0.81 0.82 0.77 0.80 0.76 0.79 0.73 0.72 0.77 0.83 0.76 0.76 0.84 0.78 0.70 0.76 0.80 0.81

1.10 0.93 0.84 0.82 0.83 0.83 1.05 1.00 0.92 0.83 0.94

St. Triphon-Massongex TR 1 65.17 0.76 TR2 60.98 0.84 TR3 66.04 0.78 TR4 63.83 0.80 TR5 56.37 0.78 TR6 63.16 0.79 TR7 63.72 0 . 8 1 TR8 64.33 0.79 TR9 55.25 0.79 TR 10 59.57 0.71 TR 11 55.53 0.78 TR 12 68.72 0.74 TR 13 61.55 0.78 TR 14 66.84 0.82 TR 15 65.66 0.80

3.67 3.58 3.45 3.62 3.47 3.37 3.76 3.02 3.70 1.82 1.51 3.38 3.75 3.11 3.14 2.61 3.75 3.51 3.39 3.65 3.44

3.09 3.16 3.21 3.10 3.20 3.13 3.02 3.12 2.95 2.31 2.22 2.94 3.09 2.82 2.96 2.52 3.00 2.79 3.01 3.09 3.16

0.15 0.14 0.16 0.16 0.10 0.20 0.16 0.21 0.21 0.53 1.01 0.14 0.28 0.13 0.10 0.23 0.17 0.41 0.30 0.11 0.15

100.74 100.58 100.12 100.42 100.31 100.67 100.79 100.82 100.08 100.66 100.59 100.49 100.31 100.38 100.20 100.46 100.08 99.54 100.40 100.42 99.59

H200.13 0.78 0.00 0.19 0.00 0.28 0.10 0.57 1.14 0.67 0.77 0.21 0.41 0.38 0.18 3.28 0.15 0.23 0.28 0.18 0.21

(Continued)

LATE LA TI~NE POTTERY

69

Table 1. Continued No.

SiO2 TiO2 Al203 Fe203* MnO MgO CaO Na20 K20 P205 Total

FeO

LOI H20-

Yverdon YV 1 YV2 YV3 YV4 YV5 YV6 YV7 YV 8 YV 9 YV 10 YV 11 YV 12 YV 13 YV 14 YV 15 YV 16 YV 17 YV 18 YV 19 YV20 YV21

59.06 59.50 57.52 56.71 65.71 65.95 60.80 65.52 67.46 66.74 58.42 60.55 60.08 62.71 65.60 67.04 61.13 68.07 67.01 67.24 59.46

3.75 3.82 3.30 1.12 2.90 0.91 4.08 2.92 3.85 3.49 3.08 2.17 4.01 4.14 2.54 0.32 0.42 1.46 1.02 2.29 3.93

0.94 1.72 2.41 3.96 0.69 3.67 2.07 0.83 0.99 0.72 1.17 3.08 1.93 0.96 0.68 2.81 2.38 0.62 4.50 4.20 0.56

0.86 0.80 0.86 0.85 0.79 0.7l 0.80 0.79 0.74 0.77 0.87 0.79 0.80 0.84 0.89 0.95 0.81 0.81 0.66 0.79 0.87

*Total iron is givenas

21.59 20.13 21.66 21.93 17.45 15.83 18.98 17.88 16.55 17.55 22.08 18.92 19.02 19.15 18.04 17.12 19.40 18.22 14.80 16.76 20.72

Fe203.

8.10 7.12 8.38 8.22 6.90 6.65 6.98 6.80 7.01 6.32 8.38 8.26 7.13 7.67 7.21 6.12 7.35 5.79 7.00 6.18 8.07

0.15 0.27 0.16 0.36 0.16 0.13 0.18 0.16 0.32 0.18 0.14 0.14 0.29 0.19 0.17 0.27 0.16 0.10 0.08 0.07 0.16

3.62 1.11 1.19 3.14 3.26 1.14 3.60 1 . 5 9 1.11 3.17 1.73 1.06 2.82 1 . 1 9 1.38 2.I9 1.85 1.51 2 . 9 3 3 . 3 5 1.23 2.80 1 . 0 0 1.33 2.83 1.05 1.35 2.63 1.15 1.41 3.76 1.15 1.00 3 . 0 2 1 . 9 0 1.57 2 . 8 5 3 . 3 6 1.21 3.15 1.37 1.13 2.64 1 . 2 9 1.18 1.21 2 . 0 3 0.36 2.94 1 . 8 3 1.38 2.13 1.29 1.08 1.81 2 . 5 2 1.26 2 . 8 8 1 . 4 9 0.87 3.53 1.64 1.35

4.54 4.23 4.55 4.58 3.43 3.42 3.91 3.69 2.82 3.49 4.57 3.80 3.94 3.56 3.30 3.28 4.14 2.79 2.51 3.30 4.37

0.35 0.96 1.09 1.72 0.22 2.17 0.90 0.32 0.51 0.40 0.28 1.69 1.55 0.63 0.19 1.85 1.58 0.34 2.60 0.81 0.14

100.57 100.55 100.52 100.33 100.05 100.4l 100.06 100.29 100.64 100.64 100.65 100.64 100.23 100.40 100.51 100.23 100.72 100.62 100.25 100.39 100.31

0.44 0.87 1.18 2.37 0.35 2.28 0.86 0.27 0.57 0.45 0.27 1.80 1.05 0.51 0.23 1.48 1.47 0.27 2.93 2.30 0.26

LOI, loss on ignition.

is further substantiated by the observation of sandstone fragments as a characteristic nonplastic phase. As marl BE 40 contains such sandstones, it appears that sandstone fragments must be a specific characteristic of the Bernese clays. The non-plastic elements of fabric 2 differ quantitatively from those of fabric 1. Chemically, the samples belonging to this fabric cannot be distinguished from the ceramic population of fabric 1. The samples are not unusual with regard to their archaeological typology, either. Furthermore, the microscopic image does not indicate any temper addition during manufacturing. Consequently, it may be assumed that the clay used to produce these samples originated from a lean layer embedded in the otherwise fat clay deposit of Bern. In conclusion, it may be possible for BE 26, 28, 48, 53 and 71 to have been manufactured either from a local clay with a different composition from that of the fine ceramic main group, or to have been imported to this La T~ne settlement from a geologically similar or different region. BE 68, on the other hand, can be interpreted as being of local manufacture, because its chemical composition corresponds well to that of the marl BE 40. However, to reduce the size of the reference group, BE 26, 28, 48, 53, 68 and 71 are excluded from the Bernese La T~ne reference group, which comprises therefore 79 specimens. Cluster analysis of the reference group (Ward method, not log-transformed data,

squared Euclidean distances, z-scores) shows that the wasters are distributed over the main subgroups and that the two provenances cannot be differentiated. This argues for a single or different local workshops using chemically variable clay deposits.

Selection and processing of the raw materials. BE 40 and BE 52 are marls and therefore cannot be the starting material for the bulk of the fine ceramic. This is because microscopic examination reveals that the CaO occurs as finely distributed calcite in the matrix, and not as coarse particles in a siliceous clay, which could have been eliminated mechanically during treatment. However, the two marls may well have been used in the production of the CaO-rich samples. Raw material BE 40 contains hardly any non-plastic material and completely resembles the CaO-rich ceramics in this respect. This is probably because natural, fatty raw materials were used without much processing in the Bernese La T~ne ceramic. The wide range of CaO, SiO2 and A1203 concentrations shows that the clay deposit was heterogeneous in its chemical composition, and this heterogeneity was not eliminated by specific material preparation. However, as is to be expected, the fluctuation range is significantly lower in single objects than in the whole population. Five analyses of the same specimen (BE 36, 58, 9 0 - 9 2 ) show that the variations of the major and trace

70 (a)

M. MAGGETTI & G. GALETTI 30

,

,

(b)

,

1.2

I

o

53

1.1 0

0

o

% o

°

do.9

~ 2o~ 15

o

o ~OoC~

~8 ~"o.8

0

5

10

©

15

0.7

20

o

o o o

o

o

to 2

0

MgO (~°/o)

CaO (wt%) (c)

(d) 300

,

1

1

,

,

,

500

' 0

071

400

53

250

0 26

A

E t,~ o.. .....- 200

300 0

-

0 48

0

0 0

200 150

o~

100 ~8 0

100 10

I 20

68 Oi 30

t 40

= 50

I 60

I 70

I

80

Y (ppm)

050

1 O0

150

J

200

I

250

300

Ni (ppm)

Fig. 2. Correlation diagrams of selected oxide and element pairs for 85 fine ceramic samples from Bern.

elements, with the exception of Ba and Cr, lie within the normal range. As there appears to be no correlation between archaeological type and chemical composition, a purposeful selection of a particular raw material for specific objects seems unlikely. Genkve

Sixty-two samples were examined, originating either from the town of Genrve (four excavation sites) or from the surrounding area (five excavation sites). They include two clays (GE 43, found adjoining the kiln Four Rue du Cloffre; GE 44, cob from the cathedral excavations) and 60 fine ceramic

A new reference group.

sherds, of which c. 30% are made up of pottery waste from the Four Rue du Cloi'tre kiln, partly consisting of 13 firing waste fragments (GE 1-8, 49-53). The firing waste GE 67 was found in Annecy (France). As can be observed from the A1203-CaO correlation diagram (Fig. 4a), the fragments are mainly CaO-poor (Table 1). In contrast, both clays and the fine ceramic samples GE 22, 56 and 57 are characterized by very high or at least increased CaO concentrations. Considering the SIO2-A1203 diagram (Fig. 4b), all these samples stand out from the densely packed field of the remaining 57 specimens. This is corroborated by the C r - N i correlation diagram (Fig. 4c), where these outliers plot in isolation

LATE LA TI~NE POTTERY (a)

71

(b) 26

r

24-

O0

0

ooO °O8o

20-

0 0

0

~(~)0

o

18-

•

'

oOo o,i1

'

"oo

J

2o

-

_

O

N

18 16

Geneve

O Bern 4O

I

22

0

16

I

24

BE 41

22-

14

26

T

50

60

•

14 0.6

70

I 0.7

I 0.8

I 0.9

I 1.0

Geneve

I O Bern 1.1 1.2 1.3

T i 0 2 (wt%)

Si02 (wt%) (d)

500

I

I

I

I

I

I

(c) 400

26

T

1

l

~~T--©

24

-

o

o

"~300

e~

A

~ --

O..

22

O0

200

20 18

100 16

~

~

j

~

l

• Geneve © Bern

14 6

7

8

9

10

0 11

F e 2 0 3 tot ( w t % )

0

I

I

I

50

100

150

O Geneve © Bern

200

250

3013

Ni ( p p m )

Fig. 3. Correlation diagrams of selected oxide and element pairs for the reference groups of Bern (n = 79, O) and Gen~ve (n = 57, • ). (a), (b) and (c) The Bernese outliers are otherwise not different enough to postulate a foreign origin. The trends suggest the use of illitic-chloritic clays; (d) The Genevan pottery is Cr- and Ni-richer than the Bernese ceramic.

from the bulk. The samples of the subgroup (GE 5 8 - 6 7 ) with slightly lower Cr and higher Ni than the main Genevan body belong to finds from outside the town itself (Annecy, D a r d a g n y Brive, Meinier, Mornex and VandoeuvrePressy). The following conclusions can be drawn: (1) most of the analysed samples were manufactured from CaO-poor clays; (2) neither of the two analysed marls corresponds chemically to the fine ceramic; (3) 57 fine ceramic

samples form a homogeneous, CaO-poor group; (4) three fine ceramic specimens are CaO-rich (GE 22, 56, 57). It now needs to be determined if the CaO-poor specimens form a homogeneous group when examined by themselves, or if the various specimens from the town and its surroundings differ in terms of their chemical composition. A second question arises with regard to the relationship between the three outliers and the 57 samples.

72

M. MAGGETI'I & G. GALE'Iq'I

(a)

(b) 80

20

15

~8 ' @12 @4 @ 56 @

75 57 @

@

o~" 7O

57

64 @

v

0

(:~ 65 =~ ¢,/)

10

@44@

44 60

43

22 55

I

I

I

10

15

43

20

25

5

I

I

10

15

CaO (wt%) (c)

20

A I 2 0 3 (wt%)

500

400

E 300 O.

0

@22

200 56_057

~43

100

44 0

I

0

50

I

I

100 150 Ni (ppm)

I

200

250

Fig. 4. Correlation diagrams of selected oxide and element pairs for fine ceramic samples from Genbve (n = 60) and two clays (GE 43 and 44).

Cluster analysis shows that some samples from outside Gen~ve (GE 5 8 - 6 7 ) group together. This group contains the waster from Annecy. On one hand, this proves that we are dealing with different groups or different local production centres, but, on the other hand, that a similar type of raw material was used in the production of all analysed fragments. Factor analysis corroborates this statement, despite the fact that the sample number is statistically sufficient for the kiln population only. As the number of analysed sherds is high and the main group homogeneous, the outliers cannot be explained by a non-statistically representative sampling of the Genevan production. There is

no evidence of secondary contamination for outliers GE 22, 56 and 57, because their aberrant elements belong to the so-called immobile elements. It remains unclear, however, if the pieces were imported or produced from local clay with a different composition than that of the fine ceramic main group. Because of the presence of marls, GE 22, 56 and 57 may possibly be local, despite the fact that they do not correspond to the raw materials GE 43 and 44. This could be explained by chemical fluctuations in the clay or marl pit. Based on the previous discussion, a local to regional production of La T~ne fine ceramic by more than one workshop in the area of Gen~ve is indicated and allows for the

LATE LA TENE POTTERY definition of a new reference group of 57 CaOpoor samples.

Selection and processing of the raw materials. The reference group is characterized by a lean, silicate matrix containing fine-grained non-plastic elements, i.e. predominantly quartz and rare, but characteristic ultramafic grains (serpentinite and actinolitic or chloritic-actinolitic fels or schist). The maximum diameter of the nonplastic elements is 0.71 mm, but coarse single grains can be as large as 3.99 ram. Both clay and cottage plaster also show a lean, but carbonate matrix containing many silicate and carbonate non-plastic fragments up to 2.65 mm diameter. Given the chemical and microscopic similarities, it is possible that a very CaO-rich raw material similar to GE 43, present in the immediate proximity of the Rue du Clo~tre kiln, was used as cob (GE 44). For the production of fine pottery, however, a different raw material, poor in CaO and of local but still unknown provenance, was employed. Both analysed marls are extremely rich in CaO, and there may have been too many risks associated with their use (lime spalling). A CaO-richer clay than the one employed for the reference group may have been used very sporadically for the production of some fine ceramic objects (GE 22, 56, 57). Similar to the example from Bern, this kind of clay may be an inhomogeneity within the local clay deposit, because the fabric of these three samples is identical to that of the CaO-poor samples. As shown by microscopic analysis, the Genevan fine ceramic is characterized by a homogeneous structure. Its high Cr and Ni concentrations can be linked to the presence of ultramafic non-plastic grains. In the reduced parts of the samples, these have a light brown to beige colouring, whereas in the oxidized parts they are red to auburn. The nonplastic fragments of the Genevan fine ceramic therefore point to a hinterland with acidic, granitic to gneissic, as well as ultramafic rock types. This is compatible with the catchment area of the Rh6ne and Arve rivers and their glaciers, respectively. In selected correlation diagrams, the Genevan products are significantly more homogeneous and, because the use of a lean raw material, markedly poorer in aluminium (less clay minerals) but richer in silicon (more quartz) than the Bernese reference group (Fig. 3a-c). The Bernese raw material may have contained more chlorite (higher TiO2 and FezO3tot concentrations, Fig. 3b and c), but significantly less chromium and nickel (Fig. 3d). All this is supported by the XRD results showing, for lightly fired samples (i.e. below

73

800 °C), the association of quartz, illite, plagioclase and K-feldspar. According to Figure 3a-c, chlorite could be inferred as a further primary constituent in addition to illite. Furthermore, kaolinite cannot be ruled out, because of the proximity of the excavation site to the Jura mountain belt.

Pottery from other sites Grotte du Four The three analysed fine ceramic specimens belong to the CaO-poor ceramic (Fig. 5a). They vary very significantly in their A1203 concentrations and differ in matrix as well as appearance of their non-plastic elements. GF 1 has a fatty matrix (it has the highest A1203 value) and stands out because of the use of grog, whereas GF 3 has a lean clay similar to GF 2. It contains more and coarser non-plastic elements than the latter (consequently less aluminium). The use of at least two clays (fat and lean) therefore appears likely, whereby an admixture of temper to GF 3 is probable, considering the hiatal structure as well the presence of grog.

La TOne Of the eight analysed fine ceramic specimens, four are very poor in CaO (LT 2, 3, 7 and 8), three poor in CaO (LT 1, 5, 6), and only one, LT 4, is rich in CaO (Fig. 5a). LT 3 stands out from the rest of the CaO-poor group because of its increased Fe203tot, MgO, Cr and Ni concentrations (Tables 1 and 2). There is no evidence of a relationship between typology and chemical composition.

Marin The majority of the 11 analysed fine ceramic specimens are CaO-poor, except for ME 3 and 6 (Fig. 5b). The CaO-poor specimens show, for instance in their A1203 content, a twofold clustering. All these features can be interpreted as evidence for the use of three different clay sources.

Saint-Triphon and Massongex Of the 15 analysed fine ceramic samples, TR 1-14 were found at Saint-Triphon and TR 15 was found at Massongex. Six of them (TR 5, 9, 10, 11, 13, 14) are CaO-rich (Fig. 5c) and show, in contrast to the nine CaO-poor samples, low Cr and Ni concentrations (Table 2). Sample TR 12 differs from the other CaO-poor specimens by its low Cr and Ni contents. Microscopic analyses reveal that the samples were manufactured from two different

74

M. MAGGETTI & G. GALE'Iq'I

(a) (b)

25

+

25

+ 23

23

-h4A

1

21

0 i

O4

19 -

<¢ 17 -

15

7~]3 []2

[]6 []

(3

19

4

[] 1

3 • I

2

3,6

17

[] I I 4 6 CaO (wt%)

= 8

-~ 4,5,10 15 0

(c)

= 2

0

1 I 4 6 CaO (wt%)

10

(d)

25

25

23

23

°•"

A

•~

21

5

•8

0

~

21

21

X

v

•

0

5

O3

9

~.o , I

19

19

13

/ 11~ !

17

•13

0

17 10

n14

•

15

X 2

~7

0

I

I

I

I

2

4

6

8

10

15

0

CaO (wt%)

2

I

I

I

4

6

8

10

CaO (wt%)

Fig. 5. AI203-CaO correlation diagrams of fine ceramic samples from (a) Grotte du Four (n = 3, II) and La T~ne (n = 8, [~); (b) Matin (n = 11); (c) Saint-Triphon-Massongex (n = 15); (d) Yverdon (n = 21).

raw materials, a lean and a fatty one, as evidenced by two fabrics. Fabric 1. Lean, silicate matrix. The silicate non-plastic elements are characteristic serpentinite grains, with a diameter of up to 0.7 mm, and coarse single grains up to 5.6 mm (TR 1, 3, 4, 8, 12, 13). TR 10 and 11 show hardly any non-plastic elements. Fabric 2. Fat, silicate matrix. Few silicate nonplastic elements occur as characteristic serpentinite grains with a maximum diameter of 1 mm and coarse single grains up to 2.1 mm (TR 2, 5, 6 , 7 , 9 , 14). Under the microscope, the samples appear very heterogeneous, with grain size and temper content varying significantly. The high Cr and

Ni values detected can be correlated with the serpentinite grains. XRD analysis reveals that the three fabrics are correlated with ill•tic raw materials. Chlorite may have been present initially, but would have been destroyed during firing. It appears therefore that the clays were initially very inhomogeneous, an assumption that is substantiated by varying CaO values. Yverdon With the exception of a few samples (YV 2, 7, 13), most of the 21 analysed fine ceramic objects are very poor in CaO. YV 9 has increased Cr and Ni values and thus stands out from the rest of the ware, which seems homogeneous in its

LATE LA TI~NE POTTERY

75

Table 2. Chemical analyses (ppm) No.

Ba

Cr

Cu

Ga

Nb

Ni

Pb

Rb

Sr

Th

V

Y

Zn

Zr

634 775 1023 638 834 832 701 965 834 1037 901 978 1356 1174 1065 1694 1676 1918 1571 967 961 1236 1272 1250 1356 1548 1010 1217 1113 1575 1577 1213 1587 521 370 1153 1347 1453 520 885 1107 913 1128 1501 905 267 802 1080 511 450 1220 961 864 848 1107 1789 957

168 147 185 188 166 174 187 209 238 249 194 189 170 203 195 198 118 143 129 125 480 187 166 192 170 170 136 191 211 204 206 209 187 165 110 159 194 215 191 177 207 175 285 184 129 73 165 167 152 132 184 163 169 161 165 149 171

26 26 28 27 21 16 33 23 25 26 29 41 88 25 2540 35 60 25 32 26 29 152 24 36 23 26 40 23 34 568 30 29 18 24 13 15 27 23 39 27 33 31 57 40 40 21 18 35 23 17 25 238 30 17 26 24 20

24 19 25 28 20 20 24 28 34 27 25 27 20 21 24 22 23 24 26 25 24 22 10 29 22 22 20 24 27 22 26 26 23 24 10 27 25 25 29 30 25 22 24 26 16 6 14 21 20 15 20 20 22 20 t9 13 18

19 18 18 19 21 17 22 20 21 15 23 19 19 17 20 21 21 21 21 19 18 18 7 20 19 20 15 23 20 20 22 22 21 22 16 17 22 18 19 18 19 22 16 21 19 13 18 22 18 15 19 21 19 20 21 17 19

100 80 94 96 86 82 97 105 120 133 108 95 84 99 102 101 98 107 104 98 271 103 75 118 98 97 94 108 118 103 115 108 92 106 59 97 99 97 107 92 93 85 236 98 79 52 67 111 100 88 103 105 105 98 11I 97 103

18 15 19 23 21 14 13 23 30 16 18 27 31 27 37 30 34 35 17 21 30 29 0 19 8 29 0 15 19 28 20 20 29 8 2 0 37 9 23 27 32 35 8 33 20 0 0 0 0 0 8 0 0 3 0 0 0

146 121 95 121 113 101 100 80 116 83 65 92 127 114 146 63 135 62 134 142 56 124 93 70 62 113 89 78 108 57 56 80 86 107 94 72 74 59 179 95 157 93 142 107 124 99 57 169 178 154 61 71 90 118 82 41 68

134 393 427 87 320 366 132 270 303 226 323 356 177 225 137 264 256 334 303 280 150 201 132 405 297 189 328 276 227 308 305 261 303 165 288 393 272 325 117 175 258 216 288 191 147 325 172 145 174 233 330 251 242 302 277 414 221

5 3 0 8 15 8 6 12 8 0 6 5 24 15 22 18 22 23 9 2 20 22 0 0 1 22 0 6 6 22 1 8 27 5 1 0 27 0 6 11 23 26 0 26 23 0 7 7 4 0 9 12 8 10 5 1 5

150 106 117 192 109 130 121 163 182 109 163 169 156 157 183 147 140 119 166 149 183 146 144 135 107 156 136 155 188 157 151 171 83 128 70 86 151 112 161 163 143 107 176 106 128 85 89 179 162 134 113 166 99 165 111 155 152

43 34 41 42 39 45 42 45 51 40 54 38 35 47 38 43 47 49 45 42 32 46 17 46 43 41 39 49 51 45 53 49 39 46 28 36 41 44 46 44 38 39 34 47 39 25 70 36 44 39 44 51 50 48 52 38 48

114 104 121 129 116 110 112 149 165 130 142 117 149 112 137 135 164 152 150 154 138 132 109 160 108 139 101 134 144 148 146 130 125 119 84 119 118 124 118 134 173 124 137 118 123 72 50 138 94 78 113 124 153 153 140 100 127

166 135 151 179 156 192 191 192 175 149 197 161 146 145 146 147 151 151 158 188 199 139 111 157 156 144 143 168 163 150 176 169 149 155 134 149 152 156 147 182 152 181 164 160 162 103 263 151 133 128 145 161 174 176 163 139 166

Bern BE BE BE BE BE BE BE BE BE BE BE BE BE BE BE BE BE BE BE BE BE BE BE BE BE BE BE BE BE BE BE BE BE BE BE BE BE BE BE BE BE BE BE BE BE BE BE BE BE BE BE BE BE BE BE BE BE

6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 52 53 54 55 56 57 58 59 60 61 62 63

(Continued)

76

M. MAGGETTI & G. GALETI'I

Table 2. Continued No. BE BE BE BE BE BE BE BE BE BE BE BE BE BE BE BE BE BE BE BE BE BE BE BE BE BE BE BE BE BE

64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93

Genbve GE 1 GE 2 GE 3 GE 4 GE 5 GE 6 GE 7 GE 8 GE 9 GE I0 GE 11 GE 12 GE 13 GE 14 GE 16 GE 17 GE 18 GE 19 GE 20 GE 21 GE 22 GE 23 GE 24 GE 25 GE 26 GE 27

Ba

Cr

Cu

Ga

Nb

Ni

Pb

Rb

Sr

Th

V

Y

Zn

Zr

1264 935 1132 1172 579 900 1433 784 1228 462 1205 919 1513 714 1031 896 1447 1424 792 965 1047 1301 1868 1079 1091 1053 1294 1268 992 1237

147 151 151 164 87 139 175 420 166 155 171 178 121 131 105 111 179 188 173 175 170 155 146 172 163 156 151 159 159 152

57 20 221 34 19 23 26 24 26 34 218 24 22 23 23 33 34 33 20 23 17 345 22 88 20 23 38 37 1302 22

19 16 16 18 6 13 17 17 20 23 21 21 16 20 18 18 20 18 21 22 19 21 15 20 19 18 19 20 20 19

20 16 20 18 14 17 13 14 21 17 19 18 17 19 18 20 24 19 19 21 22 19 14 22 21 22 17 27 24 20

108 90 95 102 73 87 98 251 103 102 108 109 93 106 98 86 106 98 115 111 112 110 96 106 106 108 100 114 113 109

0 5 0 0 0 0 0 I 0 0 0 3 1 0 0 0 0 0 3 0 0 0 0 0 0 0 4 0 0 2

166 80 71 149 121 120 126 69 73 175 100 143 113 125 79 113 101 91 96 96 68 154 51 114 95 166 47 61 66 123

177 321 282 182 349 306 233 139 302 145 316 302 289 158 294 171 281 296 183 190 336 241 278 201 220 194 301 350 283 216

5 9 8 9 2 3 2 4 2 1 8 7 9 7 3 7 3 5 3 6 5 6 7 7 8 5 9 6 5 8

179 115 154 178 106 120 138 181 140 156 147 126 141 161 144 127 127 149 194 166 131 174 133 192 159 143 154 164 164 182

37 39 45 43 27 37 32 35 44 46 47 44 34 46 43 38 44 40 53 49 51 48 39 49 47 50 42 57 54 48

212 120 124 151 106 126 161 106 116 116 148 138 138 146 99 101 150 161 134 112 144 161 117 135 137 135 126 159 142 129

142 140 179 152 103 141 133 140 152 138 155 150 135 146 141 170 151 151 161 150 156 144 136 163 156 152 149 171 169 159

445 494 455 371 505 438 451 437 471 446 458 533 453 449 460 424 451 494 506 479 463 391 370 513 465 366

377 420 357 384 423 361 368 387 393 374 365 352 363 382 368 386 376 376 392 367 212 376 353 381 347 364

41 50 45 29 44 38 50 34 33 25 43 47 43 41 38 41 43 41 40 36 35 51 51 48 47 25

20 19 20 19 20 22 22 21 20 22 21 2l 22 21 22 21 20 19 20 21 17 19 18 19 19 18

13 14 14 14 14 19 16 16 16 16 16 16 15 17 14 16 13 16 16 17 12 13 15 16 14 15

201 183 204 205 210 204 210 210 208 201 206 200 209 208 206 206 207 198 207 204 136 184 196 190 205 187

6 7 8 9 3 9 8 8 6 13 11 10 9 12 8 13 6 6 2 11 2 7 10 9 10 10

156 151 149 133 158 158 161 157 158 152 156 152 153 157 154 151 156 147 155 157 149 140 133 143 145 137

72 79 67 77 82 73 74 76 76 80 80 82 74 75 76 74 74 80 91 81 235 73 86 156 71 74

2 3 5 9 4 12 11 7 3 9 9 7 9 10 9 11 6 10 0 12 4 7 10 7 2 7

180 167 172 159 192 176 182 182 190 I84 177 193 192 179 186 171 187 191 186 197 145 157 165 169 172 161

29 35 31 31 29 33 32 30 29 32 32 29 32 32 27 32 29 28 27 30 30 25 29 31 28 29

ll6 124 114 104 187 118 121 120 121 123 119 123 115 120 117 118 119 118 119 120 97 104 107 125 117 113

159 168 168 166 160 165 166 170 164 167 163 160 162 169 165 175 160 163 159 163 143 166 174 153 165 169

(Continued)

LATE LA TI~NE POTTERY

77

Table 2. Continued No.

Ba

Cr

Cu

Ga

Nb

Ni

Pb

Rb

Sr

Th

V

Y

Zn

Zr

GE 28 GE 29 GE 30 GE 31 GE 32 GE 33 GE 34 GE 39 GE 40 GE 41 GE 42 GE 43 GE 44 GE 45 GE 46 GE 47 GE 48 GE 49 GE 50 GE 51 GE 52 GE 53 GE 54 GE 55 GE 56 GE 57 GE 58 GE 59 GE 60 GE 61 GE 62 GE 63 GE64 GE 65 GE 66 GE 67

430 390 425 532 473 430 456 454 455 482 499 206 185 395 479 503 496 452 510 471 434 485 486 517 371 499 448 505 406 385 496 502 425 467 459 399

387 364 385 372 420 365 351 370 386 377 366 107 95 351 438 427 411 379 423 345 404 401 373 475 140 122 325 302 321 306 332 342 299 329 304 316

40 33 37 38 38 35 33 34 37 39 41 22 25 40 49 49 42 46 49 53 36 42 39 53 17 13 33 35 34 30 32 37 23 35 35 35

19 19 19 19 20 17 19 21 22 22 21 6 5 23 20 22 21 22 23 23 20 21 19 20 10 9 17 18 19 18 17 17 14 16 17 19

13 15 15 13 16 13 15 15 17 14 16 8 8 17 14 18 15 16 17 16 15 13 14 14 7 7 11 12 14 12 17 15 14 13 16 14

174 174 213 196 219 176 200 202 210 202 204 55 50 196 226 212 209 217 198 211 206 211 195 215 66 55 225 228 234 220 220 246 224 229 235 244

6 11 6 9 10 8 8 12 6 6 7 0 0 7 18 8 6 13 13 11 11 7 11 5 0 0 1 0 0 0 0 0 0 0 0 0

133 143 140 151 146 133 139 154 153 160 161 58 50 161 144 157 156 154 160 163 149 146 142 147 94 87 125 138 135 140 108 146 126 141 153 159

79 77 67 82 82 80 80 76 74 78 94 301 301 73 90 73 81 71 81 71 72 75 80 95 149 173 70 89 63 65 66 80 92 83 76 85

14 12 6 10 14 7 11 13 8 2 7 2 1 10 6 9 5 19 10 9 8 7 9 1 0 1 7 3 2 3 6 2 7 2 4 3

164 171 170 189 178 162 178 180 192 184 191 39 42 188 177 190 191 184 183 189 177 182 171 175 130 123 166 182 180 168 188 175 146 171 174 172

22 21 27 25 32 23 29 29 31 29 30 21 17 30 28 35 28 32 32 30 28 31 32 30 22 23 27 28 27 25 32 30 32 31 30 35

101 108 108 114 118 101 117 117 122 126 129 110 48 118 119 125 117 120 124 121 111 117 117 119 71 74 97 114 101 100 109 110 105 111 115 110

151 159 164 158 161 153 167 166 168 160 160 155 145 160 151 167 172 174 166 162 170 159 161 157 164 169 147 145 151 148 170 145 164 160 151 145

GroneduFour GF 1 430 GF 2 660 GF 3 735

152 137 183

27 46 26

24 25 19

19 16 15

63 83 98

25 34 14

133 172 137

123 74 93

19 18 8

180 161 145

35 32 28

117 120 104

218 150 198

LaTkne LT 1 LT 2 LT 3 LT 4 LT 5 LT 6 LT 7 LT 8

467 804 488 681 593 667 555 505

121 171 380 120 170 146 168 185

25 28 53 27 33 32 28 26

14 17 20 17 20 19 18 16

18 19 15 17 21 18 17 18

62 90 379 77 100 86 93 116

7 10 1 0 10 9 2 2

141 116 149 140 96 157 159 139

152 111 75 178 121 120 86 93

18 24 15 15 28 19 17 15

135 166 196 132 174 157 164 151

30 30 35 35 38 31 38 40

99 102 106 97 119 108 116 98

187 171 146 140 182 156 161 168

642 1128 1293 1321 1637

212 208 122 206 175

27 23 35 29 19

26 22 17 17 13

22 19 14 17 16

115 122 119 121 131

30 4 0 35 0

97 125 149 101 104

85 106 123 94 105

39 19 7 24 10

146 177 166 141 166

38 46 31 31 38

149 128 136 116 104

161 168 134 227 206

Ma Hn ME 1 ME 2 ME 3 ME 4 ME 5

(Continued)

78

M. MAGGETTI & G. GALE'Iq'I

Table 2. Continued No. ME ME ME ME ME ME

6 7 8 9 10 11

Ba

Cr

Cu

Ga

Nb

Ni

Pb

Rb

Sr

Th

V

Y

Zn

Zr

1053 427 671 1480 1144 591

121 173 199 107 176 153

37 20 28 21 20 25

17 29 23 24 15 21

16 23 20 19 18 21

96 112 147 79 140 81

0 18 16 16 9 12

151 83 124 168 103 56

132 75 82 139 80 95

2 21 9 10 25 23

156 209 154 165 161 202

41 65 60 48 33 44

ll8 150 146 149 102 114

138 175 161 141 207 191

St. Triphon-Massongex TR 1 TR2 TR 3 TR 4 TR 5 TR 6 TR 7 TR 8 TR 9 TR 10 TR 11 TR 12 TR 13 TR 14 TR 15

500 818 456 473 717 795 648 577 627 923 545 643 607 638 519

323 283 311 325 144 326 333 328 137 141 136 185 170 129 353

42 39 42 46 34 36 34 39 29 27 35 28 32 26 43

19 22 21 19 21 22 22 20 21 17 22 19 20 16 19

12 12 17 15 14 15 15 12 14 15 15 14 15 11 17

232 197 222 244 88 229 234 224 90 81 91 96 101 72 247

8 7 15 7 9 9 7 2 14 7 7 5 0 14 0

142 157 149 147 158 147 153 140 166 135 163 149 165 113 157

103 306 85 98 422 242 122 121 314 388 340 131 216 243 80

7 8 7 0 2 6 5 5 5 5 5 3 1 11 0

167 187 170 178 151 175 192 173 156 125 143 135 138 126 179

28 30 31 29 25 27 28 27 30 33 30 29 26 25 33

126 156 124 118 144 134 134 129 146 120 127 107 120 118 126

159 144 169 160 135 151 163 159 124 138 136 203 160 145 156

760 882 946 1194 561 1192 917 60I 711 605 724 1004 1027 842 562 1085 1149 604 1552 990 614

144 139 153 151 125 108 133 131 318 125 154 143 134 212 166 128 129 139 177 139 129

28 29 25 30 26 27 25 35 26 29 29 31 22 38 31 21 47 27 36 31 23

27 24 27 25 22 17 21 22 19 20 27 22 21 22 23 19 22 21 17 19 25

16 17 14 14 14 14 15 |6 12 15 15 13 15 9 18 19 8 17 12 15 13

86 85 94 94 75 70 82 81 205 80 94 84 81 102 104 69 67 77 109 78 76

16 21 18 16 12 15 23 15 8 18 15 18 24 20 13 18 31 18 12 14 16

215 188 208 210 153 142 176 161 134 156 215 174 173 161 167 120 168 158 121 145 195

93 177 182 234 82 236 180 83 116 95 76 249 237 123 98 251 189 125 289 179 70

11 13 10 9 6 13 12 10 5 11 6 9 9 11 3 5 12 12 12 11 0

192 169 199 192 144 143 164 161 177 152 202 160 166 176 170 136 149 171 151 172 160

32 35 34 24 33 25 38 33 23 36 32 27 34 30 29 53 21 31 33 33 30

151 146 176 220 121 128 143 123 130 117 144 147 164 132 116 114 165 130 170 114 128

|29 138 127 115 171 177 150 173 154 181 125 139 146 132 168 286 129 154 180 175 134

Yverdon YV 1 YV 2 YV 3 YV4 YV5 YV 6 YV7 YV 8 YV 9 YV 10 YV 11 YV 12 YV 13 YV 14 YV 15 YV 16 YV 17 YV 18 YV 19 YV 20 YV 21

chemical composition. Microscopically, two fabric types can be identified, indicating the use o f two different clays. Fabric 1. Lean, silicate matrix. Silicate non-plastic e l e m e n t s occur, with a m a x i m u m d i a m e t e r o f 0.9 m m and as single coarse single grains up to 1.2-2.1 m m (YV 5, 8, 9, 10, 15, 16, 18, 19, 20 and 21). Fabric 2. Fat, silicate matrix. Silicate, as well as carbonate (YV 2, 7, 13) non-plastic e l e m e n t s

occur, with a m a x i m u m d i a m e t e r o f 1 m m (YV 1, 2, 3, 4, 6, 7, I l, 12, 13, 14, 17). As expected, the samples belonging to fabric 1 are mostly poorer in a l u m i n i u m than those o f fabric 2. Microscopic analysis reveals large fluctuations in nonplastic e l e m e n t s and carbonate contents, as well as in granulometry. A c c o r d i n g to X R D results, the use o f illitic clays is most likely. Chlorite m a y have been present and was d e s t r o y e d during firing.

LATE LA TENE P O T T E R Y

Discussion and conclusions The chemical analysis has revealed clearly that the majority of the sherds are CaO-poor. A further chemical and geographical subdivision can be obtained based on the chromium and nickel concentrations. The studied 70 samples from the Rh6ne Valley and Gen~ve, i.e. the products from Gen~ve (without GE 22, 56, 57), Saint-Triphon-Massongex (without TR 12), as well as the samples BE 26, 48, 71, LT 3 and YV 9, are distinguishable from the rest of the ceramic because of their high Cr and Ni values. The question is therefore whether in the Rh6ne Valley-Lake Geneva corridor ceramic was produced in several places or exclusively in one place, i.e. Gen~ve, and if ceramic was exported from this region to Bern, La T~ne, Saint-Triphon-Massongex and Yverdon. A similar question can be asked for the Cr- and Ni-poor pottery, concentrated in the region between the three lakes and Bern: is there evidence of imports from Bern or of several production centres?

Cr- and Ni-rich pottery: ceramic import from Genkve to Bern, La Tkne, Saint-TriphonMassongex and Yverdon ? According to the factor analysis, the specimens from Bern, La Tbne and Yverdon differ from the Genevan population, whereas the samples from Saint-Triphon-Massongex can be integrated into this group, with the exception of TR 2, 6, 7 and 8 (Fig. 6). Mahalanobis distances of the Genevan reference group (n = 57), calculated by discriminant analysis, range from 0.000 to 4.579. The distances of the Saint-Triphon-Massongex specimens TR 1, 3, 4 and 15, when

79

compared with the centroid of the Genevan group, range from 0.005 to 1.118. This argues for a provenance of these objects from the Genevan region. Microscopic analysis provides further evidence of such a origin, because the thin-section specimens TR 1, 3 and 4 show a lean matrix with serpentinite non-plastic elements (but not the actinolitic or chlotiticactinolitic fragments characteristic of some Genevan samples). Because of their different chemical composition, the other Cr- and Ni-rich samples, BE 26, 48, 71, LT 3 TR 2, 6, 7, 8 and YV 9, cannot be attributed to the Genevan area and probably belong to different production sites. Their provenance remains unknown, because the known Swiss reference groups are all Cr- and Ni-poor and cannot be used as a tool for comparison. However, based on the microscopic analysis (sandstone fragment) it is possible that BE 26 originates from Bern. This sample would consequently have been manufactured from local clay, rich in Cr and Ni.

Cr- and Ni-poor pottery: one or several workshops? Discriminant analysis of 130 Cr- and Ni-poor fine ceramic La T~ne products shows, apart from a wide scatter, a relatively clear separate grouping of the 79 Bernese reference samples from the specimens from Genbve, Matin, Yverdon and the other sites (Fig. 7). Attempts to solve the overlap of the samples from La T~ne, Grotte du Four and Saint-TriphonMassongex on the Yverdon scatter are prevented by the wide spread of the Yverdon samples, which rules out the possibility of establishing a statistically valid reference group of this pottery

BEo48

TR=2 "(b -

2

o 1 --

,o

BE 71 O

t~_0

• %-• TRs• • omO=I~ZiI~= ot °

0

e,.

Y~°

u.I

+¢÷ oo R~o ,.,.9 o o~O 7 ~ oO

T..7

. .'.,"

l--

--2

liB ~

I

• ..-z

n3 -

t2 11 1 0 REGR factor score 2

n 1

--6

+ Marin

41.°

• Grotte aux E •

Geneve

O Bern

Fig. 6. Factor analysis of 70 Cr- + Ni-rich potteries from five sites in western Switzerland.

--8 --4

St Triphon/Mass.

a La Tene

Y~7

St Triphon/Mass.

[ ] La T~ne

ill

-3

•~ Yverdon •

YV 14 -,X-

X Yverdon •

•

-I-

•

01

-1

-2

+ +

I

I

I

I

-2

0

2

4

Geneve

O Bern

Discriminant Scores from Function 2

Fig. 7. Discriminant analysis of 130 Cr- + Ni-poor potteries from seven sites in western Switzerland.

80

M. MAGGETTI & G. GALETFI

production. Only GF 2 is chemically similar enough to the majority of the Yverdon samples for a provenance from this region to be likely. Summarizing, the chemical patterns of these ceramics samples argue for geographically distinct pottery workshops. These cannot be described in more detail because of a lack of analyses from the studied sites as well as of clearly defined CaO-poor reference groups in this region. Conclusion In conclusion, most of the samples from each site differ in their chemical composition from the pottery of the other sites. It is therefore likely that the Late La T6ne ceramic of western Switzerland correspond to one or several local pottery productions, as was demonstrated by Maggetti & Galetti (1981) and Maggetti et al. (1988) for the Late La T/me ceramic production in northwestern Switzerland. The number of sherds analysed from the sites of Bern and Gen~ve is statistically sufficient; consequently, two new Swiss reference groups can be defined. We thank Ph. Curdy, G. Kaenel, P. Kohler and D. Paunier for their help in providing the material and for helpful discussions. References

ARNOLD, B. 1992. Le site hallstattien et l'enceinte quadrangulaire lat6nienne de Marin-Les Bourguignonnes (canton de Neuchhtel). Cahiers d'archdologie romande, 57, Lausanne, 309-315. BACHER, R. 1989. Bern-Engemeistergut 1983. Staatlicher Lehrmittelverlag, Bern. BONNET, C. 1997. Le d~veloppement urbain jusqu'au dfbut du Bas empire. La Gen~ve sur l'eau. Monuments d'art et d'histoire du Canton de Genbve, 1, 24-30. BRUNETTI, C. 2005. L'oppidum d'Yverdon-les-Bains au ler si~cle av. J.-C. Cahiers d'archdologie romande, 101, Lausanne, 19-27. COLLOMB, P. & MAGGETTL M. 1996. Dissolution des phosphates presents dans des c~ramiques contamin6es. Revue d'Archdomdtrie, 20, 69-75. CURDY, P., KAENEL, G. & ROSSl, F. 1992. Yverdonles-Bains (canton de Vaud) h la fin du Second age du Fer: nouveaux acquis. In: L'6ge du Fer dans le Jura, 285-300. CURDY, P., BESSE, M. & MARIETHOZ, F. 1997. Le rituel fun~raire en territoire s6dune (fin du 2~me fige du fer). Nouveaux acquis. Bulletin d'dtudes prdhistoriques et archgologiques alpines, 5 - 6 (1994-1995), Soci6t6 Vald6taine de pr6histoire et d'arch~ologie, Aoste, 169-187. CURDY,P., FLUTSCH,L., MOULIN,B. & SCHNEITER,A. 1995. Eburodunum vu de profil: coupe

stratigraphique /l Yverdon-les-Bains VD, Parc Piguet, 1992. Jahrbuch Schweizerische Gesellschaft fiir Ur- und Friihgeschichte, 78, 7-56. EGLOFF, M. 1991. L'artisanat celtique d'apr6s les trouvailles de La T~ne. Celti 1991, 369-371. HALDIMANN, M.-A., CURDY, PH., GILLIOZ, P.-A., KAENEL, G. & WIBLI~, F. 1991. Aux origines de Massongex VS. Jahrbuch Schweizerische Gesellschaft fffr Ur- und Friihgeschichte, 74, 129-182. KAENEL, G. 1990. Recherches sur la p6riode de La T/me en Suisse occidentale. Analyse des s6pultures. Cahiers d'archdologie romande, Lausanne, 50. KAENEL, G. 1991. La Grotte du Four (Boudry, canton de Neuchfitel): "temple helv~te" ou habitat-refuge de la fin de La T/me? Les Celtes dans le Jura, MusSes de Pontarlier, Yverdon-les-Bains, Lonsle-Saunier et Lausanne, Lausanne, 111 - 113. KOHLER, P. 1991. Bern-Heiligkreuzkirche. Eine mittel- und spiitlatbnezeitliche Siedlungsstelle. Diploma thesis, University of Bern. KORITNIG, S. 1978. Phosphorus. hT: WEDEPOHL,K. H. (ed.) Handbook of Geochemistr)." 11-2. Springer, Berlin, 15 k 1-15 k 5. LANGE, B. & VEJDELEK, Z. J. 1980. Photometrische Analyse. Verlag Chemie, Altenberg. MAGGETTI, M. & GALETTI, G. 1981. Arch~iometrische Untersuchungen an sp/itlatenezeitlicher Keramik von BaseI-Gasfabrik und Sissach-Briihl. ArchiioIogisches Korrespondenzblan, 11(4), 321-328. MAGGETTi, M., GALETTI, G. & SCHNEUWLY, R. 1988. Die sp/itlat6nezeitliche feinkeramische Referenzgruppe Sissach-Bri.ihl. Archiiologie und Museum, Liestal, 13. MAGGETTI, M., GALETTI, G. & PAUNIER, D. 1989. R6ntgenographische Phasenanalyse schweizerischer antiker Keramik. In: SCHWEIZER, F. & VILLIGER, V. (eds) Methoden zur Erhaltung yon Kulturgiitern, P. Haupt, Bern, 209-214. MULLER, F. 1990. Der Massenfund vonder Tiefenau bei Bern. Antiqua 20. MI~ILLER, F. 1996. Lat~nezeitliche Grabkeramik aus dem Berner Aaretal. Jahrbuch Schweizerische Gesellschaft fiir Ur- und Frtihgeschichte, 76, 43 -66. MI~LLER-BECK, HJ. 1963-1964. Die Erforschung der Engehalbinsel in Bern bis zum Jahre 1965. Jahrbuch Bernisches Historisches Museum, 43-44, 375 -400. PAUNIER, O. 1981. La c6ramique gallo-romaine de Gen~ve. Mdmoires et Documents, X. SCHNEIDER, G. 1989. Chemische Zusammensetzung. In" SCHNEIDER, G. (ed.) Naturwissenschaftliche Kriterien und Verfahren zur Beschreibung von Keramik. Acta Praehistorica et Archaeologica, 21, 28-29. SCHWAB, H. 1990. Les Celtes sur la Broye et la Thielle. Arch~ologie de la 2e correction des eaux du Jura, 1989. (Arch~ologie fribourgeoise, 5). Editions Universitaires, Fribourg. ST,~.HLI, B. 1977. Die Latbnegriiber yon Bern-Stadt. Schriften Seminar fiir Urgeschichte, Universit/it Bern.

Characterization of Maltese pottery of the Late Neolithic, Bronze Age and Punic Period by neutron activation analysis H. M O M M S E N 1, A. B O N A N N O 2, K. C H E T C U T I B O N A V I T A 2, I. K A K O U L L I 3, M. M U S U M E C I 4, C. S A G O N A 5, A. S C H W E D T 1'6, N. C. V E L L A 2 & N. Z A C H A R I A S 7

1Helmholtz-Institut fiir Strahlen- und Kernphysik, University of Bonn, Nussallee 14-16, 53115 Bonn, Germany (e-mail: [email protected]) 2Department of Classics and Archaeology, University of Malta, Tal-Qroqq MSD 06, Malta 3The UCLA/ Getty Archaeological and Ethnographic Conservation Program, A410 Fowler, Los Angeles, CA 90095-1510, USA 4Academic Division, MATSEC Support Unit, University of Malta, Tal-Qroqq, Msida MSD 06, Malta 5School of Fine Arts, Classical Studies and Archaeology, University of Melbourne, Melbourne, Vic. 3010, Australia 6present address: Equip de Recerca Arqueomktrica de la Universitat de Barcelona, Departament de Prehistbria, Histbria Antiga i Arqueologia, C/de Baldiri Reixac, s/n, 08028 Barcelona, Spain 7Laboratory of Archaeometry, Institute of Materials Science, NCSR Demokritos, Aghia Paraskevi, 15310 Attiki, Greece Abstract: A set of 41 samples from Tas-Silg, Malta, has been analysed by neutron activation. It contained nine ware groups formed by visual examination covering the Late Neolithic, Bronze Age and Punic Periods (c. 3000-218 Be). Despite this diversity and long time range, seven of these ware groups, including the 'Thermi Ware', all have a similar chemical composition and, therefore, have been made from the same clay. This points most probably to a local origin. One group from the Punic Period, containing only Bricky Red cooking ware, is chemically separate and represents a second distinct pattern probably assignable to a local production. Five amphora sherds also from the Punic Period, and consisting of a micaceous fabric, all have different chemical characteristics and are probably imports from overseas production sites of unknown location.

Since 1996 the University of Malta, through the Department of Classics and Archaeology, has conducted archaeological excavations at the ancient sanctuary site at Tas-Silg, Malta. A substantial amount of ancient ceramics from stratified deposits has been recovered in the course of these excavations. From the outset, one of the principal aims of the project has been to identify ceramic ware types at the site, ranging from its prehistoric origins (c. 3000 Be) into the Punic Period, and beyond to more recent periods. In 2001, after five seasons of scientific excavations and rigorous extensive analysis, a point has been reached at which the range of wares at the site had been classified at the macroscopic level by visual inspection. Details about the site at Tas-Silg and its archaeological context

are available in a preliminary report (Bonanno

et al. 2000), where a detailed description of the various types of pottery, its fabrics, styles and decorations is given by C. Sagona. The next stage in the pottery study was the chemical classification of the various types of wares using neutron activation analysis (NAA). As is well known, the minor and trace element concentrations in pottery reflect mainly the composition of the clay paste prepared by the potters, which depends mainly on the geochemical composition of the clay beds exploited (e.g. Perlman & Asaro 1969; Jones 1986; Mommsen 2001). Therefore, the concentration patterns in pottery point to their production places, assuming that the raw clays available in the region have been used by the potters to prepare their pastes and

From:MAGGETTI,M. & MESSIGA,B. (eds) 2006. Geomaterialsin CulturalHeritage. Geological Society, London, Special Publications, 257, 81-89. 0305-8719/06/$15.00 © The Geological Society of London 2006.

82

H. MOMMSEN E T AL.

that clays have not been traded. The aim of this study was to determine if the pottery ware types were produced from the same or different clay pastes and, if possible, if types of local production could be distinguished from imported types from foreign, overseas workshops. It was anticipated that this might give some insight into Malta's trade relations, especially during the Bronze Age and the Punic Period.

Sample choice and description An initial set of 40 pottery sherds has been selected for chemical analysis by NAA. Having identified visually and haptically a number of ware types, C. Sagona, responsible for the TasSilg typology, proceeded to select eight of the most common or characteristically distinct ware type groups excavated at Tas-Silg for further comparative analyses. Five sherds of vessels of different shapes from each ware type were selected as being representative of that ware type and examined with the aid of a magnifying glass. Details of the fabric, shape, inclusions, surface treatment and colour were noted for each sherd. In addition, one sherd of a 'Thermi Ware' bowl as a ninth ware type excavated at Tas-Silg was available for NAA (Sagona in Bonanno et al. 2000, p. 87, fig. 9.4). This ware is assigned to the Early Bronze Age of the Aegean. Identified as 'Thermi Ware' are bowls and pedestal bowls with an internally thickened lip decorated with dot-filled incised triangles, for which a close parallel was identified in Early Bronze Age Thermi (Lesbos). Traditionally thought to be imported from the east, it occurs also in Castelluccio levels in eastern Sicily. Matters are further complicated by the fact that some sherds were discovered in Temple Period contexts. An overview of the sample set is given in Table 1. Although all the ware type groups classified by visual examination are distinct from other ware types, there are notable differences between sherds within any one group, such as extent and nature of inclusions, i.e. vessels of the same ware type were produced from clays with different non-plastic admixtures. This is particularly the case for the Temple Period Wares, Borg in-Nadur Wares, and micaceous Wares. But because such inclusions are usually poorer in trace elements compared with the higher concentrations in clays, NAA could reveal if the same clay bed was exploited for the preparation of these different pastes. The nine ware types selected are attributed to five time periods. The Late Neolithic Temple Period is represented by sherds Silg 16-20, the

Early Bronze Age Period by the 'Thermi Ware' sherd Silg 43, and the Late Bronze Age Borg in-Nadur phase by sherds Silg 1-5. One sample, Silg 42, is from an early modern tile and was included as a possible reference piece. The remaining 30 sherds all belong to the Punic Period and comprise the following ware types: Crisp Ware (Silg 6-10), Soft Brown Ware (Silg 11-15), Biscuit Ware (Silg 21-25), Bricky Red Ware (Silg 31-35), Drab Coarse Ware (Silg 36-40), and Micaceous Wares (Silg 26-30), which have been described by Sagona in Bonanno et al. (2000). Although clay deposits with a rare scattering of biotite (brown mica) are found on the island (M. Pedley, pers. comm.; Digeronimo et al. 1981), the five amphora sherds (Silg 26-30) have muscovite (transparent mica) inclusions, each one having a characteristically distinct fabric. This, together with the fact that only a small number of Micaceous sherds excavated at Tas-Silg can be identified, suggests that they are imports.

Analytical procedure and data evaluation Neutron activation analysis (NAA) is a method very well suited to classify pottery by chemical means, as it is multi-elemental, sensitive down to the trace element level, and precise (e.g. Perlman & Asaro 1969; Mommsen 2004). At the Helmholtz-Institut ftir Strahlen- und Kernphysik, University of Bonn, only 80 mg of pottery powder are needed for an analysis. The sample is taken by drilling with a pointed sapphire-drill of 10 mm diameter, usually on the internal surface of the sherds, leaving only a shallow depression. The neutron irradiations are carried out at the research reactor in Geesthacht. The analytical procedure has been described by Mommsen et al. (1991). It is a modified version of that given by the former Berkeley group (Perlman & Asaro 1969). As the Bonn pottery standard is calibrated against the Berkeley standard, our concentration data can be compared directly with the values of this group. A calibration to other standards in use has been given by Hein et al. (2002). The comparison of the measured elemental data and the formation of groups of samples having a similar composition are done by a multivariate statistical filtering procedure developed in Bonn (Beier & Mommsen 1994a,b) using a Mahalanobis distance search (Harbottle 1976). In contrast to the common statistical methods such as cluster analysis or principal component analysis, it is able to take into consideration two important features: (1) individual experimental errors; (2) a possible constant

83

CHARACTERIZATION OF MALTESE POTTERY Table 1. List of sherds from Tas-Silg, Malta, studied by NAA Sample number

Excavation inventory number

Ware type name, time period

Fabric description: fabric texture/surface treatment/inclusions

Silg 1 Silg 2 Silg 3 Silg 4 Silg 5 Silg 6 Silg 7 Silg 8 Silg 9 Silg 10 Silg 11 Silg 12 Silg 13 Silg 14 Silg 15 Silg 16 Silg 17 Silg 18 Silg 19 Silg 20 Silg 21 Silg 22 Silg 23 Silg 24 Silg 25 Silg 26 Silg 27 Silg 28

TSG96/2112/5 TSG96/2112/15 TSG96/2097/6 TSG96/2101/5 TSG96/205/47 TSG96/2019/28 TSG96/2061/24 TSG96/1010/14 TSG96/2097/17a TSG96/29/32 TSG96/2054/15 TSG96/210/4 TSG96/23/8 TSG96/2077/14 TSG96/2092/33 TSG96/9/1 TSG96/11/63 TSG96/11/57 TSG96/210/9 TSG96/220/27 TSG96/34/26 TSG96/205/19 TSG96/205/49 TSG96/105/9 TSG96/2016/54 TSG96/205/54 TSG96/204/28 TSG96/220/29

Borg in-Nadur Ware, Bronze Age Borg in-Nadur Ware, Bronze Age Borg in-Nadur Ware, Bronze Age Borg in-Nadur Ware, Bronze Age Borg in-Nadur Ware, Bronze Age Crisp Ware, Punic Period Crisp Ware, Punic Period Crisp Ware, Punic Period Crisp Ware, Punic Period Crisp Ware, Punic Period Soft Brown Ware, Punic Period Soft Brown Ware, Punic Period Soft Brown Ware, Punic Period Soft Brown Ware, Punic Period Soft Brown Ware, Punic Period Temple Period Ware, Late Neolithic Temple Period Ware, Late Neolithic Temple Period Ware, Late Neolithic Temple Period Ware, Late Neolithic Temple Period Ware, Late Neolithic Biscuit Ware, Punic Period Biscuit Ware, Punic Period Biscuit Ware, Punic Period Biscuit Ware, Punic Period Biscuit Ware, Punic Period Micaceous Ware, Punic Period Micaceous Ware, Punic Period Micaceous Ware, Punic Period

Silg 29 Silg 30 Silg 31 Silg 32 Silg 33 Silg 34 Silg 35 Silg 36 Silg 37 Silg 38 Silg 39 Silg 40 Silg 42 Silg 43

TSG96/220/23 TSG96/34/42 TSG96/34/36 TSG96/34/33 TSG96/34/37 TSG96/2026/22 TSG96/34/34 TSG96/1033/1790 TSG96/1033/1375 TSG96/1033/800 TSG96/1033/1229 TSG96/1033/1291 TSG96/2061/9

Micaceous Ware, Punic Period Micaceous Ware, Punic Period Bricky Red Ware, Punic Period Bricky Red Ware, Punic Period Bricky Red Ware, Punic Period Bricky Red Ware, Punic Period Bricky Red Ware, Punic Period Drab Coarse Ware, Punic Period Drab Coarse Ware, Punic Period Drab Coarse Ware, Punic Period Drab Coarse Ware, Punic Period Drab Coarse Ware, Punic Period Tile of structure, Early Modern Thermi Ware, Early Bronze Age

Coarse/burnish/quartzite grit Coarse/burnish/sparse grog Fine/burnish/moderate grog Coarse/quartzite grit Fine/burnish/moderate grog Fine/slip/moderate grog Fine/slip/moderate grog Coarse/slip/grog Fine/slip Fine/thick slip/brown rock? grit Fine/self-slip/grog specks Fine/self-slip/Ca specks Refined/self-slip Coarse/self-slip/grog, Ca grit Fine/slip/grog specks Coarse drab/slip/moderate grog Fine/burnish/red specks Refined/burnish Refined/burnish/Ca specks Refined/burnish/Ca specks Coarse/slip/Ca grit Fine/slip/Ca grit Fine/slip/Ca specks Fine/slip/Ca specks Coarse/slip/Ca grit Refined/muscovite mica Refined/slip/muscovite mica Refined/ mica-dusted/muscovite mica Refined/self-slip/muscovitemica Refined/slip/muscovite mica Coarse/cream slip/Ca grit Coarse/self-slip/Ca grit Coarse/self-slip/Ca grit Fine/self-slip/sparse Ca grit Coarse/self-slip/Ca grit Coarse/slip/frequent grog Very coarse/slip/frequent grog Coarse~slip~frequent grog Very coarse/slip/frequent grog Coarse/slip/moderate grog

shift of the data caused by diluting compounds in the samples. Examples of this evaluation procedure have been presented by Mommsen (2001) and Akurgal et al. (2002). The neglect of experimental errors during the groupforming procedure is dangerous in the case of elements measurable only with a large margin of error and might lead to wrong classifications. With regard to NAA as performed at Bonn, this mainly concerns the elements Ca, Ti and Ni.

Thickened lip with dotted decoration

For these elements, only isotopes excited by fast neutrons have long enough half lives to be measurable 5 days after the irradiations as done in Bonn. As a result of a substantial reduction of the fast neutron fraction in the irradiating neutron beam since an upgrading of the reactor in Geesthacht in the year 2000, these elements are currently determined only with large errors. However, consideration of errors allows their inclusion during grouping. In addition, possible

84

H. MOMMSEN ET AL.

constant shifts of all concentration values should be corrected during group formation to reduce the spreads of the average concentration values of the pottery groups obtained. Levigating the raw clays or adding tempering material such as sand or calcite in varying amounts will enrich or dilute the clay fraction of the paste, often changing all the concentration values by a constant factor. Besides such pottery-making practices, experimental errors, e.g. weighing errors and/or neutron flux heterogeneities, may also be the cause of such constant shifts. Therefore, constant shifts should be corrected by either using concentration ratios (Buxeda i Garrigos 1999) in statistical cluster analyses or, to be independent of the single concentration value chosen as denominator, performing a best relative fit of the individual datasets to the group mean values (Harbottle 1976). The reduction of the spreads of the elemental patterns observed in all our studies considering dilutions demonstrates that it is mainly the clay part of the paste that determines the provenance. The element concentrations of As, Ba, Ca and Na are found to scatter strongly in pottery groups made at the same workshop (see Table 1) and should be considered with care during the group-forming procedure (Mommsen 2004). In particular, As and Na are not and should not be included in the filtering procedure, as they are measured with small errors (see Table 4). Therefore, they will strongly influence the value of the calculated dilution factor and may result in an inappropriate factor.

Results and discussion Some of the results of this study were rather unexpected. The evaluation of the NAA data revealed that most sherds of the set of samples from Tas-Silg have a well-defined and similar chemical composition, forming a group here named SILA. Its average concentrations M and spreads o-(root mean square deviations) are listed in Table 2. For many elements (Ce, Eu, Fe, La, Sc, Th, Yb) spread values < 5 % are found; nine more elements (Co, Cr, Hf, Lu, Nd, Rb, Sin, Ta, Tb) scatter by < 10% around their mean values M. This result was obtained by employing the method of a best relative fit of the concentrations of each sample to the group mean values with the fit factors given in Table 3. Very high factors are found for the two Bronze Age samples Silg 1 and 4 (1.55 and 1.68, respectively), i.e. the concentrations of these samples have to be increased by 55% and 68%, respectively, to match the mean group pattern. A cross-section of one of the two

sherds, sampled as Silg 1, is shown in Figure 1 (top). Silg 1 and 4 both have a high Ca content (18.7% and 19.9%, respectively), which points to Ca as at least one of the diluents. By visual examination, both samples were noted for their moderate amount of coarse quartzite inclusions; this also may account for their strong dilution. On the other hand, the concentrations for three of the oldest samples from the Temple Period (Silg 18-20) have to be lowered by more than 20% to match the group mean values; these samples have the highest concentrations of trace and minor elements in the whole group of samples. The cross-section of sherd sampled as Silg 18 is also shown in Figure 1 (bottom) for comparison. All three samples belong to thinwalled vessels having a fine-textured fabric, in contrast to sample Silg 16, which is a typical example of the Temple Period coarse ware with the usual concentrations of group SILA (fit factor-----0.99). Therefore, it may be concluded that samples Silg 18-20 are made from a refined clay paste compared with sample Silg 16, where some non-plastic parts with low element concentrations have been taken out, giving no contribution to the measurable NAA data obtained at Bonn. Despite this unusually large range of fit or dilution factors, all members of the group SILA can be assumed to have been made from clay of the same geological formation, which was refined by different 'recipes', resulting in different absolute compositions. Their common origin is confirmed by the dilution-corrected NAA compositional data. Unexpectedly, also the 'Thermi Ware' sample Silg 43, diluted by 16%, belongs to group SILA. The fact that all these different wares are members of group SILA would not have been obtainable using standard statistical cluster analysis without corrections for possible inclusions. As group SILA contains vessels of different time periods (Temple Period, Bronze Age, Punic Period) and of different archaeological types and shapes, the probability that this group represents the local workshop(s) in the region of Tas-Silg is very high. Therefore, the pattern SILA can be taken as a reference pattern for a pottery production centre possibly in the area of Tas-Silg, Malta. The possibility that all these different ware types during this long time period were imported from the same workshop somewhere overseas is extremely remote, but cannot be totally excluded, especially as, on the basis of plate tectonics, the southern part of Sicily and Malta belonged to the same Pelagian sea basin during the Miocene. More light will be shed on this issue only once reference material from the Sicilian coast facing Malta has been analysed.

CHARACTERIZATION OF MALTESE POTTERY

85

Table 2. Element concentrations of groups SILA (most probably Malta), SILB (probably Malta) and Seg2 (Segesta, Sicily) SILA 27 samples M _+ tr (%) As Ba Ca % Ce Co Cr Cs Eu Fe % Ga Hf K% La Lu Na % Nd Ni Rb Sb Sc Sm Ta Tb Th Ti % U W Yb Zn Zr

7.50 250 10.2 89.9 13.7 111 5.98 1.44 4.34 25.6 4.59 2.44 42.0 0.38 0.56 30.5 92.0 108 0.56 15.4 5.50 1.21 0.79 11.8 0.55 2.97 1.75 2.63 125 106

15 20 23 2.4 8.9 7.4 13 3.2 4.6 15 6.7 13 4.4 5.3 40 6.4 40 6.9 14 3.0 7.1 6.1 6.1 2.5 17 12 13 3.5 15 25

SILB 5 samples M + ~r (%) 11.2 389 3.49 112 17.5 110 6.24 2.03 5.11 27.3 9.45 2.48 54.2 0.60 0.97 43.8 104 124 0.71 17.6 8.10 1.93 1.20 15.2 0.63 2.49 2.42 4.28 100 286

9.9 7. I 6.0 1.0 2.3 2.2 7.6 2.3 1.7 10 4.5 5.9 1.3 2.9 5.3 4.0 34 6.7 5.2 0.5 7.1 2.1 7.5 1.3 12 4.1 16 2.0 4.8 9.9

Seg2 5 samples fit factor 0.93 M +_ or (%) 4.33 228 7.33 89.8 15.1 117 7.42 1.40 4.65 27.1 5.07 2.30 43.3 0.41 0.43 29.2 81.2 124 0.49 16.8 5.18 1.26 0.77 12.5 0.43 3.05 2.17 2.79 118 243

42 44 16 2.1 9.7 4.1 4.0 1.6 2.2 8.8 13 8.2 2.7 6.2 52 11 37 3.2 8.8 1.4 6.5 3.1 9.1 2.6 14 11 12 2.7 8.0 13

Average concentrationvaluesM in ~g g-~ (ppm). if not indicatedotherwise,and spreads tr in % of M; group Seg2 correctedfor dilution with respect to SILA.

c o m m o n wares from a H e l l e n i s t i c - R o m a n kiln site at Segesta and, therefore, m a d e there locally with a high probability, form this group, n a m e d Seg2 (Montana et al. 2003). If the values of group Seg2 are adjusted by a fit factor of 0.93, group SILA is very close to

W h e n the databank of patterns in Bonn is searched for groups of similar compositions, the closest group to SILA is a group from Segesta, Sicily. This is m e n t i o n e d here to stress again the importance of high precision. Five sherds, all kiln wasters of amphorae and

Table 3. List of group members and their individual fit factors (in parentheses) to the mean group values calculated by using all elements given in Table 2 except As, Ba, Ca and Na (see text)

Group SILA, 27 samples Silg 1 (1.55) 3 (1.07) 12 (1.06) 13 (1.03) 22 (0.89) 23 (0.96)

4 (1.68) 14 (1.14) 24 (0.84)

5 (0.93) 15 (1.13) 25 (0.88)

6 (0.95) 16 (0.99) 37 (1.05)

Group SILB, 5 samples Silg 31 (0.98) 32 (1.00)

33 (0.98)

34 (1.03)

35 (1.00)

27

28

Chemical singles, 10 samples Silg 2 10 17

26

7 (0.95) 18 (0.71) 38 (1.02)

8 (0.86) 19 (0.79) 39 (1.03)

9 (0.95) 20 (0.72) 40 (1.06)

11 (1.21) 21 (1.08) 43 (1.16)

29

30

36

42

86

H. MOMMSEN ET AL.

Fig. 1. Cross-sections of the coarse Borg in-Nadur sherd (sample Silg 1, above) and of the fine-textured sherd of the Temple Period (sample Silg 18, below). Both are made from the clay with pattern SILA. The concentrations of sample Silg 1 and Silg 18 have to be corrected by a fit factor of 1.55 and of 0.71, respectively, to match pattern SILA.

group Seg2, as can be seen from Table 2. Figure 2 shows a graphical comparison between groups SILA and Seg2 made by plotting the concentration differences for all 30 elements in units of the average spread values o'ave as a bar diagram. The only clearly differing element is Zr, which is difficult to measure at Bonn, but because

three elements measured with high precision, Cs, Rb and Sc, differ by slightly more than 2o'.~ve, the two patterns can be chemically separated. Without this high precision of the chemical patterns, which is obtained here only by the correction of the unusually large dilution effects, the groups of Tas-Silg, Malta, and Segesta, Sicily, could not have been distinguished. The local wasters from Segesta are obviously made using a clay formation that can be distinguished from the clay of group SILA only by using appropriate evaluation procedures on the data. A second chemical group SILB is formed by five sherds of the type Bricky Red Cooking Ware, dated to the Punic Period. The mean values and spreads are also listed in Table 2. This group has a very different chemical composition from group SILA with on average 21% higher concentrations in nearly all elements except for Ca, Cr, U and Zn. This may be due to the fact that the local potters, in considering the special requirements for cooking ware, have produced a different paste deliberately by using different clays or clay mixtures for the production of these wares (see Kilikoglou et al. 1998). In this case, pattern SILB may form a second local pattern for the site at Tas-Silg. This assumption is strengthened by the comparatively large presence of this ware and by its typological similarity to the local Biscuit Ware of group SILA (Bonanno et al. 2000). However, as only five pieces with pattern SILB were identified, the assumption of a local production is not very conclusive; all five vessels of this group might have been imported from a production site with this as yet unknown pattern. On the other hand, several sherds of this ware type carry inscriptions to Ashtart, which were inscribed before firing. It is most unlikely that this was done abroad for Maltese consumption. No other pattern in our databank in Bonn is close to SILB. During a chemical classification of pottery several chemically single samples are usually found. One cannot conclude much about these outliers. A chemically single sample might have been contaminated in the laboratory, during burial or already in the production workshop, or it might represent the first member of a still unknown production series with a different paste. The individual NAA data for all singles found in this study are listed in Table 4. In the set of non-micaceous wares from Tas-Silg five such singles (Silg 2, 10, 17, 36, 42) were identified. Silg 42 is the Early Modern tile. Its composition does not help to localize any of the patterns found at Tas-Silg. By visual inspection Silg 17 has been noted as having a unique fabric

CHARACTERIZATION OF MALTESE POTTERY SILA

-

Seg2

(factor

87

0.93)

qo z_

O. (R

> O p, t~ W "O . m

-2

-4

-6

-8

elements Fig. 2. Normalized differences (in units of average spread value) of average concentrations of the 30 elements given in Table 2 of the pottery groups SILA (27 samples) and Seg2 (5 samples). The concentrations of group Seg2 are multiplied by a fit factor of 0.93 to obtain an optimal agreement.

differing from that of the other Temple Period sherds. As such, it has been given the name Sandy Pink Ware because of its fine-textured red clay with small red inclusions. Its wellburnished exterior surface, however, is typical of the Temple Period. This may point towards a different production group. With regard to Silg 10, whereas Crisp Ware fabrics are known to have a number of inclusions such as calcium grit or grog, Silg 10 is unusual for its matt brown, coarse inclusions, which may account for its unique chemical characteristics. Finally, the whole group of Micaceous amphorae of the Punic Period (Silg 26-30), believed to be imported to Malta, consists of chemical singles; they all have different compositions. The element analysis permits the conclusion that all five pieces were produced with different pastes, presumably in different workshops. These amphorae seem to have been used as transport containers, which would explain the presence

of vessels from different production sites in this group. As the five patterns of the amphorae are unknown to us, we are not able to assign them to production workshops. More work on vessels from different overseas sites is necessary to locate the origin of these chemical outliers. Conclusion NAA of 41 sherds excavated at Tas-Silg consisting of nine ware types was able to show the unexpected result that only one characteristic clay with a well-defined, sharp concentration pattern of many elements, SILA, was used for seven of these ware types, including the 'Thermi Ware'. The dating of these ware types defines the time range of the use of this clay as beginning with the Late Neolithic Temple Period (about 3000 Be) and lasting through the Bronze Age (Tarxien Cemetery and Borg in-Nadur phases, 2 5 0 0 - 7 0 0 BC) until the Punic Period (550-218

1.31 I. 19 0.59 1.34 0.30 1.01 0.97 1.20 1.32 1.03 0.032 3. I

0.018 0.3

Av. meas. error in %

Ta

Sm

5.40 6.96 3.90 5.69 2.67 6.33 4.62 5.71 6.60 5.88

2 10 17 26 27 28 29 30 36 42

•0 .0 •0 •0 •0 •0 1.0 1.0 1.0 1.0

Silg Silg Silg Silg Silg Silg Silg Silg Silg Silg

0.032 1.4

3.01 2.41 1.03 2.22 0.86 3.03 2.16 3.18 2.01 3.18

K %

Fit factor

4.58 8.48 2.10 9.02 2.26 6.83 5.14 4.82 5.69 4.52

Sample

1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0

Fit factor

0.065 1.2

2 10 17 26 27 28 29 30 36 42

30 8.3

271 345 116 400 217 666 286 377 310 632

Ba

Av. meas. error in %

Silg Silg Silg Silg Silg Silg Silg Silg Silg Silg

Hf

10.6 14.4 29.7 39.7 19.2 15.1 8.89 25.6 8.15 7.34

Sample

1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0

As

0.12 0.7

2 10 17 26 27 28 29 30 36 42

Fit factor

Av. meas. error in %

Silg Silg Silg Silg Silg Silg Silg Silg Silg Silg

Sample

0.049 5.9

0.76 1.10 0.72 0.96 0.43 1.05 0.77 0.90 1.01 0.71

Tb

0.15 0.4

34.8 44.2 22.9 38.9 17.2 40.4 39. I 38.7 46. I 40.7

La

0.19 2.5

3.53 6.77 21.9 3.96 8.48 1.93 12.9 3.94 8.07 4.91

Ca %

0.068 0.5

11.7 12.8 7.38 15.2 5. I 0 17.2 13.3 17.1 12.3 13.5

Th

0.016 3.8

0.36 0.46 0.25 0.56 0.36 0.60 0.38 0.46 0.47 0.37

Lu

0.80 1.0

77.1 97.8 53.6 97.0 29.1 109 80.4 82.5 93.8 82.6

Ce

0.073 14

0.45 0.49 0.54 0.53 0.30 0.68 0.59 0.56 0.48 0.59

Ti %

0.005 0.6

0.65 0.62 0.23 0.72 I. 14 1.23 0.40 0.31 0.48 2.19

Na %

0,12 0.6

12. I 11.2 9.44 24.4 36.4 23.1 I 1.4 29.2 15.2 17.8

Co

0.093 3.3

2.52 3.39 2.69 2.81 1.09 2.76 2.68 4.83 2.74 2.66

U

1.5 5.4

27.4 37. I 19.0 30.0 14.3 32.6 24.9 31.2 34.8 32.5

Nd

1.2 0.6

122 10l 90.8 300 685 148 59.0 283 103 57.0

Cr

0.16 7.9

1.67 1.48 1.05 2.63 1.12 2.16 2.12 4.08 1.69 2.13

W

34 23

128 47.7 54,7 125 544 43.9 I(X) 312 142 37.4

Ni

0.087 1.4

6.19 5.62 3.03 8.69 3.59 8.19 7.80 8.76 5.61 5.43

Cs

0.053 1.8

2.57 3.29 1.79 3.72 1.83 3.79 2.70 3.04 3.28 2.54

Yb

2.4 2.1

I 18 I 15 54.0 120 36.6 159 I 14 179 103 141

Rb

0.026 1.9

1.23 1.77 1.19 1.47 0.79 1.60 1.19 1.40 1.67 1.26

Eu

2.4 2.2

154 98.2 1() I 80.3 82. I 112 86.4 108 111 130

Zn

0.020 1.8

0.83 (I.52 1.49 1.21 0.40 1.31 I. 15 3.05 0.74 0.48

Sb

0.017 0.4

4.88 4.02 4.22 4.81 5.34 5.66 3.22 5.70 4.37 3.45

Fe %

T a b l e 4. Unchanged (fit factor = 1) element concentrations C in tzg g - l (ppm), if not indicated otherwise, and average errors, also in % of C, for the nine ungrouped samples (chemical singles)

27 20

1(15 239 34.2 265 83.3 144 10 I 90.7 122 182

Zr

0.022 0. I

16.3 14.4 8.74 20.5 22.2 21.4 I 1.8 18.6 15.3 12.5

Sc

2.4 9.9

27.2 21.7 13.4 23.7 18.9 36.7 24.9 30.4 23.4 26.6

Ga

CHARACTERIZATION OF MALTESE POTTERY BC). This large time span together with the presence of different ware types in the group argues for a local production of all these pottery vessels, although, because of the lack of reference material, no definite proof is obtained. A second paste with pattern SILB was in use during the Punic Period for the Bricky Red cooking ware and was probably also locally produced. In contrast, the group of five Micaceous amphorae found in Tas-Silg all have different chemical patterns and are most probably imports from five different workshops outside Malta. We would like to thank the staff of the research reactor at Geesthacht, Germany, for the neutron irradiations, and Y. Taniguchi (formerly of the Malta Centre for Restoration) and M. H. Pedley (University of Hull, UK) for their advice and comments.

References AKURGAL, M., KERSCHNER, M., MOMMSEN, H. & NIEMEIER, W.-D. 2002. TSpferzentren der Ostdgdis, Archdometrische und archiiologische Untersuchungen zur mykenischen, geometrischen und archaischen Keramik aus Fundorten in WestkIeinasien (mit einem Beitrag yon S. Ladstdtter). 3. Erg~inzungsheft der Jahreshefte des Osterreichischen Archiiologischen Institutes, Wien BEIER, TH. & MOMMSEN, H. 1994a. Modified Mahalanobis Filters for grouping pottery by chemical composition. Archaeometry, 36, 287-306. BEIER, TH. & MOMMSEN, H. 1994b. A method for classifying multidimensional data with respect to uncertainties of measurement and its application to archaeometry. Naturwissenschaften 91, 546-548. BONANNO, A., FRENDO, A. J. t~z VELLA, N. C. (eds) 2000. Excavations at Tas-Silg, Malta: a preliminary report on the 1996-1998 campaigns conducted by the Department of Classics and Archaeology of the University of Malta. Mediterranean Archaeology, 13, 67-114

89

BUXEDAI GARRIGOS,J. 1999. Alteration and contamination of archaeological ceramics: the perturbation problem. Journal of Archaeological Science, 26, 295- 313. DIGERONIMO, I., GRASSO, M. & PEDLEY,H. M. 1981. Palaeoenvironment and palaeogeography of Miocene marls from southeast Sicily and the Maltese Islands. Palaeogeography, Palaeoclimatology, Palaeoecology, 34, 173-189. HARBOTTLE, G. 1976. Activation analysis in archaeology. In: NEWTON,G. W. A. (ed.) Radiochemistry 3. Chemical Society, London, 33-72. HEIN,A.,TSOLAKIDOU,A., ]LIOPOULOS,L MOMMSEN,H., BUXEDA ! GARRIGOS, J., MONTANA, G. & KILIKOGLOU, V. 2002. Standardisation of elemental analytical techniques applied to provenance studies of archaeological ceramics--an interlaboratory calibration study. Analyst, 127, 542-553. JONES, R. E. 1986. Greek and Cypriot Pottery. British School at Athens, Fitch Occasional Papers, 1. KILIKOGLOU, V., VEKINIS, G., MANIATIS, Y. t~z DAY, P. M. 1998. Mechanical performance of quartztempered ceramics: Part I, Strength and toughness. Archaeometry, 40, 261-280. MOMMSEN, H. 2001. Provenance determination of pottery by trace element analysis: problems, solutions and applications. Journal of Radioanalytical and Nuclear Chemistry, 247, 657-662. MOMMSEN, H. 2004. Short note: provenancing of pottery--the need for an integrated approach? Archaeometry, 46, 267-271. MOMMSEN, H., KREUSER, A., LEWANDOWSKI, E. & WEBER, J. 1991. Provenancing of pottery: Status report and grouping. In: HUGHES,M. COWELL, M. & HOOK, D. (eds) Neutron Activation and Plasma Emission Spectrometric Analysis in Archaeology. British Museum Occasional Papers, 82, 57-65. MONTANA, G., MOMMSEN, H., ILIOPOULOS, I., SCHWEDT, A. ~z DENARO,M. 2003. Petrography and chemistry of thin-walled ware from a HellenisticRoman site at Segesta (Sicily). Archaeometry, 45, 375-389. PERLMAN, I. t~ ASARO, F. 1969. Pottery analysis by neutron activation. Archaeometry, 11, 21-52.

Cobalt blue painted pottery from 18th Dynasty Egypt A. J. S H O R T L A N D l, C. A. H O P E 2 & M. S. TITE 3

1Centre for Archaeological and Forensic Analysis, Department of Materials and Medical Sciences, Cranfield University, Shrivenham, Swindon SN6 8LA, UK 2Centre for Archaeology and Ancient History, Building 11, Clayton Campus, Monash University, Clayton, Vic. 3800, Australia 3Research Laboratory for Archaeology and the History of Art, 6 Keble Road, Oxford OX1 3QJ, UK (e-mail: Michael. tite @rlaha.ox.ac, uk) Abstract: Cobalt blue painted pottery was produced in New Kingdom Egypt, with the heyday for its production being from about 1400 BC tO 1200 BC. Previous scientific examination has established that the cobalt blue pigment was a CoAl-spinel, which it was suggested was produced from cobaltiferous alums from the Western Desert of Egypt. In the present paper, quantitative analyses of a range of cobalt blue painted pottery have confirmed the Western Desert as the source of the cobalt blue pigment but suggested that the cobaltiferous alums used for the pottery differed in composition from those used in the production of contemporary cobalt blue glass. The pottery bodies were produced using either non-calcareous Nile silt or calcareous clay. Before being painted, the Nile silt bodies were first coated with pale firing calcareous clay slip to which gypsum had probably been added.

Throughout the course of the dynastic period (c. 3000-332 Bc) ancient Egyptian pottery was predominantly undecorated. Intermittently decoration did occur (Arnold 1993, pp. 9 5 - 1 0 0 ; Aston 1998, pp. 5 3 - 5 7 ) but not in significant quantities. An exception to this occurred during the New Kingdom, 18th-20th Dynasties (1550-1087 BC), when a wide range of forms received decoration prior to firing in which the dominant colour was a pale blue, supplemented by red and reddish-brown to black. The ware is now commonly termed blue painted pottery, the most detailed study of which remains the unpublished PhD thesis of Hope (1980). However, quantification of its frequency within an assemblage studied at one site (Malkata in Thebes) showed that it accounted for less than 4% (Hope 1989a, p. 12). Despite this, because of its limited period of occurrence and the evolution of design features, it can be used as a convenient dating tool. In addition, it provides an indicator of Egyptian taste in decorative arts during the New Kingdom, can show the probable existence of specialist workshops or potters responsible for its manufacture, and is a tool for use in gauging those potters' technological achievements. In contrast to blue painted pottery, faience glazes and glasses in brilliant blues are

common from the third millennium BC onwards. Part of the reason for the rarity of blue painted pottery is undoubtedly technical. Glazes and glasses are usually coloured blue by the deliberate addition of copper, but compounds of this element are very difficult to apply as a paint to a ceramic in such a way that the resultant ceramic will be robust enough for regular use. Very rarely, the man-made pigment Egyptian blue was applied to ceramics in Egypt, but this flakes off readily. Instead of copper, the colorant used on the great majority of Egyptian blue painted pottery is cobalt. The technology for using cobalt colorants in glass, glaze and ceramics was apparently discovered in the reign of the Egyptian king Tuthmosis III (1479-1425 BC), when it is used first in dark blue glasses (Lilyquist & Brill 1993). It becomes fairly common during the reigns of his son, Amenhotep II (1427-1400 BC), and grandson, Tuthmosis IV (1400-1390 Be). However, it is during the reigns of his successors Amenophis III and Akhenaten (1390-1336 BC) that cobalt blue in glass, glaze and ceramics is most abundant (Kozloff & Bryan 1992). In this period, especially at the palace site of Malkata and city of Amarna (Fig. l), it is a frequent part of the glass and ceramic assemblage; indeed, walking now over either site, it is only a matter of minutes before

From: MAGGETTI,M. & MESSIGA,B. (eds) 2006. Geomaterialsin CulturalHeritage. Geological Society, London, Special Publications, 257, 91-99. 0305-8719/06/$15.00 © The Geological Society of London 2006.

92

A.J. SHORTLAND ET A L

T

o/

Fig. 1. Map of Egypt showing location of places mentioned in text.

several sherds of blue painted pottery or cobalt blue glass or faience can be collected. The last examples of cobalt colorants both in glazes and glasses and in blue-painted pottery appear to be dated to the 20th Dynasty (1186-1069 BC), after which it ceases to be used (Arnold & Bourriau 1993). The fact that it appears to cease to be used simultaneously in all forms might suggest that perhaps there was a problem in the supply of the cobaltiferous alum, which, as discussed below, was used to make the colorant. Previous scientific investigations of cobalt blue painted pottery (Riederer 1974; Noll & Hangst 1975; Bachmann et aI. 1980; Noll 1981) established that the cobalt blue pigment was in the form of a CoAl-spinel containing significant amounts of zinc and nickel, and that the pigment was applied over a thin slip containing gypsum. On the basis of scanning electron photomicrographs, Noll (1981) showed that the pigment was a loose aggregate of roundish particles with diameters typically less that about 0.5 ~m. Bachmann et al. (1980) first suggested that the cobaltiferous alums from the Kharga and Dakhla Oases of the Western Desert in Egypt were the possible source of the cobalt pigment. Kaczmarczyk (1986) subsequently presented analyses of alums collected from the Dakhla Oasis, which established them as being magnesium aluminium sulphates with trace amounts of cobalt, zinc, nickel, manganese and iron, and thus echoing the trace elements present in the cobalt blue pigment. Noll (1981)

suggested that the first step in the production of the pigment was to convert the sulphates to hydroxides by the precipitation from an aqueous solution of the cobaltiferous alum through the addition of an alkali such as the natron, a complex evaporite consisting mostly of sodium carbonate. He thus suggested that the CoAl-spinel was not a natural mineral but was formed by firing the hydroxide mixture in the range 800-1000 °C. Warachim et al. (1985) investigated the co-precipitation of cobalt, aluminium and magnesium hydroxides, and their use in the subsequent production of cobalt spinels. Using highly alkaline ammonia, they achieved complete precipitation of all three hydroxides. However, Rehren (2001) suggested that, with the less strongly alkaline natron, differential precipitation would occur, with aluminium and transition element hydroxides precipitating before magnesium hydroxide. Subsequently, in laboratory experiments using sodium carbonate as the alkali, Shortland et al. (2006) confirmed that aluminium hydroxide precipitated first, followed by cobalt and manganese hydroxides, and then magnesium hydroxide. Warachim et al. (1985) also established that the blue (Mg,Co)Al-spinel was formed by heating the hydroxide mixture in the range 900-1200 °C. However, when fired below about 900°C, various cobalt oxides (COO, Co203, Co304) and MgCo-spinel were instead formed, resulting in a dark grey to black pigment. The aim of the research reported in the present paper has been to obtain quantitative data on the compositions of the cobalt blue pigment layer and the underlying slip for a range of cobalt blue painted pottery from New Kingdom Egypt, and so to extend our understanding of production and application of the pigment.

Archaeological context There are two main categories of Egyptian blue painted pottery, the heyday for the production of which was during the reign of Amenhotep III (1390-1352 Be) in the 18th Dynasty until that of Ramesses II (1279-1213 Be) in the 19th Dynasty. In one the forms and decoration differ from the classic blue painted pottery and the blue is not dominant (Hope 1987). Examples come from Thebes (modern Luxor) but are mostly without provenance. The decoration has several features in common with examples of pottery decorated in bi-chrome red and brown/black from the same period; this latter type, along with monochrome decoration in red or brown/black, has its antecedents in the

COBALT BLUE PAINTED POTTERY earlier 18th Dynasty and late Second Intermediate Period. The other category is characterized by the dominant use of blue, with examples from Giza possibly indicating that the style was an innovation of the potters of the capital at Memphis (now under modern Cairo; Hope 1997). The greatest quantity was produced during the 18th Dynasty, with some reduction during the early 19th Dynasty; following Ramesses II there was significant reduction until manufacture ceased during the reign of Ramesses IV in the 20th Dynasty (Aston 1998, p. 56; Hope 1989b, p. 56). It may be noted that contemporary with its manufacture was a post-firing decorative technique most commonly found upon amphorae, which employed a wider colour palette that included the copper-based Egyptian blue rather than cobalt blue pigment (Bell 1987; Hope 1989c, 1991). This may be termed polychrome or post-firing decorated pottery and it appears to have been manufactured from a slightly earlier date within the reign of Thutmose III (14791427 Bc; Hope 1987). It occurs at some settlement sites (Malkata, Amarna) but mostly in cemeteries (Deir el-Medineh (near Malkata) and Gurob). The decoration upon blue painted pottery during each of the two major phases of its production, the late 18th Dynasty (1390-1295 Bc) and the early 19th Dynasty (1295-1213 Bc), shows a remarkable homogeneity of layout and composition irrespective of provenance. In the former, motifs are laid out within panels that emphasize the characteristics of the vessel profile and are dominated by floral elements (Hope 1989a, 1991). The compositions range from those with simple repeat motifs to extremely elaborate ones with numerous elements and complex structures. The floral elements, predominantly based upon the blue lotus flower but with numerous other sources of inspiration, are supplemented with faunal, human, divine, hieroglyphic and abstract motifs. Most commonly the decoration is over a pale background coating, either cream or pinkish-orange; rarely is it on a red ground or directly upon the uncoated surface, except when the vessel is made from marl (calcium-rich) clay. This light coating upon vessels made in Nile alluvial silt clay, which was most regularly utilized, provides a suitable contrast to the blue pigment and highlights the design. In terms of techniques, painting is the most common, but modelling from the vessel wall, application of finger-modelled or mould-made elements, incision, impression and fenestration also occur. This range of techniques continued into the next phase as did the variety of motifs, but the designs became less complex

93

and there was a growing tendency to use bands of blue without elaboration (Hope 1989b, p. 56-57; Aston 1998, p. 57). As previously indicated, Memphis may have led the way in the development of blue painted pottery, and of the little well-provenanced and dated material that survives for the period prior to Amenhotep III, Thebes is another early find spot. Thereafter, during the 18th Dynasty the largest deposits occur at Thebes (Malkata, Karnak), Amarna and in the Memphite region to the Fayum (Hope 1989a, 1991). For the 19th Dynasty, Qantir, Memphis and its environs, the Fayum and Thebes (Deir el-Medineh) have yielded the most substantial quantities, but possibly less than earlier; the latest well-dated material of the 20th Dynasty derives from Thebes (Valley of the Kings). It can be seen clearly that, if find spot and quantity indicate place of manufacture, the production centres coincided with the major administrative or royal residence sites, whence distribution on a smaller scale to a greater number of locations occurred. Blue painted pottery has been found throughout the Nile Valley in Egypt, and some also in Nubia to at least the Third Cataract, in the Eastern Desert, the Oases of Dakhla and Kharga in the Western Desert, and in the Levant, where the quantity is greater in the 19th Dynasty, with the most northerly find at Tell Nebi Mend (ancient Kadesh-on-the-Orontes). The largest quantity of blue painted pottery was used within a domestic context by the elite, and some appears to have had a distinctly ornamental value. It was also included within their burials on a smaller scale; such contexts have produced a more limited range of forms than the former but they include types rarely, if at all, found within houses or palaces, sometimes in sets. This indicates specialist production of mortuary wares; the most distinctive examples derive from the tomb complexes of Horemheb and Maya at Saqqara (near Memphis; J. D. Bourriau, pers. comm.). The products of the reigns of Amenhotep III and his son Akhenaten from Thebes (Malkata and Karnak) and Amarna are remarkably similar, prompting the suggestion that the same potters might have been responsible for their manufacture. Contemporary and slightly later material from Memphis shows significant differences in detail whereas the general aspect is very similar. Access to limited quantities was gained by those of lesser rank for whom it may have been a desirable product, a status marker. The quality of blue painted pottery varies, and although that from elite contexts is often of a higher standard of execution, this is not always the case. In addition

94

A.J. SHORTLAND ETAL.

to finds from domestic and funerary contexts are those from temples. These are far fewer, but the material discovered at the temples of Karnak proves the use of blue painted pottery within ritual contexts, something confirmed by representations in temple reliefs. The manufacture of this material is characteristic of the taste for the luxurious and elaborate that is a feature of a more wealthy, New Kingdom society.

Experimental procedures Eight sherds of blue painted pottery were selected for analysis; four from Malkata and four from Amarna. All the samples were surface finds, those from Malkata dating to the later part of the reign of Amenhotep III, and those from Amarna to between post the fifth year of the reign of Akhenaten and the early years of the reign of Tutankhamun. The majority of the sherds are from necked jars of various sizes (Fig. 2) with the decoration occurring on the exterior. All showed blue paint with varying degrees of preservation from quite fresh to rather eroded, plus the presence in places of black, red and white paints or slips.

Fig. 2. Large water container in blue painted ware, currently in the Agyptisches Museum, Berlin. Vessel is about 70 cm in height.

Details of the shape, fabric and decoration of the associated vessels are given in the Appendix. Polished cross-sections through the cobalt blue pigment layer, slip and body were prepared and examined in a scanning electron microscope (SEM; JEOL JSM-840A) at the Department of Earth Sciences, University of Oxford. The pigment layers were analysed using wavelengthdispersive spectrometry (WDS) associated with a JEOL 8800 SEM at the Department of Materials, University of Oxford, which had been calibrated using appropriate primary standards. This machine was run at 15 kV and 10 nA, with a spot size of 5 Ixm and 20 s count time per element, giving a detection limit of around 0.05 wt% and a relative error on analyses of minor elements of the order of 5%. Analyses were made on between three and five points on each pigment layer and the results averaged. In the case of PAL 4, the pigment layer was too weathered to obtain valid analyses. As a result of porosity, the analytical totals were typically in the range of 6 0 - 7 0 wt%, and therefore, for ease of comparison, the analytical data were normalized to 100% totals. The body and slip compositions were determined by energy-dispersive spectrometry (EDS) associated with the JEOL JSM-840A SEM using Oxford Instruments Isis 300 software in conjunction with appropriate primary standards. This second machine was run at 20 kV and 6 nA, giving a detection limit for most elements heavier than silicon of around 0.05-0.10 wt%, and 0.10-0.30 wt% for lighter elements. Typically the error on analyses of major elements is of the order of 1.0 wt%. The areas analysed were associated with magnifications of x 100 for the bodies, and with x2000x4000 for the slips. Again, as a result of low analytical totals resulting from porosity, the analytical data were normalized to 100% totals. Small samples that, because of the thinness of the layers, tended to contain both pigment and slip were taken from selected sherds, and the mineral phases present were identified by X-ray diffraction (XRD). This was carried out at the Department of Materials, University of Oxford, using a fully automated Siemens D500 powder diffractometer employing CuK~ radiation (wavelength 0.15406 nm) and a secondary monochromator. The samples were continuously spun during data collection and scanned using a step size of 0.05 ° 20 in the range of 5 - 7 5 ° 20 and a count time of 8 s per step.

Results and discussion Examination of the cross-sections through the cobalt painted pottery in the SEM indicated

COBALT BLUE PAINTED POTTERY

95

~

p

~

.....

Fig. 3. Scanningelectron photomicrograph of a crosssection through a typical vessel, showing pigment layer (1), slip (2) and body (3). Phases present in the body include quartz (Q), biotite (B), potassium feldspar (P) and iron oxides (bright backscatter contrast). The apparent backscatter contrast observed in the pigment layer reflects surface irregularities, which are the result of difficulties in polishing this soft, friable layer, rather than being due to the presence of different phases. Scale bar represents 50 ~m.

z

~

~

~

-

~ . . ~ % ~ . ~

i

ddddddd~ddddodd

iddddd2~dd~d~MN E

that, except for the Malkata sherds, PAL 4 and PAL 5, the pigment layer overlies a slip layer (Fig. 3). The pigment layer can be as much as 30 p~m in thickness but, as a result of weathering and abrasion during burial, the surviving pigment layer is frequently as little as 10 txm thick. Compared with the bodies, which are coarse textured with abundant quartz, potassium feldspar and biotite inclusions, some of which are up to 50 Ixm across, the slip layers, which are some 8 0 - 1 0 0 p~m in thickness, are much finer textured. On the basis of the interconnecting vitreous phase linking the inclusions in the slip layer and body, a firing temperature in the range of about 9 0 0 - 9 5 0 °C can be inferred (Maniatis & Tite 1981). The compositions of the cobalt blue pigment layers are characterized by high aluminium, magnesium, nickel, zinc, iron and manganese contents (Table 1). This result is as previously observed (Bachmann et al. 1980; Noll 1981), and, therefore, again consistent with the use of the cobaltiferous alums from the Dakhla and Kharga Oases (Kaczmarczyk 1986; Shortland et al. 2006). The bulk of the silica, lime, soda and potash contents of the pigment layers is most probably the result of interaction with and contamination from the underlying slips or, when no slip layer is present, the underlying bodies. In addition, because the iron oxide contents of the slips or bodies were higher than those of the pigment layers, it is possible that a

ddddddddddd~d£

d2dd2~22~M2~d~ <

..= e..

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

~" .~ .~ ~ "~ ~ ce

~. ~,

.~ ~. '~ [-

2~

b~ ©< .

-4--

A.J. SHORTLAND ETAL.

96

proportion of the iron oxide in the pigment layers was similarly the result of interaction with and contamination from the underlying slips or bodies. The XRD analyses confirmed that Al-spinels, such as the CoAl-spinel previously identified by Noll (1981), were present in the pigment layers. Because the XRD spectra for Al-spinels tend to be very similar and because only very small samples were available for analysis, it was not possible to identify specific spinels with any certainty. However, on the basis of the bulk compositions of the pigment layers, it is possible that Al-spinels incorporating cobalt, zinc, magnesium, iron and nickel were present. In addition, the calcium sulphates gypsum and anhydrite were identified in the slip layer that was sampled together with the pigment layer, again both these phases not always being present. Comparison of the compositions of the pigments with those of contemporary cobalt blue glass and of the coarse-textured cobalt blue frits found at Amarna in association with the shallow 'fritting pans', first described by Petrie (1894) and further studied by Tite & Shortland (2003), reveals both similarities and differences. Thus, the aluminium and nickel to cobalt ratios for the pigments are comparable with those for both the glass and the frits. However, the magnesium, manganese and iron to cobalt ratios for the pigments are all lower than those for both the glass and the flits, whereas the zinc to cobalt ratios are higher. Furthermore, in the case of iron, because of a possible contribution to the pigment layer from the underlying slip or body, its ratio to cobalt in the original pigment could have been even lower. In addition to these differences between the pigment layers

and the glass plus flits, the magnesium, manganese, iron, nickel and aluminium to cobalt ratios for the pottery from Malkata tend to be lower than those for the pottery from Amarna. Lower magnesium to cobalt ratios in the pigment layers might be expected in comparison with the glass, as there will be a contribution to the total magnesium content of the glass from the plant ash, which it has been suggested was added to a frit plus quartz mixture in its production (Tite & Shortland 2003). However, this is not the case for the frit itself, where natron is the probable source of the alkali (Tite & Shortland 2003). Because, on the basis of laboratory experiments, it has been shown that magnesium is the last hydroxide to precipitate (Shortland et al. 2006), the lower magnesium to cobalt ratios in the pigment could be the result of the precipitation reaction not being taken to 100% completion. However, in view of the fact that the manganese, iron and zinc to cobalt ratios for the pigments also differ from those for the glass and flits, it seems probable that the compositions of the alums used to produce the pigments differed from those used in the production of the glasses and frits. The compositions of the body clays fall into two groups (Table 2). The first group, consisting of two Malkata bodies (PAL 1 and PAL 2) and all the Amarna bodies, were made using noncalcareous clays that are similar in composition to Nile silt which, other than changes as a result of pre-treatment, tends to be uniform in composition throughout the Nile Valley (Bourriau et al. 2000, pp. 133-135). Thus, the compositions of this first group are similar to those of the two previously analysed Nile silt samples

Table 2. Body compositions (oxides in wt%) Sherd no.

Site

SiO2 A1203 CaO MgO Na20

K20

FeO TiO2 MnO SO3

Non-calcareous clays--NC PAL 1

Malkata

58.4

16.4

4.4

3.3

3.1

2.8

PAL 2 PAL 6 PAL 7 PAL 8 PAL 9

Malkata Amarna Amarna Amarna Amarna Average--NC Nile silt Nile silt

60.2 65.1 61.9 62.6 62.7 61.8 59.7 62.8

16.2 13.3 15.9 15.8 14.7 15.4 14.2 15.8

3.8 4.2 3.7 3.7 4.6 4.1 5.2 3.3

3.4 3.0 2.7 3.3 2.8 3.1 3.4 3.1

3.3 2.6 2.2 2.4 3.1 2.8 1.6 1.1

2.3 2.8 2.8 1.8 2.2 2.5 1.2 1.0

Malkata Malkata Average--C

57.9 50.8 54.3

1 4 . 2 10.5 3.9 1 1 . 8 17.8 5.2 1 3 . 0 14.1 4.5

3.3 2.5 2.9

1.8 1.6 1.7

PN 6A* PN 13A*

1.8

0.2

0.8

8.6 1.4 7.2 1.3 8.9 1.4 8.4 1.5 7.7 1.6 8.2 1.5 12.0 2.8 11.2 1.7

8.8

0.1 0.1 0.2 0.2 0.1 0.1

0.6 0.4 0.4 0.3 0.6 0.5

0.1 0.1 0.1

0.5 0.3 0.4

Calcareous clays--C

PAL 4 PAL 5

*Analyses taken from Shortland (2000).

6.8 8.1 7.5

1.2 1.9 1.5

COBALT BLUE PAINTED POTTERY from Amarna (PN6A and PN13A) (Shortland 2000), the principal differences being the high,Jr iron oxide and lower soda contents of the Nile silt samples. The second group, consisting of the remaining two Malkata bodies (PAL 4 and PAL 5), were made using calcareous clays containing 10-18% CaO (i.e. marls). The slip layers have the composition of calcareous clays containing variable amounts of sulphur (Table 3). On the basis of the XRD analyses, the sulphur is most probably associated with calcium sulphates (i.e. gypsum and anhydrite). However, the amounts of calcium sulphate are not sufficient to account for all the lime present in the clays, and after subtraction of the calcium sulphate and renormalization to 100% totals, the slip clays, with the exception of PAL 2, are still calcareous. Compared with the calcareous clays used for two Malkata bodies (PAL 4 and PAL 5), the silica contents of the slip clays are lower, but the alumina, lime and magnesia contents are higher. This difference is consistent with the fact that the slip clays are much finer textured with no large non-plastic inclusions (Fig. 3). In the case of the calcareous slip clays, the iron oxide will be incorporated into silicate phases, and therefore will fire to a pale buff colour, rather than the red colour of the non-calcareous Nile silt. These slips will therefore provide the pale background wash appropriate for the cobalt blue painted decoration. The PAL 2 slip with the lowest lime content (5% CaO as compared with 17-22% CaO) also has the lowest iron oxide content (2% FeO as compared with 6 - 1 0 % FeO), and therefore will again be paler than the underlying Nile silt body. Finally, the

97

two bodies (PAL 4 and PAL 5) on which no slips were detected were made using calcareous clays (10-18% CaO), and therefore are themselves buff firing. The origin of the calcium sulphates in the slips is not entirely clear. It is possible that, as suggested by Bachmann et al. (1980), the calcium sulphates are the result of contamination during burial and are, therefore, secondary. However, if this was the explanation, one would expect to observe similar SO3 levels in the pigment layers and the bodies whereas, compared with the 2 - 2 4 % SO3 contents in the slips, EDS analyses of the pigment layers indicate SO3 contents in the range 2 - 3 % , and the SO3 contents of the bodies are less than 1% (Table 2). Therefore, it seems probably that at least a significant proportion of the calcium sulphate in the slips was added as gypsum, perhaps to make the slip layer even paler. Firing the pottery would have resulted in dehydration of the gypsum but the resulting anhydrite will survive as its decomposition requires temperatures in excess of 1100 °C even in the presence of clay and silica (Searle & Grimshaw 1959). The gypsum observed in some pigment layers could then be the result of rehydration during burial.

Conclusions The new quantitative analyses of the cobalt blue painted pottery have confirmed that the cobalt pigment was almost certainly produced using cobaltiferous alums from the Dakhla and Kharga Oases in the Western Desert of Egypt. However, it appears that the alums used for the

Table 3. Slip compositions (oxides in wt%) Sherd no.

Site

SiO2 A1203 CaO

MgO

Na20

K20

FeO

TiO2

CoO

MnO SO3

As measured PAL 1 PAL 2

Malkata Malkata

41.3 48.1

13.8 12.1

23.0 11.2

4.9 5.3

2.3 9.0

0.9 1.4

9.7 1.9

1.9 1.2

0.1 0.1

0.3 0.1

1.9 9.8

PAL 6 PAL 8 PAL 9

Amarna Amarna Amarna Average

43.3 29.0 23.3 37.0

16.4 13.2 12.2 13.5

23.1 24.2 27.8 21.8

5.0 4.3 4.7 4.8

2.4 4.3 2.4 4.1

0.4 1.5 1.2 1.1

5.9 5.2 3.6 5.3

0.7 0.7 0.4 1.0

0.1 0.6 0.1 0.2

0.4 0.3 0.3 0.3

2.3 16.6 24.1 10.9

After subtraction of 'calcium sulphate' PAL PAL PAL PAL

1 2 6 8

Malkata Malkata Amarna Amarna

42.6 57.7 45.1 40.4

14.3 14.4 17.1 18.4

22.4 5.2 22.3 17.5

5.1 6.3 5.2 6.0

2.4 10.8 2.5 6.0

0.9 1.7 0.4 2.0

10.0 2.3 6.1 7.3

2.0 1.4 0.7 0.9

0.1 0.1 0.1 0.9

0.3 0.1 0.5 0.4

PAL 9

Amarna Average Average

39.4 45.1 54.3

20.7 17.0 13.0

18.4 17.2 14.1

8.0 6.1 4.5

4.1 5.1 2.9

2.0 1.4 1.7

6.2 6.4 7.5

0.6 1.1 1.5

0.1 0.3

0.4 0.4 0.1

Body--C*

*Analytical data for calcareous clay body taken from Table 2.

0.4

98

A.J. SHORTLAND ET AL.

pigments were somewhat different in composition from those used in the production of cobalt blue frits and glasses. This may be due to a different alum source being used for different materials, or may instead suggest a sorting of alum from one source into grades, each, in the mind of the worker, more suitable for one use than another. There were also differences in composition between the alums used at Malkata and Amarna, but these latter differences are significantly less than those between the cobalt blue paint and the glass and frit, and when the different states of preservation of the pigment layers are taken into account, may not be significant. Both non-calcareous Nile silt and calcareous clay (i.e. marl) were used in the production of the pottery bodies, the former being used for all the analysed pottery samples from Amama whereas both clays were used for the analysed samples from Malkata. Prior to painting, the Nile silt bodies were first coated with a calcareous clay slip to which gypsum had probably been added. In contrast, no slip was applied to the calcareous clay bodies, the pigment being painted directly onto the body.

Appendix: Description of pottery samples The designations of pottery shape and fabric employed here are those of the so-called Vienna System, which has been discussed by Nordstrrm & Bourriau (1993) and Bourriau et al. (2000, pp. 130-132). P A L 1: Malkata Shape: medium to large jar, type unknown. Fabric: Nile silt B2, with a cream surface coating outside. Decoration: narrow black and broader blue horizontal bands. P A L 2: Malkata Shape: a very tall necked jar with ovoid body (see Hope 1989a, fig. 1 la); rim fragment. Fabric: Nile silt B2, with a cream surface coating inside rim top and outside. Decoration: wide blue petals tapering upwards outlined in red, with red intervening strokes (stamens).

Fabric: marl A4; the thin cream surface on the exterior resembles a distinct coating. Decoration: blue overlapping petals outlined with thick black lines--gap--blue band with an upper black line. P A L 5: Malkata Shape: shallow bowl with very short convex sides, compare Hope (1989a, fig. 5f); rim fragment. Fabric: marl A4. Decoration: blue inside and outside. P A L 6: Amarna Shape: a medium-sized jar, type unknown; fragment from the upper body. Fabric: Nile silt B2 with whitish-cream surface coating outside. Decoration: part of a frieze of blue petals tapering downwards, outlined in brown, the upper part over a blue band outlined in brown. P A L 7: Amarna Shape: a medium-tall- to tall-necked jar with ovoid body, compare Hope (1991, fig. 6e). Fabric: Nile silt B2, with a pinkish-orange surface coating outside. Decoration: gap---two bands of blue outlined in reddish brown, the upper with a red line, over which is a frieze of blue petals tapering downwards outlined in reddish-brown, with thin red strokes (stamens) between the lower part of the petals, which terminate at a brown line. P A L 8: Amarna Shape: short- or medium-tall-necked jar with ovoid body, compare Hope (1991, figs 5j and 6c); part of neck and upper body extant. Fabric: Nile silt B2, with a pinkish-orange surface coating outside. Decoration: a single wide blue band from base of the neck over upper body, over which details in red and brown are added to produce a design of overlapping petals below an upper composite band, and below which is a brown line--gap. P A L 9: Amarna

P A L 4: Malkata Shape: medium to large jar, type unknown but possibly compare Hope (1989a, fig. 6a); fragment from the upper body.

Shape: a short-necked jar, compare Hope (1991, fig. 5j); fragment from the upper body. Fabric: Nile silt B2, with cream surface coating outside.

COBALT BLUE PAINTED POTTERY Decoration: fugitive; traces of blue with the outlines of overlapping petals terminating at a brown line.

References ARNOLD, D. 1993. Techniques and traditions of manufacture in the pottery of ancient Egypt. In: ARNOLD, O. & BOURRIAU,J. (eds) An Introduction to Ancient Egyptian Pottery. Philipp von Zabern, Mainz am Rhein, 9-141. ARNOLD, D. & BOURRIAU,J. (eds) 1993. An Introduction to Ancient Egyptian Pottery. Philipp von Zabern, Mainz am Rhein. ASTON, D. A. 1998. Die Keramik des Grabungsplatzes QI, Teil 1: Corpus of Fabrics, Wares and Shapes. Philipp von Zabern, Mainz am Rhein. BACHMANN, H. G., EVERTS, H. & HOPE, C. A. 1980. Cobalt blue pigment on 18th Dynasty pottery. Mitteilungen der Deutschen Archaologischen Instituts Abteilung Kairo, 36, 33-37. BELL, M. R. 1987. Regional variation in polychrome pottery of the 19th Dynasty. Cahiers de ta Cdramique Egyptienne, 1, 49-76. BOURRIAU,J. D., NICHOLSON,P. T. & ROSE, P. J. 2000. Pottery. In: NICHOLSON, P. T. & SHAW, I. (eds) Ancient Egyptian Materials and Technology. Cambridge University Press, Cambridge, 121-147. HOPE, C. A. 1980. Blue-painted pottery of the XVlllth Dynasty. PhD thesis, University College London. HOPE, C. A. 1987. Innovation in the decoration of ceramics in the mid-18th Dynasty. Cahiers de la Cdramique [~gyptienne, 1, 97-122. HOPE, C. A. 1989a. The XVIIIth Dynasty pottery from Malkata. In: HOPE, C. A. (ed.) Pottery of the Egyptian New Kingdom: Three Studies. Victoria College Press, Burwood, 3-44. HOPE, C. A. 1989b. Pottery of the Ramesside Period. In: HOPE, C. A. (ed.) Pottery of the Egyptian New Kingdom: Three Studies. Victoria College Press, Burwood, 4 7 - 84. HOPE, C. A. 1989c. Amphorae of the New Kingdom. In: HOPE, C. A. (ed.) Pottery of the Egyptian New Kingdom: Three Studies. Victoria College Press, Burwood, 87-126. HOPE, C. A. 1991. Blue-painted and decorated pottery from Amarna: a preliminary corpus. Cahiers de la Cdramique [~gyptienne, 2, 17-92. HOPE, C. A. 1997. Some Memphite blue painted pottery of the mid-18th Dynasty. In: PHILLIPS, J. (ed.) Studies in Honour of Martha Rhoads Bell. van Sichen Books, San Antonio, Texas, 249-286.

99

KACZMARCZYK, A. 1986. The source of cobalt in ancient Egyptian pigments. In: OLIN, J. S. & BLACKMAN, M. J. (eds) Proceedings of the 24th International Archaeometry Symposium. Smithsonian Institution Press, Washington, DC, 369-376. KOZLOFF, A. P. & BRYAN, B. M. 1992. Egypt's Dazzling Sun: Amenhotep III and his World. Cleveland Museum of Art, Cleveland, OH. LILYQUIST, C. & BRILL, R. H. 1993. Studies in Ancient Egyptian Glass. Metropolitan Museum of Art, New York. MANIATIS, Y. & TITE, M. S. 1981. Technological examination of Neolithic-Bronze Age pottery from Central and South-east Europe and from the Near East. Journal of Archaeological Science, 8, 59-76. NOLL, W. 1981. Mineralogy and technology of the painted ceramics of ancient Egypt. In: HUGHES, M. J. (ed.) Scientific Studies in Ancient Ceramics. British Museum Occasional Papers, 19, 143-154. NOLL, W. & HANGST, K. 1975. Zur Kenntnis der alt~igytische Blaupigments. Neues Jahrbuch Mineralogische, Monatshefte, 209-214. NORDSTROM, H.-A. & BOURRIAU, J. 1993. Ceramic technology: clays and fabrics. In: ARNOLD, D. & BOURRIAU, J. (eds) An Introduction to Ancient Egyptian Pottery. Philipp von Zabern, Mainz am Rhein, 168-182. PETRIE, W. M. F. 1894. Tell el-Amarna. Methuen & Co., London. REHREN, T. 2001. Aspects of the production of cobaltblue glass in Egypt. Archaeometry, 43, 483-490. RIEDERER, J. 1974. Recently identified Egyptian pigments. Archaeometry, 16, 102-109. SEARLE, A. B. & GRIMSHAW,R. W. 1959. The Chemistry and Physics of Clays and other Ceramic Materials. Interscience, New York. SHORTLAND, A. J. 2000. Vitreous Materials at Amarna: the Production of Glass and Faience in 18th Dynasty Egypt. British Archaeological Reports. International Series, $827. SHORTLAND, A. J., TITE, M. S. & EWART, I. 2006. Ancient exploitation and use of cobalt alums from the Western Oases of Egypt. Archaeometry, 48, 153-168. TITE, M. S. & SHORTLAND, A. J. 2003. Production technology for copper- and cobalt-blue vitreous materials from the New Kingdom site of Amarna--a reappraisal. Archaeometry, 45, 285-312. WARACHIM, H., RZECHULA, J. & PIELAK, A. 1985. Magnesium-cobalt(II)-aluminium spinels for pigments. Ceramics International, 11, 103-106.

The ceramic technology used in the manufacture of Iron Age pottery from Galilee S H L O M O S H O V A L 1, P I R H I Y A H BECK 2'* & ESTHER Y A D I N 2

1Geology Group, Department of Natural Sciences, The Open University of Israel, The Dorothy de Rothschild Campus, 108 Ravutski Street, Raanana, Israel (e-mail: [email protected]) 2lnstitute of Archaeology, Tel-Aviv University, Tel-Aviv, Israel (*deceased) Abstract: The Iron Age ceramic technology used in the manufacture of functional pottery

from Galilee was studied. Applied methods included petrography, X-ray diffraction, infrared spectroscopy and chemical analyses. The results demonstrate that the potters in biblical times had knowledge of raw materials and manufacturing technologies, enabling them to select suitable ones, according to their advantages, for the manufacture of cooking pots, storage jars and tableware vessels. The paper describes the petrography of the pottery, the composition of the ceramic matrix, the firing temperature, the tempering of the cooking pots, the processes that allow consolidation of the ceramic and the origin of the pottery. The results are placed in an archaeological context.

Iron Age pottery was excavated at Tel-Hadar, on the eastern shore of the Sea of Galilee. The excavations were carried out in the framework of The Land of Geshur Archaeological Project, of The Institute o f Archaeology, Tel Aviv University (Kochavi 1989). A large pillared building from the 1 lth century BC was excavated on the site. The ground floor of the building was used for storage and marketing and included storage rooms, granary rooms and a columned hall (Kochavi 1989). Many storage and tableware vessels and some cooking pots were found in the storage rooms. The pillared building was destroyed by conflagration at the end of the l lth century BC and the site was subsequently abandoned. The extreme intensity and high temperature of the conflagration demonstrate that it was the grain-filled granary rooms that bumed (Shoval et al. 1989). No signs of any unusual intensity of conflagration were observed in the storage rooms containing the pottery (Kochavi 1989). As the event occurred in the late eleventh century Bc, it may have happened during the wars of King Saul against enemies of the Israelite settlers in Gilead (II Samuel 2:9). A small agricultural and fishing village was built on the destroyed site in the ninth century BC (Kochavi 1989). A study of the production technology of ancient pottery using mineralogical and petrographic methods can focus on the type of raw material and on the firing process (Maggetti 1982, 1994; Maggetti et aI. 1984). The ancient

potters usually used local raw materials suitable for pottery production. Two main types of raw materials were used, calcareous and noncalcareous clays (Maniatis & Tite 1981; Shoval 2003), each of which has different advantages for pottery production (Maggetti 1981; Maggetti et al. 1984). Improvements in the properties of the local raw materials, in terms of forming, firing and the resultant physical properties of the ceramic were obtained by sorting, tempering by adding non-plastic coarse particles, and mixing of clays (Hein et al. 2004). The firing process can be assessed from the phase assemblage in the ceramic (Maggetti 1982; Shoval 1994, Mirti et al. 1999). The reactions that occur during firing of clay follow two main mechanisms: continuous reactions producing only compositional variations of phases, and discontinuous reactions leading to nucleation and growth of new mineral phases (Duminuco et al. 1998; Riccardi et al. 1999). The evolution of the phase assemblages depends on the mineralogy and chemistry of the raw material, their particle-size distribution, maximum firing temperature, heating rate, duration of firing, and the oxygen fugacity of the kiln atmosphere (Moropoulou et al. 1995; Livingstone Smith 2001). Different newly formed phases are observed when firing typical raw materials (Maggetti 1981, 1984; Shoval 1988). The firing of noncalcareous material can be approximated by firing of kaolinitic clay. Kaolinite dehydroxylates at

From: MAGGETTI,M. & MESSIGA,B. (eds) 2006. Geomaterialsin CulturalHeritage. Geological Society, London, Special Publications, 257, 101-117. 0305-8719/06/$15.00 © The Geological Society of London 2006.

102

S. SHOVAL ETAL,

400-650 °C and a pseudo-amorphous phase, metakaolinite, is formed (Mackenzie & Rahman 1987; Mackenzie et al. 1988). The thermal transformation from metakaolinite to mullite starts with short-range ordered material. According to Freund (1974), at about 900 °C the metakaolinite transforms into a cubic phase described as spinel, which is poorly crystallized. Gualtieri & Belloto (1998) stated that after the loss of the residual hydroxyls, metakaolinite transforms into three phases: ",/-alumina (which has a spinel structure), mullite (AlgSiOs) and silica. Well-crystallized phases of mullite and cristobalite (SiO2) are observed at about 1200 °C (Dong-Li & Thomson 1991; Belloto et al. 1995). On firing smectitic (montmorillonitic) clay, the smectite dehydroxylates at 600-800 °C, and with increasing temperature meta-smectite is formed (Shoval 1988). The major phases formed by the heating of smectite have been identified as metasmectite, spinel, mullite, cordierite and cristobalite (Seyama & Soma 1986). After firing to 800 °C, some rehydroxylation of the clay minerals may take place within the fired material in ambient conditions (Heller et al. 1962; Shoval et al. 1991). When firing calcareous raw materials, the dehydroxylation of the clay stimulates earlier decarbonation of calcite (Heller-Kallai et al. 1986, 1987). The microcrystalline calcite of the calcareous raw material decomposes above 600 °C (Shoval et al. 1993), Carbon dioxide gas (CO2) is released, and quicklime (CaO) is formed. Above 900 :C, the quicklime reacts with amorphous phases resulting from the dehydroxylation of clays and Casilicates; gehlenite (Ca2AI2SiOT), plagioclase anorthite (CaAl2Si208) and pyroxene wollastonite-diopside (CaSiO3) are formed (Maggetti 1981, 1982; Maggetti et al. 1984; Shoval 1988). Under ambient conditions, newly formed calcite may crystallize from non-reacted quicklime (Shoval et al. 1993, 2003; Shoval 2003). During this process, the hygroscopic quicklime picks up moisture from the air, forming calcium hydroxide (Ca(OH)2) (Shoval 1988; Shoval et al. 1993). The Ca(OH)2 slowly reacts with atmospheric CO2 and calcite is formed. Some calcite may also crystallize during cooling (Deutsch & Heller Kallai 1991). Newly formed calcite is usually characterized by small grain size and a low degree of crystallinity (Maciejewski & Relier 1989). The formation of microcrystalline calcite during the recarbonation process may lead to cementation of the material, like the process in lime-based plasters and mortars (Kingery 1988; Moropoulou et al. 2005). On firing above 900 °C, the formation of Ca-silicates reduces the amount of free lime

and thus the amount of the newly formed calcite. It should be noted that Ca-silicate phases may contain some carbonate, although it is not so likely in dense ceramic material. The purpose of the present research is to study the Iron Age ceramic technology used in the manufacture of functional pottery (cooking pots, storage jars and tableware vessels) from Galilee. E x p e r i m e n t a l methods Ceramics

The studied pottery was excavated by archaeologists at Tel Hadar from strata of the eleventh and the ninth centuries BC (Kochavi 1989). The pottery from the 1 lth century BC stratum was excavated from a large pillared building and was found in storage rooms located on the ground floor of the building. Pottery from the ninth century BC stratum was excavated from the site of a small village built on the destroyed site. R a w materials

Some alluvia and soils, which are equivalent to the raw materials used for the production of the investigated pottery, were also analysed. These raw materials include basaltic soil of Golan, Terra Rossa of Galilee, alluvium from Wadi Samakh, alluvium soil of the Jordan Valley and rendzina soil of Galilee. Methods

The pottery was investigated using the following methods. Petrography. Thin sections of the pottery were analysed under a polarizing microscope. Fourier transform scopy. A Nicolet

infrared

FTIR

spectro-

FTIR spectrometer and 'Omnic' software were used. Spectra from powdered samples of the ceramic matrix (excluding the temper particles) were obtained. Samples of the bulk pottery (including the temper particles) were also analysed. The powders were diluted in KBr and pressed into discs. These were heated to 110 °C and 350 °C to remove water bands. Before measurement, the discs were repressed to improve the resolution of the spectra.

(XRD). XRD was performed on powder samples with a Philips PW3710 diffractometer using C u I ~ radiation at 35 kV and 40mA, and a curved graphite

X-ray diffractometry

CERAMIC TECHNOLOGY OF IRON AGE POTTERY monochromator. Diffractograms from powdered samples of the ceramic were obtained. For the identification of the types of clay minerals in the raw materials, oriented clay fractions, after carbonate removal with diluted HC1, decantation and glycolation, were employed.

Chemical analyses. The major element contents in the ceramic matrix (excluding the temper particles) were determined by energy dispersive spectrometry (EDS) using a LINK-10000 EDS system (Oxford ISIS) attached to a JEOL (JSM-840) scanning electron microscope (SEM) and ZAF4/FLS program. The reference standard for X-ray microanalyses of Microanalyses Consulted Ltd. (Registered standard 1691) was used for the EDS analysis. The analyses were carried out on 'fresh' sections of the pottery. The samples were carbon coated. At least three measurements were obtained from each sample. For deeper penetration an 18 kV electron beam was used. The beam width was 15 nm and the resolution 1 Ixm. Accumulations of 6 0 s ( 3 - 5 iterations) were applied. The analytical error is _+2%. The ceramic matrix area to be analysed was marked by scratching around it, and was located using the SEM. A back-scattered electron (BSE) detector attached to the SEM was used to ensure that calcite temper was far from the analysed area. Results and discussion

Petrography of the pottery Petrographic groups of the pottery, types of pottery and a list of the investigated pottery are shown in Table 1. Examples of cooking pots and storage jars are shown in Figure 1. The pottery was divided into petrographic groups according to the characteristics of the ceramic matrix and the nature of the temper particles. The groups differ in the type of raw material used in their manufacture.

Identification of the raw materials. Potential raw materials available in the region are clayey soils, alluvia and sediments. Each raw material has typical petrographic characteristics, which allow unambiguous identification. Alluvia and soils usually contain attached coarse particles and thus the use of these raw materials does not require tempering with particles from an additional source. Various types of unsorted and rounded grains are an indication that calcareous alluvium was used as raw material, whereas a single type of fragments attached to the raw material indicates that soil was used. High

103

amounts of iron oxide and quartz silt in the ceramic matrix are an indication of soil-derived raw material. The soils are enriched in quartz silt because of the contribution of dust. On the other hand, clayey sediments usually lack coarse particles that could act as temper and therefore require the addition of temper particles from some other source. The potter obviously preferred to add coarse particles of a local material, as was corroborated in ethnological studies (Maggetti 1982).

Petrographic groups of cooking pots. The cooking pots were divided into three major petrographic groups. Group 1: manufactured from basaltic soil and tempered with calcite crystals. This group consists of cooking pots from the l lth and ninth century BC strata. The presence of some basalt fragments within a noncalcareous or slightly calcareous blackish or dark matrix, rich in iron oxide and quartz silt (Fig. 2a), indicates that the raw material was basaltic soil. No limestone particles are found. The pots were tempered with broken pieces of large calcite crystals.

Group 2: manufactured from Terra Rossa and tempered with calcite crystals. This group consists of cooking pots from the 1 lth and ninth century BE strata. The presence of some limestone fragments within a noncalcareous or slightly calcareous brownish-red matrix, rich in iron oxide and quartz silt, indicates that the raw material was Terra Rossa. No basalt particles are found. Similar to Group 1, the pots were tempered with calcite crystals.

Group 3: manufactured from basaltic soil not tempered with calcite crystals. This group consists of cooking pots from the ninth century BE stratum. Similar to Group 1, this group was manufactured from basaltic soil, but not tempered with calcite crystals (Fig. 2b).

Petrographic groups of storage jars and tableware vessels. The storage jars and tableware vessels were divided into six major petrographic groups. The pottery in these groups was excavated from the l lth century B¢ pillared building.

Group 4: manufactured from calcareous alluvium. This group consists of storage jars, pithoi, kraters, bowls, jugs, a chalice, a lamp, and amphora. The presence of unsorted and rounded grains of basalt, chalk, and limestone and quartz sand within a calcareous light matrix (Fig. 3a) indicates that the raw material was calcareous alluvium.

104

S. SHOVAL ET AL.

Table 1. Petrographic groups of the pottery, types of pottery and list of the investigated Iron Age pottery excavated from Tel-Hadar Group

Types of pottery

List of pottery

Group 1: manufactured from basaltic soil and tempered with calcite crystals la Cooking pots, 1lth century Bc 1574/1 (TH-7) 1574/3 (TH-27) 1783/3 (TH-31) 1834/2 (TH-8) lb Cooking pots, 9th century Bc 110/158 (TH-118) 119/158 (TH-117) Group 2: manufacturedfrom Terra Rossa and tempered with calcite crystals 2a Cooking pots, 1lth century Bc 1734/1 (TH-2) 1756/1 (TH-6) 2264/1 (TH-259) 2b Cooking pots, 9th century Bc 166/181(TH-110) 1115/252 (DTH-9) 1117/256 (TH-103) 1127/252 (DTH-12) 1250/256 (TH-104) 1672/295 (TH-47) 1672/295 (TH-51) 1244/248 (TH- 124) Group 3: manufacturedfrom basaltic soil but not tempered with calcite crystals Cooking pots, 9th century Bc 11/106 (TH-115) 171/185 (TH-119) 1115/252 (TH-112) 1177/256 (TH-106) 1244/248 (TH-127) 1244/248 (TH-125) Group 4: manufactured from calcareous alluvium Storage jars 1277/1 (TH-202) 1562/1(TH-24) 1757/1 (TH-20) 1783/2 (TH-28) 1810/1 (TH-18) 1823/1 (TH-11) 1834/1 (TH-200) 1864/1(TH-35) 1873/1 (TH-16) 2139/1 (TH-309) Pithoi 1302/1 (TH-15) 2238/1 (TH-264) Kraters 1740/1 (TH-14) 1828/2 (TH-32) Bowls 1746 (TH-98) 1844/2 (TH-40) Jugs 1790/1 (TH-9) 2105/1 (TH-261) Lamp 1868/2 (TH-44) Chalice and amphora 2149/1 (TH-36) 2217/2 (TH-251) Group 5: manufactured from calcareous alluvium soil Storage jars 1783/1 (TH-30) 2039/1 (TH-313) 1866/1 (TH-43) 1867/2 (TH-204) Krater 1267/1 (TH-250) Jug and amphora 1861/1 (TH-315) 2340/1 (TH-265) Group 6: manufactured from calcareous rendzina soil Storage jar 2060/1 (TH-302) Bowl and amphora 2152/1 (TH-310) 1803/1(TH-303) Group 7: manufactured from calcareous marl Storage jars 2105/2 (TH-326) 2217/1 (TH-263) Jugs 1565/1 (TH-206) 2207/1 (TH-317) Group 8: manufactured from calcareous sediment of the coastal plain Storage jars 2225/1 (TH-301) 2220/1 (TH-270) Jugs 1702/1 (TH-316) 2348/1 (TH-260) Group 9: manufacturedfrom calcareous sediment from a foreign coastal area 'Greek' bowl 1842/1 (TH-325)

1778/1 (TH-37) 1844/3 (TH-42) 1117/256 (TH-105) 2021/1 (TH-5) 1115/252 (TH-113) 1227/256 (TH-107) 1672/295 (TH-50) 1108/252 (DTH-I1) 1244/248 (TH-126) 1734/2 1809/1 1828/1 1865/1 2148/1

(TH-3) (TH-4) (TH-33) (TH-209) (TH-308)

2108/1 (TH-13)

2044/1 (TH-324)

2374/1 (TH-99)

The listof pottery includesthe excavationsymbol(e.g. 1574/1)and the samplenumberusedfor the experiments(e.g. TH-7,whereTH indicatesTel Hadar).

Group 5: manufactured from calcareous alluvium soil. This group consists of storage jars, kraters, jugs and amphora. The temper particles are smaller and rounder than those in Group 4 and consist of highly weathered basalt grains and some iron oxide concretions. The matrix is richer in iron oxide and quartz silt. This composition indicates that the raw material was calcareous alluvium soil.

Group 6: manufactured from calcareous rendzina soil. This group consists of storage jars, amphora and a bowl. The presence of chalk fragments within a calcareous matrix rich in quartz silt (Fig. 3b) indicates that the raw material was calcareous rendzina soil. The matrix contains relicts of basaltic soil in the form of mud balls, as well as a few basalt fragments (Fig. 4a), indicating that the calcareous

CERAMIC TECHNOLOGY OF IRON AGE POTTERY

105

(a)

Fig. 2. Thin-section photomicrographs of cooking pots. (a) Group 1: manufactured from basaltic soil and tempered with calcite crystals, and characterized by the presence of some basalt fragments within noncrystallized blackish or brownish-red noncalcareous matrix (a basalt fragment is observed at the upper right corner, and a rhombohedral calcite crystal in the middle). (b) Group 3: manufactured from basaltic soil but not tempered with calcite crystals (a basalt phenocryst with ophitic texture is observed in the middle). (40x magnification, cross-polarized light.)

Fig. 1. Examples of 1lth century ac pottery studied in the present research. (a) Cooking pot. (b) Storage jar. The pottery was excavated by archaeologists at Tel Hadar (Kochavi 1989).

rendzina soil raw material was mixed with noncalcareous basaltic soil.

Group 7: manufactured from calcareous marl. This group consists of storage jars and jugs. The composition of the calcareous matrix is almost lacking in quartz silt, and the presence of Foraminifera fossils (Fig. 4b) indicates that the raw material was selected from marl. No basalt fragments are found. The matrix contains relicts of Terra Rossa in the form of mud balls, as well as a few limestone fragments (Fig. 4b), indicating that the calcareous marl raw material was mixed with noncalcareous Terra Rossa.

Group 8: manufactured from calcareous sediment of the coastal plain. This group

consists of storage jars and jugs. The presence of a few rounded grains of quartz and some skeletal fragments of Mediterranean fauna within a calcareous matrix rich in quartz silt (Fig. 5a) indicates that the raw material was selected from calcareous sediment of the coastal plain.

Group 9: manufactured from calcareous sediment from a foreign coastal area. Group 9 consists of a single 'Greek' bowl. The matrix is calcareous, rich in Foraminifera fossils and poor in temper particles (Fig. 5b). A large fragment of coarse crystalline limestone or marble is observed. The presence of few rounded grains of quartz and some skeletal fragments indicates that the raw material was selected from calcareous sediment near a coast. The cooking pots were manufactured from noncalcareous or slightly calcareous soil. The soil was tempered with calcite crystals from an additional source. On the other hand, the

106

S. SHOVAL ETAL.

(A)

(a)

(b}

(bl

Fig. 3. Thin-section photomicrographsof storage jars. (a) Group 4: manufactured from calcareous alluvium and characterized by the presence of various types of unsorted and rounded grains within crystallized calcareous light matrix (basalt grain is observed in the middle and at the lower right comer, chalk grains are dispersed throughout the section). (b) Group 6: manufactured from calcareous rendzina soil and characterized by the presence of chalk fragments attached to the soil raw material within partly crystallized calcareous matrix. (Chalk grains are dispersed throughout the section, 40 × magnification, cross-polarized light.)

Fig. 4. Thin-section photomicrographsof tableware vessels. (a) Group 6: made by the mixing of rendzina soil raw material with basaltic soil and characterized by the presence of relicts of noncalcareous soil in the form of mud balls (a large dark ball is observed in the middle). (b) Group 7: made by the mixing of foraminiferousmarl raw material with Terra Rossa soil and characterized by relicts of noncalcareous soil in the form of mud balls. (Foraminifera fossils are dispersed throughout the section, and a dark ball appears in the middle left comer; 40x magnification,cross-polarized light.)

storage jars and most of the tableware vessels were manufactured from calcareous alluvium or calcareous soil. The largest group (Group 4) of storage jars and tableware vessels was manufactured from calcareous alluvium, as identified by the presence of various types of unsorted and rounded grains in a light ceramic matrix. In petrographic Groups 6 and 7, highly calcareous rendzina soil or marl was mixed with noncalcareous Terra Rossa or basaltic soil, respectively, as observed from the presence of relicts of noncalcareous material appearing in the form of mud balls. Similar mixing of rendzina soil or foraminiferous marl with Terra Rossa was previously reported in the region in the manufacture of Roman pottery (Wieder & Adan-Bayewitz 1999) and Iron Age pithoi from Tel Sasa (Cohen-Weinberger & Goren 1996),

respectively. The mixing was performed to enrich the clay component of the raw material, and to improve the properties of the raw material for pottery manufacture.

The composition of the ceramic matrix: the mineral and phase composition X-ray diffraction.

XRD is a useful method for estimating the mineral composition of pottery fired at a high temperature (Maggetti 1981, 1982; Maggetti et al. 1984; Bertolino & Fabra 2003; Eramo et al. 2004). However, this method is suitable for the determination of crystalline materials, but not amorphous phases. The XRD results of this study demonstrate that peaks of Ca-silicates that may relate to the firing process are missing or very weak.

CERAMIC TECHNOLOGY OF IRON AGE POTTERY

The rise of the baseline in the diffractograms of the examined pottery is related to the presence of a pseudo-amorphous phase of fired clay. Absence of newly formed Ca-silicates indicates that the pottery was fired below 900 °C. XRD was also used here as accurate evidence for the presence of calcite in the ceramic matrix. The strong peaks of calcite in the diffractograms of matrix samples of the storage jar and tableware vessels are in agreement with their calcareous composition observed through petrographic analysis. Weak or missing peaks of calcite in the diffractograms of matrix samples of the cooking pots confirm their slightly calcareous or noncalcareous composition. In the alluvium and soil raw materials, which are equivalent to those used for the production

107

of the investigated pottery, the major type of the clay mineral was also analysed by XRD. The results demonstrate that smectite is the dominant clay mineral in the basaltic soil of Golan, in the Terra Rossa of Galilee, in the alluvium of Wadi Samakh, in the alluvium soil of the Jordan Valley and in the rendzina soil of Galilee. The raw materials contain some kaolinite mixed with the smectite. The calcareous sediments in the region are also dominated by smectitic IS (illite-smectite) accompanied by palygorskite (Shoval 2004). FTIR spectroscopy. F'I'IR spectra of the ceramic matrix of representative cooking pots and storage jars are shown in Figure 6. The FTIR data are given in Table 2 and arranged according

o.~oi

(a)

0.35

(a)~l,,3,

]

,o41

0.3O 0,25. 0.2O 0.15.

0.3

I= j ~ 0 .2

(b)

<

(C)

142o

713

0.7

0otl.I

0..51 •

0.3

I

t

1040

77

463 "/'/8 713

0,2

Fig. 5. Thin-section photomicrographs of tableware vessels. (a) Group 8: manufactured from calcareous sediment of the coastal plain and characterized by the presence of few rounded grains of quartz and some skeletal fragments of Mediterranean fauna within partly crystallized calcareous light matrix (a skeletal fragment is observed in the middle). (b) Group 9: manufactured from calcareous sediment of a foreign coastal area and characterized by a calcareous matrix which is rich in Foraminifera fossils and poor in temper particles. (Foraminifera fossils are dispersed throughout the section; a large fragment of coarse crystalline limestone or marble is observed in the lower left corner; 40× magnification, cross-polarized light.)

Wave.umber (era"1) Fig. 6. FTIR spectra of representative pottery. (a) Ceramic matrix of a cooking pot of Group 1 showing the main S i - O stretching band of fired clay with one maximum at 1041 cm -1. (b) Ceramic matrix of a cooking pot of Group 1 showing splitting of the main S i - O stretching band into bands at 1053 and 1079 c m - 1. (c) Bulk pottery material of a cooking pot of Group 1 showing the main CO3 band of the calcite temper at 1426 c m - i. (d) Ceramic matrix of a storage jar of Group 4 showing the main CO3 band of newly formed calcite at 1438 cm-1.

108

S. SHOVAL ET A L

Table 2. The range of the band maxima in F'FIR spectra and the percentages of calcite in the ceramic matrix of Iron Age pottery excavated from Tel-Hadar

Group Group Group Group Group Group Group

Pottery type la lb 2a 2b 3 4

Group 5

Group 6 Group 7 Group 8 Group 9

Cooking pots Cooking pots Cooking pots Cooking pots Cooking pots Storage jars Pithoi Kraters Bowls Jugs Chalice Amphora Lamp Storage jars Krater Jug Amphora Storage jar Bowl Storage jars Jugs Storage jars Jugs 'Greek' bowl

Si-O str. (I) (cm - J) 1039-1055 1053-1055 1050-1056 1045-1060 1045-1059 1035-1045 1037-1045 1031 - 1039 1034-1035 1033-1034 1039 1041 1043 1050-1054 1061 1061 1050 1041 1045 1037-1047 1038-1045 1037-1040 1038-1040 1040

Si-O str. (II) (cm- i ) 1077-1080 1074-1083 1076-1082 1078-1083 1078-1081 1082 1082 1080 -

Si-O, AI-O def. (cm- i) 467-475 463-467 461-467 461-468 461-465 461-469 470-475 459-463 463-464 458-459 467 463 466 462-466 467 466 465 471 467 466-470 468-470 471-473 470-473 470

CO3 (cm - i) 1430-1439 1430-1432 1430-1432 1430-1441 1432-1435 1430-1441 1430-1431 1430-1448 1431 - 1 4 3 2 1434-1436 1430 1432 1431 1430-1436 1435 1432 1440 1430 1430 1430-1435 1435-1440 1430-1432 1431 - 1 4 3 2 1432

% Calcite 7-13 5-12 3-14 7-15 0-10 33-57 39-40 30-55 50-55 39 47 30 37 33-50 36 35 30 55 35 50-55 45-75 40-55 30-55 45

The FFIR data are arranged accordingto the petrographic groups and the potterytype. str., stretching;def., deformation. to the petrographic groups. FTIR spectroscopy is a useful method for the phase analysis of pottery (Shoval 1994, 2003; De Benedetto et al. 2002), as this technique allows the identification of both the amorphous and the crystalline materials (Shoval et al. 2001, 2002). To identify the mineral and phase composition of the ceramic matrix of the pottery, the spectra of the pottery are compared with those of the individual minerals and phases expected; for example, fired kaolinite or fired smectite (both fired at 750 °C), quartz and calcite (Fig. 7). The spectra of the ceramic matrix of the cooking pots are indicative of noncalcareous ceramic, showing strong S i - O vibrations characteristic of fired clay (Fig. 6a and b). Those of the storage jars are indicative of calcareous ceramic, showing strong CO3 vibrations characteristic of newly formed calcite, and weaker S i - O vibrations characteristic of fired clay (Fig. 6d). The main S i - O stretching mode appears in the pottery at 1035-1083 cm -1 and the combined S i - O , A 1 - O bending mode at 4 5 9 - 4 7 5 cm -1 (Table 2). The quartz doublet at about 779 and 798 cm -1 (Fig. 7c) appears weak in the spectra of the pottery (Fig. 6) and demonstrates that some quartz is present (the main S i - O band of quartz at 1085 cm - l

is distorted by the main band of the fired smectite). In the ceramic matrix of the pottery, the main CO3 band is located above 1430 cm - l (Table 2) and commonly characterized by a broad shape at the top (Fig. 6d), whereas in the spectra of natural calcite it appears at 1422cm -~ (Fig. 7d). The higher frequency of the main CO3 band in the ceramic matrix can be related to the presence of newly formed calcite (Shoval et al. 1993). This shift may result from the low degree of crystallinity of the newly formed calcite and the existence of impurities of magnesium and iron. It seems that the microcrystalline calcite of the ceramic matrix was fully decarbonated during firing and calcite was newly formed in the ceramic after firing. On the other hand, in the spectra of calcite temper of the cooking pots, the main CO3 band appears at about 14251429 cm -~ and is sharp at the top (Fig. 6c). The lower frequency in the calcite crystal is in accordance with its higher degree of crystallinity and lower amount of impurities. It seems that the location of the calcite temper inside the clay matrix shield allowed the decomposed calcite to be recarbonated and reconstructed almost to its original form (Shoval et al. 1993). However, in low-temperature firing the calcite in temper

CERAMIC TECHNOLOGY OF IRON AGE POTTERY

o ]l.I

t

A"

0.4

17473

0,3

O,2 1A 121.0":

o~ 0.8-" ~

0.4.

(c)

<

0,15

(d) o.e o.6 oA o2

Wavenumber

(cm-!)

Fig. 7. FTIR spectra of individual phases expected in pottery composition. (a) Fired smectite (at 750 °C). (b) Fired kaolinite (at 750 °C). (c) Quartz. (d) Calcite. may remain as primary calcite. The relatively smooth faces of the calcite temper seen in the thin sections confirm these observations. FFIR spectroscopy was also used for quantitative analysis of calcite in the ceramic matrix (Table 2). These amounts were calculated according to the intensities of the main CO~ band of calcite at about 1 4 3 0 - 1 4 4 8 c m relative to the main S i - O band of the fired clay at about 1035-1083 cm -~ (Fig. 6d). Calibration curves were used for this determination. The results demonstrate that the amounts of the calcite component in the ceramic matrix (excluding the temper particles) range between 0 and 15% in the cooking pots and between 30 and 57% in the storage jars and the tableware vessels.

The chemical composition of the ceramic matrix The content of the major elements in the ceramic matrix was obtained in the selected pottery by point chemical analysis by EDS-SEM. The compositions of the ceramic matrix and the

109

fired clay (calculated without the CaO component of the ceramic matrix) are shown in Tables 3 and 4, respectively. The percentages of calcite in the ceramic matrix, calculated according to the CaO portion, are given in the tables. The compositions of the examined raw materials are shown in Table 5. The amounts of calcite in the pottery are similar to those in the corresponding raw materials. The amounts of calcite in the ceramic matrix (line a in Tables 3 and 4) are similar to those observed by F f I R spectroscopy (Table 2) and confirm the quantitative analysis by the latter method. The composition of the fired clay in the ceramic matrix (without the microcrystalline calcite component of the ceramic matrix; line b in Tables 3 and 4) is used here to characterize the raw material. In most of the pottery, the amounts of A1203 in the fired clay are around 15-20%. These amounts are in accordance with the smectitic composition of the examined raw materials (Table 5). It should be mentioned that, by XRD analysis, smectite was found to be the major clay mineral in these raw materials. In most of the pottery the amounts of SiO2, around 55-60%, are also in accordance with the smectitic composition of the raw materials. On the other hand, in kaolinitic raw material the amounts of A1203 and SiO 2, around 40% and 50%, respectively, are significantly different (e.g. kaolin Ramon in Table 5). The amounts of K20 and MgO are also in agreement with a smectitic composition of the examined raw materials. Larger amounts of A1203 in some of the pottery are related to the presence of some kaolinite in the raw material whereas larger amounts of SiO2 are related to the presence of quartz silt. The large amount of FeO is indicative of pottery manufactured from soil raw material.

The firing temperature of the pottery The firing process of ancient pottery is commonly assessed by investigating the thermal phases that are temperature dependent (Maggetti 1982; Shoval 1994). In this study, the firing temperature is estimated using FTIR spectroscopy by comparing the spectral properties of the pottery with those of the raw materials fired at various temperatures (Shoval 1994, 2003). The estimation is based on several observations, as follows.

The main Si-O

stretching

band.

spectra of the storage jars of Group 4, S i - O stretching band appears in the 1031-1045 cm -1 (Table 2), and this close to that observed after firing

In the the main range of range is the raw

S. SHOVAL ET AL.

110

Table 3. The major element composition (normalized to 100%) obtained by point analysis in selected Iron Age cooking pots from Tel-Hadar Sample number

Group 1 1574/1 (TH-7) 1778/1 (TH-37) 1783/3 (TH-31) 1834/2 (TH-8) 1844/3 (TH-42)

Group 2 1734/1 (TH-2) 1756/1 (TH-6) 2021/1 (TH-5) 2264/1 (TH-259)

Group 3 11/106 (TH-115) 171/185 (TH-119)

Ceramic matrix and fired clay composition

% Calcite

SiO2

A1203

MgO

FeO

Ti02

K20

CaO

a b a b a b a b a b

56.13 62.92 58.95 62.96 58.96 61.04 58.46 60.90 56.38 59.18

17.84 20.19 12.35 13.05 19.45 19.96 21.25 21.88 20.12 20.84

2.56 2.33 4.77 5.00 2.00 2.04 2.68 2.73 2.42 2.49

11.38 10.65 15.03 15.84 11.15 11.47 10.37 10.71 12.32 ! 2.78

2.00 1.22 1.22 1.28 1.69 1.74 1.19 i.22 2.15 2.22

2.37 2.69 1.43 1.88 3.36 3.75 2.18 2.55 2.09 2.49

7.72 6.25 3.39 3.87 4.52 -

a b a b a b a b

61.13 66.77 62.78 64.57 64.88 66.24 60.49 63.17

17.78 18.97 17.76 18.12 17.24 17.50 19.57 20.18

1.52 1.61 1.41 1.44 1.39 1.40 1.38 1.43

8.55 9.21 10.58 10.83 10.03 10.20 10.88 11.26

1.31 1.39 2.96 3.01 2.54 2.58 2.09 2.14

1.47 2.06 1.79 2.04 1.91 2.09 1.50 1.83

8.24 2.72 2.01 4.09 -

a b a b

60.28 63.51 58.39 61.96

17.82 18.85 19.94 21.29

1.83 1.93 2.51 2.66

10.22 10.83 9.73 10.42

1.55 1.64 1.12 1.19

3.08 3.24 2.05 2.48

5.22

9

6.26 -

11

14 11 6 7 8

15 5 4 7

a, The compositionof the ceramicmatrix;b, the compositionof the firedclay (normalizedwithoutthe amountof CaO). The percentages of calcite in the ceramicmatrix,calculatedaccordingto the CaO portion,are given.

materials between 700 and 800 °C (Shoval 2003). In the spectra of some of the cooking pots of Groups 2 and 3, this band splits in the ranges of 1045-1060 and 1 0 7 6 - 1 0 8 3 c m -1 (Table 2), and these ranges are close to those observed after firing the raw materials between 800 and 900 °C (Shoval 2003).

The S i - O and A l - O deformation bands. The combined S i - O , A 1 - O bending mode appears with one peak in the range of 4 5 9 - 4 7 5 c m - l (Table 2), which is typical of dehydroxylated fired clay. Thus, the firing was performed above the dehydroxylation temperature of smectite. The main C03 band. The appearance of the main CO3 band above 1430 c m - l in the spectra of the matrices of most storage jars can be related to the presence of newly formed calcite. In the calcareous ceramics the calcite is newly formed after the firing (Shoval et al. 1993). Thus, the firing was performed above the decarbonation temperature of the microcrystalline calcite in the ceramic matrix. In mixtures of clay and microcrystalline calcite,

decarbonization occurs at about 600 °C (Shoval et al. 1993), indicating that the storage jars were fired above this temperature. Similar results were observed in slightly calcareous matrix of cooking pots.

The OH stretching band of rehydroxylated clay. After removing the water bands, a weak OH band is observed in the spectra of cooking pots at 3 6 2 0 - 3 6 3 2 cm -1 (Shoval et al. 1991). Because smectite was found to be the main clay mineral in the raw material, the 3 6 2 0 3632 cm -1 band is related to the presence of some rehydroxylated smectite in the ceramic. The weak intensity of this band indicates that small amounts of rehydroxylated smectite are present in the ceramic matrix. Thus, the firing was performed below the temperature for nonreversible dehydroxylation of the smectite (Shoval et al. 1991). In smectite, rehydration and rehydroxylation occur at room temperature, after dehydration and dehydroxylation by heating to temperatures as high as 800°C (Heller et al. 1962).

111

CERAMIC T E C H N O L O G Y OF IRON AGE P O T T E R Y Table 4. The major element composition (normalized to 100%) obtained by point analysis in selected Iron Age storage jars and tableware vessels from Tel-Hadar Sample number

Ceramic matrix and fired clay composition

% Calcite

SiOz

A1203

MgO

FeO

TiO2

K20

CaO

a b a b a b a b a b

49.91 67.00 41.01 60.48 41.20 59.28 42.16 59.94 37.84 55.63

11.39 14.82 14.27 20.05 13.54 18.71 14.78 20.13 13.56 19.54

2.50 3.20 2.30 3.17 4.75 6.41 2.76 3.69 2.43 3.44

8.86 11.62 6.40 9.12 7.01 9.74 7.44 10.22 8.64 12.52

0.95 1.22 1.55 2.14 1.05 1.41 1.49 1.99 2.47 3.46

1.07 2.14 2.58 5.04 2.33 4.46 2.11 4.02 2.76 5.41

25.33 31.89 30.12 29.26 32.29 -

45

a b a b a b

49.71 67.42 42.37 62.81 42.51 64.59

11.38 14.94 12.81 17.91 11.82 14.28

2.23 2.88 3.29 4.86 2.80 4.01

7.13 9.48 5,47 8.40 6.30 9.33

1.52 1.97 1.12 1.69 1.56 2.24

1.81 3.31 2.90 4.33 2.73 5,55

26.22 32.03 32.26 -

47 57 58 -

a b a b a b

43.21 63.88 51.89 64.00 58.82 68.25

11.98 16.94 13.81 16.64 11.55 15.85

1.85 2.56 3.10 3.69 2.22 0.91

6.57 9.45 8.84 10.70 7.54 5.28

1.45 2.02 1.50 1.77 1.35 1.47

2.64 5.15 2.06 3.20 1.32 8.23

32.30 18.81 20.20 -

58

a b a b a b a b a b

42.49 62.27 48.78 67.60 46.15 62,14 34.48 64.27 34.48 59.86

14.93 20.83 11.62 15.54 16.03 20.75 11.70 20.24 15.51 22.22

1.06 1.46 2.90 3.81 1.77 2.27 1.46 2.46 1.70 2.90

6.77 9.63 5.79 7.86 6.74 8.84 3.90 6.94 6.03 7.38

1.17 1.62 0.90 1.19 1.66 2.13 0.62 1.07 0.67 1.49

2.09 4.19 2.14 3.99 2.18 3.88 1.71 5.01 2.83 6.14

31.50 27.87 25.48 46.13 38.79 -

56

a b a b a b a b

43.01 60.85 49.74 60.99 40.28 59.05 51.18 59.83

15.62 21.10 17.45 20.81 15.66 21.81 22.36 25.45

1.22 1.62 2.39 2.81 2.19 2.98 1.06 1.19

7.59 10,41 6.41 7.73 5.48 7.78 7.66 8.80

1.67 2.23 0.85 1.01 0.51 0.72 2.20 2.49

1.98 3.79 4.72 6.66 4.21 7.67 1.53 2.23

28.91 18.44 31.67 14.01 -

a b

43.42 60.14

17.46 23.10

1.76 2.30

5.83 7.85

0.52 0.69

3.45 5.91

27.56 -

Group 4 1562/1 (TH-24) 1783/2 (TH-28) 1828/2 (TH-32) 1746 (TH-98) 1844/2 (TH-40)

57 54 52 58

Glvup 5 1783/1 (TH-30) 2039/1 (TH-313) 2044/1 (TH-324)

Group 6 2060/1 (TH-302) 1803/1 (TH-303) 2152/1 (TH-310)

34 36 -

Group 7 2 1 0 5 / 2 (TH-326) 2217/1 (TH-263) 1565/1 (TH-206) 2207/1 (TH-317) 2374/1 (TH-99)

50 45 82 69

Group 8 2225/1 (TH-301) 2220/1 (TH-270) 1702/1 (TH-316) 2348/1 (TH-260)

52 33 57 25 -

Group 9 1842/1 (TH-325)

49

a, The composition of the ceramic matrix; b, the composition of the fired clay (without the amount of CaO). The percentages of calcite in the ceramic matrix, calculated according to the CaO portion, are given.

F r o m t h e s e o b s e r v a t i o n s , it s e e m s that the c o o k i n g p o t s w e r e fired at a b o u t 7 5 0 - 8 5 0 °C a n d the s t o r a g e j a r s a n d the t a b l e w a r e v e s s e l s at a b o u t 6 5 0 - 7 5 0 °C.

Tempering of the cooking pots The cooking pots were tempered with broken p i e c e s o f c a l c i t e c r y s t a l s that w e r e a d d e d to the c l a y e y r a w m a t e r i a l f r o m an a d d i t i o n a l s o u r c e

112

S. SHOVAL ET AL.

Table 5. The major element composition (normalized to 100%) of local alluvia and soils Sample

Basaltic soil of Golan Terra Rossa of Galilee Alluvium of Wadi Samakh Alluvium soil of Jordan Valley Rendzina soil of Galilee Bentonite Ramon Kaolin Ramon

Ceramic matrix and fired clay composition

a b a b a b a b a b

% Calcite

SiO2

A1203

MgO

FeO

TiO2

K20

CaO

54.69 59.90 61.88 64.42 38.29 59.35 42.78 63.90 31.36 58.35 62.28 51.14

16.42 17.77 18.72 19.89 14.98 18.97 11.86 16.68 9.03 15.94 22.10 38.40

4.87 5.23 1.88 1.89 3.27 4.50 2.83 3.54 2.56 4.42 2.75 0.13

12.48 12.33 10.34 10.45 9.64 12.53 6.51 9.30 8.38 14.72 6.74 1.77

1.98 2.12 1.63 1.64 1.71 2.32 1.43 2.53 1.46 2.44 1.04 8.56

2.11 2.65 1.63 1.72 0.99 2.32 2.62 4.06 1.38 4.13 5.09 0

7.44 3.93 31.11 31.98

13 7 56

45.84

82 -

-

57

a, The compositionof the bulk material; b, the compositionof the clay component(normalizedwithout the amount of CaO). The percentages of calcite in the ceramic matrix, calculated according to the CaO portion, are given. Analysesof two referenceclays, kaolin (kaolinite)Ramonand bentonite(smectite)Ramon.are addedfor comparison. (Shoval et al. 1993). The cooking pots have high concentrations of calcite temper. The amounts of the calcite temper, calculated by subtracting the percentages of calcite in the bulk pottery from the percentages of the microcrystalline calcite in the ceramic matrix, are up to 32%. According to Tite et al. (2001), the production of pottery with high toughness and thermal shock resistance requires low firing temperatures and high temper concentrations. The high temper content used in the production of cooking pots suggests that the requirement for high thermal shock resistance was a factor that at least influenced the technological choice in this case (Tite et al. 2001). In the investigated cooking pots, the high temper concentration may be related to the use of smectitic clay raw material for their production. Swelling smectitic clay is less suitable for ceramic production than kaolinitic clay. The temper particles were necessary during the manufacture of the pottery to reduce high plasticity and the collapse of the smectitic clay during the shaping of the vessels, as well as to prevent shrinking and cracking during the drying of the pots. Most of the cooking pots were tempered with broken pieces of large calcite crystals that were added to the clayey raw material from an additional source. Three types of calcite crystals were found in the pottery (Fig. 8). The rhombohedral calcite and the coarse crystalline calcite were probably collected from veins, druses and speleothems. The large calcite crystals were crushed and mixed for tempering during production (Porat 1989). Large calcite crystals are not readily available and therefore the potters must have made great efforts to collect them. According to Arnold (1985), tempering with

calcite crystals prevents owing to thermal shock cooling of the vessels. the similarity between

cracking and fracturing during rapid heating or This was explained by the thermal expansion

(a)

(c)

Fig. 8. Thin-section photomicrographsof calcite temper types in cooking pots. (a) Temper of rhombohedral calcite (in the middle left corner). (b) Temper of coarse crystalline calcite (drusy calcite, in the middle right corner). (c) A large fragment of speleothem (in the middle, 40x magnification, cross-polarized light).

CERAMIC TECHNOLOGY OF IRON AGE POTTERY coefficient of the calcite crystals and the surrounding fired clay during firing. However, this is true only at low firing temperatures, at which calcite crystals do not decompose. If the firing temperature is high enough for decomposition of the calcite crystals to quicklime, the thermal expansion coefficient would no longer be similar to that of the clay (Broekmans et al. 2004). Limestone particles composed of polycrystalline calcite should have similar thermal expansion coefficients to those of large calcite crystals (Shoval et al. 1993). Although limestone is much more readily available than large calcite crystals, potters usually rejected it for the preparation of cooking pots (Arnold 1985). Alternative tempering with coarse limestone particles is inappropriate as it brings about earlier and intense decarbonation during firing, which causes defects in the pots (Shoval et al. 1993). Limestone has small calcite crystals with large specific surface area, more crystalline defects and impurities, which lead to earlier intensive thermal decarbonation compared with large calcite crystals. The rapid release of carbon dioxide gas (CO2) is accompanied by blowing and fracturing, which induces mechanical damage in the pots. Furthermore, after firing, the hygroscopic quicklime (CaO) picks up moisture from the air, forming calcium hydroxide (Ca(OH)z). The formation of calcium hydroxide is accompanied by volume expansion, which induces stresses in the surrounding clay body, causing defects in the vessels (Arnold 1985). Tempering with broken pieces of large calcite crystals prevents cracking and fracturing during the firing as it provides higher resistance to thermal decomposition and this seems to be the reason for their use in cooking pot production (Shoval et al. 1993). Processes that allow consolidation o f the ceramic

Although the cementation of the ceramic body by sintering of the clay usually occurs upon firing above 900-1000 °C (Rice 1987; Traore et al. 2000), lower firing temperatures were found here for the cooking pots and for the storage jars and tableware vessels. Consolidation o f the noncalcareous ceramic. The relative low firing temperature of the cooking pots, at about 750-850 °C, suggests that the cementation to ceramic was obtained by lowtemperature sintering of the clay. It seems that the use of soil raw material composed of

113

smectitic clay allowed the low-temperature sintering. The clay from soil is relatively poorly crystallized and rich in natural iron oxide, both of which induce earlier sintering. In very finetextured raw material (such as soil), the sintering begins earlier than in coarse raw material (Rice 1987). Moreover, very fine ferrous oxide (FeO) in a reduced state may act as a flux at temperatures between 800 and 900°C (Grimshaw 1971). It seems that the high iron oxide content in the basaltic soil or in the Terra Rossa acts as an efficient flux material, which reduces the sintering temperature during firing. The potters used noncalcareous raw material for the preparation of cooking pots to produce dense ceramic, impermeable and stable on cooking directly over fire, and able to withstand repeated heating and cooling. Consolidation o f the calcareous ceramic. The lower firing temperature of the storage jars and tableware vessels, at about 650-750 °C, was obtained by using calcareous raw material. Lime is used as a flux, and vessels prepared from this clay are sintered at lower temperatures. However, in storage jars and tableware vessels containing a large amount of calcite in the ceramic matrix (Tables 2 and 4) it seems that, instead of sintering the clay, the potters used lime technology to achieve consolidation of the vessels (Shoval 2003). It should be noted that Tables 2 - 4 represent the amounts of microcrystalline calcite mixed with the clay in the ceramic matrix, but not that in the calcite temper or other carbonate particles. Firing the calcareous raw material above 600 °C was sufficient for decomposition of the microcrystalline calcite to quicklime (Shoval et al. 1993). The consolidation of the storage jars was achieved by recrystallization of microcrystalline calcite during the recarbonation process. Consolidation by cementation with calcite required lower firing temperatures than those necessary to complete the sintering of the clay. Indeed, in these storage jars, a micrographic texture of microcrystalline calcite (micrite) is observed in the ceramic matrix and such a texture, with a low degree of crystallinity, is typical of newly formed calcite (Maciejewski & Relier 1989). The use of lime technology for cementation of lime-based plasters and mortars (Kingery 1988; Moropoulou et al. 2005) and for production of Vaiselle Blanche ('Whiteware', Goren 1991) was already known from the Neolithic period. It seems that after firing, the storage jars were carefully stored for a period of time to complete the solidification by recrystallization of the

114

S. SHOVAL ET AL.

newly formed calcite. It should be noted that some pozzolanic activity was found in mixtures of quicklime and meta-clay, and such a process was used for the cementation of ancient plasters and mortars (Moropoulou et al. 2004, 2005).

The origin of the pottery The place of production of each petrographic group of the pottery is related to the area in which the raw material is common. Local ceramic is recognized by temper particles collected near the site, whereas imported ceramic is identified by temper particles not found in the vicinity.

The cooking pots. The cooking pots of Groups 1 and 3 were manufactured from basaltic soil, which is common on the Golan basalt plateau, east of the excavated area. Group 2 was produced from Terra Rossa, which is common on Upper Cretaceous limestone and dolostone of the Galilee Mountains, west of the excavated area. The storage jars and the tableware vessels. Group 4, the largest petrographic group, was manufactured from calcareous clayalluvium. Such alluvium, containing both basalt and carbonate grains, is present in the vicinity of the Tel Hadar site only in the drainage system of the Golan slopes, toward the shore of the Sea of Galilee (in the river beds of Wadi Samakh and the Yarmouk river). Group 5 was produced from calcareous alluvium soil. Such soil, containing small amounts of highly weathered basalt grains, is common in the Jordan Valley south of the Sea of Galilee. Group 6 was manufactured from calcareous rendzina soil mixed with noncalcareous basaltic soil. Rendzina soil is common on Eocene chalk, and basaltic soil is common on Neogene basalt flows, both of which appear nearby in the Galilee Mountains. The presence of Foraminifera fossils of Eocene age within the chalk fragments supports this observation. Group 7 was produced from calcareous marl mixed with noncalcareous Terra Rossa. Marls are common in Maastrichtian and Paleocene formations, and Terra Rossa is common on Upper Cretaceous limestone and dolostone, both of which appear nearby in Western Galilee-Lebanon. Group 8 was manufactured from calcareous sediment of the coastal plain. Such sediment is common in the northern Mediterranean shore. Group 9 was produced from calcareous sediment from a foreign coastal area and seems to be imported. The

large fragment of coarse crystalline calcite or marble may suggest an origin on a Mediterranean island, where marble is exposed. Indeed, the pottery is fine and thinner than in the other groups. It is reasonable to assume that this particular pottery was imported, rather than to suppose a sudden change in the local technique.

Summary The manufacturing technology The results demonstrate that the potters in biblical times had knowledge of raw materials and temper particles, thus allowing them to select suitable ones, according to the advantages of each, for the manufacture of cooking pots, storage jars and tableware vessels. ( 1) The cooking pots were manufactured from noncalcareous or slightly calcareous raw material, whereas the storage jars and most of the tableware vessels were manufactured from calcareous raw material. (2) This tradition of pottery making was applied during the Iron Age by using noncalcareous and calcareous raw materials available near each production site; this is the reason for the variability in raw materials. (3) The potters preferred to use clayey soils and alluvium as raw materials, rather than clayey sediment for this purpose. The clay from soil is relatively poorly crystallized and rich in natural iron oxide, both of which induce earlier sintering. Alluvium and soil are available in the vicinity of the site and there was no need to bring the raw material from a distance. (4) Basaltic soil and Terra Rossa were used as noncalcareous or slightly calcareous raw materials for the preparation of the cooking pots. It should be noted that the common noncalcareous geo-material in the region is clayey soil, whereas sediments are usually calcareous, particularly appearing as marls. (5) Alluvium, rendzina soil and clayey sediment were used as calcareous raw materials for the preparation of the storage jars and tableware vessels. (6) To reduce the high calcareous properties of the rendzina soil or marl, the raw materials were mixed with noncalcareous Terra Rossa or basaltic soil, as observed from the presence of relicts of noncalcareous material appearing in the form of mud balls. (7) The potters kept the attached coarse particles of the alluvium and soil raw materials as natural temper. Tempering by adding large calcite crystals from an additional source was

CERAMIC TECHNOLOGY OF IRON AGE POTTERY used only for the production of cooking pots from soil-derived raw material. The raw materials are rich in smectite and the temper particles were necessary during the manufacture of the pottery to reduce the high plasticity and prevent collapse of this clay. (8) Alternative tempering of the cooking pots with coarse limestone particles composed of polycrystalline calcite is inappropriate as it brings about earlier and intense decarbonation during firing, which causes defects in the pots. (9) The cooking pots were fired at about 750-850 °C and the cementation to ceramic was obtained by low-temperature sintering of the clay. The use of soil raw material composed of smectitic clay allowed low-temperature sintering. (10) The storage jars and most of the tableware vessels were fired at about 650-750 °C. The large amount of calcite in the ceramic of Group 4 indicates that instead of sintering the clay, the potters used lime technology to achieve consolidation of the vessels. Consolidation by cementation with calcite required lower firing temperatures than those necessary to complete the sintering of the clay. (11) A higher quality of noncalcareous or slightly calcareous raw material was necessary for the manufacture of cooking pots to produce dense ceramic that was impermeable and stable in cooking directly over fire and able to withstand repeated heating and cooling.

The archaeological context The results are placed in an archeological context, as follows. (1) The various petrographic groups of pottery excavated in the pillared building of Tel Hadar reflect the trade connections of this site with the surrounding region. These groups demonstrate that the pottery was manufactured in different locations and imported to Tel Hadar. (2) The petrographic groups demonstrate that the pottery was manufactured in the vicinity of the site, as well as in the Golan, the Jordan Valley, Galilee, Western Galilee-Lebanon and the northern shore. Thus, Tel Hadar had connections with these regions. (3) The many storage and tableware vessels found in the storage rooms of the pillared building and their various origins indicate that the site was used as a trading post. It seems that columned hall of the pillared building was used for marketing. The large granary rooms of the ground floor were also used for this purpose. The movement of goods seems to have covered a large area.

115

(4) The large storage jars of Group 4 are of local production. It seems that for production of large vessels designed for storage of dry foods, a coarse local calcareous raw material produced by using lime technology was sufficient. These vessels were probably used as containers for storing the food intended for marketing in the columned hall. It is reasonable to assume that large jars of local production were not transported over large distances. (5) Tableware vessels of Group 4 are also of local production. These vessels, which were designed for everyday use, are coarse and thick and usually not burnished or decorated. They served as tools for preparing foods and for eating. (6) The cooking pots of Groups 1-3 were imported to Tel Hadar from Golan and from Galilee. For their preparation, a higher quality raw material was needed. They served as implements for heating foods. However, Group 2 includes some large cooking pots that had been not used, and these pots were probably intended for marketing in the columned hall. (7) The pottery of Groups 6 - 8 was imported to Tel Hadar from far away. These vessels are burnished or colour-slipped and part of them was decorated with brown-red bands. They were imported for marketing because of the quality of the ware and probably subsequently used for storing oils or other expensive products. (8) The imported storage jars of Groups 6 - 8 were probably brought as containers for the traded goods. (9) The 'Greek' bowl of Group 9 was possibly imported from a Mediterranean island. This pottery is fine and thinner than that of the other groups and decorated inside with a red cross. (10) The variability in the raw materials indicates that during the Iron Age the pottery needed at each site was prepared by the local potters. There was no pottery-making centre for regional production. (11) The major technological evolution in the Iron Age, relative to that in previous periods, is the specification in the use of raw materials and manufacturing technologies for the production of different functional vessels. The transformation from hand-made to wheel-made pottery becomes widespread during the Bronze Age 2a and production in workshops began in this period. (12) The technology used by the potters of the ninth century is similar to that in the 1 lth century, indicating a similar tradition of pottery production. (13) The major technological evolution in later periods relative to that of the Iron Age is the establishment of pottery-making centres for regional production. These centres specialized

116

S. SHOVAL ETAL.

in the manufacture of particular types of vessels. For example, the bulk of the c o m m o n cooking ware of Roman Sepphoris was supplied by the Kefar Hananya site located in the Galilee region (Adan-Bayewitz & Perlman 1990). The excavations at Tel Hadar were carried out in the framework of the Land of Geshur Archaeological Project, of The Institute of Archaeology, Tel Aviv University, by M. Kochavi (director) and E. Yadin (field director at Tel Hadar). The supply of pottery and fruitful discussions are gratefully acknowledged. This research was supported by the Israel Science Foundation administered by the Israel Academy of Sciences and Humanities and completed through the support of the Open University of Israel Research Authority. Both are gratefully acknowledged.

References ADAN-BAYEWITZ, D. & PERLMAN, I. 1990. The local trade of Sepphoris in the Roman Period. Israel Exploration Journal, 40, 153-172. ARNOLD, D. 1985. Ceramic Theory and Cultural Process. Cambridge University Press, Cambridge. BELLOTO, M., GUALTIERI, A., ARTIOLI, G. & CLARK, S. M. 1995. Kinetic study of the kaolinite-mullite reaction sequence. Parts I, II. Physics and Chemistry of Minerals, 22, 207-222. BERTOLINO, S. R. & FABRA, M. 2003. Provenance and ceramic technology of pot sherds from ancient Andean cultures at the Ambato valley. Applied Clay Science, 24, 21-34. BROEKMANS, T., ADRIAENS, A. & PANTOS, E. 2004. Analytical investigations of cooking pottery from Tell Beydar (NE Syria). Nuclear Instruments and Methods in Physics Research, B226, 92-97. COHEN-WE1NBERGER, A. & GOREN, Y. 1996. Petrographic analysis of Iron Age I pithoi from Tel Sasa. Atiqot, XXVIII, 77-83. DE BENEDETTO, G. E., LAVIANO,R., SABBATINI,L. & ZAMBONIN, G. 2002. Infrared spectroscopy in the mineralogical characterizing of ancient pottery. Journal of Cultural Heritage, 3, 177-186. DEUTSCH, Y. & Heller KALLAI, L. 1991. Decarbonation and recarbonation of calcites heated in CO2. Thermochimica Acta, 182, 77-89. DONG-LI., X. & THOMSON, W. J. 1991. Tetragonal to orthorhombic transformation during mullite formation. Journal of Materials Research, 6, 819-824. DUMINUCO, P., MESSIGA, B. & RICCARDI, M. P. 1998. Firing process of natural clay: some microtexture and related phase compositions. Therrnochimica Acta, 321, 185-190. ERAMO, G., LAVIANO,R., MUNTONI,I. M. & VOLPE, G. 2004. Late Roman cooking pottery from the Tavoliere ara (Southern Italy): raw materials and technological aspects. Journal of Cultural Heritage, 5, 157-165. FREUND, F. 1974. Ceramics and thermal transformations of minerals. In: FARMER, V. C. (ed.)

The Infrared Spectra of Minerals, Mineralogical Society, London, Monograph, 4, 465-482. GOREN, Y. 1991. The beginnings of ceramic production in Israel, technology and typology of proto-historic ceramic assemblages in Eretz Israel (6th-4th millennia B.E.C.). PhD thesis, The Hebrew University of Jerusalem. GRIMSHAW, R. W. 1971. The Chemistry and Physics of Clays and other Ceramic Materials. Wiley, Chichester. GUALTmRI, A. & BELLOTO, M. 1998. Modelling the structure of the metastable phases in the reaction sequence kaolinite-mullite by X-ray scattering experiments. Physics and Chemistry. of Minerals, 25, 442-452. HEIN, A., DAY, P. M., CAU ONTIVEROS, M. A. & KIL1KOGLOU, V. 2004. Red clays from central and eastern Crete: geochemical and mineralogical properties in view of provenance studies on ancient ceramics. Applied Clay Science, 24, 245-255. HEELER, L., FARMER, V. C., MACKENZIE, R. C., MITCHELL, B. D. & TAYLOR, H. F. W. 1962. The dehydroxylation and rehydroxylation of triphormic dioctahedral clay minerals. Clay Minerals Bulletin, 5, 56-72. HELLER-KALLAI,L., I~JLOSLAVSKI, I. & AJZENSHTAT,Z. 1986. 'Dissolution' of calcite by steam derived from clay minerals. Naturwissenschaften, 73, 615-616. HELLER-KALLAI,L., ]VIILOSLAVSKI, I. ~ mIZENSHTAT,Z. 1987. Volatile products of clay mineral pyrolysis revealed by their effects on calcite. Clay Minerals, 22, 339-348. KINGERY, W. D. 1988. The beginnings of pyrotechnology, part II: Production and use of lime and gypsum plaster in the pre-ceramic Neolithic Near East. Journal of Field A rchaeology, 15, 219- 244. KOCHAVI, M. 1989. The Land of Geshur project: regional archaeology of the Southern Golan (198788 seasons). Israel Exploration Journal, 39, 1- 17. LIVINGSTONE SMITH, A. 2001. Bonfire II: the return pottery firing temperatures. Journal of Archaeological Science, 28, 991 - 1003. MACIEJEWSKI, M. & REELER, A. 1989. Formation of amorphous CaCO3 during the reaction of CO2 with CaO. Thermochimica Acta, 142, 175-188. MACKENZIE, R. C. & RAHMAN,A. A. 1987. Interaction of kaolinite with calcite on heating. I. Instrumental and procedural factors for one kaolinite in air and nitrogen. Thermochimica Acta, 121, 51-69. MACKENZIE, R. C., RAHMAN, A. A. & MOIR, H. M. 1988. Interaction of kaolinite with calcite on heating. II. Mixtures with one kaolinite in carbon dioxide. Thermochimica Acta, 124, 119-127. MAGGETTI, M. 1981. Composition of Roman Pottery from Lausanne (Switzerland). British Museum Occasional Paper, 19, 33-49. MAGGETTI, M. 1982. Phase analysis and its significance for technology and origin. In: OLIN, J. S. & FRANKLIN, A. D. (eds) Archaeological Ceramics, Smithsonian Institution Press, Washington, DC, 121-133. MAGGETTI, M. 1994. Mineralogical and petrographical methods for the study of ancient pottery.

CERAMIC TECHNOLOGY OF IRON AGE POTTERY

In: BURRAGATO,F. GRUBESSI,O. & LAZZARINI,L. (eds) First European Workshope on Archaelogical Ceramics, Rome. 23-35. MAGGETTI, M., WESTLEY, n. & OLIN, J. S. 1984. Provenance and technical studies of Mexican majolica using elemental and phase analysis. In: Archaeological Chemistry. American Chemical Society, Washington, DC, 151 - 191. MANIAT1S, Y. & TITE, M. S. 1981. Technological examination of Neolithic Bronze Age pottery from central and southeast Europe and from the Near East. Journal of Archaeological Science, 8, 59-76. MIRTI, P., APPOLONIA, L. t~ CASOLI, A. 1999. Technological features of Roman Terra Sigillata from Gallic and Italian centers of productions. Journal of Archeological Science, 26, 1427-1435. MOROPOULOU, A., BAKOLAS, A. & BISBIKOU, K. 1995. Thermal analysis as a method of characterizing ancient ceramic technologies. Thermochimica Acta, 269-270, 743-753. MOROPOULOU,A.,BAKOLAS,A.&ANAGNOSTOPOULOU, S. 2004. Evaluation of pozzolanic activity of natural and artificial pozzolans by thermal analysis. Thermochimica Acta, 420, 135-140. MOROPOULOU,A.,BAKOLAS,A.&ANAGNOSTOPOULOU, S. 2005. Composite materials in ancient structures. Cement and Concrete Composites, 27, 295-300. PORAT, N. 1989. Petrography of pottery from southern Israel. In: MIROSCHEDJI,P. (ed.) L'Urbanisation de la Palestine a l'Age du Bronze Ancien. British Archaeological Reports, 169-188. RICCARDI, M. P., MESSIGA,B. & DUMINUCO,P. 1999. An approach to the dynamics of clay firing. Applied Clay Science, 15, 393-409. RICE, M. P. 1987. Pottery Analysis. A Sourcebook. University of Chicago Press, Chicago, IL. SEYAMA, H. & SOMA, M. 1986. X-ray photoelectron spectroscopic study of the effect of heating on montmorillonite containing sodium and potassium cations. Clays and Clay Minerals, 34, 672-676. SHOVAL,S. 1988. Mineralogical changes upon heating calcitic and dolomitic marl rocks. Thermochimica Acta, 135, 243-252. SHOVAL, S. 1994. The firing temperature of a PersianPeriod pottery kiln at Tel Michal, Israel, estimated from the composition of its pottery. Journal of Thermal Analysis, 42, 175-185.

117

SHOVAL, S. 2003. Using FTIR spectroscopy for study of calcareous ancient ceramics. Optical Materials, 24, 117-122. SHOVAL,S. 2004. Deposition of volcanogenic smectite along the southeastern Neo-Tethys margin during the oceanic convergence stage. Applied Clay Science Journal, 24, 299-311. SHOVAL, S., EREZ, Z., IORSH, Y., DEUTSCH, Y., KOCHAVI, M. & YADIN, E. 1989. Determination of the Intensity of an Early-Iron-Age conflagration at Tel-Hadar, Israel. Thermochimica Acta, 148, 485 -492. SHOVAL, S., BECK, P., KIRSH, Y., LEVY, D., GAFT, M. & YADIN, E. 1991. Rehydroxylation of clay minerals and hydration in ancient pottery from the Land of Geshur. Journal of Thermal Analysis, 37, 1579-1592. SHOVAL, S., GAFT, M., BECK, P. & IORSH, Y. 1993. The thermal behavior of limestone and monocrystalline calcite temper during firing and their use in ancient vessels. Journal of Thermal Analysis, 40, 263-273. SHOVAL, S., YARIV, S., MICHAELIAN, K. H., BOUDEULLE, M. & PANCZER, G. 2001. MicroRaman and FTIR spectroscopy study of the thermal transformations of St. Claire dickite. Optical Materials, 16, 319-327. SHOVAL, S., MICHAELIAN, K. H., BOUDEULLE, M., PANCZER, G. LAPIDES, I. t~ YARIV, S. 2002. Study of thermally treated dickite by infrared and Micro-Raman spectroscopy using curve-fitting technique. Journal of Thermal Analysis and Calorimetry, 69, 205-225. SHOVAL, S., YOFE, O. & NATHAN, Y. 2003. Distinguishing between natural and recarbonated calcite in oil shale ashes. Journal of Thermal Analysis, 71, 883-892. TITE, M., KILIKOGLOU, V. • VEKINIS, G. 2001. Strength, toughness and thermal shock resistance of ancient ceramics, and their influence on technological choice. Archaeometry, 43, 301-324. TRAORE, K., KABRE, T. S. t~z BLANCHART, P. 2000. Low temperature sintering of a pottery clay from Burkina Faso. Applied Clay Science, 17, 279-292. WIEDER, M. & ADAN-BAYEWITZ,D. 1999. Pottery manufacture in early Roman Galilee: a micromorphological study. Catena, 35, 327-341.

Fibre-tempered pottery of the Stallings Island Culture from the Crescent site, Beaufort County, South Carolina: a mineralogical and petrographical study M I C H A E L S. SMITH 1 & M I C H A E L B. T R I N K L E Y 2

1Department of Earth Sciences, University of North Carolina, Wilmington, NC 28403-5944, USA (e-mail: smithms @uncw.edu) 2Chicora Foundation, Inc., PO Box 8664, Columbia, SC 29202, USA Abstract: Late Archaic to Early Woodland (4500-3000 years BP) Stallings Island Culture fibre-tempered plainware pottery is found from northern North Carolina to NW Florida and is often separated into two temper groups (Stalling and Orange series) based upon fibre type. Thirty-four sherds were studied to determine textural or mineralogical characteristics to assist in form and type separation. This study finds that only a few of the sherds were dominated by fibre. The fibre is visible as secondary porosity (voids) with some carbonized remains and exhibits specific shape and orientation at different locations within the sherd. Only two sherds had carbonized stem fragments that allowed identification of the Spanish moss (Tillandsia usnedoides). The remainder of the Crescent site plainware pottery has such low (to no) fibre contents as to be indistinguishable from similar age sand-tempered Thorn' s Creek waxes. The identity and textural features of the fine-grained aplastic minerals (quartz, feldspar, biotite and opaque minerals) in the paste are similar in both fibre- and non-fibre-tempered sherds, and this suggests that the materials used are consistent with sediment extraction from a fluvial coastal plain or estuarine setting. These observations indicate that the degree of fibre incorporation in these sherds may be related to the specific clay source (or manufacturing location) or represent examples of an evolving pottery manufacturing process within the Stallings Island Culture.

Archaeological ceramics can provide important insights into the culture, technology, and development of a civilization. By examining the forms and the methods of decoration, clues to cultural evolution or trade among cultures are suggested. However, many prehistoric ceramics do not allow easy or definitive separation by these methods. When this situation arises, analytical investigation via a variety of techniques (Rice 1987; Sinopoli 1991) such as petrographic investigation is used to identify the plastic (paste) and the aplastic (temper) components of the pottery. This study examines 34 sherds of late Archaic to Early Woodland pottery recovered from the Crescent site (Stallings Island Culture) in Beaufort County, South Carolina. Trinkley & Hacker (1986, p. 166) and Trinkley (1998) have previously discussed the form and type of this pottery, its 14C and supplied information concerning the macroscopic investigation of similar sherds. This study was undertaken to determine if any textural or mineralogical characteristics would allow further form and type separation of this pottery as well as

clarify the identity and amount of the fibre temper.

Archaeological context Around 4500 years BP, the Stallings Island Culture established a Late Archaic seasonally occupied shellfish-collecting society along the southeastern Atlantic coastal environments with a material cultural assemblage equivalent to that found on Archaic sites (Sassaman 1993, pp. 16-19). Their occupation sites are represented by large shell middens that range from the Tar River drainage in North Carolina, southward to NW Florida. Stallings Island fibretempered ceramics (4500-3000 years SP) were named after a major shell midden site on an island in the Savannah River near Augusta, Georgia (Fig. 1) and represent the oldest documented ceramics in the southeastern USA (Stoltman 1966; Calmes 1968). The ceramics were generally simple shallow bowls and large, wide-mouthed bowls, with some deeper jar forms. Although most of these ceramics were plain, some with punctuated (drag and jab)

From: MAGGETTI,M. & MESSIGA,B. (eds) 2006. Geomaterialsin CulturalHeritage.

Geological Society, London, Special Publications, 257, 119-125. 0305-8719/06/$15.00 © The Geological Society of London 2006.

120

M.S. SMITH & M. B. TRINKLEY

Fig. 1. Map of the southeastern US Atlantic coastal region with locations of major Late Archaic pottery styles. Adapted from Sassaman(1993). Location of Crescent site in Beaufort County, South Carolina is designated by filledpentagon. It shouldbe noted that the Stallings and Thom's Creek ceramic styles overlap along the Coastal Plain of South Carolinabetween the Santee and SavannahRivers. surface decoration were found (Griffin 1943; Trinkley & Hacker 1986, pp. 160-162). Temporally overlapping the Stallings Island series are the sand-tempered Thom's Creek pottery (4000-2900 years BP; Calmes 1968). This series is found with the Stallings wares along the coastal areas of South Carolina between the Savannah and Santee Rivers and into the interior Coastal Plain (Sassaman 1993, p. 20; Fig. 1). Trinkley (1980, 1983) indicated that the fibre-tempered Stallings ware preceded the sand-tempered Thom's Creek ware by a few centuries, even though they share many design attributes and co-occur at many sites north of the Savannah River (Sassaman 1993). Contemporary with Stallings Island fibretempered pottery along the southeastern Atlantic coast are fibre-tempered Orange ware from sites in NE Florida and SE coastal Georgia (40003000 years BP; Sassaman 1993, pp. 20-21). Orange period sites have been located at Canaveral National Seashore, Fort Matanzas National Monument, and Timucuan Ecological and Historic Preserve (Fig. 1). The earliest

Orange period ceramics are shallow, fiat-based and straight-sided circular bowls and rectangular trays that are often undecorated (Sassaman 1993). The later Orange ware is more decorated, with less rectangular trays, some mixture of sand and fibre-tempering as well as the appearance of coiling manufacture in the final Orange period (c. 3250-3000 years BP; Sassaman 1993, p. 20). By 3000 years BP (Early Woodland period), fibre-tempered ceramic technology appears to have spread throughout much of the Deep South from the southeastern Atlantic coast to the Okeechobee Basin area of South Florida (Sassaman 1993). During the early Gulf Formational period (c. 4500-3000 years BP; Jenkins 1982) of Alabama, middle Tennessee, and eastern Mississippi, fibre-tempered ceramic technology (Wheeler Culture) was acquired as a by-product of trade between the Stallings Island and Orange Cultures of the southeastern Atlantic coast and the Poverty Point Culture of the lower Mississippi River Valley (Sears & Griffin 1950; Jenkins et al. 1986). However, fibretempered ceramics were eventually replaced by plain, fabric-impressed, and cord-marked sandtempered Alexander ceramics during the later Gulf Formational period.

Petrographic techniques Petrographic analysis is the principal method of identifying minerals (and other substances) in archaeological pottery (Rice 1987). Standard (27 mm × 46 ram) petrographic thin sections were prepared. All of the thin sections were epoxy impregnated because of the friable nature of the sherds. Grain-size values are very fine (<0.0625 ram), fine (0.0625-0.25 mm), medium (0.25-0.49mm), coarse (0.50-1.0ram) and very coarse ( > 1.0 ram). Colour identification was based upon the Munsell colour chips (Geological Society of America (GSA) 1991), and colours were observed under fluorescent lamps and described from a dry surface. The thin sections were point-counted using the techniques of Stoltman (1989a,b) and Stoltman et al. (1992). The point step was 0.1 mm so as to allow statistically significant counts (>300 points per thin section) and also to overlap with the macroscopic evaluation of the size distribution of paste and aplastic materials (Trinkley & Hacker 1986; Trinkley 1998). In any pointcounting technique the assumption is that the component has a nearly spherical grain shape. With this pottery, the evidence for the presence and abundance of fibre temper is provided by secondary porosity (voids) that contains some carbonized remnants. As the fibre voids are of

CRESCENT SITE FIBRE-TEMPERED CERAMICS two different shapes and orientations, the influence of this orientation may account for some of the percentage differences (and ranges) that were observed. The point count categories used were paste (clay minerals and amorphous phases), quartz (separated by grain size), fibre, feldspar (noted as feldspar unless optical characteristics allowed specific designation as either plagioclase or alkali feldspar), opaque minerals, other (includes epidote/clinozoisite, biotite and amphibole) and ACF (argillaceous clots or fragments of air-dried clay; see Whitbread 1986). Several features are observed in both hand specimen and thin section that may represent compositional or mineralogical changes caused by the firing conditions as well as by use or burial. Oxidation features (commonly a red to red-orange colour) were observed on both the inner and outer sherd surfaces, extending inward for several millimetres. The region between these oxidized zones (called the core) is generally reduced and is either black or smoky grey. Lastly, some of the sherds show secondary carbonate infilling in the fibre void spaces. This mineralization may have resulted from burial and interaction with ground water or may be a result of use.

Petrographic results Table 1 summarizes the petrographic results and textural characteristics of the samples. The rivers of the study area (Fig. 1; e.g. Savannah, Santee, and Altamaha Rivers) drain the igneous and metamorphic basement rocks of the Piedmont region to the west and downcut through the Late Cretaceous to Pleistocene aged sediments of the Coastal Plain (Horton & Zullo 1991; Nystrom et al. 1991; Ward et al. 1991). These Piedmont-draining rivers have a larger capacity and carry relatively large amounts of sediment that is less mature than the reworked (and heavily weathered) sediment carried by Coastal Plain draining rivers (Soller & Mills 1991, p. 299). The Crescent site is located in Pliocene to Pleistocene sediments (and terrace deposits) that were deposited through either the reworking of Piedmont-derived quartz- and feldspar-rich sediments or during transgressive-regressive cycles caused by eustatic sea-level fluctuations (Nystrom et al. 1991; Soller & Mills 1991). In addition, although these sherds are from the southeastern Atlantic coastal region, there was no petrographic or textural evidence for any type of shell material or carbonate rock fragments.

121

A p l a s t i c (temper) c o m p o n e n t s

The dominant aplastic mineral is quartz with a very coarse to very fine grain size. The very coarse to coarse crystals are angular to subangular quartz rock fragments with sutured grain boundaries, undulatory extinction, and little rounding of the corners or edges. The medium grain size quartz is monocrystalline (single crystals) with a blocky (rectangular) to elongate slivers grain shape (Fig. 2). The shape and texture suggest that the mineral grains were broken away from the parent rock fragment. The fine to very fine quartz is also monocrystalline and subangular to blocky in shape. There was very little rounding of the edges of these crystals and no well-rounded crystals were observed. Feldspar is found in very small modal abundance (<5 vol%) with only sherd JBM-6 having more feldspar (<15 vol%). The grain size ranges from medium to fine and the grain shape is blocky and rectangular to subrounded, primarily controlled by the two cleavage directions (Fig. 2). Most feldspar is untwinned, yet some sherds have potassium feldspar (microcline; tartan albite-pericline twinning) in addition to plagioclase feldspar (polysynthetic twins; JBM-30). The feldspars all exhibit some form of alteration (e.g. sericite and argillite) reflecting the deep weathering that is common in the Coastal Plain sediments (Nystrom et al. 1991; Ward et al. 1991). In the fine to very fine grain size fraction opaque minerals (probably iron oxides) along with mica (biotite), epidote, and an amphibole (hornblende) were found (i.e. JBM-20, JBM29). These minerals were probably part of the sediment (clay component) used as paste (similar in grain size) and are found in very low abundance (< 1 to 3 vol%) similar to that described by Nystrom et al. (1991) and Ward et al. (1991) for the sedimentary rocks deposited in this region. A few argillaceous inclusions (ACF) were found (Whitbread 1986). They are fine- to medium-grained (0.1-0.5mm) clots, which have a red (brick red to red black) colour and are ellipsoidal to spherical in shape, with irregular surfaces that feather into the surrounding paste. The ACF are either composed totally of clay minerals or dotted with inclusions of quartz and feldspar (similar in size to the aplastic grains in the paste), and may represent air-dried clay fragments. Although they are not found in all of the sherds, they are distinctive at both the macroscale and the microscale and were probably formed (or incorporated) during vessel

122

M.S. SMITH & M. B. TRINKLEY

Table 1. Petrographic summa~ for Crescent site sherds Sample

Quartz

JBM- I JBM-2 JBM-3 JBM-4 JBM-5 JBM-6 JBM-7 JBM-8 JBM-9 JBM- 10 JBM- 11 JBM-12 JBM-13 JBM-14

med-coarse* med-coarse v. coarse-fine v. coarse-fine med-fine v. coarse-fine med-fine v. coarse-fine med-fine med-fine coarse-fine v. coarse-v, fine

JBM-15 JBM-16 JBM-17 JBM-18

fine-v, fine fine-v, fine v. coarse-fine v. coarse-fine

JBM-19 JBM-20 JBM-21

coarse-v, fine med-fine fine

JBM-22 JBM-24 JBM-25 JBM-26 JBM-27 JBM-28 JBM-29 JBM-30 JBM-31 JBM-32

med-fine v. coarse-v, fine v. coarse-v, fine v. coarse-fine fine-v, fine fine-v, fine med-v, fine v. coarse-v, fine fine-v, fine coarse-v, fine

5% (K-spar/plag) 5% (plag) 2-3% (plag)

<1-2% 5-10% 5-10% 10-20% (core) > 30% >30% 5-10% 5-10% None 5-10%

JBM-33 JBM-34 JBM-35

reed-fine med-v, fine v. coarse-med

1-2% (plag) 1% (plag) 3-5% (plag)

5-10% 5-10% None

fine-v, fine

Feldspar

2-3% (plag) <2% <15% <2% < 1% ,-~ 1% < i% <1% <1% '~ <1%

<1% <1% !%

<1% 1 grain

Fibre Voids only 5-10% 5-10% 5-10% 5-10% 5-10% 10-20% (core) 5-10% Void only Voids only 1-2% 5-10% 5-10% >30% >30% 5-10% None None

Mica (biotite)

Other Opaque minerals Hornblende Hornblende ACF

<<1% 2% 2% <1%

Hornblende Opaque minerals Voids infilled with carbonate ACF

3% Opaque minerals Quartz -opaque rock fragment

<5% <5% <5%

Hornblende Hornblende, zircon, opaque minerals

1-3% 3-5% 2-3% 2 grains <1%

1%

Epidote Hornblende Epidote Voids infilled with carbonate ACF Quartz rock fragment

*Grain-size values are very fine (<0.0625 mm), fine (0.0625-0.25 mm), medium (0.25-0.49 mm), coarse (0.50-1.0 mm) and very coarse (> 1.0 mm). ACF, argillaceousclay fragment; K-spar. alkali feldspar.

formation and are not considered to be tempering material introduced by the potter. Further evidence for this interpretation is that the argillaceous clots do not contain inclusions that differ from those found in the paste.

Fibre (temper) component The fibre (temper), which is often only visible as secondary porosity containing some carbonized remains, is found as either elongate or oval to round voids. Only a few sherds were dominated by fibre (i.e. JBM-14, JBM-27, JBM-28; Table 1). A n u m b e r of the sherds without visible evidence for fibre (i.e. JBM-17, JBM-18,

JBM-3 l, JBM-35; see Fig. 3) have mineralogical and textural characteristics more similar to Early W o o d l a n d s (c. 3000 years BP) sand-tempered Thorn's Creek style pottery (Trinkley 1980, 1983; Smith 1997), if not for the provenance of the archaeological site. The fibre component, based upon carbonized secondary porosity, makes up (at least) 3 5 - 4 0 vol% in the fibre-dominated sherds (Table 1). Based upon the m o r p h o l o g y and textures of the stem fragments in thin section, the likely identity of the fibre is Tillandsia usnedoides (Spanish moss) and not pounded palmetto frond as found in the northern Florida and southern Georgia Orange series ceramics (Simpkins & Allard

CRESCENT SITE FIBRE-TEMPERED CERAMICS

Fig. 2. JBM-30 (crossed-polarized light). Very coarse to fine-grained aplastic components with 5-10 vol% fibre (as voids and carbonized remnants). The large grain near the centre of the sherd is polygranular quartz with sutured grain boundaries. These rock fragments are subangular to subrounded and many of the monocrystalline quartz grains (the blocky to sliver-like medium-sized grains) probably have a similar origin. Also visible are blocky feldspars, polysynthetic twinned plagioclase (upper left) and mica laths (centre). It should be noted how the outer surface appears to have had a floated surface treatment, possibly a self-slip. Scale bar represents 0,4 mm.

1986; Fig. 4). The cross-sections of the stem (oval voids) are generally more abundant than the stem fragments (elongate voids) and are found to exhibit specific orientation within the sherds (Fig. 4). In general, the cross-sections of

123

Fig. 4. Sherd JBM-8 (plane-polarized light). Crosssection of carbonized fibre stem fragment showing vascular traces and stem wall texture. Photomicrograph from core region of sherd. Scale bar represents 0.4 mm. stem fragments (round voids) are concentrated in the reduced core whereas the stem fragments (elongate voids) are found in the regions near both the interior and exterior surfaces. The regularity of this orientation suggests this was the result of the manufacturing process. The outer surfaces of the sherds (those that contain fibre) are devoid of fibre. This absence might be the result of floating or smoothing practices that were applied by the potter to the ceramic to bring up fine clays to cover the fibre, although one sample was found to have what appears to be a self-slip (Fig. 2). These observations generally concur with the fibre analysis study by Simpkins & Allard (1986) on 60 Stallings sherds from Florida, Georgia and South Carolina. Lastly, several sherds (JBM-14, JBM-32) were observed to have infilling (or rimming) of the secondary porosity by calcite mineralization. This is probably a result of post-depositional groundwater interaction rather than a response to cooking procedures (Sassaman 1993, pp. 186-188).

Paste components

Fig. 3. Sherd JBM-35 (plane-polarized light). Very coarse to medium-grained aplastic material and lack of fibre temper (or voids; black circles are epoxy bubbles). The blocky grain (upper left) is feldspar and subrounded grain (lower right) is quartz. Polygranular quartz rock fragments with sutured grain boundaries and undulatory extinction, as well as smaller blocky feldspar, rare biotite, and monocrystalline quartz grains (the blocky to sliver-like medium-sized grains) from the remainder of the temper. Scale bar represents 0.4 mm.

The petrographic investigation of the paste reveals a nearly isotropic (at 6 0 x objective magnification) matrix with a reddish to reddishblack hue. This material is probably a mixture of extremely fine-grained clay minerals and amorphous phases. Resolvable within this matrix are fine to very fine aplastic grains. These grains (mainly quartz, some feldspar, and opaque minerals) are monocrystalline, subangular to blocky in shape, and with very little rounding of the edges. These results are consistent with the macroscale examination by Trinkley (1998).

124

M.S. SMITH & M. B. TRINKLEY

Oxidation features (red to red-orange colour) were observed extending inward (to several millimetres) on both the inner and outer sherd surfaces, whereas the core region was generally reduced and was black to smoky grey.

Conclusions Although the Stallings Island Culture pottery is identified macroscopically as fibre tempered, the results of this study indicate that both fibre and non-fibre-tempered wares coexisted at the Crescent site. Regardless of whether they were fibre- or non-fibre-tempered wares, the mineralogical and textural characteristics of the aplastic (temper) components are consistent with the macroscale typological analysis (Trinkley & Hacker 1986; Trinkley 1998). The dominant aplastic (temper) components are quartz (mineral and rock fragment) with minor alkali feldspar, plagioclase and carbonized fibre. Quartz is either angular to subangular monocrystalline mineral crystals or polycrystalline rock fragments. The identity and textural features of the fine-grained aplastic minerals found in the paste are also consistent in both the fibre- and non-fibre-tempered sherds. Quartz, possibly feldspar, very fine opaque minerals, mica (biotite), epidote, and an amphibole (hornblende) are found in the paste. A few brick-red argillaceous inclusions (ACF; 0.1-0.5 mm) were found in some sherds. These observations indicate that the materials used in the pottery formation are not inconsistent with sediment (and clay) extraction from a fluvial Coastal Plain or estuarine setting in the southeastern USA (Nystrom et al. 1991; Ward et al. 1991). In addition, there were no indicator minerals or substances (e.g. shell material or carbonate clasts) that would point to a marine source for the clay minerals. The very coarse to coarse-grained aplastic components in these sherds are often quartz rock fragments with sutured grain boundaries and undulatory extinction. The texture and optical characteristics of this material, as well as the small amount of feldspar, biotite, and lack of carbonate or shell material, is consistent with sediments derived from the metamorphic terranes of the eastern Piedmont and western Coastal Plain (Fig. 1) by the Piedmont-draining rivers (Nystrom et al. 1991; Soller & Mills 1991). The rare instances of carbonized fibre temper that have extant textural features (Fig. 4) suggest that the identity may be Spanish moss (Tillandsia usnedoides) rather than the pounded palmetto frond found in similar fibre-tempered Orange wares of northern Florida (Simpkins & Allard 1986). Simpkins & Allard (1986) used

palaeobotantical techniques that required complete destruction of the sherds to separate out the fibre components. In that study they found very little of the diagnostic-scale botanical component that would allow specific identification of the Spanish moss species. The presence of the fibre is an important temper component in the majority of the samples and suggests that it was either actively incorporated into the pottery materials or was commonplace to the clay source materials. The orientation of the fibre reveals that its presence was part of the manufacturing process and not a casually incorporated substance. Why fibre was an important component may be related to its potential to cushion thermal shock. Thermal shock resistance is affected by the paste composition, as well as the distribution and uniformity of aplastic components and vessel shape (Rye 1981). The use of an organic temper would result in a secondary porosity left behind after firing that would help arrest crack development (Rye 1981; Bronitsky & Hamer 1986; Skibo et al. 1989). A mixture of fibre and quartz aplastic temper could represent possible attempts to improve the thermal properties of the pottery (Rye 1981; Bronitsky & Hamer 1986; Skibo et al. 1989). Furthermore, Sassaman (1993, pp. 161-163) reported that Stallings fibre-tempered pottery may have evolved from cooking vessels heated by indirect methods (stone boiling using soapstone) to later, more directly heated vessels (over a fire) where better thermal shock resistance would be an important property. Skibo et al. (1989, p. 140) suggested that although an organic-tempered pot may be more easily abraded and have slightly less thermal shock resistance, its lightness would allow its easy transportation in a mobile, foodgathering society, as well as allowing the manufacture of this pottery in one sitting, without the need for intermediate drying stages. However, some of the pottery at this site lacks fibre and more closely resembles the sandtempered Thom' s Creek wares (Fig. 1). Although all of the material factors proposed for the production of fibre-tempered pottery (Rye 1981; Bronitsky & Hamer 1986; Skibo et al. 1989) may have been important considerations, the incorporation of fibre may have been simply related to the specific clay source (or manufacturing location) or it may represent examples of a continually evolving pottery manufacturing process in the Late Archaic to Early Woodland Period in this area. The authors would like to thank M. Maggetti and B. Messiga for the opportunity to report on the Late Archaic to Early Woodland Stallings Island Culture

CRESCENT SITE FIBRE-TEMPERED CERAMICS ceramics; we also acknowledge the insightful and helpful comments and suggestions by our reviewer that improved the manuscript.

References BRONITSKY, G. & HAMER, R. 1986. Experiments in ceramic technology: the effects of various tempering materials on impact and thermal shock resistance. American Antiquity, 51, 89-101. CALMES, A. 1968. Test excavations at three Late Archaic shell-ring mounds on Hilton Head Island, Beaufort County, South Carolina. Southeastern Archaeological Conference Bulletin, 8, 45-48. Geological Society of America 1991. Rock Color Chart. Geological Society of America, Boulder, CO. GRIFFIN, J. B. 1943. An analysis and interpretation of the ceramic remains from two sites near Beaufort, South Carolina. Bureau of American Ethnology, Bulletin, 133, 155-168. HORTON, J. W., JR & ZULLO, V. A. 1991. An introduction to the geology of the Carolinas. In: HORTON, J. W., JR & ZULLO, V. A. (eds) The Geology of the Carolina: Carolina Geological Society Fiftieth Anniversary Volume. University of Tennessee Press, Knoxville, 1 - 10. JENKINS, N. J. 1982. Archaeology of the Gainesville Lake Area, synthesis. In: Office of Archaeological Investigations (ed.) Archaeological Investigations in the Gainesville Lake Area of the TennesseeTombigbee Waterway. Report of Investigations, 23(5). JENKINS, N. J., DYE, D. H. & WALTHALL, J. A. 1986. Early ceramic development in the Gulf Coastal Plain. In: FARNSWORTH, K. B. & EMERSON, T. E. (eds) Early Woodland Archaeology. Kampsville Seminars in Archaeology, 2, 546-563. NYSTROM, P. G., WILLOUGHBY, R. H., JR & PRICE, L. K. 1991. Cretaceous and Tertiary stratigraphy of the upper Coastal Plain, South Carolina. In: HORTON, J. W., JR & ZULLO, V. A. (eds) The Geology of the Carolina: Carolina Geological Society Fiftieth Anniversary Volume. University of Tennessee Press, Knoxville, 221-240. RICE, P. M. 1987. Pottery Analysis. A Sourcebook. University of Chicago Press, Chicago, IL. RYE, O. 1981. Pottery Technology: Principles and Reconstruction. Taraxacum Press, Washington, DC. SASSAMAN, K. E. 1993. Early Pottery in the Southeast: Tradition and Innovation in Cooking Technology. University of Alabama Press, Tuscaloosa. SEARS, W. H. & GRIFFIN, J. B. 1950. Fiber tempered pottery of the Southeast. In: GRIFFIN, J. B. (ed.) Prehistoric Pottery of the Eastern United States. Museum of Anthropology, University of Michigan, Ann Arbor, 2-20. SIMPKINS, D. L. & ALLARD, D. J. 1986. Isolation and identification of Spanish Moss fibre from a sample of Stallings and Orange Series ceramics. American Antiquity, 51, 102-117.

125

SINOPOLI, C. M. 1991. Approaches to Archaeological Ceramics. Plenum, New York. SKIBO, J. M., SCHIFFER, M. B. • REID, K. C. 1989. Organic-tempered pottery: an experimental study. American Antiquity, 54, 122-146. SMITH, M. S. 1997. Petrographic and mineral characterization of Thom's Creek plain sherds. In: TRINKLEY, M. & HACKER, D. (eds) Excavations at a Portion of the Secessionville Archaeological Site (38CH1456), James Island, Charleston County, South Carolina. Chicora Foundation, Columbia, Research Series, 52, 171-174. SOLLER, D. R. & MILLS, H. H. 1991. Surficial geology and geomorphology. In: HORTON, J. W., JR & ZULLO, V. A. (eds) The Geology of the Carolina: Carolina Geological Society Fiftieth Anniversary. Volume. University of Tennessee Press, Knoxville, 290-308. STOLTMAN, J. B. 1966. New radiocarbon dates for Southeastern fibre-tempered pottery. American Antiquity, 31, 872-874. STOLTMAN, J. B. 1989a. A quantitative approach to the petrographic analysis of ceramic thin sections. American Antiquity, 54, 147-160. STOLTMAN, J. B. 1989b. Ceramic petrography as a technique for documenting cultural interactions: an example from the Upper Mississippi Valley. American Antiquity, 56, 103-120. STOLTMAN, J. B., BURTON, J. H. & HAAS, J. 1992. Chemical and petrographic characterizations of ceramic pastes: two perspectives on a single data set. In: NEFF, H. (ed.) Chemical Characterization of Ceramic Pastes in Archaeology. Monographs in World Archaeology, 7, 85-92. TRINKLEY, M. B. 1980. A typology of Thorn's Creek pottery for the South Carolina coast. South Carolina Antiquities, 12, 1-35. TRINKLEY, M. B. 1983. Ceramics of the Central South Carolina coast. South Carolina Antiquities, 15, 43-54. TRINKLEY, M. B. 1998. Intensive Archaeological Survey of Crescent Plantation, Beaufort County, South Carolina. Chicora Foundation, Columbia, Archaeological Report, 226. TRINKLEY, M. B. & HACKER, D. 1986. Indian and Freedman Occupation at the Fish Haul Site (38BU805), Beaufort County, South Carolina. Chicora Foundation, Columbia, Research Series, 7. WARD, L. W., BAILEY, R. H. & CARTER, J. G. 1991. Pliocene and Early Pleistocene stratigraphy, depositional history, and molluscan paleobiogeography of the Coastal Plain. In: HORTON, J. W., JR & ZULLO, V. A. (eds) The Geology of the Carolina: Carolina Geological Society Fiftieth Anniversary Volume. University of Tennessee Press, Knoxville, 274-289. WHITBREAD, |. K. 1986. The characterization of argillaceous inclusions in ceramic thin sections. Archaeometry, 28, 79-88.

Chemical-mineralogical characterization of historical bricks from Ferrara: an integrated bulk and micro-analytical approach G I A N L U C A BIANCHINI, E L E N A M A R R O C C H I N O , ALESSANDRO MORETTI & CARMELA VACCARO

Dipartimento di Scienze della Terra, Universitgl degli Studi di Ferrara, Via Saragat 1, 1-44100 Ferrara, Italy (e-mail: [email protected]) Abstract: In this paper we present bulk X-ray fluorescence-X-ray diffraction (XRF/XRD) and microanalytical scanning electron microscope-electron microprobe analysis (SEMEMPA) data on historical bricks from Medieval or Renaissance buildings of Ferrara (NE Italy) to provide insights into the nature and provenance of the raw material as well as clues on the sintering techniques. Chemical data indicate that the starting materials were obtained by mixing high Cr-Ni clay and subordinate sand (both quarried from the Po river alluvial deposits) with the possible introduction of a Na-rich flux component. Thinsection observation, XRD and micro-analytical data indicate the presence of key accessory phases such as pyroxene, amphibole, epidote and rare olivine in the pre-fired mineral assemblage, confirming the utilization of the Po river sediments. Recognition of neo-formation firing phases (e.g. melilite, wollastonite), together with composition of micas, amphiboles and interstitial glasses, indicate kiln temperatures between c. 800 and 1000 °C. This provides guidelines for making new compatible and durable bricks to be utilized for restoration, and contributes to the preservation of historical masonry.

Ancient bricks from Medieval and Renaissance buildings of Ferrara (NE Italy) have been studied using bulk techniques that analyse a few grams of homogenized (powdered) sample, and microanalytical techniques that allow composition analysis of single phases of materials that are heterogeneous at a micrometre scale. This characterization provides useful clues to constrain the provenance of the original raw materials and the historical firing techniques. These data provide information useful to understand the physico-chemical behaviour of the studied bricks that will be helpful in future preservation and restoration strategies.

Historical background The Medieval city of Ferrara (NE Italy, Fig. 1), is located in the eastern part of the Po alluvial plain, where at that time the Primaro river branched off from the main course of Po river (Bondesan et al. 1995; Ferri & Giovannini 2000). This geographical position, situated on a major waterway at a natural crossroads between the Adriatic Sea and the Po alluvial plain, contributed to the flourishing of the city, which peaked during the Renaissance. In this period, the Este family promoted a further fortification of the wall ring and embellished the city centre in a grandiose

style with numerous sumptuous palaces. This building boom utilized bricks (cotto ferrarese) and mortars as the dominant building materials. The use of these construction materials is related to the geographical position and the geomorphological framework of the Ferrara area, which is characterized by the widespread presence of silico-clastic sediments (and the lack of rock outcrops; Amorosi et al. 2002). Bricks and a few cotto decorative elements were collected from some important historical monuments of the city. These included the Monastery of Sant'Antonio in Polesine (built in several phases during the 12th-16th centuries), the Church of Santa Mafia in Vado (founded in the 10th century and extensively modified in the 15th-16th centuries), the Church of Santo Stefano (founded in the 10th century, but rebuilt in the 15th-16th centuries), the Cathedral of Ferrara (apse, 15th- 16th centuries), the Schifanoia Palace ('Hall of Stuccoes', 15th- 16th centuries), and the surrounding city walls (15th century).

Bulk chemical composition of historical bricks from Ferrara Chemical composition (major and trace elements) of terracottas sampled from the

From: MAGGETTI, M. & MESSIGA, B. (eds) 2006. Geomaterials in Cultural Heritage.

Geological Society, London, Special Publications, 257, 127-140. 0305-87191061515.00 © The Geological Society of London 2006.

128

G. BIANCHINI E T A L .

(a)

(b)

Ferrara

Ci~' Walls

Fig. 1. Sketch map showing (a) the geographical position of Ferrara; (b) the locations of the studied historical buildings. S. Stef, Church of Santo Stefano; Schifa, Schifanoia Palace; SMV, Church of Santa Maria in Vado; StAPol, Monastery of Sant'Antonio in Polesine.

historical buildings of Ferrara and its city walls was determined by X-ray fluorescence (XRF) using a Philips PWI400 spectrometer. The method was calibrated with a wide range of natural standards, and the precision and accuracy were better than 5%. Representative analyses of bricks and cotto decorative elements from historical buildings and the city walls are presented in Table 1. Major element analyses have been recalculated and summed to 100% including LOI (loss of ignition at 1100 °C), and trace element data are expressed in parts per millions (ppm). To investigate the nature of the raw material, these results were compared with data that included natural sediments from the Po alluvial plain around Ferrara (Bianchini et al. 2002a, plus our unpublished data; Table 2) and plotted using bivariate diagrams with SiO2 wt% as a variation index (Fig. 2). The relatively restricted CaO range typical of the studied terracotta (in contrast to the high CaO variability recorded in the local clays) suggests that the most carbonaterich pelitic sediments were preferentially quarried as raw materials for brick production. This means that the Ferrara brick-makers noticed

that CaO-rich clays ultimately led to bricks characterized by better physico-mechanical properties (e.g. higher compressive strength, higher durability) because carbonates have a positive influence on brick textures by promoting a higher degree of vitrification (Elert et al. 2003). The A1203 content of the studied bricks is generally lower (and the SiO2 content slightly higher) than that of the local clays, showing a rough trend towards the composition of the local sands. The occurrence of coherent trends in other plots (see the K20 v. SiO2 diagram) indicates that such sands were used to temper the original clay body. This is confirmed by thin-section analysis, in which we observed that several samples are not homogeneous and are characterized by distinct textural domains (e.g. portions characterized by different grain sizes; presence of detrital grains). This suggests that the original raw materials were often mixed to obtain a suitable composition. On the other hand, the excess of Na20 recorded in the studied materials (compared with the local clay composition) cannot be ascribed to the introduction of sand in the starting body clay. This feature could be induced by

FERRARA HISTORICAL BRICKS

129

Table 1. Major (wt%) and trace element (ppm) composition of bricks and terracotta samples from the historic

buildings and city walls of Ferrara. Sant' Antonio in Polesine Church (StAPol)

SiO2 TiO2 A1203 Fe203 MnO MgO CaO NaeO K20 P205 LOI Total

EMT 54.80 0.71 14.91 5.78 0.12 4.79 10.28 1.91 2.04 0.20 4.46 100

19MT 52.71 0.74 15.21 6.48 0.12 5.17 10.97 1.57 2.29 0.19 4.55 100

Ni Co Cr V

152 21 239 89

167 20 252 127

32MTint 54.04 0.64 13.34 5.32 0.15 3.51 13.66 1.48 2.09 0.23 6 100 156 21 202 114

104MT 52.03 0.75 14.52 6.16 0.14 4.20 13.80 0.97 2.35 0.18 4.90 100

105MT 54.11 0.68 15.14 5.79 0.11 5.06 9.94 1.95 2.08 0.20 4.95 100

107MT 53.11 0.71 15.35 6.22 0.12 5.26 10.21 1.65 2.73 0.23 4.43 100

122MT 55.27 0.67 14.77 5.76 0.11 4.90 9.42 1.90 2.24 0.18 4.80 100

154MT 52.55 0.73 15.11 6.56 0.13 4.86 13.14 1.20 2.27 0.18 3.27 100

127 18 169 116

153 21 231 110

165 21 222 118

150 19 221 98

166 21 241 121

Cathedral

SiO2 TiO2 A1203 Fe203 MnO MgO CaO Na20 K20 P205 LOI Total

C1 57.76 0.65 13.64 5.44 0.11 4.14 9.22 1.47 2.13 0.25 5.2 100

C8 54.91 0.64 13.96 5.50 0.12 4.18 9.13 1.28 2.39 0.27 7.63 100

C9 53.90 0.63 14.28 5.53 0.11 4.22 9.02 1.12 2.54 0.21 8.45 100

MT2 55.90 0.68 13.86 5.76 0.12 4.69 11.49 1.51 2.20 0.25 3.54 ! 00

MT3 53.89 0.62 12.60 5.20 0.10 4.42 11.67 1.48 2.42 0.26 7.34 100

MT6 54.59 0.64 13.82 5.40 0.10 4.05 8.48 1.08 2.30 0.21 9.32 100

MT14 52.20 0.62 12.55 5.15 0.11 4.25 11.33 1.52 2.66 0.28 9.32 100

MT15 54.48 0.65 13.64 5.50 0.11 4.07 10.22 1.40 2.35 0.25 7.33 100

Ni Co Cr V

143 17 201 92

148 14 187 104

148 19 194 102

161 21 216 100

143 14 229 133

145 17 188 112

133 17 199 98

142 17 201 102

Santo Stefano Church (S.Stef)

SiO2 TiO2 A1203 Fe203 MnO MgO CaO Na20 K20 P205 LOI Total

SL36 55.69 0.72 14.32 6.42 0.13 4.14 10.96 0.87 2.31 0.43 4.02 100

SL37 56.30 0.67 13.37 5.90 0.12 4.65 12.36 0.99 2.05 0.48 3.11 100

SL39 56.50 0.72 15.07 6.33 0.12 4.54 8.47 0.72 2.46 0.34 4.73 100

SL40 49.03 0.68 14.51 6.27 0.11 3.84 10.81 0.68 2.30 0.47 11.31 100

SL41 56.61 0.71 15.47 6.07 0.12 5.56 10.29 1.21 2.08 0.35 1.52 100

SL46 51.07 0.67 15.57 5.85 0.08 5.00 9.40 0.62 2.48 0.28 8.97 100

Ni Co Cr V

144 19 252 100

140 20 239 95

151 22 233 98

152 19 218 100

155 22 242 93

159 21 304 124 (Continued)

G. BIANCHINI ETAL.

130

Table 1. Continued City Walls *

•

.

:,

,

*

*

*

SiO2 TiO2 A1203 Fe203 MnO MgO CaO Na20 K20 P205 LOI Total

1EG 54.70 0.74 16.16 6.62 0.11 4.68 9.76 1.24 2.48 0.29 3.23 100

7EG 53.37 0.70 14.47 5.96 0.11 4.32 13.26 1.23 2.09 0.28 4.20 100

9EG 52.06 0.76 16.80 7.18 0.12 4.93 10.67 1.04 2.66 0.40 3.38 100

17EG 53.03 0.73 15.94 6.70 0.13 5.13 11.89 1.16 2.38 0.28 2.62 100

19EG 54.87 0.79 16.78 7.10 0.12 5.14 9.51 1.23 2.55 0.25 1.66 100

22EG 53.36 0.69 14.92 6.03 0.12 4.40 12.94 1.27 2.24 0.30 3.74 100

33EG 51.73 0.64 13.95 5.46 0.12 4.72 11.92 0.84 2.59 0.33 7.70 100

35EG 47.80 0.69 14.96 6.34 0.12 4.74 11.43 1.26 2.52 0.29 9.85 100

37EG 51.11 0.73 15.80 6.46 0.13 5.61 8.87 1.39 2.50 0.21 7.18 100

40EG 53.54 0.70 15.26 5.99 0.12 4.82 11.16 1.33 2.22 0.33 4.52 100

Ni Co Cr V

179 19 264 104

159 18 200 97

199 19 288 118

197 22 246 102

195 25 248 116

157 22 204 80

144 16 188 91

166 25 238 87

185 22 253 131

165 20 231 82

Santa Maria in Vado Church (SMV) *

*

,

*

*

*

SiO2 TiO2 A1203 Fe203 MnO MgO CaO Na20 K20 P205 LOI Total

SL25 56.50 0.70 16.08 6.08 0.11 4.75 8.12 0.89 2.47 0.28 4.01 100

SL26 53.54 0.74 16.95 6.78 0.09 5.15 8.79 0.88 2.72 0.28 4.08 100

SL27 53.10 0.68 14.30 5.90 0.12 5.54 11.86 1.07 2.02 0.38 5.03 100

SL28 50.78 0.65 14.92 5.69 0.11 4.64 11.15 0.73 2.63 0.35 8.35 100

SL29 54.58 0.70 15.46 6.24 0.13 4.77 10.34 1.10 2.14 0.36 4.19 100

N4 54.46 0.72 16.34 6.20 0.08 4.98 7.72 0.93 2.76 0.17 5.63 100

N 13 50.58 0.67 14.66 5.91 0.12 4.92 10.86 0.88 2.48 0.22 8.72 100

N20 49.28 0.60 12.78 5.04 0.10 3.57 12.00 1.12 2.06 0.23 13.22 100

Ni Co Cr V

148 22 210 114

187 25 262 135

156 21 243 117

150 22 207 114

160 22 259 110

164 23 233 120

159 19 240 117

112 15 170 81

*

*

.

,

*

*

*

*

SiO2 TiO2 A1203 Fe203 MnO MgO CaO Na20 K20 P205 LOI Total

SE 15P 51.66 0.70 14.77 6.08 0.13 4.87 11.80 1.16 2.34 0.17 6.31 100

SE9P 53.11 0.76 15.49 6.69 0.13 4.71 11.46 1.34 2.42 0.21 3.68 100

SE 19P 53.46 0.73 15.30 6.37 0.12 5.17 11.87 1.33 2.19 0.20 3.28 100

SN6P 54.05 0.68 14.35 6.04 0.13 5.00 10.94 1.53 2.31 0.20 4.77 100

SE20P 54.34 0.71 14.83 6.06 0.12 4.88 i 1.01 1.56 2.10 0.19 4.20 100

SS5P 54.61 0.75 15.26 6.50 0.13 5.28 10.19 1.61 2.36 0.19 3.12 100

SS4P 54.89 0.76 15.40 6.61 0.13 5.31 11.24 1.36 2.43 0.19 1.7 100

SEIOP 58.23 0.80 14.78 6.89 0.14 3.56 10.88 1.47 2.59 0.28 0.38 100

Ni Co Cr V

155 22 227 112

162 23 243 123

172 21 250 92

154 20 224 103

153 19 229 112

169 22 251 120

181 24 262 107

152 20 221 111

Schifanoia Palace (Schifa)

Symbols: *, bricks; = , cotto decorative elements. Labels: StAPol, bricks and terracotta elements from the Monastery of Sant'Antonio in Polesine; SMV, bricks and terracotta elements from the Church of Santa Maria in Vado; S.Stef, bricks and terracotta elements from the Church of Santo Stefano; Schifa, bricks and terracotta elements from Schifanoia Palace; Cathedral, bricks and terracotta elements from the Cathedral of Ferrara.

FERRARA HISTORICAL BRICKS

131

Table 2. Major (wt%) and trace element (ppm) composition of natural sediments from the Ferrara surrounding Low-Cr clay* SL8

SL9

SL10

SiO2 TiO2

High-Cr clay* SL14

SL4

SL16

SLI7

Po river sands SL18

ME1

F5C2

F6C4

F8C4

49.63 49.00 47.88 0.75 0.75 0.72 A1203 1 9 . 0 5 1 8 . 4 6 1 7 . 6 3 Fe203 6.95 7.09 6.49 MnO 0.10 0.12 0.10 MgO 3.15 3.03 3.01 CaO 6.47 7.65 7.93 Na20 0.35 0.36 0.39 K20 3.38 3.37 2.99 P205 0.18 0.24 0.24 LOI 9.98 9.95 1 2 . 6 2 Total 100 100 100

47.20 0.68 15.90 6.76 0.14 2.99 9.72 0.48 2.58 0.28 13.27 100

53.49 45.57 5 1 . 1 1 51.27 69.83 58.14 60.85 62.05 0.72 0.67 0.70 0.74 0.36 0.39 0.41 0.41 20.96 1 4 . 9 4 1 7 . 0 7 17.94 9.51 1 0 . 7 2 10.37 9.96 6.06 6.09 6.14 6.51 2.66 3.52 3.64 3.34 0.04 0.12 0.07 0.07 0.07 0.06 0.08 0.08 3.20 4.32 4.69 4.23 3.85 4.23 3.47 3.31 1.32 11.04 6.37 4.82 6.10 10.31 8.79 9.03 0.71 0.53 0.67 0.48 1.98 1.53 1.79 1.79 3.38 2.22 2.51 2.67 1.98 2.07 2.03 1.91 0.00 0.33 0.17 0.15 0.08 0.11 0.12 0.12 1 0 . 1 4 1 4 . 1 6 10.5 11.12 3.58 8.91 8.46 8.00 100 100 100 100 100 100 100 100

Ni Co Cr V

89 18 122 144

112 17 268 174

82 20 136 163

82 20 143 155

80 18 134 148

145 21 210 124

155 29 288 146

162 32 281 158

83 13 138 40

96 14 132 59

87 12 138 56

81 10 124 55

*Analysesfrom Bianchiniet al. (2002a).

soluble salts precipitated from Na-bearing aqueous solutions. However, the common occurrence of such Na20 excess appears to be unrelated to the spatial distribution of the sampled bricks, some of which were located indoors and far from the floor (i.e. in a position not accessible to rain and/or ground water). Therefore, we suspect the addition of an 'exotic' Na-rich component as a fluxing agent incorporated during the firing phase. In this regard, excluding feldsparrich lithologies (no enrichment in A1203 is observed) and salts such as Na2CO3 (not available in the area and too expensive in Medieval times), we speculate that either vegetable ash (obtained by burning seaweed?; see Chapman & Chapman 1980; Stiaffini 1999) or common marine salt (NaCI) were used. Introduction of the latter component would cause acceleration of mineral decomposition and subsequent crystallization of newly formed Ca-silicates. This would lower the prograde chemical reactions by c. 100-200 °C compared with those in NaCl-free ceramic systems (Maggetti & Von Der Crone 2004). To address the provenance of the materials used, trace elements such as Ni and Cr (Fig. 3) indicate that the clay sediments around Ferrara can be separated into two groups characterized by high (Cr > 1 8 0 p p m and Ni > 1 0 0 p p m ; High-Cr) and low (Cr <180ppm and Ni < 1 0 0 p p m ; Low-Cr) contents. Bianchini et al. (2002a) showed that the Low-Cr sediments are characterized by a higher proportion of clay minerals in which smectite + mixed layers are

more abundant than chlorite. The High-Cr sediments have a coarser grain size and a lower abundance of clay minerals, with chlorite (Mg-rich chlorite in this group of samples) predominating over s m e c t i t e + m i x e d layers. Most of the studied Medieval or Renaissance bricks and terracotta elements seem to have been prepared starting from High-Cr sediments, as these clays show a chemical affinity with the present-day Po river sediments. The Low-Cr compositions (recorded in only a few samples) are analogous to the sediments of rivers flowing from the Bolognese Apennines. This probably means that Low-Cr clays were not available in the Ferrara area at that time. It is plausible to assume that Low-Cr clays were introduced in the area only after important hydraulic works in the 14th-16th centuries (Bondesan et al. 1995) resulted in the diversion of some Apenninic torrent-rivers (e.g. the Reno river) into the southern branches of the Po river (which was flowing south of Ferrara at that time). The only two brick samples characterized by a Low-Cr affinity have to be considered as outliers, possibly representing 'allochthonous' bricks probably made in neighbouring areas where Reno river sediments were available.

Mineralogical composition of historical bricks from Ferrara The mineralogical composition of the terracottas was investigated by X-ray powder diffraction (XRD) carried out using a Philips P W I 8 6 0 / 0 0

132

G. BIANCHINI E T A L ++

+

• StAPol raSMV

200.

+

180.

~.~<x+x

°SSmf 0 Schifa

++

+ + x .~-w ~

160

+

120

+,"

+ l"figh-Cr clay

80

X

60 50

55

60

65

70

75

20

16

0

quadrant 0

14 12 •

o

Low-Cr clay

40

Ig

.-

xx

+

4

+++

2

.,

0

. . . .

;

50

. . . .

;

55

;

70

o

J3

....

: ....

: ....

55

~

xx

....

[

65

x I''''I

~

75

~" 1.0

0.0

....

50

• SU~Pol

'. . . . .

55

I ....

60

: ....

65 Si02 Ovt%)

I ....

70

0 Still

I

75

}

0 SMV eSS~f

+H~Cf~ } Bricks

120 160 2o0 240 280 320 Cr (m~)

75

+

o~

8o

; " " , , I

. . . .

65

+

~x

35

. . . .

60

4o

Fig. 3. Binarydiagram for Ni v. Cr. Symbols and labels as in Figure 2. Quadrants delimited by dotted lines are useful to identify the nature of the clay raw materials (High-Cr v. Low-Cr),

$

~ 6

+ •

X Po Sands × Low-Cr clay

100

x X

10

..

0 Cathedral ° City Warts

140

and

decorative elements

x Low-Or d~ ~Po~

Natural

sediments

• Cry Wds

Fig. 2. Major element oxide bivariate diagrams. Analyses recalculated to 100% on an H20-CO2 (LOI) free basis. Low-Cr clay, clays characterized by 'Apennine' affinity; High-Cr clays, clays closely comparable with the present-day fine sediments of the Po river; other abbreviations as in Figure 1.

diffractometer with graphite-filtered Cu-Ka radiation (.A = 1.54 A). The diffraction patterns were collected in the 20 angular range from 5 to 50=, with 5 seconds per step (0.02: 20). Most samples can be attributed to two distinct mineral parageneses (Bianchini et al. 2002b), as follows.

(1) Carbonate-beating mineral assemblages (calcite _+ dolomite) with ubiquitous quartz, illite/muscovite, biotite, iron oxides, alkaline feldspar and plagioclase. Amphibole has been detected, but it is not ubiquitous. M e l i l i t e _ wollastonite may also be present. These mineral associations have been recognized in samples from the Monastery of Sant'Antonio in Polesine, the Church of Santa Maria in Vado, and the city walls. (2) Clinopyroxene-bearing (carbonate-free) mineral assemblages, with quartz, alkaline feldspar, plagioclase, iron oxides, melilite and wollastonite. Biotite and illite/muscovite are not ubiquitous and amphibole is rare. These mineral associations are typical for samples from the Church of Santo Stefano. Both mineral parageneses can be observed in the samples from the Ferrara Cathedral, and mineral analyses are not yet available for Schifanoia samples. These mineral parageneses do not appear to be consistent with the (CaO + MgO)-A1203-SiO2 phase diagram (Fig. 4) in which the stability fields of firing phases are reported (Duminuco et al. 1998; Riccardi et al. 1999; Artioli et al. 2000). In Figure 4 the historical terracottas and the local clay composition plot mainly in the subtriangle Q z - W o ( D i ) - A n that corresponds to the quartz-wollastonite (diopside)plagioclase equilibrium paragenesis. However, the common presence of carbonates and/or melilite (recorded by XRD in several samples) suggests that equilibrium conditions were not attained. This can be ascribed to the inhomogeneity of the original starting material, which consists of several micro-domains of different composition, and indicates a limited mobility of the chemical species during the firing processes (Duminuco et al. 1998; Riccardi et al. 1999).

FERRARA HISTORICAL BRICKS

133

SiO 2 (wt%) Bricksand 'cotto' decorative e l e m e n t s o f F e r r a r a historical buildings ~

//~ / t\ ,

, "

Wo" Di

CaO+MgO(wt%)

ii~

..... ~l~errara lays ~

A120 3 (wt%)

Fig. 4. (CaO + MgO)-AI203-SiO2 phase diagram with compositions of historical bricks and terracotta elements from Ferrara. Compositions of firing phases, such as wollastonite, diopside, melilite (gehlenite) and plagioclase (anorthite), as well as relative tie-lines, are reported.

For the same reason, XRD bulk analyses cannot be considered completely reliable for the estimation of firing temperatures (Duminuco et al. 1998; Riccardi et al. 1999). This is evident when we consider some samples apparently characterized by 'misleading parageneses' containing carbonate minerals (an indicator of low firing temperature, i.e. < 850 °C) and clinopyroxene and/or plagioclase (usually considered high-temperature products, i.e. > 950 °C). Microanalysis of constituent phases The apparent incongruence in the recorded mineral parageneses can be examined using petrographic thin-section observation and in situ analytical techniques capable of investigating the reactions occurring in the various micro-domains. Microanalytical data have been obtained by: (1) scanning electron microscope (SEM) at the University of Ferrara using a Cambridge Stereoscan S-360 that provided semiquantitative analyses and sample images at a scale of a few tens of micrometres; (2) electron microprobe analysis (EMPA) using a CAMECA SX-50 at the CNR-IGG Institute of Padova with natural silicates and oxides as standards to provide quantitative analyses of the constituent phases. Microprobe analyses were focused on coarse detritic grains possibly representing traces of the sand fraction introduced as temper. Among the mineral phases that constitute the sand we

ignored common felsic minerals (i.e. quartz and feldspars) and focused investigation on accessory minerals, which provide more reliable insights into the raw material provenance as well as clues to the firing temperatures. In particular, micas (biotite and muscovite; Table 3) and amphiboles (Table 4) were examined and compared with data on their firing behaviour (i.e. temperature stability fields; Maggetti et al. 1984; Capel et al. 1985; Brindley & Lemaitre 1987; Dumunico et al. 1998; Riccardi et al. 1999; Cultrone et al. 2000; Bianchini et al. 2002b). The K-poor composition of biotites (Table 3) does not reflect a mineral destabilization that occurs during the firing process. Similar compositions are commonly observed in a wide variety of sedimentary environments where pristine biotite crystals progressively release potassium, ultimately leading to the formation of hydrobiotite (Blum & Erel 1997; Murphy et al. 1998). In sample MT3 muscovite is colourless, whereas hydrobiotite is nearly opaque but still homogeneous in composition, and amphibole is characterized by a wide compositional range, which includes coarse and homogeneous crystals of orange hornblende and edenite and brown pargasite. Actinolite microcrystals have been found within the finer matrix. The orange colour of hornblende suggests oxidizing conditions and firing temperatures around 800 °C, and the absence of destabilization

MT3

Hydrobiotite

Muscovite

MT 15

Hydrobiotite

53.23 0.08 27.53 0.00 3.48 0.01 3.22 0.10 1.41 8.68 97.73

2.27 0.00 1.78 0.00 1.82 0.00 ! .78 0.02 0.15 0.18 8.00

29.36 0.04 19.48 0.05 28. I 1 0.05 15.40 0.22 0.97 1.81 95.48 2.23 0.00 ! .74 0.00 1.84 0.01 ! .90 0.03 0.13 0.18 8.05

28.70 0.03 19.00 0.05 28.33 0.08 16.40 0.38 0.87 i.80 95.64

Atomic formula units (a.f.u.) on the basis of 11 oxygens,

3.23 0.01 2.52 0.00 0.10 0.00 0.19 0.01 0.1 i 0.73 6.88

48.51 0.10 32.09 0.04 1.93 0.00 1.95 0.11 0.85 8.57 94.14 2.11 0.00 2.06 0.00 1.76 0.01 1.81 0.00 0.I0 0.12 7.97

27.62 0.04 22.87 0.02 27.62 0.14 15.96 0.03 0.66 1.18 96.14 2.14 0.00 2.05 0.00 2.06 0.02 1.28 0.09 0.17 0.24 8.04

26.30 0.02 21.36 0.03 30.3 i 0.28 10.53 0.97 1.05 2.27 93.13 3.01 0.01 2.90 0.00 0.02 0.00 0.03 0.02 0.24 0.86 7.08

46.58 0.1 I 38.05 0.00 0.33 0.07 0.26 0.32 1.90 10.46 98.09 3.04 0.00 2.91 0.00 0.02 0.00 0.02 0.01 0.15 0.86 7.01

47.71 0.08 38.86 0.00 0.34 0.02 0.20 0.18 1.18 10.55 99.13 3.31 0.00 2.35 0.00 0.07 0.01 0.27 0.06 0.37 0.53 6.96

50.23 0.06 30.21 0.00 1.25 0.09 2.71 0.8 i 2.91 6.27 94.54 3,09 0.02 2.81 0.00 0.05 0.00 0.05 0.00 0.12 0.83 6.97

47.75 0.39 36.88 0.04 0.86 0.00 0.53 0.03 0.98 10.06 97.53 2.43 0.00 1.93 0.00 i .44 0.02 1.38 0.03 0,29 0.48 7,99

32.10 0.01 21.60 0.00 22.71 0.29 12.22 0.37 !.96 4.02 95.27

2.20 0.06 2.03 0.00 1.55 0.00 ! .55 0.03 0.29 0.33 8.04

28.66 0.96 22.41 0.01 24.12 0.02 13.52 0.32 1.93 3.41 95.37

MT3p3c MT3p3d MT3p4c MT3p4d MT3p4f MT3pld MTI5 2a MTI5 2b MTI5 3a MTI5 3a* MTI5 lb MTI5 2 h

Muscovite

Cations/I 1 oxygens Si 3.44 Ti 0.00 AI 2. i 0 Cr 0.00 Fe 0.19 Mn 0.00 Mg 0.3 ! Ca 0.01 Na 0.18 K 0.72 Total 6.95

SiO2 TiO2 A1203 Cr203 FeO MnO MgO CaO Na20 K20 Total

Mineral:

Sample:

Table 3. Representative compositions of mica (muscovite and hydrobiotite) in historical bricks from Ferrara

2.19 0.03 2.05 0.00 1.61 0.00 1.49 0.04 0.33 0.35 8.09

27.95 0.52 22.13 0.05 24.56 0.04 12.75 0.42 2.18 3.47 94.08

84.99 0.05 7.12 0.00 1.04 0.07 0.67 0.03 1.05 4.31 99.32 Non-stoichiometric

59.52 0.43 15.21 0.06 9.98 0. I 0 3.66 0.06 !.59 5.95 96.55

MT15 21 MTI5 4a MTI5 4c

Destabilization products on hydrobiotite

.r-,

t~

> 7: ¢3

135

FERRARA HISTORICAL BRICKS Table 4. Representative compositions of amphibole in historical bricks from Ferrara

Experimental trial Mineral: Actinolite Hornblende Pargasite Edenite Actinolite actinolite (fired at 800 °C) MT3ple MT3plf MT3p3a MT3p3b MT3p2c MT3p2d MT3plb MT3plc clc Sample:

MT3

SiO2 58.56 5 7 . 0 1 48.52 47.73 43.02 TiO2 0.00 0.03 0.23 0.20 1.36 A1203 0.08 0.14 9.70 9.59 11.84 Cr203 0.00 0.06 0.09 0.20 0.07 FeO 3.78 3.46 11.85 11.83 15.06 MnO 0.10 0.10 0.19 0.30 0.18 MgO 22.20 21.58 1 3 . 8 6 1 3 . 7 5 10.35 CaO 13.33 1 3 . 5 7 1 2 . 7 9 1 2 . 5 5 1 1 . 8 2 Na20 0.21 0.11 0.98 1.03 1.30 K20 0.03 0.04 0.30 0.31 1.57 Total 98.30 9 6 . 1 1 98.50 9 7 . 5 1 96.58 Cations/24 oxygens Si 7.59 7.99 6.97 6.95 6.50 Ti 0.00 0.00 0.03 0.02 0.16 A1 0.01 0.02 1.64 1.65 2.11 Fe 0.41 0.41 1.42 1.44 1.90 Mn 0.01 0.01 0.02 0.04 0.02 Mg 4.28 4.51 2.97 2.98 2.33 Ca 3.08 2.04 1.97 1.96 1.92 Na 0.05 0.03 0.27 0.29 0.38 K 0.01 0.01 0.06 0.06 0.30 Total 15.44 1 5 . 0 2 1 5 . 3 5 1 5 . 3 8 1 5 . 6 3

C1

43.20 1.33 11.79 0.04 15.27 0.20 9.94 11.50 1.42 1.57 96.26

48.72 0.10 8.22 0.00 13.81 0.34 12.74 9.92 2.45 0.28 96.57

47.79 0.15 10.08 0.00 14.81 0.39 11.95 9.49 2.53 0.29 97.47

57.90 0.01 1.12 0.11 5.85 0.22 20.83 12.01 0.65 0.52 99.20

56.03 0.00 2.75 0.29 5.94 0.15 19.69 12.27 0.95 0.03 98.11

6.56 0.15 2.11 1.91 0.03 2.25 1.87 0.42 0.30 15.60

7.18 0.01 1.43 1.70 0.04 2.80 1.57 0.70 0.05 15.48

7.00 0.02 1.74 1.82 0.05 2.61 1.49 0.72 0.05 15.50

7.45 0.01 1.09 0.69 0.02 3.60 1.97 0.27 0.13 15.21

7.79 0.00 0.45 0.69 0.02 4.08 1.83 0.26 0.01 15.12

Atomicformulaunits on the basis of 24 oxygens.Amphibolenomenclatureaccordingto Leakeet al. (1997).

evidence indicates a temperature lower than 900 °C (Maggetti et al. 1984). Moreover, the persistence of actinolite (well equilibrated up to 800 °C in our firing experiments of the local

Fig. 5. Scanning electron micrograph of an actinolite amphibole in an experimental trial obtained by firing a local clay at 800 °C. Chemical analysis of the same crystal by EMPA is also reported.

clays, but not recorded at 900 °C; Fig. 5) would confirm that sample MT3 has been fired at a temperature between 800 and 900 °C. The interpretation of the observed anhedral crystals of epidote (approaching the ideal Ca2(Fe,A1)3(SiOa)3(OH) composition; Table 5) is not straightforward. This mineral is not reported as a common firing phase in ceramics. Its stability field appears to be confined at temperatures lower than 550 °C, as suggested by natural analogues in contact metamorphic parageneses (albite + epidote hornfels facies; Lentz et al. 1995; Moor & Gunderson 1995; Singoyi & Zaw 2001 ; Martfnez-Serrano 2002). Therefore the observed epidote microcrystals may represent relicts of coarser crystals present in pre-fired mineral assemblages. This agrees with the presence of epidote in the sand fraction of sediments from the Po river (Marchesini et al. 2000). In sample MTI5 amphibole and epidote are not detected, muscovite is still colourless and hydrobiotite is totally destabilized. Microprobe investigation of hydrobiotite crystals indicates that they are not homogeneous and stoichiometric balanced compositions of hydrobiotite coexist with non-stoichiometric SiO2-A1203-K20-rich

136

G. BIANCHINI ETAL.

Table 5. Representative compositions of epidote in historical bricks from Ferrara

The observation of crystal habits (euhedral v. anhedral) is useful to discriminate between phases originally present in the pre-fired mineral Sample: C1 C1 C1 MT3 assemblage and neoformed phases (Fig. 6). epidote epidote Within the fine matrix of sample MT15, Mineral: Cla cld clg MT3p4e SEM analysis recorded extremely variable compositions (in terms of Si, Ca, A1 and Mg; SiO2 38.06 38.94 39.31 37.40 Figs 7 and 8). These analyses do not fit stoichioTiO2 0.06 0.14 0.08 0.01 metrically with potential firing phases (e.g. wolA1203 22.41 28.08 28.28 21.04 lastonite, melilite, clinopyroxene) and could be FeO 13.00 5.25 5.60 13.55 interpreted as the compositions of glassy blebs. MnO 0.09 0.06 0.15 0.19 The different glass compositions within the MgO 0.07 0.06 0.07 0.00 same sample may be related to lack of equiliCaO 23.93 2 4 . 5 8 24.58 22.68 brium melting, and result in the coexistence of Na20 0.05 0.04 0.03 0.00 K20 0.01 0.05 0.04 0.00 different micro-domains that were not homogenTotal 97.68 97.20 98.13 94.87 ized during the firing processes. This implies that during the incipient partial melting, the compoCations/25 oxygens sition of the neoformed melt phase (glass) is Si 6.24 6.14 6.14 6.33 strongly influenced by the particular paragenesis Ti 0.01 0.02 0.01 0.00 of each micro-domain. AI 4.33 5.22 5.21 4.20 Microanalysis also revealed, in sample MT 15, Cr 0.00 0.00 0.00 0.00 the presence of euhedral crystals of olivine Fe 1.78 0.69 0.73 1.92 (Fig. 9). This may represent evidence for the Mn 0.01 0.01 0.02 0.03 origin of the raw materials. The persistence of Mg 0.02 0.01 0.02 0.130 olivine (a mineral very susceptible to weathering Ca 4.20 4.15 4.12 4.11 processes) in the mineral assemblage of the Total 16.59 1 6 . 2 4 16.24 16.58 starting sediments, together with the high Atomic formulaunits on the basis of 25 oxygens. N i - C r clay content, suggests that sediments from the Po (a river hydrological basin that contains mafic and ultramafic rocks) may be the compositions (Table 3). This sample is also source. This hypothesis is strengthened by the characterized by the widespread presence of presence of olivine (Fig. 10) in experimental glassy films within the fine matrix that brick trials manufactured in our laboratories suggest a higher firing temperature compared using alluvial clays of the Po river. with sample MT3 (probably approaching 1000 °C). The persistence of carbonate cannot be used as C o n c l u s i o n s an indicator of low firing temperature because In this paper, chemical and mineralogical the reactivity of large crystals of carbonate (poss- characterization of bricks and terracotta elements ibly introduced as temper) is often limited to from historical buildings and city walls of surfaces and would allow carbonate preservation Ferrara has allowed us to evaluate the nature of as a metastable phase at a higher temperature the original raw material (i.e. clay-rich sediment than expected. Moreover, experiments show of local provenance) and to define technological that not all CaO could combine with other information regarding the manufacture (i.e. firing oxides to form new calc-silicate phases during temperatures between c. 800 and 1000°C). firing. This would lead to rehydration of the This information provides guidelines for the free CaO and the formation of secondary production and use of new bricks, tiles and terracalcite (Maggetti et al. 1984). Furthermore, cotta elements that are durable and compatible carbonate could also be related to weathering with the historical materials. The proper preprocesses that form 'secondary' calcite in servation of ancient architectural heritage is microcracks and pores (Lopez-Arce & Garcia- extremely important and should be taken into Guinea 2005). account during the restoration of damaged hisClinopyroxene and plagioclase cannot be con- torical masonries. This study shows that to sidered as indicators of high firing temperature characterize the old materials and to establish because they occur as accessory minerals in the causes of the decay, as well as to plan for natural sediments of the Ferrara surroundings suitable protective or restoration treatment, it is (Marchesini et al. 2000) and could have been necessary to identify and characterize the present in the pre-fired mineral assemblage. materials that were used to make these

2

~

4

8

8

'O~c~mtuge t

~1$c~e 2324 (:Is Ox$~: 9~198key (6 cts)

)

'

t0

12

I~I

/ 10~

¢

"m~mmm~

Spectrum 4

Anhedral clinopyroxene

~'ig. 6. Scanning electron micrographs (and EDS spectra) of pyroxene crystals recorded in sample MT15; (a) euhedral orthopyroxene plausibly present in the pre-firing mineral lssemblage; (b) anhedral clinopyroxene microcrystal in the matrix (plausible neoformation).

Ill

~ec~

2

6 ; 0 . . . . . . . .

'l~c:lNnlm~ 1

3 6 s Curs~. O.c40~,v~)ds)

4CIn

N

QZ

""

"

gl ~'1

~i~

"llllln ~

! Z 3 4 $ 'd'8o~lOOec~ Curoct $~Jrl I~,,'~ms)

/

6

I

T

O

9I

~'ig. 7. Scanning electron micrographs (and EDS spectra) of silica-rich glassy patches within the matrix of sample MTI5.

i

%'-

*.',,,,;,. •

:>I-: :.,: t

,m

°'*

..w

"%4.

"•

t 3 • tl OJeo¢,W~V

$ cb

,lleci~ I,inlo !

FERRARA HISTORICAL BRICKS

~

'a~m twpt

•

lq~

139

'ea:mn~ t

Fig. 8. Scanning electron micrographs (and EDS spectra) of calcium-rich glassy patches within the matrix of sample MT15.

Fig. 10. Scanning electron micrographs of an olivine crystal in an experimental trial obtained by firing a local clay at 800 °C. Chemical analyses of the same crystal by EMPA are also reported.

1 2 3 4 5 6 ~e ~g61 ¢tS Cursor~ ;'o314 ~ V ( g d , )

7

8

9

10

Fig. 9. Scanning electron micrographs (and EDS spectrum) of a euhedral olivine crystal in sample MTI5.

architectural elements. This gives the restorer adequate information for choosing suitable new materials when replacement is necessary and to avoid incorrect restoration materials, such as were used in the Piazza Municipale of Ferrara, where new floor tiles were totally damaged and broken by f r e e z e - t h a w cycles during the first winter.

140

G. BIANCHINI ETAL.

L. Beccaluva and F. Siena are kindly acknowledged for their preliminary review of the manuscript. The authors are also grateful to R. Tassinari (Universit/a di Ferrara) and R. Carampin (CNR-IGG, Padova) for their analytical assistance, and to the reviewers for their constructive comments.

References AMOROSI, A., CENTINEO, M. C., DINELLI, E., LUCCmNI, F. & TATEO, F. 2002. Geochemical and mineralogical variations as indicators of provenance changes in Late Quaternary deposits of SE Po Plain. Sedimentary Geology, 151, 273-292. ARTIOLI, G., BAGNASCO GIANNI, G., BRUNI, S., CARIATI, F., FERMO, P., MORIN, S. & RUSSO, U. 2000. Studio spettroscopico della tecnologia di cottura di ceramiche etrusche dagli scavi di Tarquinia. In: Atti del I Congresso di Archeometria. Patron Editore, Bologna, 335-349. BIANCHINI,G., LAVIANO,R., LOVO, S. & VACCARO,C. 2002a. Chemical-mineralogical characterization of clay sediments around Ferrara (Italy): a tool for an environmental analysis. Applied Clay Science, 21, 165-176. BIANCHINI, G., MARTUCCI, A. &VACCARO,C. 2002b. Petro-archaeometric characterization of 'cotto ferrarese': bricks and terracotta elements from historic buildings of Ferrara. Periodico di Mineralogia, 71, 101-111. BLUM, J. D. & EREL, Y. 1997. Rb-Sr isotope systematics of a granitic soil chronosequence: the importance of biotite weathering. Geochimica et Cosmochimica Acta, 61, 3193-3204. BONDESAN, M., FERRI, R. & STEEANI, M. 1995. Rapporti fra lo sviluppo urbano di Ferrara e l'evoluzione idrografica, sedimentaria e geomorfologica del territorio in Ferrara nel Medioevo. In: Topografia storica e archeologia urbana. Casalecchio di Reno, Bologna, 27-42. BRINDLEY, G. & LEMAITRE, J. 1987. Thermal, oxidation and reduction reactions of clay minerals. In: NEWMAN, A. C. D. (ed.) Chemistry of Clays and Clay Minerals. Mineralogical Society, London, Monograph, 6, 319-370. CAPEL, J., HUERTAS, F. & LINARES, J. 1985. High temperature reactions and use of Bronze Age pottery from La Mancha, Central Spain. Mineralogica et Petrographica Acta, 29-A, 563-575. CHAPMAN, V. J. & CHAPMAN, D. S. 1980. Seaweeds and their Uses. Chapman and Hall, London. CULTRONE, G., SEBASTIAN-PARDO,E., CAZALLA, O., RODRIGUEZ-NAVARRO, C. & DE LA TORRE, M. J. 2000. Mineralogical changes during brick production in laboratory experiments. In: QuarryLaboratory-Monument International Congress, Pavia 2000, Proceedings Volume 1, 253-258. DUMINUCO, P., MESSIGA, B. & RICCARDI, M.P. 1998. Firing processes of natural clays. Some microtextures and related phase compositions. Thermochimica acta, 321, 185-190. ELERT, K., COLTRONE, G., NAVARRO,C. R. & PARDO, E. S. 2003. Durability of bricks used in the conservation of historic buildings--influence of

composition and microstructure. Journal of Cultural Heritage, 4(2) 91-99. FERRI, R. & GIOVANNINI, A. 2000. Analisi dello sviluppo urbanistico della citth di Ferrara nel quadro dell'evoluzione geomorfologica del territorio circostante. In: GALLINA, M. (ed.) Dal Suburbium al Faubourg: evoluzione di una realt~ urbana. ET, Milan, 9-24. LEAKE, B. E., WOOLLEY, A. R., ARPS, C. E. S., et al. 1997. Nomenclature of amphiboles: Report of the Subcommittee on Amphiboles of the International Mineralogical Association, Commission on New Minerals and Mineral Names. Mineralogical Magazine, 61, 295-321. LENTZ, D. R., WALKER, J. A. & ST1RL1NG, J. A. R. 1995. Millstream Cu-Fe skarn deposit: an example of a Cu-bearing magnetite-rich skarn system in northern New Brunswick. Exploration and Mining Geology, 4, 15- 31. LOPEZ-ARCE, P. 8£ GARCIA-GUINEA,J. 2005. Weathering traces in ancient bricks from historic buildings. Building and Environment, 40, 929-941. MAGGETTI, M. t~ WON DER CRONE, M. 2004. Mineral reactions in synthetic clay NaCl system. Abstracts from the 32nd International Geological Congress, Florence, 2004; Session Tl6.01--GeoarcheometO': geomaterials in cultural heritage. MAGGETTI, M., WESTLEY, H. & OLIN, J. S. 1984. Provenance and technical studies of Mexican majolica using elemental and phase analysis. In: LAMBERT, J. B. (ed.) Archaeological Chemistry III. American Chemical Society, Advances in Chemistry Series, 205, 151-191. MARCHESINI, L., AMOROSI, A., CIBIN, U., ZUFFA, G., SPADAFORA, E. & PRETI, D. 2000. Sand composition and sedimentary evolution of a Late Quaternary depositional sequence, Northwestern Adriatic coast, Italy. Journal of Sedimentary Research, 70, 829-838. MARTJNEZ-SERRANO, R. G. 2002. Chemical variations in hydrothermal minerals of the Los Humeros geothermal system, Mexico. Geothermics, 31, 579-612. MOOR, J. N. & GUNDERSON, R. P. 1995. Fluid inclusion and isotopic systematics of an evolving magmatic-hydrothermal system. Geochimica et Cosmochimica Acta, 59, 3887-3907. MURPHY, S. F., BRANTLEY, S. L., BLUM, A. E., WHITE, A. F. & DONe, H. 1998. Chemical weathering in a tropical watershed, Luquillo Mountains, Puerto Rico: II. Rate and mechanism of biotite weathering. Geochimica et Cosmochimica Acta, 62, 227-243. RICCARD1, M. P., MESSIGA, B. & DUMINUCO,P. 1999. An approach to the dynamics of clay firing. Applied Clay Science, 15, 393-409. SINGOYI, B. & ZAW, K. 2001. A petrological and fluid inclusion study of magnetite-scheelite skarn mineralization at Kara, Northwestern Tasmania: implications for ore genesis. Chemical Geology, 173, 239-253. STIAFFINI, D. 1999. 1l vetro nel Medioevo. Fratelli Palombi, Rome.

Golden mica cooking pottery from Giikeyiip (Manisa), Turkey M f J M T A Z (~OLAK 1, M A R I N O M A G G E T T I 2 & G I U L I O G A L E T T I 3

1Dokuz Eyliil University, Department of Geological Engineering, 35100 Bornova, Izmir, Turkey (e-mail: mumtaz, colak @deu. edu. tr) 2University of Fribourg, Department of Geosciences, Mineralogy and Petrography, Ch. du Mus~e 6, CH-1700 Fribourg, Switzerland Abstract: Grkeytip cooking pottery is a particular type of pottery produced according to ancient craft tradition in western Turkey. It is made by mixing 75 wt% of local red and green smectitic clays with 25 wt% of local gneissic temper. Both temper and tempered objects are rich in MgO, as can be seen from XRF analyses. The vessels are coated with a sheet-silicate enriched layer, corresponding to the <2 mm sieved fraction of the crushed gneissic temper. The pottery is fired for 45 min using the bonfire technique. Apart from the dehydroxylation of the smectites, no clear mineralogical difference can be observed between the unfired and fired products. The reduction factor FeO/FeOtot reveals no significant oxidizing or reducing firing conditions. As evidenced by SEM-EDS analyses, there is no chemical difference between the unfired and externally fired micas of the coating. The golden colouring is therefore due to the oxidation of the biotites during firing.

A type of traditional cooking pottery is made in the small village of G6keyiip, located c. 50 km NE of Salihli (Manisa) in western Turkey (Fig. l a). This unique ceramic possesses many features of a cooking pot, i.e. a round shape (Fig. 2a), a rather open texture, and a fabric with many large inclusions (see Fig. 6c and d, below). Grkeytip cooking vessels are sold in different parts of Turkey under various local names, such as Menemen pottery. Previous studies by Giiner (1988) concentrated on the formal classification of the Anatolian pottery and the recording of shaping techniques, surface treatment and firing methods. Despite its low price, this kind of pottery is becoming increasingly outdated. The manufacturing tradition will therefore be lost very soon, as the youth of the village does not appear to be interested in maintaining their ancestors' skills. This partially ethnographic study focuses (1) on the mineralogical and chemical characterization of the raw materials and products and (2) on the cause of the colour change of the sheet silicates.

Geology The village of Grkeyiip is situated on basement rocks of the Palaeozoic Menderes Massif

(Fig. lb). The augengneisses, banded gneisses and biotite gneisses are covered by Neogene sediments and alternating volcanic rocks, as well as alluvium. The SE part of the study area is characterized by Quaternary basaltic lavas of the last stage of the important Kula volcanism. The related volcanic ashes contain human footprints (c. 12ka, Tekkaya 1976). The Neogene sedimentary rocks to the NE of the Grkeytip village can be divided into a fluviatile and a lacustrine unit (Ayan 1973). The base of the fluviatile unit is characterized by gneiss and schist boulders, embedded in an abundant clayey and sandy matrix. The upper parts have rounded to angular gneiss and schist pebbles in a silty to clayey matrix. The pebbles are commonly coated with a film of dark green clay (Yflmaz 1979). The lacustrine unit, including mudstone and cherty limestone, overlies the fluviatile unit concordantly. Palaeochannels are very common in the fluviatile unit to the NW of Grkeytip village (Fig. lb) with rounded quartzite and gneiss pebbles (diameter 4 m m - 1 5 cm). The top 1-2 m of basement rock are strongly weathered and show a random mixture of altered gneiss, from single grains to blocks, and rounded quartz pebbles of centimetre dimensions, embedded in a red or green clayey

& MESSIGA,B. (eds) 2006. Geomaterialsin CulturalHeritage. Geological Society, London, Special Publications, 257, 141-150. 0305-8719/06/$15.00 © The Geological Society of London 2006. From: MAGGETTI,M.

142

M. (~OLAK ET AL.

(a) , 50 ,

¢ (b)

.~;'~.~ Ko~;~kh~

~ , , ~ . ~ , , ~ . ~ , :" z " , ' -

/,,.J/t:~ll

.......

>~.

,.la~ i

r - ,1,- t ~ t , - t

- - I

liar ,~.0.(;).,',,,," ~

- ;, ~ -,t-//~"

. . . . . .

/

~

', ~.~".~ ~ / ) '

I

~£

~"

?'@~ >""1"">':

.;~

:..,.//.~,,/.,

.

! .....~..~..

.../.~;z

"~"

"

~" -

"~;¢~-..-,,,.~o°~"

II

.........

Banded

Y

.

/'~"

~/'L,'~ .....

~.'.,,, 4.'.,,,.~ ~ ~ " - . ~ ' . ' 4 ~ - - ~

!~,~S.~;~,~ ~""~ /i;~;"~,,".......

xt<',,.ys;y z;'~',',,.(,%~'..j;~

: ..... ;~;x,:

"~2~.", ";~.','(@~

:'7;,'~.~",.;~,

~ne~ss

'~

"-

t--

..

1

"~K." ~ ; ~.'.

Fig. 1. (a) Geographicaland (b) geological location.

matrix (Fig. 3). The clay passes laterally and vertically into sand and to gravel. Pottery production

Production and selling of the G6keytip pottery is strictly gender-related. Men provide the raw materials and sell the final products, whereas women are the actual potters. A detailed discussion of the pottery production has been given by Gtiner (1988). The raw materials, i.e. the gneisses and the clays (Fig. 3), are extracted separately from the weathered basement during the summer, using only

shovels, and subsequently stored in nearly every house of the village. The gneisses are crushed with hammers and/or by running over with a tractor, to obtain small temper grains. By sieving, a mica-rich gneiss fraction < 2 mm is obtained, which is used as a surface coating. The women first prepare a paste by mixing manually about 75 wt% of clay and 25 wt% of gneiss. Next, shaping and forming of the vessels occur on a hand wheel, placed in a hole in the ground. The plasticity of the paste is continuously controlled visually and corrected by adding crushed gneiss or water. The

GOLDEN MICA COOKING POTTERY, GOKEY(SP

"! •

. ~ , ~

143

good fire, wind must come from one direction only. The whole firing process takes c. 45 min. The hot pieces are taken from the fire and placed upside down on straw. In a last stage, dried and crushed goat manure is scattered over the still hot objects. The mica coating then turns golden (Fig. 2b) and the bottom as well as the core of the pottery become dark grey. Customers prefer dark coloured pottery, because they believe that only dark grey vessels have been fired to a satisfactory quality standard.

'.~ ~,~ ~.~-~

(b) Materials and methods

•

,

•

Fig. 2. (a) Firing of the Grkeytip pottery; (b) surface of the G6keyiip pottery after firing, with the characteristic golden mica (sample TU 5).

hand-formed pottery is then dried in the sun and later covered with the coating material. Firing is by a simple bonfire technique using brushwood, either in the garden or the field. To obtain a

Five clays from two sites (red clays from the northern and green clays from the southern part of the fluviatile unit; Fig. lb), two gneisses, a mica coating, a ready-to-use paste, five unfired and four fired objects were collected (Table 1). Thin sections were made for the gneisses and the fired or unfired objects and studied under a polarizing microscope. Powders for the mineralogical and chemical analyses were obtained by grinding in a tungsten carbide mortar. Mineral analyses were performed by powder X-ray diffraction (XRD) using CuK~ radiation (30 kV, 40 mA) on a Philips PW 1800 at the Department of Geosciences of the University of Fribourg (Switzerland). The < 2 la,m size fraction was separated and oriented clay aggregate samples were prepared after centrifuging (Gibbs 1965, 1968). A semi-quantitative analysis of the < 2 FLm size fraction of the clays was obtained by multiplying the intensities of the basal reflections of each

~ ~

Clay lenses uartzpebbles

.~, Altered gneissblocks ~

~

~

~

~

~

~

~

~

4

~

~

~

~

~

~

~

~

~

4

~

Clayeysand

~:l Gneissbasement

0 Fig. 3. Schematic generalized section of the clay deposits.

5m

M. (~OLAK ET AL.

144 Table 1. Sample description Analysis no.

Provenance

Description

Fired vessel Clay Clay Clay Clay Clay Gneiss

Grkeyiip Grkeyfip Adem Ablak field Galip Yilmaz f i e l d Ayse Ozeren workshop Ayse Ozeren workshop Adem Ablak pit

Tu 13

Gneiss

Ayse Ozeren pit

Tu Tu Tu Tu Tu Tu Tu Tu Tu Tu

Mica fraction Paste Unfired vessel Unfired vessel Unfired vessel Unfired vessel Unfired vessel Fired vessel Fired vessel Fired vessel

Tahibe Giiler workshop Tahibe Gtiler workshop Tahibe Giiler workshop Tahibe Giiler workshop Ayse Ozeren workshop Ayse Ozeren workshop Yilmaz Alpay workshop Tahibe Giiler workshop Tahibe Gtiler workshop Ayse Ozeren workshop

Plate Green 86925/08200, green 87045/08250, green Red Green Altered gneiss from a fractured zone, 85375/11750 Altered gneiss from a fractured zone, 86150/10175 Sieved to <2 mm Ready mix Without coating Without coating Without coating Without coating Without coating With coating With coating With coating

Tu Tu Tu Tu Tu Tu Tu

5 6 8 9 10 11 12

14 15 16 17 18 19 20 21 22 23

Type

clay phase by suitable factors according to unpublished internal reports of the Institute of Mineralogy and Petrography of the University of Bern (Switzerland). Reproducibility and error are _ 5% and c. 20%, respectively. Chemical analyses were performed using a Philips PW2400 X-ray spectrometer with a Cr anode at the Department of Geosciences of the University of Fribourg (Switzerland). FeO was determined following the Dipyridilic method (Lange & Vejdelek 1980) and using a Philips Pye Unicam PU8650 spectrophotometer. Based on reproducibility tests, the average error of the FeO analysis is +0.02 wt%. Electron microscopical analyses were carried out on polished thin sections of gneisses, and on unfired and fired vessels with a FEI XL30 Sirion FE6 scanning electron microscope (SEM) at the Department of Geosciences of the University of Fribourg (Switzerland). Quantitative analyses of sheet silicates were performed by energy-dispersive spectrometry (EDS) and standardizing with a biotite standard (ASTIMEX Scientific Limited, MINM25-35, Special Mineral Standard Mount, Serial No. 04-034). The measured surface was 5 ~m × 7 ~m.

Results

Optical microscopy The biotite gneisses are fine- to medium-grained (diameter 0.5-3 mm) and possess a heterogranular texture. They contain, in order of

abundance, plagioclase, quartz, muscovite, biotite, K-feldspar and garnet, and, in some instances, kyanite. Principal accessory minerals are ilmenite, magnetite, zircon, pyrite and apatite. Plagioclase has an average composition of oligoclase (An30) and is partly to completely altered to illite and chlorite (Ydmaz 1979). Quartz shows anhedral forms and cataclastic textures. K-feldspar is less altered to illite than plagioclase. Biotite is highly altered to chlorite or has lost its original colour as a result of iron leaching. All these effects are due to retrograde processes in the Menderes Massif. In the fired objects quartz, K-feldspar, biotite, chlorite and kyanite could be detected. The grain size of the temper (quartz, K-feldspar, kyanite) varies between 0.5 and 3 mm and the biotite temper is 4 mm long and 0.1 mm across.

X-ray diffraction Clay. The green and red clays have the same mineralogical composition, i.e. smectite, illite, chlorite and/or kaolinite, chlorite-smectite mixed layer as the main clay minerals, and quartz, plagioclase, K-feldspar, + calcite, _+magnetite, ___hematite, _+anatase, + serpentine, _+rutile as the other components. In the red clays, the major clay mineral is a chlorite-smectite mixed layer, with a very sharp d(001) reflection at 14.4 A. Semi-quantitative analysis of the clay fraction of such a smectite-rich clay

Fe203tot ,

0.5 0.09

389 121 29 17 73 21 130 115 101 49 81 186

total iron given as

wt% FeO FeO/FeOtot

Ba Cr Cu Nb Ni Pb Rb Sr V Y Zn Zr

Fe203.

0.58 0.11

368 97 41 22 69 34 167 177 75 39 84 176

0.37 0.03

374 226 70 11 119 7 29 149 169 48 93 185

LOI, loss on ignition.

0.59 0.08

302 150 37 14 89 33 112 137 92 41 84 204 0.55 0.09

387 100 23 17 50 23 138 113 89 44 75 259 4.52 0.41

389 254 <2 22 126 <7 197 52 78 44 57 399

54.32 0.95 16.29 7.08 0.06 11.88 0.81 1.28 4.08 0.45 2.74 99.94

3.7 0.46

90 21 <2 <5 30 <7 90 9 81 18 36 125 5.51 0.44

414 105 4 26 46 <7 153 42 132 34 52 433 1.96 0.23

414 104 41 17 55 17 147 86 110 43 63 249

46.06 45.00 62.19 0.27 1.33 1.03 26.27 19.29 15.51 4.59 7.55 7.05 0.03 0.05 0.04 13.18 15.30 4.64 0.62 0.71 1.35 1.32 1.83 1.25 1.85 3.55 2.93 0.41 0.40 0.20 5.05 5.02 4.21 99.64 100 100.4

TU 10 TU 11 TU 12 TU 13 TU 14 T U 1 5 Clay Clay Gneiss Gneiss Mica Paste

64.83 61.89 64.52 53.69 64.35 0.70 0.95 0.71 1.69 0.91 15.88 15.17 15.89 16.87 15.57 5.24 6.65 4.78 10.94 5.75 0.05 0.07 0.05 0.09 0.02 2.29 2.38 2.46 3.29 2.07 1.51 3.97 2.63 4.73 2.20 1.35 1.19 1.84 0.95 1.52 1.97 1.40 2.08 0.48 2.42 0.03 0.08 0.53 0.13 0.97 6.10 6.39 4.53 7.66 4.42 99.95 100.14 100.02 100.52 100.2

TU9 Clay

wt% SiO2 TiO2 A1203 Fe203tot MnO MgO CaO Na20 K20 P205 LOI Sum ppm

TU8 Clay

TU6 Clay

Analysis no.: Type:

Table 2. Chemical analyses

1.58 0.20

291 116 25 16 59 11 69 97 113 40 56 279

57.12 1.14 16.89 6.83 0.04 5.77 2.15 2.60 1.38 0.36 5.46 99.74

1.75 0.19

303 151 32 18 81 10 64 102 132 44 65 296

54.10 1.35 17.73 7.93 0.07 6.48 2.55 2.11 1.27 0.27 6.07 99.93

1.59 0.20

422 101 27 17 57 15 143 89 112 44 64 263

62.18 1.06 15.48 6.80 0.04 4.27 1.21 1.27 2.83 0.20 4.12 99.46

1.6 0.20

284 111 24 17 61 13 71 99 115 41 57 292

56.53 1.15 17.29 6.94 0.04 5.86 2.18 2.56 1.40 0.37 5.35 99.65

2.76 0.29

397 90 17 19 38 50 113 70 92 47 57 273

60.60 0.83 17.54 7.37 0.04 3.75 0.89 2.34 2.69 0.25 4.01 100.3

2.44 0.27

434 99 23 16 52 14 137 99 115 49 59 281

62.18 1.10 16.54 7.32 0.04 5.16 1.06 1.36 2.85 0.21 1.70 99.52

T U 1 6 TU 17 TU 18 TU 19 T U 2 0 TU21 Unfired Unfired Unfired Unfired Unfired Fired

2.7 0.28

447 102 25 17 52 12 136 105 112 57 60 266

61.76 1.09 16.70 7.46 0.04 5.16 1.12 1.25 2.89 0.17 2.07 99.7

TU22 Fired

1.92 0.24

322 91 37 18 41 20 103 91 108 36 55 293

59.99 0.92 17.85 6.58 0.03 6.03 1.20 1.95 2.11 0.27 2.82 99.76

TU23 Fired

1.53 0.21

475 248 30 20 91 16 146 89 113 41 69 228

60.82 0.79 17.04 6.23 0.05 5.31 1.27 1.27 2.82 0.23 4.16 99.99

TU5 Fired

,-< ~"

©:

. I

©

2: Cl

©

>-

© t" t::7

M. ~OLAK ETAL

146 (a) 7o

Whole-rock chemistry Clays. The red clay can be distinguished •

from the green clays by, for example, its higher TiO2 and Fe203tot and lower SiO2 values (Table 2). Both clay types have low MgO as well as reduction factors FeO/FeOtot (Fig. 4).

EEl []

60

•

B TU 10 •

TU 12 +

D

50 • clay 4- gneiss Zk mica O paste [] unfirec

Gneisses and mica coating. Both are characterized by very high MgO and a high reduction factor (Fig. 4). Gneiss TU 13 stands out because of its very high Al203.

TU 13 +

40 • fired 0

Paste, unfired and fired objects.

In bivariate plots (Fig. 4a and b), the three types group very closely. As is to be expected, the loss on ignition (LOI) of the fired objects is lower than that of the unfired ones. The reduction factor is higher compared with that of the clays.

4

MgO(wt%) (b)

8 7 []

6

Scanning electron microscopy (SEM)

~5

A+

00

0

~-] 4

jm©

[]

• clay 3 4- gneiss I

A mica I1 O paste 2 [] unfired] • fired { I 0 .1 !

2

3

.4

.5

FeO/FeOto t Fig. 4. (a) Si02-MgO binary plot; (b) LOI-reduction factor binary plot.

(TU 6) shows 58 wt% smectite, 36 wt% illite, 4wt% kaolinite and 2wt% chlorite. These amounts vary from clay to clay.

Mica coating.

This consists of mica, Mg-chlorire, quartz and K-feldspar. Mica and chlorite minerals predominate over the quartz. Feldspars are present in low quantities.

Unfired paste. This is mostly quartz, mica (biotite, illite), Mg-chlorite, smectite, plagioclase, K-feldspar, Mg-chlorite-smectite mixed layer, +_anatase. Mg-chlorite and mica are the dominant minerals in most of the mixed paste. Fired objects.

These are characterized by quartz, mica, Mg-chlorite, plagioclase, K-feldspar, _ anatase.

The analysed sheet silicates of a gneiss sample (TU 12), an unfired object (TU 20) and a fired object (TU 23) are biotites (totals around 95 wt% in Table 3, e.g. TU 12-1), Mg-chlorites (sums close to 85 wt%, e.g. TU 20-10), kaolinites (TU 20-5, 6 in Table 3), smectites (e.g. TU 12-3, 4) and all degradation states between biotites, chlorites and smectites. The discrimination of 2:1, 2:1:1 and 1:1 phases was carried out on the basis of SEM-EDS analyses, plotted in a ternary MR3-2R3-3R2 diagram according to Velde (1985; see Fig. 5). This diagram shows that the gneiss sample TU 12 has chlorites (analyses 13, 14, 15 and B in Table 3), muscovites (analyses 5, 6 and 7), montmorillonites (analyses 3 and 4) and kaolinites (analysis 9). Many analyses plot in the field between biotites and montmorillonites. Sheet silicates of the unfired sample TU 20 classify as kaolinites (analysis 3, 5 and 6) and chlorites (analysis 10), but also lie in the field between biotite and montmorillonite. The sheet silicates of the fired sample TU 23 show, as for the other samples, a large variation on the MR3-2R3-3R2 ternary diagram: chlorites (analyses 7A, 7B, 7C, 16 and 19), montmorillonites (analyses 6A, 6B, 6C and 1 5 ) , biotites to montmorillonites (analyses 13, 14 and 18) and I/S mixed layer (analyses 10, 11 and 12). As evidenced by Table 3 and Figure 6a and b, the transformation process proceeds by leaching of iron, magnesium and potassium from the initial biotite. There are no major differences between the sheet silicates of the unfired TU 20 and with those in the core (TU 23-10, 11 and 12) or in the mica coating (TU 23-6, 7 and 13-19) of a fired object.

1

38.76 1.68 15.11 19.48 0.10 0.03 10.62 9.91 95.70

1

38.33 1.45 19.36 13.71 0.03 0.04 15.74 9.36 98.02

6A

32.62 0.82 15.02 15.49 0.10 0.04 5.25 4.46 73.78

TU12

SiO2 TiO2 A1203 MgO CaO MnO FeO K20 Total

TU20

SiO2 TiO2 A1203 MgO CaO MnO FeO K20 Total

TU23

SiO2 TiO2 A1203 MgO CaO MnO FeO K20 Total

33.11 0.75 15.02 15.37 0.01 0.03 5.65 5.37 75.31

6B

37.52 1.35 18.46 14.29 0.04 0.04 14.39 8.76 94.85

2

38.76 1.78 15.13 19.50 0.10 0.04 10.78 9.96 96.05

2

33.11 0.72 15.57 16.12 0.04 0.00 5.35 5.35 76.25

6C

46.29 0.20 38.25 0.78 0.06 0.00 0.89 1.51 87.97

3

31.91 0.67 14.02 17.33 0.08 0.01 4.81 3.60 72.43

3

21.58 0.12 18.34 20.01 0.01 0.01 8.98 0.06 69.12

7A

37.33 1.25 18.32 14.19 0.03 0.00 13.75 9.02 93.89

4

32.15 0.62 14.15 17.57 0.07 0.01 5.03 3.70 73.30

4

21.52 0.05 18.49 19.60 0.00 0.03 8.81 0.04 68.53

7B

44.13 0.18 36.50 0.93 0.06 0.00 1.03 1.39 84.21

5

37.54 0.70 27.11 2.04 0.01 0.01 1.04 7.71 76.16

5

21.73 0.05 18.29 19.13 0.00 0.00 8.74 0.23 68.17

7C

44.34 0.23 36.25 0.73 0.06 0.00 0.82 1.39 83.82

6

37.8 0.65 26.97 1.99 0.01 0.01 0.98 7.91 76.33

6

36.88 2.72 18.66 9.70 0.04 0.03 18.39 7.54 93.96

10

36.45 1.52 17.64 13.11 0.06 0.01 13.90 8.16 90.85

7

38.52 1.00 27.28 2.07 0.03 0.01 1.00 7.58 77.49

7

36.81 2.35 18.30 10.26 0.04 0.01 18.08 8.08 93.95

11

34.84 1.45 17.17 13.48 0.01 0.03 11.92 8.59 87.50

8

30.93 0.63 14.45 15.62 0.04 0.00 5.48 5.23 72.38

8

36.83 2.09 18.57 10.40 0.03 0.01 17.84 8.19 93.95

12

36.00 1.57 17.61 13.61 0.04 0.01 14.72 8.12 91.68

9

35.04 0.20 29.22 0.71 0.03 0.03 0.53 1.24 66.99

9

40.26 0.75 18.06 20.49 0.01 0.03 6.21 8.76 94.57

13

26.82 0.03 22.16 27.80 0.01 0.01 6.76 0.07 83.68

10

30.03 1.07 15.07 13.50 0.00 0.00 6.93 6.73 73.33

A

40.13 0.83 18.12 20.39 0.01 0.00 6.31 8.70 94.50

14

20.83 0.07 18.30 19.25 0.01 0.03 9.13 0.05 67.67

B

42.91 0.72 18.98 11.95 0.24 0.01 7.99 4.67 87.47

15

38.91 0.97 18.70 18.20 0.01 0.01 8.28 8.84 93.93

10

Table 3. EDS analyses of sheet silicates for gneiss (TU 12), unfired pottery (TU 20) and fired and coated pottery (TU 23) 11

27.34 0.08 22.04 26.49 0.03 0.03 7.91 0.11 84.03

16

38.20 0.97 18.19 18.01 0.01 0.00 7.90 8.34 91.61

12

38.63 0.63 17.61 19.40 0.07 0.03 6.40 7.96 90.73

18

30.42 0.70 14.70 14.04 0.00 0.03 6.07 6.59 72.54

13

40.34 0.72 17.70 20.84 0.27 0.04 5.49 5.16 90.55

19

26.50 0.03 21.76 26.13 0.04 0.04 7.33 0.06 81.90

14 27.51 0.10 22.35 25.52 0.03 0.03 6.74 0.35 82.61

15 29.75 0.30 19.36 23.86 0.06 0.04 6.97 1.84 82.18

©:

©

© ©

),

m

© t"

148

M. (~OLAK E T A L . MR3

40%

20%

2R3

40%

Rif~tite~0%

60%

8~'o

3R2

Fig. 5. EDS analyses plotted on an MR3-2R3-3R2 diagram (Velde 1985), where MR3 is (Na + + K + + 2Ca2+), 2R3 is (AI 3+ + Fe 3+ - MR3)/2 and 3R2 is (Mg 2+ + Mn2+)/3; 22 oxygen atoms (O, TU 12 gneiss; 0 , TU 20 unfired; C], TU 23 fired).

(a)

Discussion The chemical analysis of clays and temper shows that the final products lie, as expected, on straight mixing lines in bivariate plots (Fig. 4a). The firing process did not significantly affect the initial mineralogical composition. Only the smectites of the clay fraction show a dehydration reaction in response to increasing temperature, but no new firing minerals could be observed by XRD (Fig. 7). Astonishingly, in the coarse-grained fraction of the mica coating, no clear thermal effect on one single crystal could be detected, as can be seen from analyses TU 23-6A to C and TU 23-7A to C (Fig. 6f). The short firing span is likely to explain this obvious absence of any dehydration, even of the outermost smectitic sheet silicates. Regarding the firing atmosphere, it is tempting to characterize it, overall, as more reducing than oxidizing. In fact, tempering of clays with a rock possessing a higher reduction factor will

(b)

11

12

B A

10

50 m i c r o n (c)

(d)

Fig. 6. SEM backscattered electron photomicrographs of polished cross-sections. (a) Gneiss sample (TU 12). (b) Gneiss sample (TU 12). (c) Matrix of the unfired sample TU 20 (bar represents 200 p.m). (d) Matrix of the fired sample TU 23 (bar represents I00 p.m). (e) Border of the fired and mica-coated sample TU 23 (bar represents 200 ~m). (f) Fired mica coat of sample TU 23. Numbers and letters correspond to analyses in Table 3. E, epidote; P, plagioclase; Q, quartz; R, rutile.

149

GOLDEN MICA COOKING POTTERY, GOKEYUP

(1)

(e)

C

B

:, ,

~

Fig. 6. Continued.

reducing conditions. It can therefore be inferred that the whole cycle did not significantly modify the initial FeO/FeOtot ratios. The colour change of the sheet silicates in the outermost mica coating from brown to golden is therefore

result in intermediate reduction factors for the unfired vessels, as is the case for the studied material. Interestingly, the fired objects do not have higher reduction values, as would be expected of a firing cycle under predominantly

(a)

tx~ata/s ..................................................................................................................................................................................

'~

Q

Ch

i

3soo~

TU16

Q

i

3000.-!

el

Ch

2500-i

i

Ol

2o~i

i

lsoo-

Ch

~

,=o :=

__~: ~-~= i Oi

A

j

i~i

Q

7=L

: '=-l, Jil ,!

Q

=dl)

:

,

T

=

10

20

30

40

c,o m l t a / ~ t

Q

L 60O2Thet a

50

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

(b) ~ooo]

o

t

i

i 5ooo-i

J

TU22

Ch'i 4ooo.i 3000- i

i

IPl

Ch

li"

Ch

,oooi I

lo

20

I

t

iI

M

I~

M

Q

t!l !~,orQ A /~, ;, f , k.~.~,~kG"~'~j ~u~,j~,,~__>]~..~Jk~.~

30

4o

5o

6o °2Theta

Fig. 7. X-ray diffractograms of (a) an unfired (TU 16) and (b) fired object (TU 22). A, anatase; Ch, chlorite; Ch/Sm, chlorite-smectite mixed layers; Kf, K-feldspar; M, mica; P1, plagioclase; Q, quartz.

150

M. (~OLAK ET AL.

due to an oxidation process. As shown by the experiments of Hogg & Meads (1975), the oxidation of ferrous to ferric iron in biotites proceeds with a colour change from brownish below 350 °C, to light brown at 400 °C and to orange brown at 900 °C.

Conclusion The authors conclude that the golden colouring observed is caused by the oxidation of the biotites during firing. The authors are grateful to the Grkeyiip potters, especially to A. Ablak. We thank O. Marbacher for invaluable help in the whole-rock chemical analyses, C., Neururer for assistance in the SEM analyses, and J.-P. Bourqui for the careful making of the thin sections.

References AYAN, M.

1973. Salihli-Krpriibasl qevresindeki uranyum zuhurlan olusumu ve prospeksiyonu. Prospektrr, 2, 37-52.

GIBBS, R. S. 1965. Error due to segregation in quantitative clay mineral X-ray diffraction mounting techniques. American Mineralogist, 50, 741-751. GIBBS, R. S., 1968. Clay mineral mounting techniques for X-ray diffraction analysis: a discussion. Journal of Sedimentary Petrology, 38, 242-244. GONER, G. 1988. Anadolu'da Yasamakta Olan ~rmlekgilik. Ak Yaymlan, Kiilt/Jr Serisi. HOGG, C. S. & MEADS,R. E. 1975. A Mrssbauer study of thermal decomposition of biotites. Mineralogical Magazine, 40, 79-88. LANGE, B. & VEJDELEK, Z. J. 1980. Photometrische Analyse. Chemie, Altenberg. TEKKAVA, I. 1976. Insanlara ait fosil ayakizleri. Yeryuvan ve lnsan, 1(2), 8-10. VELDE, B. 1985. Clay Minerals. A Physico-chemical Explanation of their Occurrence. Elsevier, Amsterdam. YILMAZ, H. 1979. Genesis of uranium deposits in Neogene sedimentary rocks, Menderes Metamorphic Massif Turkey. PhD thesis, University of Western Ontario, London, Ontario, Canada.

Chemical and mineralogical investigations of majolicas (16th-19th centuries) from Laterza, southern Italy C. D E L L ' A Q U I L A 1, R. L A V I A N O 2 & F. V U R R O 2

lDipartimento di Informatica, Universitgt degli Studi di Bari, Via E. Orabona 4, 70125 Bari, Italy 2Dipartimento Geomineralogico, Universith degli Studi di Bari, Via E. Orabona 4, 70125 Bari, Italy (e-mail: [email protected]) Abstract: Laterza (southern Italy) was the most important town for the manufacture of

Apulian majolica ceramic from the 16th century until the end of the 18th century. The Laterza majolicas have previously been subjected to only preliminary analyses. This study extends the archaeometric knowledge of the Laterza productions with mineralogical, petrographical and chemical characterizations of ceramic body, glazed coating and pigments of the majolica. A number of 16th to 19th century pottery and tile fragments of majolica have been studied and compared with clay from local and surrounding deposits. Analyses were carried out by optical microscope, scanning electron microscope, energy-dispersive spectrometry, X-ray powder diffraction, X-ray fluorescence and inductively coupled plasma. A purification process of the raw material is suggested and for some fragments, doubtfully attributed to Laterza, a different place of manufacture. Slip ('ingobbio') was never found under the glaze. Si, Pb and Sn are confirmed as the principal elements in the tin-glazed coatings. The differences in the glaze opacity were attributed to different manufacturing techniques and not simply the quantity of tin. The orange-yellow colour is due to a Sb-Pb compound; black to Ni with a lower amount of Co, Fe and Sb; blue to Co, As, Fe and Ni; and Mn is the pigment of the violet-brown.

Majolica is a glazed ceramic, distinguished by its soft earthenware paste covered by an opaque vitreous enamel of glaze. Modern Laterza (Taranto province, southern Italy) majolica production may have begun in the late 15th or in the early 16th century (dell'Aquila & dell'Aquila 1988). At present, however, no dated and/or signed majolica of that period, which can be certainly ascribed to the Laterza production, is known (dell'Aquila & dell'Aquila 1990). In contrast, the 17th-18th baroque majolica wares are well documented: many artistic faiences, especially in 'turchino' monochrome 'istoriato' style, are conserved in the main Italian and foreign museums as well as in private collections. Archive documents and craftsmens' signatures on plates and tiles allow these wares to be attributed to Laterza productions (Donatone 1980, plates 3, 25, 28 and 29; dell'Aquila & dell'Aquila 1988). From the end of the 18th century, quality and quantity progressively diminished and finally the majolica production ended in Laterza in the middle of the 19th century. This drastic negative trend in ceramic production could be explained by the competition from less expensive industrially produced

enamel metallic and glass wares. A decline caused by a significant variation in the technological requirements for the clay raw material, such as that proposed by Dondi (1999) for the tile production in the Sassuolo ceramic district in the 1960s to 1970s, is not a feasible proposition in the Laterza context. The Laterza majolicas are well known from an artistic point of view, but as regards their archaeometric study, only some isolated and non-systematic studies (dell'Aquila et al. 1994, 1995a, b) have so far been carried out. For this reason, the aim of this work is to delineate a coherent and general picture of majolicas made at Laterza from the 16th to the early 19th century AD. To cover the whole of this time span, the chemical data, already published (dell'Aquila et al. 1995b), for non-coated pottery belonging to the first quarter of the 19th century have been included in this paper also. In the context of Laterza majolica production, the ceramic body, the glazed coating and the pigments have been characterized. A comparison between the present results concerning the pigments used at Laterza and earlier results for other localities (Gratuze et al. 1996; Casadio

From: MAGGETTI,M. & MESSIGA, B. (eds) 2006. Geomaterials in Cultural Heritage.

Geological Society, London, Special Publications, 257, 151-162. 0305-8719106/$15.00 © The Geological Society of London 2006.

152

C. DELL'AQUILA ET AL.

et aL 1999; Padilla et aL 2005) has also been carried out.

Materials The majolicas studied have been subdivided into six groups according to the date of their production and occurrence (Fig. 1). These groups include pottery and tile samples. In addition, another two groups, consisting of some fragments of non-coated pottery, have been included here to give an additional contribution to the knowledge of the chemical composition of the Laterza majolica. Samples (a few square millimetres) have been selected and subjected to laboratory investigations.

Group A: polychrome majolica ware from a well close to the Chiesa de/Purgatorio Many fragments of majolica ware in 'compendiary' style were found in a well inside a house close to the Purgatorio church at Laterza. The material of the fragments and the decorative palettes indicate that they were mainly locally manufactured (as the kiln spacers found at the same site confirm) in the late 16th century or early 17th century. A set of five faience fragments has been analysed. In addition, another three majolica fragments, found in the same well, have been analysed to verify the possible local production of these not well-documented typologies: two samples (A.1 and A.2) are of

'majolica turchina' with blue glaze ground, whereas another sample (A.3) shows a monochrome blue 'medaglione a scaletta' decoration. Of the latter three samples, only sample A. 1 is shown in Group A of Figure 1.

Group B: turchino decorated majolica tile dated 1591 The tile (135 × 135 × 25 mm) comes from a tiled floor, now lost, of a monastery at Lecce (southern Italy). The decorative drawings of tiles show a single cell containing animals, birds or floral compositions inside a square frame. One of the tiles shows a cartouche with the date 1591. Initially, the tile had been attributed to Laterza production on the basis of its decorative style, as the monochrome turchino decoration is typical of the Laterza majolicas. The tile was examined to verify this uncertain attribution following the hypotheses of dell'Aquila & dell'Aquila (1990).

Group C: polychrome majolica signed 'd'Alessandro - Laterza 1678' The heraldic tile (430 × 400 x 10 mm) bears the coat of arms of Francesco Antonio Gallo, a Laterza born bishop of Bitonto, near Bari. The tile is signed and dated as follows: D. aNG(elu) S. ANt(oniu) S.II De aLexa(n)dro/ P(inxi)t/A Latertia 16718].

(e)

(c) :

10 cm

Fig. 1. Examples of the investigated pottery and tile fragments of Laterza majolica. The majolicas have been subdivided into six groups according to the date of their production and occurrence.

LATERZA MAJOLICA INVESTIGATION The restored tile was found in about 15 large fragments, alongside many others of a smaller size (dell'Aquila & dell'Aquila I988); nine of them have been used for the present study. The priest Angelo Antonio d'Alessandro (1641-1711) is considered to have been the major faience painter in Laterza (Donatone 1968; dell'Aquila & dell'Aquila 1980).

Group D: turchino monochrome decorated majolica ware The fragment, from a glazed plate decorated blue on white, belongs to a well-known Laterza majolica production, dated to the first quarter of the 18th century (dell'Aquila & dell'Aquila 1980, plate LXXXIIIa; Donatone 1980, plates 32a-c, 34c).

Group E: majolica-tiled floor of the Palagianello castle chapel The castle at Palagianello, near Taranto, was built in the 16th century and underwent various alterations up to the 18th century. The majolicatiled floor of the chapel dates back to the second quarter of the 18th century. The floor, produced at Laterza, is almost entirely lost and today only a few fragmentary tiles remain there. The decorative drawings of the tiles show a single cell consisting of central circular medallions with angular leaves that combine with the contiguous tiles. The medallions contain animals in a landscape, floral compositions and, most of all, human busts in blue, yellow and green. Four fragmented tiles have been analysed.

Group F: polychrome majolica ware from Cantina Spagnola at Laterza Many ceramic fragments from vessels and plates belonging to the first quarter of the 19th century come from the well of a cave-house which is commonly called the 'Spanish cellar' because of some wall-paintings, dated 1664, representing nobles and soldiers in Spanish costume (dell'Aquila 1998). Four fragments from different sherds have been analysed.

Group G: non-coated pottery from Cantina Spagnola at Laterza The fragments, belonging to the same occurrence as Group F, were used for the chemical analyses of the ceramic body. These 'biscotto' sherds had been subjected to the first firing only. These

153

samples have been chosen to avoid any possible contamination between the pottery body and the glazed coating in preparing the samples for the analyses. The chemical analyses of 16 fragments, already published (dell'Aquila et al. 1995b) as noted above, have been used in this work.

Group H: non-coated pottery from a well close to the Chiesa del Purgatorio The fragments, belonging to the late 16th century or the beginning of the 17th century, were found with the samples of Group A. Chemical analyses of eight fragments have been carried out. To provide information about the source of the clay raw material used by Laterza potters, chemical compositions, already published, of Pleistocene clays from the Lucanian area deposits (Dell' Anna & Laviano 1991) and Pleistocene clays from a quarry (Cava Galante) in the territory of Ginosa, a town near Laterza (dell'Aquila et al. 1995b), are also reported in this study. All these clay deposits belong to the 'Argille Subappennine' formation, which forms the top of the Plio-Pleistocene sedimentary sequence of the Lucanian basin. Pleistocene clays occur both in the Lucanian area and in the Apulian area, where the town of Laterza lies. All the sedimentary samples of the Argille Subappennine consist of clay minerals, carbonates, quartz and feldspars, small quantities of Fe-oxides (hematite, ilmenite and magnetite) and Fe-hydroxides (goethite), muscovite, biotite and gypsum. The clay minerals are a mixture of 2M illite, Mg-bearing smectite, Fe-bearing chlorite, kaolinite and randomly interstratified illite-smectite with 30-70% of montmorillonitelike layers. The carbonate consists of calcite, dolomite and rare aragonite; the feldspars are represented by orthoclase, microcline and Naplagioclase (Dell'Anna & Laviano 1991).

Experimental The ceramic body, the glazed coating and the pigments of the majolica samples were subjected to mineralogical, petrographical and chemical analyses. Raw materials from the Laterza area were also analysed. Morphological observations of the ceramic body were carried out using both a polarized light microscope on thin sections and a scanning electron microscope (SEM). Mineralogical studies of the ceramic body were carried out by X-ray powder diffraction (PXRD): a Philips powder diffractometer

C. DELL'AQUILA ET AL,

154

Table 1. Major element composition (by XRF) of the ceramic body of the majolicas and of the Argille Subappennine clays (Pleistocene clays) Group

SiO2

TiO2

A1203 Fe203

MnO

MgO

CaO

Na20

K20

P205

B (n = 1)

56.23

0.63

12.77

5.03

0.I1

2.53

18.87

1.17

2.30

0.36

51.31 0.48 53.51 1.20 53.36 0.59 51.20 0.83 53.31 0.44 54.9 1.9 54.8 0.7 55.2 2.2 52.3 0.7

0.74 0.03 0.74 0.03 0.76 0.02 0.74 0.01 0.89 0.03 0.8 0.1 0.8 0.0 0.8 0.1 0.7 0.1

14.75 0.08 15.07 0.62 15.34 0.52 14.93 0.21 19.55 0.45 16.5 0.8 16.6 0.3 16.4 0.8 13.9 1.2

6.96 0.30 6.73 0.30 6.99 0.20 6.50 0.31 8.70 0.52 6.9 0.3 6.6 0.3 6.6 0.7 5.0 1.1

0.11 0.01 0.11 0.01 0.10 0.01 0.11 0.01 0.08 0.01 n.d.

3.79 0.20 3.30 0.29 3.82 0.49 5.43 0.46 3.52 0.53 3.4 0.2 3.3 0.1 3.0 0.1 3.8 0.5

18.67 0.62 16.43 1.92 15.20 1.37 17.47 0.98 10.54 1.19 14.0 1.9 14.4 0.9 14.5 1.0 21.0 2.2

1.23 0.28 1.02 0.12 1.01 0.08 1.00 0.18 0.29 0.02 0.9 0.1 0.9 0.1 1.2 0.2 1.0 0.3

2.33 0.14 2.87 0.41 2.86 0.12 2.51 0.05 2.99 0.02 2.6 0.1 2.6 0.1 2.3 0.0 2.3 0.2

0.13 0.01 0.24 0.05 0.57 0.29 0.11 0.01 0.12 0.02 n.d.

E (n = 4) G (n = 16) H (n = 8) Clays* Galante Quarry (n = 8) Clays Galantet Quarry (n = 8) Clays$ Timmari deposit (n = 5) Clays:I: Miglionico deposit (n --- 4) Clays:l: Pomarico deposit (n = 8) Clays$ Montescaglioso deposit (n = 9)

M s M s M s M s M s M s M s M s M s

n.d. n.d. n.d.

n.d. n.d. n.d.

The values are given in wt%: M, meancomposition;s. standarddeviation:n, numberof samples used for the analysis; n.d., not detected. *Raw material; normalizeddata from dell'Aquila et al. (1995b). t Clay fraction (< 2 Ixm grain size). $ Raw material; normalizeddata from Dell'Anna & Laviano (1991 I, where original values were given with only one decimal place.

(PW 1710) with Ni-filtered Cu K~ radiation was used. P X R D patterns o f pottery samples, fired for 10 h at 1000 °C, were also r e c o r d e d to verify the o c c u r r e n c e o f further mineralogical changes. Major and trace elements o f the ceramic bodies o f majolicas (Groups B, E, G and H) and clays from a quarry near Laterza (Tables 1 and 2) were determined by X-ray fluorescence (XRF), on a Philips P W 1480/10 spectrometer (Cr anticathode for major and minor elements; Rh anticathode for Rb, Sr, Y, Zr and Nb; W anticathode for Ce, La,

Ba, Ni, Cr and V), following the analytical techniques outlined by Franzini et al. (1972, 1975) and Leoni & Saitta (1976). T w o reference standards (AGV-I o f U S G S - U S A , and N I M - G o f NIM, South Africa) were used to c h e c k the accuracy o f the analytical data. Water content was determined as loss on ignition (LOI) between 100 :C and 1000 :C. Analysis o f other majolica samples (Groups A, C, D and F) was not possible because not e n o u g h material was available.

Table 2. Trace element composition (by XRF) of the ceramic body of the majolicas and of Pleistocene clays from a quarry near Laterza Group

Ce

La

Ba

Ni

Cr

V

Rb

Sr

Y

Zr

Nb

B (n = 1)

47

29

284

114

209

108

87

161

1

67

6

68 2 63 8 67 8 55 3

34 1 34 3 36 3 31 2

238 13 236 16 258 20 208 11

87 4 80 7 82 7 104 14

141 5 121 12 133 9 137 11

121 3 103 9 99 7 108 4

i 17 5 120 20 123 13 111 2

367 16 313 42 349 43 351 55

20 2 18 3 20 5 20 2

139 5 125 17 132 20 120 6

6 2 11 2 12 5 12 0

E (n -- 4) G* (n = 16) H (n = 8) Clays* Galante Quarry (M = 8)

M s M s M s M s

The values are given in ppm: M, mean composition; s, standard deviation. n, numberof samples used for the analysis. *Data from dell'Aquila et al. (1995b). For the other Pleistoceneclay deposits reported in Table 1 no literature data are available.

LATERZA MAJOLICA INVESTIGATION

Table 3. Colouring element contents in wt% (by ICP) of blue pigment of Group A Sample A.1 A.2 A.3

As

Co

Cu

Fe

Ni

0.10 0.86 1.25

0.41 0.30 0.51

0.03 0.01 0.03

0.72 1.83 1.52

0.11 0.12 0.26

The major element compositions (qualitative investigations) of the tin-glazed coatings of the majolica were determined by SEM using a Cambridge Instruments apparatus (model $360) equipped with an Oxford-Link microanalyser (energy-dispersive spectrometry; EDS). Chemical analyses for the blue pigments of the turchino majolica ware (A.1 and A.2) and the blue decorated majolica ware (A.3) from Group A (Table 3) were performed at the ENEA Trisaia Laboratories by inductively coupled plasma (ICP), using a Thermojarrel ASH Atom Scan 25 emission spectrometer. This analysis was carried out on a few milligrams of the coloured glazed coating, scraped from the bulk ceramic, to verify the occurrence of Co in the blue decoration, which is not well detected by the S E M - E D S qualitative investigation.

Results and discussion The ceramic body Macroscopic observations of the ceramic body show for all the investigated samples a homogeneous and purified matrix, which is fine grained with small vacuoles and a few inclusions. Colours range from 0.3Y 7.1/3.2 to 4.2YR 5.6/5.7 (Munsell 1947). Thin sections (cut across the thickness of the sherds) show the presence of micas in the matrix of the ceramic body. Quartz, feldspars, calcite and poorly crystallized Fe-oxides or Fehydroxides have also been identified. As a rule, the mineral inclusions, identified in thin sections with a polarizing microscope, occur as small granules. SEM backscattered electron images and EDS analyses show that the ceramic body has a fine texture, with inclusions of quartz, K-feldspars and N a - C a plagioclases (Fig. 2). The results of PXRD analyses confirm the presence of abundant quantities of quartz and feldspars. Other minerals such as diopsidic pyroxenes, gehlenite and hematite have also been revealed. The occurrence of these minerals allows us to estimate firing temperatures ranging from 850 to 1050 °C (Heimann & Maggetti 1981). Small amounts of goethite and calcite

Palagianello 3

• ie,;

EHT = 15.0 KV 100gm ~

155 WD = 21turn =

R= 4QBSD Dip. Geomineralogico - Bad

~i+~%.~,

Fig. 2. Backscattered electron micrograph of a section of tile (sample E.3, Group E). Inclusions are silt size (<63 Ixm).

are also present in some samples. The small amounts of hematite suggest that the firing was carried out under an oxidizing atmosphere. Chemical analyses were carried out by XRF on the tile of Group B, on four sherds of the majolica-tiled floor from the Palagianello castle (Group E), on 16 sherds from Cantina Spagnola (Group G) and on eight sherds from the well close to the Purgatorio church (Group H). As noted above, analysis of other majolica samples (Groups A, C, D and F) was not possible because not enough material was available. Analytical data for major and trace components are reported in Tables 1 and 2, respectively. Because of the notable variation in LOI, the composition of major oxides is normalized to 100% without LOI. An inspection of Table 1 shows significant differences in composition of the ceramic body between Group B and the other groups. These differences are better shown by Figures 3 and 4, in which the SiO2 and CaO contents of all samples are respectively plotted v. MgO. The same consistent difference between the sample of Group B and those of Groups E, G and H is shown by Figure 5, in which Rb v. Ni is plotted. Figures 3 and 4 reveal also clear MgO differences between the sherds of Groups E, G and H and the clayey raw material from the quarry (Cava Galante) of Ginosa. Cava Galante was one of the quarries used by the later Laterza craftsmen of the first half of the 20th century; the quarry, long used by a brickmaking factory, is now exhausted. The differences in quantities of MgO may be due to the lack in the sherds of the sand fraction, owing to a purification process, as shown by the backscattered electron micrograph of a section of sample E.3 (Group E, Fig. 2).

156

C. DELL'AQUILA E T A L . 200 180 160 [ 140 120 ~ 11111 80

57 56

o o 'oO o

55

~ 53 ~ s2

cb c9 oe

08 O0

o,

51

*

0

o

20 0

,~

6O 2.5

o d~ o o o

O00

50 49

0

,

,

|

|

i

3.5

4.5

5.5

6.5

7.5

MgO (w1%)

Fig. 3. Bivariant plot of SiO2 v. MgO in the ceramic body of majolicas and the Pleistocene clays. ©, sherds of Groups E, G and H; II, sherd of Group B; :~, clays from quarry near Laterza (delrAquila et al. 1995b); l , clays from Lucanian Basin (DelrAnna & Laviano 1991). In fact, the chemical composition of the clay fraction ( < 2 ~m) of the Cava Galante samples, also quoted in Table 1, shows a decreased MgO with values close to those of the Laterza ceramic body. Therefore, if the raw material from Cava Galante quarry was used by Laterza potters, it had undergone a purification process. The hypothesis that the 16th-19th centuries Laterza craftsmen may not have used clays from the Ginosa quarry is also possible. Although the Laterza territory is abundant in clay deposits, unfortunately no historical references indicate the location of the quarries used in the past and no clear evidence is known for abandoned quarries. Nevertheless, the clay deposits of the Laterza area belong to the Argille Subappennine formation. Table 1 and Figures 3 and 4 also report the mean data for Argille Subappennine clays from other localities (normalized data from Dell'Anna & Laviano 1991). The close compositional analogies with the Laterza sherds are shown.

70

80

90

100

110

120

Ni (ppm)

Fig. 5. Bivariant plot of Rb v. Ni in the ceramic body of majolicas and the Laterza clays, o, sherds of Groups E, G and H; II, sherd of Group B; :~, clays from quarry near Laterza (dell'Aquila et al. 1995b). When plotted in ternary diagrams, as proposed

by Sandrolini & Polmonari (1974), the data for the major element composition in Table 1 show that all the studied samples occupy the field of 'majolica clays' (Fig. 6; area 3). The same results are shown by the Vincenzini & Fiori (1977) ternary diagrams. Only Group B is outside the cluster the other samples, because of different chemical composition. The glazed coating

Si, Pb and Sn are the principal elements in the tin-glazed coatings. On macroscopic observation, as a rule, all the samples show a white glazed coating of great aesthetic value (i.e. high level of thickness, opacity and whiteness), with the exception of the samples of Group F, which are characterized by a lower opacity level. F + CaO + MgO (60%)

22 o

•

20 0 0 00000

18

0 0

9 o

0 o

o~

0~0000 0

~ e ~ Oo

14

0

12 10

2.5

3.5

4.5

5.5

6.5

7.5

MgO (wt%)

Fig. 4. Bivariant plot of CaO v. MgO in the ceramic body of majolicas and the Pleistocene clays. ©, sherds of Groups E, G and H; I, sherd of Group B; #, clays from quarry near Laterza (dell'Aquila et al. 1995b); e, clays from Lucanian Basin (Dell'Anna & Laviano 1991).

S (100%)

A (60%)

Fig. 6. Temary diagram SIO2-(A1203+ TiO2)-Fe203-{CaO + MgO + Na20 + K20 of the ceramic body of majolicas and the Pleistocene clays. Grey area, Groups E, G and H, and clays; ~, Group B. Numbered fields: 1, red stoneware clay; 2, Cottoforte clay; 3, majolica clay.

LATERZA MAJOLICA INVESTIGATION EHT = 15.0 KV Cast Palagianello (TA)

WD = 21rnm

2.00 mm

R= 4QBSD Dip. Geornineralogico - Bad

Fig. 7. Backscattered electron micrograph of a crosssection of tile (sample E.3, Group E). Microscopic observations show a very uniform thickness of the coating layer, applied directly to the surface of the body. No slip (ingobbio) was applied between the body and the glazed coating, as SEM observations of cross-sections confirm (Fig. 7). However, the thicknesses of the coating layers vary greatly, ranging from 150 ~m (sample A.1, Group A) to 500/xm (Group C). For common wares, thickness ranges from 220 to 310 Ixm, and a thickness of 240 t~m was measured for the majolica floor tile (Group E). The coating layer of Group B is particularly thin, and ranges from 20 to 30 Ixm thickness. Such a low value could be due to wear on the investigated tile. On the other hand, the maximum value of the thickness of Group C (Fig. 8) was presumably due to the display purpose of the heraldic tile and to its artistic value. The low thickness of sample A. 1 of turchino majolica ware (Group A), on the contrary,

]I

L=SE1 EHT = 15.0 KV WD = 21turn Cefamica D'Alessandro, 1678 Latefza 200 prn t-........ r

~

-

~

together with other distinctive features (hardness of glazed coating, very low Sn content), seems to point to a non-local production. A similar very low Sn/Pb ratio for turchino majolica wares has been reported in the literature, as in the production of Castelli (Maldera et al. 1989) for example. The samples of Groups A, B and E show a heterogeneous structure of the white ground glaze with many inclusions of quartz and Kfeldspar and minimal interstitial glass; only a few vacuoles are present (Fig. 9). The samples of Group F show a homogeneous glass with many vacuoles and very few inclusions of quartz. The sizes of the spherical gas bubbles normally range from 20 to 80 I~m; in the tile of Group E some bubbles reach 150 I~m diameter. Gas bubbles formed during the firing process of the glass layer and may have been due to organic matter burning in the liquid glaze (Padilla et aL 2005). EDS analyses show that Si, Pb and Sn are the main elements in the tin-glazed coatings of the majolica. As a rule, Sn detection by SEM-EDS proved difficult, because of the lead shielding effect. Sn is clearly visible as granule clusters in a S i - P b phase (Fig. 10) over limited areas, where the glazed layer was fractured or removed. Sn is particularly evident in the spectrum of Group B, where wear has significantly consumed the glaze coating. Unlike Mexican majolica from Puebla (Padilla et al. 2005), Sn was never found on the inner surfaces of bubbles. The Sn/Pb ratio is variable in the samples examined. In addition to quartz and feldspar inclusions, Na and C1 occur in all samples. Ca phosphates have been found only in a few samples, conjointly with alteration or microfractures of the

EKI- = 150 KV Dip, Geomineralogico - Ban

157

Pozzo Purgatorio

WD = 21ram

100 pm ~

R= 4QBSD Dip. Geomineralogico- Bari

~

•

~

,

,

,

Fig. 8. Secondary electron micrograph of a crosssection of the d'Alessandro tile glaze (Group C).

,

Fig. 9. Backscattered electron micrograph of the glaze coating (Group A).

C. DELL'AQUILA ETAL.

158 EHT = 150 KV

WD = 21mm

Ceramica D'Alessandro, 1678 Latefza 200 IJm ~

R= 4QBS/] i

Dig Geommeralocj¢o- Ban

Fig. 10. Backscattered electron micrograph of the d'Alessandro tile glaze (Group C), showing bright tin dioxide crystals on a Pb-Si area.

surfaces, and seem to be of secondary origin (i.e. caused by burial of the samples). The powder diffraction spectra of the glazed coating, particularly on samples of Groups B and C, show that very few lines are always present, but among these those of cassiterite are the strongest lines. Some differences of glazed coating occur between the earlier samples (16th-18th century) and the later ones (19th century). These differences concern the opacity level and the morphologies of the glazed coating. The low opacity level of the 19th century samples (Group F) is not due to a low content of the tin in the glaze coating; in fact, tin is found to be always present as granule clusters dispersed in a homogeneous S i - P b mixture, similar to the earlier Laterza occurrences (Fig. 10). The opacity level differences seem to be due to the glazed coating structure, which in earlier samples (Groups A and E) shows many sharpedged inclusions of quartz and K-feldspar (Fig. 9); these are almost absent in the 19th century samples. These last mineral inclusions in the glazed coating may, according to Piccolpasso (1559), be proof of the use of the 'marzacotto' technique in the earlier majolica manufactories. Therefore, the gradual decline in the quality of the glazed coatings may be explained by a loss of craftsmen's skills in the manufacturing techniques.

The pigments At the macroscopic level the polychrome majolicas studied are characterized by blue, orange, yellow, black, green and violet-brown colours on a white tin-glazed ground. The decorations are in white only on the blue tin-glazed ground

(majolica turchina samples A.1 and A.2, Group A). On all the samples studied, qualitative analyses of colouring pigments were carried out by EDS. In addition, a quantitative analysis on three samples (A.1, A.2 and A.3; Group A), not certainly ascribable to the Laterza manufactures, was also carried out by ICP. The main results for each group examined are reported below. Group A. The samples of polychrome majolica ware show, under SEM-EDS, a Pb-Si matrix, common for all the colours. In particular, on yellow-orange Sb also always occurs; so antimony is to be considered the colouring pigment. This element, associated with Pb, shows, under SEM, a pseudo-hexagonal morphology. Chemical analyses, carried out by ICP on turchino and blue decorated majolica samples, have shown that the pigments consist of Co, As, Fe and Ni (Table 3). Sample A.1 shows a remarkable difference in Fe and As contents. Group B. The sample of turchino decorated majolica tile dated at 1591 shows a blue colour on white glaze. Co, Fe and As have been identified by SEM-EDS analyses on a cross-section of the sample. These elements occur as traces with significant amounts of Si, Pb, K, Na, A1, Sn and Mg. The pigments were found in a very narrow layer ( 1 - 2 txm) between the glaze and the body. Group C. The polychrome majolica heraldic tile signed d'Alessandro shows the following colours: blue, yellow, orange, black and green. Also for this sample, the yellow and orange colours are due to Sb. Sb, Pb and Si are concentrated in pseudo-hexagonal crystals and show euhedral morphology with a grain size of about 2 - 3 txm (Fig. 11, white hexagons). These crystals are laid upon a S i - P b matrix, which also often has a pseudo-hexagonal (about 8-15 ~m) morphology (Fig. 11, grey hexagons). PXRD analysis of yellow pigment allows the chemical compound bindheimite (PbeSb206(O,OH)) to be recognized. Black is due to Ni and to a lesser amount of Sb, Co and Fe embedded in a Pb-Si matrix. Similarly to the yellow, a pseudo-hexagonal morphology has been observed in the black. The blue colouring pigment is due to Co together with Fe, As and, to a lesser extent, Ni. Under an optical microscope the green appears to be due to the overlapping of blue and yellow. Group D. The sample of turchino decorated monochrome majolica ware has shown, by S E M - E D S analysis, the following elements: Pb, Si, P, Sn, As, Fe and Ca. Co was not found probably because it is below the detection limit of the apparatus used.

LATERZA MAJOLICA INVESTIGATION EHT = 1 5 0 KV Ceracnica

D'Aiessandro,

1878 laterza

WD = 21ram

R= 4QBSD

20.0 pm t

I

Dip. Geomineralogico - Barf

Fig. 11. Backscattered electron micrograph of a yellow-orange colour pigment of the d'Alessandro tile (Group C), showing bright Sb-Pb crystals on a Pb-SiSb matrix (light grey). The Sb/Pb ratios increase from black (spl), to grey (sp2, sp3, sp4), to white pseudohexagons.

Group E. The samples of majolica-tiled floor from Palagianello castle show the colours blue, yellow and green on white glaze. The blue colouting pigment is due to Co, Fe and As, and minor amounts of Ni. It is very interesting to observe that EDS analyses did not show As, associated with the other pigments; in addition, As, associated with Pb, Ca and Si, forms very long prismatic crystals deposited inside the surface of degassing bubbles (Fig. 12). On the other hand, Co, Fe and Ni are dispersed, as a rule, inside the upper layer of the colouring cover. Yellow is related to Sb, which appears, also for these samples, as pseudo-hexagonal crystals. SnO2 grains on a P b - S i matrix constitute the white tin glaze.

EHT = 15.0 KV Cast. Palag~anello (TA)

50.0prnl

WD = 22ram

R = 4QBS{:) I

ip

i

oQ'

-

'

Fig. 12. Backscattered electron micrograph of a blue colour pigment of a tile (Group E), showing long prismatic crystals where As, Pb, Ca and, to a lesser extent, Si are concentrated inside degassing bubbles.

159

Group F. The 19th century samples show decorations coloured in light turchino (with a grey tone), yellow, green and violet-brown. The violet-brown pigment is due to Mn; the light turchino pigment to Co, Ni and Mn; the yellow to Sb. Also for this sample group, the green might be due to the overlapping of yellow and blue. Groups G and H. These groups are non-coated pottery. Table 4 summarizes the nature of colours found in the samples studied. Sb, again with a pseudo-hexagonal morpholgy, is the yellow or yellow-orange pigment, according to Piccolpasso (1559) and Marmi (1636), both of whom quoted 'antimonia' (perhaps Sb2S3) as the main source of the yellow colour. PXRD analysis of Group C allowed us to recognize bindheimite to be the colouring substance. Also in Spanish majolicas (Santovenia polycromo, 1780-1825) it was found that the 'yellow glaze contains pigment grains rich in Pb and Sb embedded in the glaze' (Padilla et al. 2005). Unlike Laterza samples, these grains do not show a characteristic morphology and 'the colouring substance (Naples yellow, Pb3(SbO4)2) has not fused into the glaze' (Padilla et al. 2005). For the samples from Laterza Sb has also been found in green and this supports the hypothesis that in these sherds the green might be due to the overlapping of yellow and blue. The black colour on the d'Alessandro tile is due to Ni, with minor amounts of Co, Fe and Sb, in a S i - P b matrix. Perhaps Ni and Co sulphides were the main minerals used for this colour. Particularly interesting is the absence of Mn, which normally is quoted as the main colouting substance of the black (Piccolpasso 1559, p. 153; Padilla et al. 2005). On the other hand, Mn has been found in the 19th century samples, the only examined samples that show a violet-brown pigment. The technique of outlining the decorative figures with a narrow line in violet-brown began in the Laterza majolica in the second quarter of the 18th century. Co is always present in the blue (turchino) decorated coats of Laterza production, with the exception of Group D. It is often associated with Fe and As and, sometimes, with Ni. Piccolpasso (1559) and Marmi (1636) quoted 'zaffara' (zaffre: a mixture of cobalt oxide with other metal oxides) as the source of the blue colour. The characteristic of the studied majolica is a significant occurrence of As, in amounts similar to those of Fe but higher than Co and Ni, as EDS semi-quantitative and ICP analyses show (Table 3). The A s - C o - N i association, with the exception of Fe, is very common in blue

160

C. DELL'AQUILA ETAL.

Table 4. Pigments of the Laterza majolica Sample

Group A Group B Group C Group D Group E Group F

Colour* Blue

Yellow-orange

Co-As-Fe-Ni Co-Fe-As Co-Fe-As minor Ni As (no Co d e t e c t e d ) Co-Fe-As minor Ni Co-Ni-Mn

Sb-Pb . Sb-Pb

Black

Violet-brown

Green

-

-

-

Sb overlapping of blue and yellow

-

-

Mn

Sb overlapping of blue and yellow

.

.

. Sb-Pb

Ni minor Sb-Co-Fe . . -

Sb-Pb

-

. .

*Quantitative analyses by ICP on three samples (A. I, A.2 and A.3) of Group A. Semi-quantitative analyses by SEM-EDS for all other samples.

pigments (Gratuze et al. 1996; Padilla et al. 2005), but the peculiarity of the Laterza majolicas is the high As/Co ratio (about 3:1). A similar high As content was found by Casadio et al. (1999) in some majolicas from Faenza, and it was related to the nature of the raw colouring material, which was particularly poor in cobalt. Moreover, the occurrence of the coloured particles in the layers of the glaze is also variable in the sherds studied, as at least in one case As is present in long crystals associated with Pb, Ca and Si (Fig. 12). In this case, a cobalt arsenate or a calcium arsenate, together with a cobalt salt, could be the main pigments used for this colour. The late samples of Group F also show Mn in the blue pigments, associated with Co and Ni, and this may explain the grey tone of the blue. Finally, on the basis of the analyses of the colouring materials, we conclude that the 1591 tile (Group B), which significantly differs from the majolica made in Laterza as regards the chemical composition of the ceramic body, does not show any evident difference in the nature of the blue pigment. In contrast, the turchino majolica ware (sample A.1, Group A) shows, in the blue Fe and As contents, significant differences from the other samples of Group A, and this suggests that this sample was not made at Laterza. As regards the natural source of the minerals and of the pigments used at Laterza, data in the literature and archives are very sparse. There is only some information from the beginning of the 19th century (1807-1811). The raw materials came from different sources: lead, as litharge, from Venice; manganese, perhaps as pyrolusite, from Mormanno (Calabria, southern Italy); powdered cobalt, the so-called 'smaltino', imported from outside the Naples Kingdom, perhaps from Venice; yellow from roasted

stibnite and from iron rust. No direct information has been found as regards tin. In other places, such as Castelli (Abruzzo, central Italy), the tin came from Naples with a great quantity of white earth for the making of slip (ingobbio). Fabbri et al. (1996) cited imports by Venetian merchants of tin and lead from the Great Britain, from the 14th century onwards, and cobalt from Augsburg, Germany. Conclusions The ceramic body

The composition of the ceramic body of Laterza majolicas studied in this work is macroscopically fairly homogeneous. The chemical analyses carried out on the pottery samples show that no difference in composition occurs between the ceramic body of the majolicas belonging to the 16th-18th centuries and that of 19th century. In contrast, the chemical composition of the turchino decorated majolica tile dated 1591 (Group B) differs significantly from that of the other groups both in major oxides and in trace elements. Therefore, the floor of the monastery (located at Lecce), from which this sample comes, certainly does not belong to the Laterza production. Its place of manufacture is probably to be assigned to the Salento area. Also, the turchino majolica ware (sample A. 1, Group A) shows a chemical composition of the blue colouting material that is different from the others made in Laterza. Optical microscopy and SEM analyses of Laterza production show that the matrix of the pottery is homogeneous and purified, and this is proof of the high quality of ceramics that made Laterza renowned in southern Italy in the 16th- 18th centuries.

LATERZA MAJOLICA INVESTIGATION The glazed coating As regard the ground lead glazes, a qualitative difference in opacity level and morphology of the glazed coating has been established between the earlier samples (16th-18th centuries) and the later ones (19th century). The low opacity level of the 19th century samples is not due to a low content of the tin in the lead glaze coating, but possibly to the absence in the glaze of inclusions able to increase its opacity.

The pigments It is remarkable that the craftsmen at Laterza in

the 16th-18th centuries also showed a great technical ability in selecting pigments, particularly the blue, the yellow and the black, all characterized by pigment grains rich respectively in As and Co, Sb and Pb, Ni and Pb. More extensive analytical work, and quantitative information on the different elements of the colouring pigments from larger amounts of samples, could reveal clear differences between the pigments used at Laterza in different periods. The figures 8 and 11, previously published in dell'Aquila et al. 1995a, are reproduced with the permission of the editor (Techna Group Srl, Faenza, Italy). The authors thank P. Trotti for his assistance in improving the quality of the photographs. Useful discussions with O. Del Monaco are gratefully acknowledged. The constructive remarks of two anonymous reviewers helped to improve the paper.

References CASADIO, R., FABBRI, B., GUARNIERI, C. fig; MINGAZZINI, C. 1999. Analisi di scarti e prodotti intermedi di lavorazione di maiolica faentina (prima met/l del XV secolo): rivestimenti e colori. In: FABBRI, B. and LEGA, A. M. (eds) Atti della 3 ° giornata di archeometria della ceramica, Faenza, 30 marzo 1999, 37-47. DELL'ANNA, L. & LAVIANO,R. 1991. Mineralogical and chemical classification of Pleistocene clays from the Lucanian Basin (Southern Italy) for the use in Italian tile industry. Applied Clay Science, 6, 233-243. DELL'AQUILA, C. 1998. La 'Cantina spagnola' nell'insediamento rupestre di Laterza. Pro Loco di Laterza, Centro Ricerche Storiche, Rotary International 2120 ° Distretto. DELL'AQUILA, A. & DELL'AQUILA, C. 1980. La ceramica di Laterza: Angelo Antonio d'Alessandro. Faenza, 66, 343-351. DELL'AQUILA, A. & DELL'AQUILA, C. 1988. La maiolica di Laterza: sintesi dell'evoluzione storica. VIII Convegno della ceramica, Pennabilli 1987. Rimini, Annali di Studio, 4, 59-69.

161

DELL'AQUILA,A. & DELL'AQUILA,C. 1990. Ceramica pugliese del '500: documenti e frammenti da recupero. In: Castelli e la Maiolica Cinquecentesca Italiana, Pescara, 23-25 aprile 1989, 207-214. DELL'AQUILA, C., LAVIANO, R. & VURRO, F. 1994. Indagini preliminari per la caratterizzazione chimica e mineralogica di ceramiche postmedievali da Laterza (Taranto). In: BURRAGATO, F., GRUBESSI,O. • LAZZARINI,L. (eds) 1st European Meeting on Ancient Ceramics (EMAC), Universitb 'La Sapienza', Rome, 477-492. DELL' AQUILA,C., LAVIANO,R. & VURRO, F. 1995a. Coating of Laterza ceramics from the 15th to 19th centuries. Chemical and mineralogical characterization. In: VINCENZINI, P. (ed.) The Ceramics Cultural Heritage. Techna, Faenza 837-844. DELL'AQUILA, C., LAVIANO, R., VICECONTE, A. & VURRO, F. 1995b. Mineralogical and chemical characterization of clay bodies in the 18th-19th century ceramics from Laterza (southern Italy), and raw material implications. In: FABBRI, B. F. (ed.) The Cultural Ceramic Heritage, Riccione (Italy), 2-6, October 1995. Fourth Euro Ceramics, 14, 271-282. DONATONE,G. 1968. Angelo Antonio d'Alessandro e la ceramica di Laterza. Faenza, 64, 103-109. DONATONE, G. 1980. La maiolica di Laterza. Centro Studi per la Storia della Ceramica Meridionale, Bari. DONDI, M. 1999. Clay materials for ceramic tiles from the Sassuolo District (Northern Apennines, Italy). Geology, composition and technological properties. Applied Clay Science, 15, 337-366. FABBRI, B., VIALE, M. & NANNETTI, A. M. 1996. Caratteristiche chimiche per un inquadramento storico-tecnologico della majolica rinascimentale ligure. Faenza, 82, 212-226. FRANZINI,M., LEONI,L. & SAITTA,M. 1972. A simple method to evaluate the matrix effects in X-ray fluorescence analysis. X-ray Spectrometry, 1, 151-154. FRANZINI, M., LEONI, L. & SAITTA, M. 1975. Revisione di una metodologia analitica per fluorescenza X, basata sulla correzione completa degli effetti di matrice. Rendiconti della Societb Italiana di Mineralogia e Petrologia, 31, 356-378. GRATUZE, B., SOULIER, I., BLET, M. &VALLAURI, L. 1996. De l'origine du cobalt: du verre ~ la c~ramique. Revue d'Archdometrie, 20, 77-94. HEIMANN, R. B. 8¢ MAGGETTI, M. 1981. Experiments on Simulated Burial of Calcareous Terra Sigiltata (Mineralogical Changes). British Museum, Occasional Papers, 19, 163-177. LEONI, L. & SAITTA, M. 1976. Determination of yttrium and niobium on standard silicate rocks by X-ray fluorescence analyses. X-ray Spectrometry, 5, 29-30. MALDERA,R., CASADIO,R., MOLARI,F. ~ SARAGONI, D. 1989. La spettrometria RX: un contributo per l'attribuzione della provenienza della tipologia Orsini-Colonna. In: Le maioliche cinquecentesche di Castelli. Pescara, Carsa, 176-182. MARMI, D. 1636. Segreti di fornace. In: BERTI,F. (ed.) Segreti di fornace. Montelupo Fiorentino, Aedo (2003).

162

C. DELL'AQUILA ETAL.

MUNSELL, A. H. 1947. A Color Notation, Baltimore, Md., 18th edn., Munsell Color Company. PADILLA, R., SCHALM, 0., JANSSENS, K., ARRAZCAETA, R. & VAN ESPEN, P. 2005. Microanalytical characterization of surface decoration in Majolica pottery. Analytica Chimica A cta, 535, 201 - 211. PICCOLPASSO, C. 1559. Li tre libri dell'arte del vasaio. Reprint All'Insegna del Giglio, Firenze (1976).

SANDROLINI, F. & POLMONARI, C. 1974. Variazioni strutturali e dimensionali durante la cottura di argille italiane usate per materiali da costruzioni. La Ceramica, 17, 6-11. VINCENZiNI, P. & FXORI, C. 1977. Caratteristiche naturali di argille italiane e correlazione con le proprieth tecnologiche dei prodotti da esse ottenibili. Ceramurgia, 3, 119-134.

Islamic and Hispano-Moresque (mfdejar) lead glazes in Spain: a technical approach M. V E N D R E L L - S A Z l, J. M O L E R A 2, J. ROQUI~ 1 & J. P I ~ R E Z - A R A N T E G U I 3

lDepartment of Crystallography and Mineralogy, Universtitat de Barcelona, 08028 Barcelona, Spain (e-mail: marius, vendrell@ ub. edu) 2Department of Physics, Universitat de Girona, 17071 Girona, Spain 3Department of Analytical Chemistry, University of Zaragoza, 50009 Zaragoza, Spain

Abstract: Islamic and Hispano-Moresque glazes from the 10th to 15th centuries found in various archaeological sites, most of them workshops, are studied to show the technical evolution of the medieval glazing process. The technology seems to show a simplification: the early Islamic glazes were applied on prefired bodies and after fritting a lead-silica mixture, whereas for the later Islamic productions the raw materials for the lead glazes were not fritted and they were applied over unfired bodies. The same simplified technology was used in the Hispano-Moresque workshops. In the Islamic workshops lead glazes were coloured by adding elements (Fe, Cu, Mn), whereas the mfidejar technology simplified the process by using only one recipe to produce pots of different colour. This was achieved by applying the glaze in a different manner (on one side of the pot to obtain yellow or on both sides to obtain green), or using different pastes (already used to produce pottery for different uses). Finally, there are differences between Islamic and Hispano-Moresque tin glazes related to the crystal size of the opacifier (tin oxide crystals), which should indicate some technological differences in temperature, glaze composition and the process to obtain the frits because of the high dependence between viscosity, temperature and crystal nucleation and growth.

The presence of a glaze coating on a piece is for two principal purposes: (1) to make the pot impermeable so that it can contain liquids without contamination by the pore system or loss of liquid; (2) to produce a decorative finish. Both of these reasons explain the glazing of some of the ceramic productions in both the Islamic and the later Hispano-Moresque (mfidejar) productions found in the Iberian peninsula after the Islamic domination. This tradition survived until the present century, for decorative purposes and also for tourist souvenirs. The glazed ceramics produced during the Middle Ages in Spain consist of two types of pottery: pots for kitchen and storage use (stewpots, jars, candle lamps, etc.) and tableware (dishes, bowls, some jars, etc.). Most of the tableware was coated with a white opacified glaze, which is usually known as tin glaze. In all cases the glaze is a lead glass with different additives to achieve different colours and opacity. The aims of this paper are to characterize Islamic and mfidejar glazed productions of several archaeological sites excavated in Spain, from the 10th century (the workshops of Murcia, Zaragoza and Denia) to the 14th and

15th centuries (workshops of Paterna and the productions of the Catalan area). In this paper the object of the study is the glaze itself and its relationships with the ceramic body. The pottery analysed in this paper covers a wide range of periods and geographical sites, as can be seen in the map in Figure 1.

Sampling and analytical procedure As can be seen in Figure 1 the sites from which the pottery came cover a wide geographical area, mostly in the Mediterranean basin of Spain. The ancient Islamic productions studied came from a workshop named San Nicolfis in Murcia (Navarro Palaz6n 1990; Mufioz 1993). An important Islamic potters suburb dated to the 1 lth century or even to the end of the 10th century, was excavated in Zaragoza on the eastern side of the ancient town (Aguarod & Escudero 1991; Aguarod et al. 1991; Mostalac 1995; P~rezArantegui 1997; Pdrez-Arantegui & Castillo 2005; P6rez-Arantegui et al. 2005). In Palma de Mallorca waster pots dated to the 1 lth century have been found, which suggest local production (Rossell6 1995, and pers. comm.). The latest

From: MAGGETTI,M. & MESSIGA,B. (eds) 2006. Geomaterialsin CulturalHeritage. Geological Society, London, Special Publications, 257, 163-173. 0305-8719/06/$15.00 © The Geological Society of London 2006.

164

M. VENDRELL-SAZ ET AL.

t/

0 '

, dmr

100

5,0,1 o.

Im

~,

~m~

~ j ' ~ ' .

Ii,

I i

MALLORCA ~ J

m

...

Islamic b o r d e r line in the m i d - 11th c. Islamic b o r d e r line at the b e g i n n i n g o f t h e 13th c.

I

I

Fig. 1. Map showing sample locations. Islamic productions studied are from Denia, a small city that was important for pottery in medieval time (Gisbert et aL 1991). The remains of workshop rubbish with pottery tools (bars, stilts, hooks), waster pots, frits and many ceramic objects were found there, and 11 large rectangular kilns were excavated (Gisbert 1990; Gisbert eta/. 1991). The production was dated between the end of the 12th century and the Christian conquest in 1244, when the workshop was destroyed. Hispano-Moresque ceramics studied are from Paterna, an important medieval ceramic centre, which exported tin glaze and lustre ceramics around the Mediterranean basin. The ceramic fragments analysed come from two large workshops excavated at Paterna (Olleries Majors), one dated to the 13th century and the other from the 14th to the 15th century (Amigues & Mesquida 1993; Molera et al. 1996). Finally, Catalan ceramic fragments from the Barcelona and Cardona area are analysed. Table 1 shows the sites studied, the

chronology (according to archaeological criteria) and the number of the samples analysed. The analytical procedures applied to study the glazes of these ceramics were mainly based on observations by scanning electron microscope (SEM; JEOL JSM-840) with secondary and backscattered electron detectors, and by energydispersive X-ray spectrometry (EDS), providing basic images, morphology and structural and a preliminary raw chemical composition for the glazes. The observations were performed directly on glaze cross-sections of the samples, After SEM-EDS observations, more accurate chemical quantitative analyses were performed by electron microprobe (EPMA; Cameca SX-50) using standards. The measurement conditions were 20 kV and 15 nA probe current, or 2 nA probe current for Na, with a measurement spot size of about 5 p,m. Micro-analyses were performed directly on glaze cross-sections of the samples. The penetration depth was evaluated by

Table 1. List of location, time period and number of samples analysed Location Islamic productions Murcia, San Nicohis workshop Zaragoza Mallorca Denia, C/Teulada workshop Hispano-Moresque productions Paterna, Olleries Majors workshop Paterna, Olleries Majors workshop Catalonia

Time period (century)

Number of samples

10th

24 lead glaze, 13 tin glaze

11th 1l th 13th

16 tin glaze 5 tin glaze 6 lead glaze, 9 tin glaze

13th

35 lead glaze, 56 tin glaze

14th-15th

38 lead glaze, 34 tin glaze

14th- 15th

23 tin glaze

ISLAMIC AND MI~IDEJARLEAD GLAZES Kanaya-Okayanm Range (Kanaya & Okayama 1972), and was about 2.5 ixm in the glaze. The data for all the elements except tin were obtained by electronic microprobe; the tin content was analysed by X-ray fluorescence (XRF) through the production of a bead by melting 100 mg of powder scratched from the glaze. It is not possible to analyse the bulk composition of a glaze containing small particles of tin oxide by EPMA because the beam cannot be defocused to obtain a sufficiently representative sample without compromising the quality of the analysis. The glaze colour was calculated by the acquisition of the diffuse reflectance spectra in the visible range (400-700 nm) with a spectrophotometer equipped with an integrating sphere of Ba sulphate. The reflectance UV-Vis spectra from the glazes have been recorded using a Shimadzu UV-2101 PC spectrophotometer with an interval of measurement of 0.5 nm, with a 0.5 mm beam footprint and at medium speed of measurement. The reflectance with the specular component included (RSCI) and the reflectance with the specular component excluded (RSCE) spectra were acquired simultaneously. Furthermore, some laboratory re-creations of glazes with different compositions and physical characteristics (thickness, heating rate, etc.) were produced to study some features of the historical samples, such as diffusion from the paste to the glaze, and the development and position of the crystalline phases forming the interface.

Characterization of the glazes Microstructure

Through the observation of polished sections of glazes in the SEM different zones can be

recognized in the glaze; namely, the glass body itself, some crystalline phases embedded in the glass, an interface between the glass and the paste, and the paste, which acts as a support. In this section a short description of these terms is provided to clarify the following text in which these terms are used. Figure 2 represents an SEM image in which these zones are marked and identified. The ceramic body is the support on which the glaze is applied, and in this paper data on the ceramic body will be supplied only if they are necessary for better understanding the glaze behaviour and characteristics. The glass corresponds to the glaze itself and it is an amorphous phase. Inside the body of the glaze several inclusions can be identified such as incompletely reacted quartz grains, crystalline phases formed in the interface and detached during firing and trapped bubbles. Occasionally, some of the bubbles reached the glaze surface, carrying crystals (Molera et al. 1993). In some cases, grains of the temper of the paste are dragged into the glaze and appear as coarse inclusions in the glassy body. This is not common in the Islamic and mtidejar medieval glazes, except in certain cooking pots with a thin glaze. The presence of such grains results in a low-quality finish and should be avoided by the potter as far as possible. The glaze interface is a layer of different thickness formed by a network of crystals with different morphologies, which result from the reaction between the glaze and the ceramic body during firing (Raffaillac-Desfosse 1994; Molera 1996). The significance of this layer as a technological parameter will be discussed later. Table 2 gives a summary of the microstructure of the lead transparent glazes studied, and tin opacified glazes are discussed later.

glaze 6 b . .

"

....

""

165

P---

Fig. 2. SEM image of a lead glaze, showing the terms used in the text.

166

M. VENDRELL-SAZ ET AL.

Table 2. Microstructure of lead glazes Colour

Glaze thickness (~m)

Murcia, lOth centuo, Green lead glaze Yellow lead glaze Brown lead glaze Denia, 13th century Green lead glaze Paterna, 13th century Green lead glaze Yellow lead glaze Brown lead glaze Paterna, 14th- 15th centur3, Honey lead glaze Brown lead glaze

Interface thickness (~zm)

K-feldspars

Bubbles

Temper particles

100-120 100-120 100-120

5 - !0 I0 5-10

Few Medium Few

Few Few Few

Few Rare Few

110

10-40

Abundant

Abundant

Medium

60-90 60-90 100-120

10-40 10-30 10-50

Abundant Abundant Abundant

Abundant Abundant Abundant

Few Few Few

70-150 100-180

10-50 10-60

Abundant Abundant

Abundant Abundant

Few Few

Chemical composition

All Islamic and mfdejar glazes from Spain are lead glazes, that is to say, they are SiO2-PbO glazes ( > 9 0 wt%) with other elements such as A1203, K20, CaO, Fe203, etc., each in concentrations of < 5 wt% (Molera et al. 1997; P~rezArantegui 1997). Not a single case of alkaline glaze has been found, so it appears that the production of silica-lead glazes was a technology that was widely diffused in the Iberian Peninsula from the very beginning of the Arab domination. Probably, the fact that Spain has been one of the main producers of galena (PbS) since Roman times has to be considered as a possible reason for this. However, the use of lead glazes offers some technological advantages with respect to alkaline glazes, such as lower fusion temperatures and a characteristic strong shine (Tire et al. 1998). From the structural point of view, the silicalead glasses are formed by a disordered network of SiO 4 tetrahedra with some A1 in the Si structural sites. In this case the A1203 acts as a stabilizer of the glassy structure to equilibrate the electronic charges between Si and other metallic elements (K, Ca, Pb, etc.) and oxygen. The role of the PbO in these glazes is to act as a flux (instead of the alkaline elements in the early Islamic glazes; Mason & Tire 1997). Looking at the PbO-SiO2 phase diagram (Geller et al. 1934, 1943), one can see several eutectic points around 750 ~C for PbO contents between 90 wt% and 70 wt%. However, the presence of small amounts of other elements such as K20, CaO, etc. slightly modifies the eutectic points, and melting can be achieved at such temperatures with lower amounts of PbO (around 50 wt%).

Table 3 presents a summary of the chemical composition of the Islamic and mfdejar glazes studied in this paper, classified by the colour of the different glazes, and Figure 3 shows the SiO2-PbO bivariant relationships for the three time periods (Islamic 10th century, m6dejar 13th century and 14th century). The results are calculated for each sample as the mean of several measured points (7-15) following a profile from the paste to the outer surface of the glaze, and the mean of each sample is used as a single datum point to calculate the statistics of the whole production. Most of the samples studied exhibited a uniform chemical composition following the profiles measured through the glaze (except for the inclusions), but some had profiles as shown in Figure 4. The shape of these profiles suggests that there is a diffusion of elements from the paste to the glaze and from the glaze to the paste, following the chemical gradients of each element. Development o f the g l a z e - c e r a m i c body interface

From several replications of silica-lead glazes made in the laboratory under controlled conditions (Molera et al. 2001) the diffusion of components from the ceramic body to the glaze, and the diffusion of lead from the glaze to the ceramic body has been experimentally demonstrated. Diffusion is more pronounced from an unfired clay body than from already fired one, which is consistent with the fact that the phases developed in the paste after firing are more stable. Figure 5 shows two typical diffusion profiles of fired and unfired bodies with SiO2-PbO glazes of 500 ~m thickness ( 3 0 - 7 0 wt%).

I

t

I

weight %

E .~,

I

55.78 (1.69) 57.00 (4.80)

32.84 (1.58) 32.83 (0.75)

?

54.62 (11.70) 57.68 (15.52) 53.11 (5.61)

32.76 (1.53) 30.28 (1.16) 33.59 (1.09)

I

52.83 (1.38)

33.39 (0.22)

..<

54.23 (1.52) 53.01 (2.30)

PbO

34.22 (1.13) 32.54 (0.27)

SiO2

Murcia, lOth century Yellow Brown Denia, 13th century Green Paterna, 13th century Green Yellow Brown Paterna, 14th-15th century Honey Brown

Colour

Table 3. Chemical composition of lead glazes (in wt%)

~

0.63 (0.26) 0.87 (0.10)

0.93 (0.37) 0.86 (0.23) 0.84 (0.29)

1.53 (0.06)

1.78 (0.24) 1.63 (0.48)

K20

~

_

:q ~

~-~ ~'~ ~

8

"~

~u~

,.

,.--"m:~

~

,-,. o ,~,-- =" _="

_ ©

-=

=

~=<

--o"

E"~ [..~

0.58 (0.17) 0.66 (0.03)

0.49 (0.11) 0.43 (0.13) 0.70 (0.12)

0.29 (0.04)

°--'="E~ ,-,-~C~ ~

o~

B

'~ m:~ 0

MgO 0.39 (0.17) 0.38 (0.01)

"* '-" ~ ~=~ ~ :3" ~5 ~ 5 . =

~" ~

~

~"

__,.~

~'~4

_="= ~

--,, ~'~

9-9

;~

-':'~ ~ 0

'-'o ~'~'-'

~

-.~

o

0.09 (0.03) 0.17 (0.06)

0.25 (0.60) 0.10 (0.03) 0.10 (0.05)

0.85 (0.06)

0.82 (0.11) 0.52 (0.04)

Na20

1.55 (0.29) 0.89 (0.60)

2.59 (0.51) 2.41 (0.22) 1.44 (0.63)

3.04 (0.28)

3.20 (0.34) 3.50 (0.63)

CaO

5.61 (0.82) 4.88 (0.76)

5.25 (0.61) 5.44 (0.26) 5.94 (0.58)

2.92 (0.47)

2.91 (0.52) 3.31 (0.51)

A1203

--~

'~

.~%PbOo.

'

2.17 (0.56) 2.12 (0.43)

2.53 (0.50) 1.94 (0.19) 2.56 (0.50)

1.15 (0.14)

1.43 (0.66) 1.71 (0.50)

Fe20~

o~

0.00 (0.00) 0.00 (0.00)

0.00 (0.00) 0.00 (0.00) 0.00 (0.00)

2.70 (0.21)

0.04 (0.06) 0.02 (0.04)

CuO

51

;>

168

M. VENDRELL-SAZ ETAL. 10

(a)

~..

~. "~'~ _, , , ~

:~ 800°0 o 900°0 0 950°0

i

%"

°

0

--,

I

-"

-

'

T ...... r .......

1 . 0 - ,~

(b)

~" o.5

0.03 0

•

~ 1O0

,

t , 200 d (micres)

I 300

'

! 400

Fig. 5. Diffusion profiles for AI203 and K20 in lead glaze re-creations under controlled conditions.

concentration of the diffused elements is equilibrated and the diffusion profiles become flat, thus explaining why, in most samples studied, the chemical composition is relatively uniform throughout the glaze thickness. During the firing, the concentration of AI, K, Ca (in the case of Ca-rich pastes) and Fe diffused from the ceramic body to the glaze enhances the crystallization of some phases at the interface. The nucleation and growth of K-feldspar containing Pb (Molera et al. 1993, 2001 ) takes place over all the interface whereas Ca-pyroxenes are developed only around Ca-rich grains. The different specific gravities of the neoformed phases (K-feldspars containing Pb and Ca-pyroxenes) within the glass is what determines their distribution in depth with respect to the glass matrix, shifting the neoformed crystallites towards the glaze surface. However, the relative position of the glazed ware inside the kiln should be taken account because it would affect the distribution of the neoformed crystallites while they are floating within the glass matrix during firing (Molera et al. 2001). Because the supply of elements from the paste to the glaze is higher for unfired ceramic bodies than for those that are already fired, the development of a thicker interface in the case of unfired bodies should be expected. Accordingly, for single-fired glazed pottery (the glaze applied on an unfired body) the interface developed will be thicker than for double-fired glazed pottery (the glaze applied on a previously fired body). Laboratory experiments and archaeological

Fig. 6. Thin section of lead glaze that shows the glass, interface and ceramic body. (a) An Islamic glaze from Murcia (10th century), where the interface is not developed. (b) An Islamic glaze from Denia (13th century), where the interface is well developed. evidence support this idea (Molera et al. 2001), as shown in Figure 6. These recent studies confirmed that for illitic clay the interface developed on single-fired glazed pottery is expected to be around 50 txm thick whereas in the case of double-fired pottery it is between 5 and 10 txm; however, for kaolinitic clays it should be noted that the interface developed is always thinner (Molera et al. 2001). Nevertheless, the interface thickness, observed from a polished or thin section, gives valuable information that can help to determine if a glazed ceramic sample underwent a double- or a single-firing process. C o l o u r e d transparent glazes

As is shown in Table 2, the pots with a nonopacified lead glaze exhibit some colour ranging from yellow to green with occasional green and dark brown decoration lines. This means that the glazes contain some colouring element (traditionally those of the first transition row of the Periodic Table can act as colorants; Bamford 1977), which in many cases is Fe (both Fe 2+ and Fe 3+ states of oxidation), although Cu and Mn have also been detected.

ISLAMIC AND MI~IDEJARLEAD GLAZES The colour generation is complex because the lead glazes are transparent and thus the colour of the paste can be seen through the glaze. Therefore, the colour observed is the actual colour of the paste observed through a coloured glaze acting as an optical filter. In such conditions, the paste colour will influence the observed colour and we must consider both paste and glaze when discussing the colour generation. In the Islamic workshops analysed (San Nicolfis, Murcia 10th century; Zaragoza, 1 lth century; Denia, 13th century), the colours of the lead transparent glazed pottery are yellow, brown and green. The glaze was applied to a previously fired body (not for the Denia workshop) as the thin interface (5 Ixm) suggests and as much archaeological evidence demonstrates (Mesquida 1987; Navarro Palazrn 1990). The glazes were previously fritted (at least in Murcia, where frits have been found and analysed) from quartz sand and galena. The colours are due to the presence of chromophore elements such as Fe 3+ for yellow (Fe203 0.50 wt%), Cu for green (CuO 0.30 wt%) and Mn 2 + and Fe 2 + for brown (0.20 and 0.50 wt%, respectively). Therefore, different recipes were prepared for different colour generation. Because the glazes were applied to biscuit fired bodies, the diffusion of Fe from the paste is not expected to be high enough to give colour, and it seems realistic to think that for yellow glazes some addition of Fe oxides was made, as was the case for other elements for other colours. In contrast to the Hispano-Moresque pottery, in all the Islamic samples the paste is relatively Ca-rich (CaO from 10 to 15 wt%) and the paste has a buff colour; these factors do not dramatically influence the colour observed through the glaze. The mtidejar lead transparent glazed pottery shows a change between the 13th century and later productions, as has been observed in the extensively excavated workshops of Paterna (Amigues & Mesquida 1993). The cooking pots appear brown in all cases, whereas other transparent glazes are yellow and green in the 13th century productions and 'honey-like' in the 14th and 15th centuries. The colour generation of these productions has been studied (Molera et al. 1997) and it has been shown that the only colorant element is Fe (Fe 3+ for yellow, honey and brown, and Fe 2+ for green). The green glazed pots of the 13th century have a grey paste reduced during firing. The glaze was applied on both sides of the pot, so that the paste was completely sealed after the glass melting, and CO reduces the iron oxides and releases CO2 by the reaction Fe203 + CO--+ 2FeO +

169

CO2, resulting from the burning of organic matter and carbonate decomposition, so that the Fe oxides of the ceramic body are reduced to spinel phases (magnetite and herzinite) (Molera et al. 1997). In these conditions, the Fe diffused from the paste is Fe 2+, giving a green colour to the glaze. Thus, a grey ceramic paste observed through a green glaze produces a green colour. In contrast, if the glaze is applied only on one side, the ceramic body is not sealed during firing, oxygen diffuses into the body and the reduction of the iron oxides present in the it cannot be achieved. All the glazes except the green ones are yellow when observed and measured by transmittance (Molera et al. 1997), and the different colours observed are due to the use of different ceramic pastes with different Ca content, which strongly influences the colour of the paste (Kreimeyer 1987; Molera et al. 1998). The cooking pots were made with a non-calcareous clay, which after firing becomes reddish, whereas yellow and honey-coloured pots have different pastes, the yellow ones being more Ca-rich than the honey-coloured ones. Tin-glazed pottery

Some of the pots, especially those used for tableware, are coated by a glaze that appears white because the presence of small crystallites of tin oxide (SnO2) as inclusions in the glass. The origin of this technique must be placed in the Achaemian period in Iraq, in about the sixth century B¢ (Mason & Tite 1997). The first white glazes did not use tin oxide as opacifier, but a thick layer of wollastonite crystals developed on calcareous clay bodies, which is also white. The tin oxide was introduced in about the second half of the eighth century in alkaline glazes as a slip layer between the glaze and the paste. The technique was introduced in the Iberian Peninsula with the first Arab invasions and later became widely scattered through Europe and America (Caiger-Smith 1973). The chemical composition of the tin-opacified glazes in the Islamic and mfdejar world is similar to the above-described lead transparent glazes, with the addition of a SnO2 content that ranges from 5 to 10 wt% (Molera et al. 2001). To produce the white colour, the tin oxide must exist as small particles (crystals), the size of which should be of the same order of magnitude as the visible wavelength of white light. However, as scattering follows the Rayleigh formula (Jenkins & White 1987), where the scattered intensity is proportional to the fourth power of the wavelength, the scattered light

170

M. VENDRELL-SAZ ET AL.

should be bluish. This does not occur because the lead glazes absorb light in the blue region, compensating for this effect. A distribution of such crystals in the body of the glass produces the light scattering that gives the white colour observed in this pottery. An optical model to explain the optical behaviour of tin glazes and the influence of the crystal size distribution of the tin oxide and thickness of the glaze has been proposed by Vendrell et al. (1999). Furthermore, all the ceramic pastes to which the tin glazes will be applied are Ca-rich (CaO >15 wt%). If the paste has high content of CaO the resulting colour after firing is buff (Molera et al. 1998), and thus the effect of the colour of the paste when it is observed through the glaze is minimal. In fact, the whiteness of the glaze can be achieved by increasing the SnO2 content, by increasing the thickness of the glaze and by reducing the reddish hue of the paste. This last effect can be enhanced by using Ca-rich pastes. The BSE images (see Fig. 7) show that tin oxide particles exhibit crystal faces suggesting nucleation and growth during some stage of the production processes. Molera et al. (2005) have

(a)

(b)

demonstrated that tin oxide dissolves in the melting previous to the formation of the lead glaze during a first firing and recrystallizes at about 700 °C. After this, tin oxide no longer goes into solution and further firing will not modify the crystal sizes and morphologies. As tin-glaze production involves a fritting process, the recrystallization takes place during the fritting process and is retained when the flit is applied on a biscuited body. The analyses of flits found in the workshops of San Nicolfis (Islamic, 10th century) Denia (Islamic, 13th century) and Paterna (mfidejar, 13th century) showed that their composition is the same as that of the tin glazes found in the same workshop. According to ancient documentation (Allan 1973), a first step of the fritting process was to roast lead and tin together in a kiln to produce 'acerc6' (a name derived from the Arabic word for 'sand'), a mixture of tin oxide and lead oxide. No archaeological evidence has been found for this step at the sites studied here. After this, the 'acercr' was mixed with sand and some other additives and melted in cooking pots. A quartz sand layer protected the walls of these pots. When the mixture had melted, the craftsman threw it onto a cool surface and, after grinding, this frit could be applied to the ceramic bodies with some gum or clay. The chemical data for the tin-glazed pottery analysed in this paper are shown in Table 4. As can be seen, the data for the glass body are similar to those for the lead transparent glazes, except for the tin oxide content. The main difference between Islamic and mtidejar samples is the crystal size of the tin oxide crystals. In the early Islamic workshops (San Nicolfis and Zaragoza, 10th and l lth centuries, respectively) the crystal size is about 200 nm (see Fig. 7), whereas for the mtidejar samples the crystals are from 500 to 1000 nm. Furthermore, the productions of the 14th century differ from those of the 13th century by the uniformity of the tin particle distribution, which is less uniform in the 14th and 15th centuries. As can be seen in Figure 7, the crystals form clusters with a heterogeneous distribution.

Discussion and conclusions

Fig. 7. SEM image of a Hispano-Moresque tin glaze, where the crystalline morphology of SnO2 is clearly visible. In (a) tin crystals are heterogeneously distributed because they remain in phantom clusters, but in (b) their distribution is homogeneous.

On the basis of the results presented in this paper and the archaeological evidence from several sites in Spain, lead glazes were widely used to coat and finish medieval pottery in the Islamic tradition. The Arab craftsmen were the originators of the lead glaze technology and the first glaze recipes found containing lead come from 1700 Bc in the northern part of Iraq

Blue White

Catalonia

Luster Blue White

Paterna, 14th- 15th century

White Green and brown Turquoise Blue Luster Pula

Paterna, 13th century

Lustre

Granada, 4th century

Green

Denia, 13th century

Brown and green

Mallorca, l lth century

Group 1 Group 2 Second side

Zaragoza, l lth century

White

Murcia, lOth century

Tin glaze

0.87 (0.33) 1.03 (0.24)

0.91 (0.07) 1.17 (0.70) 0.72 (0.32)

1.03 (0.09) 0.53 (0.26) 0.58 (0.22) 0.9 (0.29) 0.6 (0.28) 0.62 (0.11)

1.26 (0.31)

1.80 (0.62)

1.00 (0.20)

> 1.0-2.08 < 1.0-1.98 <1.0-1.73

1.46 (0.16)

Na20 0.24 (0.14)

MgO

(0.05) (0.15) (0.13) (0.03) (0.08) (0.01)

0.35 (0.03) 0.34 (0.02)

0.29 (0.03) 0.29 (0.04) 0.22 (0.07)

0.21 0.38 0.36 0.55 0.43 0.16

0.30 (0.05)

0.22 (0.05)

0.23 (0.08)

<0.36-0.87 <0.36-0.84 <0.36-0.91

Table 4. Chemical composition of tin lead glazes

(0.74) (0.67) (0.07) (0.07) (0.10) (0.18)

2.97 (0.08) 2.94 (0.32)

3.63 (0.41) 3.01 (0.50) 2.83 (0.48)

2.89 3.51 3.43 2.45 2.33 2.77

0.88 (0.81)

0.99 (0.14)

3.23 (1.56)

0.61 (0.38) 1.02 (0.64) 2.60 (1.07)

0.48 (0.35)

A1203

(1.71) (2.11) (1.01) (3.62) (1.23) (2.04)

39.01 (1.04) 35.83 (5.72)

50.39 (3.03) 48.41 (1.61) 44.55 (4.01)

42.28 43.85 39.14 38.17 45.21 51.85

40.85 (2.89)

30.94 (1.13)

31.06 (1.38)

46.05 (1.30) 41.51 (0.70) 39.52 (1.31)

3313 (0.35)

SiO2

2.31 (0.44) 2.74 (0.76)

5.06 (1.08) 5.26 (1.19) 3.77 (1.14)

2.6 (0.43) 2.32 (0.76) 2.04 (0.51) 2.24 (0.05) 4.62 (0.06) 6.49 (0.97)

2.69 (0.30)

0.55 (0.11)

1.18 (0.50)

3.74 (0.66) 3.26 (0.50) 3.85 (1.30)

1.10 (0.34)

K20

(0.33) (0.92) (0.07) (0.13) (0.09) (0.26)

2.08 (0.42) 2.05 (0.21)

2.49 (0.33) 2.62 (0.12) 1.52 (0.31)

1.66 2.15 1.97 2.28 1.65 2.33

1.67 (0.38)

0.46 (0.23)

2.40 (0.66)

3.73 (0.89) 3.72 (0.55) 4.28 (0.86)

1.48 (0.68)

CaO

0.94 (0.11) 0.69 (0.09)

2.14 (2.20) 3.75 (1.10) 0.84 (0.34)

0.45 (0.32) 0.44 (0.15) 0.43 (0.07) 0.3 (0.14) 0.22 (0.16) 0.29 (0.16)

0.49 (0.26)

0.14 (0.11)

1.21 (0.54)

0.70 (0.40) 0.77 (0.09) 1.54 (0.61)

0.31 (0.24)

Fe~O3

5-8 5-8

5-8 5-8 5-8

7-8 6-7 6-7 8-9 8-10 5-8

7-9

6-7

7-9

5-7 9-14 4-10

7-10

SnO2

45.63 (2.05) 49.76 (8.24)

32.64 (1.96) 32.62 (0.78) 41.47 (2.81)

41.14 (2.45) 38.6 (3.72) 44.04 (1.34) 45.5 (0.45) 35.74 (0.50) 33.80 (0.79)

43.69 (1.92)

56.19 (1.14)

53.91 (1.14)

38.04 (1.65) 37.79 (2.71) 39.93 (2.09)

53.82 (2.62)

PbO

r" > N

~7

7o

>

> Z

t" >

172

M. VENDRELL-SAZ ET AL.

(Caiger-Smith 1973). However, glazed pottery and lead-glazed ware were not important in Spain until the expansion of Islam in the western Mediterranean. Following the technological parameters determined for the lead transparent glazes, the technology seems to undergo some kind of simplification. The early Islamic glazes (Murcia and Zaragoza) were applied on pre-fired bodies and after fritting the raw materials (this tradition has been demonstrated to be unnecessary). The Islamic workshop of Denia (13th century) did not frit the raw materials for the transparent glazes and they were applied over unfired bodies (and thus fired in a single operation). This technology was also used in the mddejar workshops of Paterna (13th- 15th centuries). In the Islamic workshops different recipes were used to formulate different colours; however, the mddejar technology once again simplified the process. The islamic colours were achieved by the addition of a colouring element (Cu for green, Fe for yellow, Mn and Fe for brown). In the mddejar workshops studied (particularly Paterna, but also other contemporary sites) the potters used the same glass recipe to produce ceramic glazes of different colours. They obtained the colours by applying the glaze in a different manner (on one side of the pot to obtain yellow, or on both sides to obtain green), or by using different pastes (already used to produce pottery for different uses). These later developments appear to represent a scaling-up of the process, a kind of 'industrialization', which involved a simplification of the recipes, handling of raw materials (no fritting where it could be avoided), simpler application methods, simpler processes of firing (single if possible), etc. The tin-opacified glazes are all lead glazes with tin oxide crystals as particles producing the scattering of the light. The crystal size of the opacifier has been shown to be smaller in the Islamic productions, which should mean some as yet undetermined technological difference. This is possibly a matter of temperature during the preparation of the frits, as there is a high degree of dependence between viscosity, temperature, and crystal nucleation and growth. The heterogeneity observed for the tin opacitier in the later mddejar productions seems to be related to the method of preparation and handling of raw materials; unfortunately, no frit has been found in these historical workshops of the 14th to 15th centuries, and thus for the moment the question remains unsolved. This paper has been partially developed within a project funded by a Ministerio de Ciencia y Tecnologfa (grant BQU2002-03162), and by the research project of the

Comunidad de Trabajo de los Pirineos-Diputaci6n general de Arag6n (CTPR4/2003). The authors wish to acknowledge the supply of the samples by several institutions such as the Museum of Ceramics of Paterna and the Archaeological Services of the cities of Zaragoza, Murcia and Denia.

References AGUAROD, M. C. & ESCUDERO, F. 1991. La industria alfarera del barrio de San Pablo siglos I-XIII). hi: Zaragoza: Prehistoria v Arqueolog{a. Ayuntamiento de Zaragoza, Zaragoza. AGUAROD, M. C., ESCUDERO, F., GALVE, M. P. & MOSTALAC, A. 1991. Nuevas perspectivas de la arqueolog/a medieval urbana del periodo andalus/: la ciudad de Zaragoza (1984-1991). In: Aragrn en la Edad Media IX. Universidad de Zaragoza, Zaragoza, 445-491. ALLAN, J. W. 1973. Abu'l-Qasim's Treatise on Ceramics. Iran, XI, 111 - 120. AMIGUES, F. & MESQU1DA,M. 1993. Les ateliers et la cdramique de Paterna (XIIe-XIVe sikcle). Mus~e Saint Jacques, Ville de Beziers. BAMFORD, C. R. 1977. Colour generation and control in glass. Elsevier, New York. CAIGER-SMITH, A. 1973. Tin-Glaze Potteo' in Europe and the Islamic World: The Tradition of 1,000 Years in Mayrlica, Faience and Delftware. Faber and Faber, London. GELLER, R. F. & BUNTING, E. N. 1943. Report on the systems lead oxide-alumina and lead oxidealumina-silica. Journal of Research of the National Bureau of Standards, 31, 255-270. GELLER, R. F., CREAMER,A. S. & BUNTING,E. N. 1934. The system PbO.SiO2. Journal of Research of the National Bureau of Standards, 13(2), 237-244. GISBERT, J. A. 1990. Los hornos del alfar iskimico de la Av. Montgr/Calle Teulada, casco urbano de Denia (Alicante). In: Fours de potiers et 'testares' mddidvaux en M~diterrande Occidentale. Publicaciones de la Casa de Vel~izquez, Srrie Archrologique, XIII, 75-91. GISBERT, J. A., AZUAR, R. & BURGUERA,V. 1991. La produccirn cer~imica en Daniya. El alfar isl~imico de la Av. Montg6/Calle Teulada (Denia-Alicante). In: Actas del IV Congreso A Cerfmica Medeival do Mediterrrneo Occidental, Portugal, 1987, Mertola, 247-262. JENKINS, F. A. & WHITE, H. E. 1987. Fundamentals of Optics. McGraw-Hill, New York. KANAYA. K. & OKAYAMA,S. J. 1972. Penetration and energy-loss theory of electrons in solid targets. Journal of Physics D: Applied Physics, 5, 43-58. KREIMEYER, R. 1987. Some notes on the firing colour of clay bricks. Applied Clay Science, 2, 175-183. MASON, R. B. & TITE, M. S. 1997. The beginnings of tin-opacification of pottery glazes. Archaeometry, 39(1), 41-58. MESQUIDA, M. 1987. Una terrisseria del s. XIII 1 XIV. Publicacions de l'Ajuntament de Paterna. MOLERA, J. 1996. Evoluci6 mineralrgica i interacci6 de les pastes cgdciques arab els vidrats de plom: interaccions arqueombtriques. PhD thesis, University of Barcelona.

ISLAMIC AND MUDEJAR LEAD GLAZES MOLERA, J., I~ADELL, T., MARTINEZ-MANENT, S. & VENDRELL-SAZ, M. 1993. The growth of sanidine crystals in the lead glazes of Hispano-Moresque pottery. Applied Clay Science, 7, 483-491. MOLERA, J., GARCIA-VALLI~S, M., PRADELL, Z. t~ VENDRELL, M. 1996. Hispano-moresque pottery productions of the fourteenth-century workshop of the Testar del Mol~ (Paterna, Spain). Archaeometry, 38(1), 67- 80. MOLERA, J., VENDRELL-SAZ,M., GARCIA-VALLES,M. & PRADELL,T. 1997. Technology and colour development of Hispano-Moresque lead glazed pottery. Archaeometry, 39, 23-39. MOLERA, J., PRADELL, T. & VENDRELL-SAZ, M., 1998. The colours of Ca-rich ceramic paste: origin and characterization. Applied Clay Science, 13, 187-202. MOLERA, J., PRADELL,T., SALVADO,N. & VENDRELLSAZ, M. 2001. Interactions between clay bodies and lead glazes. Journal of the American Ceramic Society, 84(5), 1120-1128. MOLERA, J., PI~REZ-ARANTEGU1,J. & VENDRELL-SAZ, M. 2005. Chemical and textural characterisation of tin glazes in islamic ceramics from eastern Spain. Journal of Archaeological Science (in press). MOSTALAC, A. 1995. Les fours islamiques de Saragosse. In: Le vert & le brun, de Kairouan Avignon, cdramiques du Xe au XVe sibcle. Rrunion des Musres nationaux, Marseille, 31-32. Muiqoz, P. 1993. Nuevos datos sohre urbanismo y alfarer~a medieval en Murcia. Verdolay, 4, 175-184.

173

NAVARROPALAZON,J. 1990. Los materiales isl~imicos del alfar antiguo de San Nicol~is de Murcia. In: Fours de potiers et 'testares' mddidvaux en Mdditerrande occidentale. Publicaciones de la Casa Vekizquez, Srrie Archrologique, XIII, 29-43. PI~REZ-ARANTEGUI, J. 1997. Les glaqures et les premiers 6maux sur la crramique islamique en al-Andalus (Espagne). TECHNE, 6, 21-24. Pt~REZ-ARANTEGUI, J. • CASTILLO, J. R. 2005. Chemical characterisation of clear lead glazes on Islamic ceramics, produced in Northern al-Andalus (Muslim Spain). Proceedings of the 31st International Symposium on Archaeometry, 1998, Budapest, Hungary. British Archaeological Reports (in press). RAFFMLLAC-DESFOSSE, C. 1994. Cdramiques glagurdes mddidvales. Recherche de donndes physiques sur les techniques de fabrication et altdration. These doctoral, Universit6 Michel de Montaigne, Bordeaux III. ROSSELLO, G. 1995. La crramique verte et brune en alAndalus du Xe au XIIIe si6cle. In: Le vert & le brun, de Kairouan gl Avignon, cdramiques du Xe au XVe sikcle. Rrunion des Musres nationaux, Marseille, 105-117. TITE, M. S., FREESTONE, I., MASON, R., MOLERA, J., VENDRELL-SAZ, M. & WOOD, N. 1998. Lead glazes in antiquity. Methods of production and reasons for use. Archaeometry, 40(2), 241-260. VENDRELL, M., MOLERA, J. t~z TITE, M. S. 1999. Optical behaviour of tin glazes. Archaeometry, 42(2), 325-340.

Archaeometric analyses of game counters from Pompeii R. A R L E T T I t, A. C I A R A L L O 2, S. Q U A R T I E R I 3, G. S A B A T I N O 3 & G. V E Z Z A L I N I ~

1Dipartimento di Scienze della Terra, Largo S. Eufemia, 19, 1-41100 Modena, Italy (e-mail: [email protected]) 2Soprintendenza Archeologica di Pompei, via Villa dei Misteri, 2, 1-80045 Pompei (NA ), Italy 3Dipartimento di Scienze della Terra, Salita Sperone, 31, 1-98166 Messina, S. Agata, Italy Abstract: Among the glass finds of the Pompeii excavations, numerous objects of opaque

and transparent glassy material of different colours were recovered and classified as game counters. The main aims of this work were to characterize these samples so as to identify the materials used as colorants and opacifying agents, and subsequently to deduce the technology used for their production. The results of the chemical and mineralogical analyses obtained for game counters were also compared with those obtained for transparent and opaque glass artefacts. The chemical analyses were carried out, using only 300 mg of sample, by both wavelength-dispersive electron microprobe and X-ray fluorescence analysis. The crystalline phases present in the opaque glass were identified using both an automatic X-ray powder diffractometer and a Gandolfi camera. Secondary and backscattered electron images were obtained to study the distribution and morphology of the opacifier particles, and qualitative chemical analyses were obtained with an energy-dispersive system. All the game counters analysed can be classified as silica-soda-lime glass. Two calcium antimonates (CaSb206 and Ca2Sb2OT) were identified in the opaque white, green and blue glass, and Pb:Sb207 particles were detected in the opaque yellow glass. Particles of metallic copper were detected by both energy-dispersive system and X-ray powder diffraction. These results support the hypothesis that transparent game counters were obtained by remelting of fragments of common transparent artefacts. In contrast, opaque finds were probably produced using the glassy paste employed in the production of mosaic tesserae.

Roman glass manufacturing reached maximum output in the first to second centuries AD. In fact, Plinius, Martial, Juvenal and other Latin authors of these centuries spoke of abundant and growing glass production, as well as improvements in recycling processes. Pompeii, smothered by volcanic ash, represents a reliable example of the use and habits for this period; only in the Pompeii and Herculaneum excavations is it possible to observe in abundant detail the results of improvements in glassblowing techniques in the first century on Roman tables. Most archaeologists have focused their attention on near eastern production centres, considering Italian production to be of a lower standard. However, several reasons suggest the presence of glass manufacturing in Campania in the first century AD. The region known as Campania felix was not only the residence of renowned philosophers and emperors, but also one of the most thriving and active regions of

the Empire. Pozzuoli harbour represented the principal centre for the supply of foodstuffs and for the transit of goods shipped from Egypt and intended for Rome. Ships loaded with glass fragments and ingots also arrived, as mentioned by Cicerone in his writings. Pozzuoli seems to have been a famous glass production centre, as proven by the discovery of a glass furnace (Gialanella 1999). The presence of a glass production centre near Pozzuoli (or in general in Campania) and the great increase and spread of glass in this period and area is attested by several historical sources (Strabo, Geographia; Petronius, Satyricon). Among the glass finds of the Pompeii excavations, some hundreds of glassy paste objects were recovered and classified as 'game counters'. Plinius, in his Historia Naturalis, defined these items as the result of recycled glass remelting. Only few of these are transparent; most are opaque in a wide range of colours. Hence, they seem to represent a broad pattern of glass

From: MAGGETTI,M. & MESSIGA,B. (eds) 2006. Geomaterials in Cultural Heritage. Geological Society, London, Special Publications, 257, 175-186. 0305-8719/06/$15.00 © The Geological Society of London 2006.

176

R. ARLETTI ETAL.

production in the Roman age and their archaeometrical study is certainly of interest, especially concerning the use of colouring and opacifying agents. Coloured opaque glass is among the earliest glass in archaeological records (Newton & Davidson 1989), but these materials did not occur in significant quantities until the middle of the second millennium BC (Mass et al. 2002). Many samples of opaque glass have been analysed recently to identify and characterize the colouring and opacifying agents used, as well as the production technology (see, e.g. Brunet al. 1991; Mass et al. 2002; Mirti et al. 2002; Shortland 2002a). However, such artefacts have never been analysed so far; hence the aim of this study is twofold: (1) to characterize these glass samples so as to define their chemical and mineralogical composition; (2) to understand the technology used for their production. Concerning the latter point, the assertion of Plinius (i.e. the use of recycled glass) is questioned by the paucity of opaque vessels and glassware in Pompeii finds. Along with the game counters, other fragments of more common translucent glass, usually employed for the production of artefacts, and the fragment of one opaque green vessel were sampled, to make a comparison with the materials possibly used to produce game counters.

Experimental methods WDS-X-ray fluorescence analysis The chemical composition of major, minor and trace elements of transparent samples was obtained by wavelength-dispersive spectrometry-X-ray fluorescence (WDS-XRF). By contrast, because of an anomalously high content of some elements such as Pb, Cu, Co and Sb, the opaque samples were studied by electron microprobe analysis (EMPA). For this study an analytical procedure was set up with the purpose of obtaining precise and accurate chemical results for major, minor and trace elements using only 300 mg of sample (Arletti 2005; Arletti et al. 2005). The data were obtained using a Philips PW1480 XRF spectrometer, at the Earth Sciences Department of the University of Modena and Reggio Emilia. The glass was carefully pulverized and mixed with one small drop of organic glue, then pellets with boric acid as the support were prepared by applying a pressure of 7 ton m -z. The major and minor element (Si, Ti, AI, Mn, Mg, Fe, Ca K, Na) concentrations were computed using a program developed by Franzini

& Leoni (1972). The trace element (Nb, Zr, Y, St, Ce, Ba, La, Ni, Co, Cr, V, Sb, Zn, Cu, As, Pb) concentrations were computed using calibration curves (103x c.p.s./element concentrations) obtained after the measurements of 11 silica glass standards (GBW 01-11) of the Institute of Geophysical and Geochemical Exploitation (Langfang, China). To correct the matrix effect of the major constituents on the trace elements, the equations of Leoni & Saitta (1976) were applied. The analytical error for major and minor elements is <3%, whereas for trace elements it can be assumed to be < 10%. Table 1 reports (in italics) the results obtained by XRF for major and minor elements for transparent glass, and the trace element composition is reported in Table 2.

EMPA analysis Electron microprobe analysis was used to determine the chemical composition of only major and minor elements of most of the samples. Small glass fragments of almost 1 mm 3 were removed from the glass artefacts and mounted in epoxy resin. After preparation, the samples were polished using a series of diamond pastes from 6 to 1 p.m. To prevent charging, a carbon coating was applied to the polished section. The analyses were carried out using an ARLSEMQ electron microprobe equipped with four scanning wavelength spectrometers. The elements analysed were Si, Ti, AI, Mn, Mg, Fe, Ca, K, Na, Co, Sb, Cu and Pb. A series of certified natural minerals were employed as standards. The analyses were performed at 15 kV and 20 nA, using counting times of 5, 10, 5 s, respectively, on background-peak-background. To prevent the known migration of alkalis under the electron beam (Rinaldi 1981), a 30 ixm defocused beam was used. Several points were analysed on each sample and the mean value of all the measurements was taken. The results were processed for matrix effects using the Probe program (Donovan & Rivers 1990) and the oxide weight percent values were computed. Table 1 reports the results for major and minor elements obtained by EMPA on opaque glass.

Scanning electron microscopy Backscattered electron images (BSE) and energy-dispersive spectrometry (EDS) data were collected on polished samples, using a Philips XL40 electron scanning microscope equipped with an OXFORD-SATW EDS system at the Centro Interdipartimentale Grandi Strumenti of the University of Modena e Reggio Emilia. The analyses were performed

Game counter

Fragment of cup Game counter Game counter Game counter

Game counter

Fragment of beaker Fragment of cup Fragment of cup Fragment of bottle Fragment of bottle Game counter

Game counter

Game counter

Game counter

Fragment of plate

PM- 11313-5

PM-35117

PM-11313-6 PM-9361A PM- 12412A I b

PM-11313-7

PM-52658

PM- l 1313-8 b

PM-11313-9

PM-I1313-4

PM-11313-10

PM-AI3d4

Opaque white Opaque white Millefiori (opaque white) Opaque yellow Opaque yellow Opaque yellow Opaque yellow-green Opaque blue-green Opaque blue-green Opaque blue Opaque red Transparent green Transparent light blue Transparent light blue Transparent dark green Transparent dark blue Transparent dark blue Transparent light blue Millefiori (transparent purple) Transparent colourless Transparent brown Transparent black Transparent yellow

Colour

68.88

66.49

70.91

71.40

65.77

71.68

72.54

67.30

66.70

74.26

69.56

66.51 64.60 67.12

61.52

69.09

62.87 54.11 65.57 65.15

65.00 62.31 60.66

SiO2

0.05

0.07

0.09

0.06

0.07

0.07

0.06

0.08

0.17

0.05

0.05

0.06 0.15 0.12

2.02

0.12

0.10 0.05 0.07 0.18

0.11 0.05 0.05

TiO2

2.26

2.28

5.09

2.08

2.77

2.37

2.02

2.51

1.66

2.12

2.40

2.49 3.50 2.20

0.17

2.24

2.32 2.02 1.78 2.02

2.49 2.13 2.12

A1203

0.34

7.00

0.38

0.36

0.37

0.45

0.79

1.83

1.25

0.30

0.38

0.75 2.17 1,01

0.48

0.88

0.86 0.91 0,71 0.86

0.73 0.38 0.34

FeO

0.87

0.19

0.09

0.02

2.27

0.42

0.65

0.46

0.43

0.08

0.29

0.55 0.40 0.68

1.15

0.38

0.36 0.37 0.26 0.64

0.45 0.32 0.90

MnO

0.74

0.45

0.42

0.40

0.60

0.61

0.54

1.14

2.39

0.37

0.64

0.57 1.08 1.77

0.98

1.14

0.45 0.42 0.43 0.74

0.66 0.53 0.58

MgO

7.53

6.99

3.13

5.64

8.11

6.29

6.19

7.86

5.83

4.86

7.07

7.57 8.44 6.43

5.90

6.57

4.47 5.76 4.16 3.45

6.02 7.06 7.06

CaO

19.39

16.27

18.74

17.34

19.05

17.54

16.80

18.12

19.90

17.46

19.01

19.01 16.21 16.84

16.45

18.47

16.49 13.30 17.16 18.81

15.83 16.40 17.29

Na20

0.52

0,62

0.95

0.80

0.78

0.52

0.51

0.58

1.02

0.32

0,50

0.77 1.45 1,51

1.58

0.84

0.68 0.59 0.71 0,73

0.78 0.46 0.67

K20

t

~"

t

1.84

0.05

t

t

t

t

t

t

1.65 0.28 0.09

2.67

1.85

1.65 1.71 1.42 0.83

5.42 10.82 8.63

Sb205

t

~"

t

0.02

0.03

t

t

t

t

t

t

0.04 0.52 0.11

3.11

0.28

8.23 18.52 5.23 5.60

0.12 0.01 0.01

PbO

t

t

t

0.01

n.d.

t

t

t

t

t

t

0.10 1.80 n.d.

3.22

0.85

0.02 0.04 n.d. 0.27

0.08 0.00 n.d.

Cu20*

The results for the opaque or small samples were obtained by EPMA, whereas those for transparent sample (in italics) were obtained by WDS-XRF. n.d., below the detection limit. *Cu is reported as Cu20, from the EPMA output. t Reported in Table 2.

PM-52527

PM-52659

PM-35094

PM-35050

Game Game Game Game

PM-11313-2 PM-11313-3 b PM-12412AI a PM-3191A

counter counter counter counter

Game counter Game counter Game counter

Type

PM- 11313-1 PM-11313-3a PM- 11313-8 a

Sample

T a b l e 1. Weight % chemical analyses o f major and minor elements

t

t

t

n.d.

n.d.

t

t

t

t

t

t

0.03 n.d. n.d.

0.01

n.d.

n,d, n,d. n,d. n.d.

n.d. n.d. n.d.

CoO

100.58

100.36

99.80

98.10

99.79

99.95

100.10

99.88

99.35

99.82

99.90

100.10 100.60 97.88

99.26

102.71

98.50 97.80 97.50 99.28

97.69 100.47 98.31

Total

R. ARLETTI ETAL.

178

Table 2. Chemical analyses of trace elements for transparent samples in ppm by WDS-XRF

Nb Zr Y Sr Ce Ba La Ni Co Cr V Sb Pb As Zn Cu

D.L.

PM11313-4

50 50 20 23 5 54 5 5 3 5 6 2 6 50 5 10

n.d. n.d. n.d. 239 12 252 6 11 4 39 n.d. 67 13 n.d. 11 163

PM11313-7 n.d. n.d. n.d. 567 9 211 6 8 9 10 10 24 41 n.d. 7 190

PM11313-10

PMAI3d4

PM35094

PM52658

PM52659

PM35050

PM52527

n.d. 57 n.d. 480 13 180 4 42 33 10 9 1426 108 n.d. 9 302

n.d. n.d. n.d. 548 9 187 6 n.d. 4 7 n.d. n.d. 23 n.d. 9 198

n.d. 53 n.d. 773 n.d. 242 n.d. 21 419 11 19 9 101 n.d. 147 1560

n.d. n.d. n.d. 348 7 165 n.d. n.d. 4 n.d. n.d. 15 156 n.d. n.d. 166

n.d. n.d. n.d. 539 n.d. 200 6 39 932 8 15 39 49 n.d. 19 1605

81 93 n.d. 461 17 165 n.d. n.d. 11 14 20 57 501 n.d. 88 8929

n.d. n.d. n.d. 521 9 204 n.d. ll 14 10 13 920 79 n.d. 15 298

D.L., detection limit; n.d., below detection limit.

using an acceleration voltage of 25 kV. The BSE images were mainly collected on opaque glasses to highlight the presence of crystalline opacifying agents in the glass matrix, and the EDS analyses were run to obtain qualitative chemical analyses of the inclusions.

X-ray powder diffraction The X-ray diffraction (XRD) experiments were performed on the powdered opaque samples to detect and identify crystalline phases dispersed in the glass matrix. The analyses were carried out on a few milligrams of glass powder with a Philips PW1729 diffractometer with Bragg-Brentano geometry 0 - 2 0 and CuK,~ radiation using a zero background quartz holder. The spectra were collected from 5 to 80 ° 20 using a 0.02 ° 0 step and counting time of 4 s for each step. The XRD experiments on some very small fragments were performed using a Gandolfi camera, which can work on sample fragments as small as few i~m 3.

Results Major element chemistry For the overall sample set, the amounts of SiO2, CaO and Na20 are in the range of 54.1174.26%, 3.13-8.44% and 13.30-19.90%, respectively (see Table 1). These values are associated with rather low amounts of K20 and MgO. In only four samples does the amount of K20 exceed 1%, and it is always associated with a higher content of MgO. However, these

values are too low to suggest the use of plant ash as a source of alkalis. We can therefore assume that all the samples analysed are s i l i c a - s o d a - l i m e glass, typical of the Roman age, produced using natron as flux (Turner 1956; Sayre & Smith 1961; Henderson 1985). The small amount of A1203 is constant over the sample set and is typical of Roman glass, deriving from the feldspars in the sands used for vitrifying. Only sample PM11313-4 has an anomalously high value of A1203 (5.09%); this fact, along with the low content of CaO, seems to suggest the use of different sands for vitrifying. The differences found in the minor elements (Table 2) are mainly related to the colouring or opacifying agents, and will be discussed in detail below.

Opaque white game counters The chemical analyses of the opaque white samples revealed a high amount of Sb205, ranging from 5.42 to 10.82%. Several studies have reported that the opacity in white and blue glass is caused by small particles of calcium antimonate (Ca2Sb207 or CaSb206) within the matrix (see, e.g. Calvi et al. 1963; Mirti et al. 2002; Shortland 2002a). The BSE images of our samples show high amounts of particles with a mean atomic number higher than that of the glass matrix. These particles, whose dimensions are around a few microns, are well distributed in the glass matrix (Fig. 1). The number of particles per volume unit is variable and agglomerates are

GAME COUNTERS FROM POMPEII

<.)

I,>l t

179

,,

*~pec,h-lml I .fl-

i-......i

1

1 _ _

20pm

'

Electron Imaqe 1

]

Fig. 1. BSE image (a) and EDS spectra (b) for white sample PM-11313-1 showing the presence of small crystals of calcium antimonate. Spectrum 1: for particles (arrow); spectrum 2: for glassy matrix (squared area).

rare. In any case, it is always impossible to distinguish the morphological shape of the crystals. Comparison of the qualitative chemical composition of the spectra collected on the particles and the glass matrix revealed, as expected (i.e. Henderson 1985; Mass et al. 2002), a high level of Ca and Sb in the crystals, and the complete absence of Sb in the matrix. With the aim of identifying the mineralogical nature of the opacifier particles, X-ray powder diffraction (XRPD) patterns were collected. The results, shown in Figure 2a and b, allowed us to identify two different phases of calcium antimonate: a hexagonal phase with formula CaSb206 (Fig. 2a, sample PM 11313-3a) and an orthorhombic phase with formula Ca2Sb207 (Fig. 2b, samples PM 11313-1 and PM 11313-8). The peaks of both phases overlap with the typical background caused by the amorphous silicate matrix.

hexagonal phase, although the occurrence of the orthorhombic one cannot be excluded. The presence of euhedral crystals (see Fig. 3) clearly indicates that calcium antimonate was not added to the batch as a crystalline phase, but that it grew inside the glass matrix. Probably, after the addition of an Sb source, calcium

~-'

(a) f~ o

m

8 ~. i0.0 2-Theta,

Opaque blue and blue-green game counters For the blue and blue-green samples the BSE images, coupled with the EDS spectra, again demonstrated the presence of small crystals of calcium antimonate, although the EMPA analyses revealed lower amounts of Sb205 (1.65-2.67). Moreover, the number of particles found in these glass samples is considerably lower than in the white glass. This low number of crystallites explains the lack of diffraction peaks in the X-ray patterns collected on the blue and blue-green samples. The lack of diffraction pattern did not allow us to identify which of the two calcium antimonates is present in these cases. However, the magnified image of PM-11313-6 (Fig. 3) clearly shows that some crystals of calcium antimonate are hexagonal. This suggests the prevalence of the

20.0 d~

,

i

I

i

i

30.0

40,0

50.0

60.0

70.0

80.0

I

(b)

J_

1

1

I

i

t

I 40.0

I 50.0

I 60,0

I 70.0

I 80.0

o g q,

o m6

~

_ ._ k i0 0 2-Theta,

I _ I 20.0 30.0 d eg

Fig. 2. XRPD pattern for white samples (PM-11313-1 and PM- 11313-3a) revealing the presence of two calcium antimonate phases: (a) a hexagonal one; (b) an orihorhombic one.

180

R. ARLETTI ETAL. i

i

i

i

1

I

i

I 80.0

o

,=

§ 8

t 10.0 2 -'/~et a,

Fig. 3. Crystals of hexagonal calcium antimonate in a blue sample (PM-11313-6). antimonate phases crystallized, drawing out Ca, which was already present in the glass as a network stabilizer. This fact would explain why the lime contents in the opaque glass are not higher than those of the transparent ones. Opaque yellow and y e l l o w - g r e e n game counters The chemical analyses of the yellow and yellow-green game counters reveal, along with a rather high amount of Sb205, a significant presence of PbO, ranging from 5.23 to 18.52%. Opaque yellow glass of the Roman age is known to owe its colour and opacity to the presence of crystals of lead antimonates dispersed in the glass matrix (Mass et al. 2002; Shortland 2002a; Galli et al. 2004). This is consistent with the XRPD analyses of the Pompeii yellow

J

I

1

l

I

I

20.0

30.0

40.0

50.0

60,0

70.0

d~:J

Fig. 4. XRPD pattern for a yellow sample (PM-11313-2) showing characteristic reflection of lead antimonate Pb2Sb207.

game counters (Fig. 4), which clearly show the presence of several reflections corresponding to a lead antimonate with stoichiometry Pb2Sb2OT. The BSE image of Figure 5a shows rather large crystal aggregates with ragged edges, as a result of partial dissolution. This feature suggests that these phases did not crystallize inside the glass but were introduced as already formed crystals. The qualitative chemical composition of these crystals and of the surrounding matrix was determined by EDS. From the spectra (Fig. 5b) it is evident that the crystals contain both Pb and Sb and that the Pb peak is also present in the glassy matrix, as confirmed by the bulk chemical analysis obtained by EMPA. In fact, the quantitative chemical analyses, reported in Table 1, indicate a PbO/Sb205 ratio higher than that required by the stoichiometry of PbzSb207, confirming the presence of a lead excess in the glass. This finding strongly supports

gpadlum 1

(b)

lb

"pt','l ru m2

.

i

/-"

t

".peel..ml -

.

L~,,.

Fe

I,

30pm

~

Electron tmaae 1

J'3C~e 21~1d~l ~

$.171 ~ 8 ~

1

Fig. 5. BSE image (a) and EDS spectra (b) for a yellow sample (PM-11313-2), showing the presence of lead antimonate crystals and of a lead-rich matrix. Spectrum 1, for particles, is from the point at the end of the arrow; Spectrum 2, for the glassy matrix, is from the squared area.

key

GAME COUNTERS FROM POMPEII the hypothesis that crystals of lead antimonate were added to the batch along with another lead-bearing phase. Several hypotheses have been proposed regarding the origin of lead antimonate in yellow opaque glass, which will be discussed below. All the analyses performed on the yellowgreen sample (PM-3191A) revealed the same features as for the yellow ones, indicating that the colour of this sample is the result of the combined presence of yellow lead antimonate as the opacifier and of Cu as the blue-green colouring agent (see Table 1). O p a q u e red g a m e c o u n t e r

The colour of sample PM-9361A is mainly due to the presence of Cu (1.80% Cu20). Rare larger spherical particles with a CuS composition were revealed by BSE images and EDS analysis (Fig. 6). They probably represent the residue, not completely reacted, of the Cu-bearing raw material used to produce this red glass. Red opaque glasses of Renaissance, Medieval, and Roman age have been widely analysed and characterized in recent years. The colour and the opacity of these glasses seem to be due to the presence of minute particles of Cu ° or Cu j+ oxide within the glass matrix (Freestone 1987; Brill & Cahill 1988; Padovani et al. 2003). Our BSE images also reveal the presence of very small spherules, <0.5 Ixm in diameter, well dispersed in the matrix (Fig. 6). These particles certainly contain Cu, but their small dimensions prevented a precise chemical analysis. Their shape, however, strongly suggests they are Cu ° particles and not cuprite crystals, as Cu20 is more usually found in dendritic

Fig. 6. BSE image of a red sample (PM-9361A) showing the presence of small clusters of metallic copper, along with a large spherule of CuS at the centre of the picture.

1

I

t

i

181 I

I

I

I

i

i

o ~

u°

10,0 20.0 2 Theta, deg

t

f

I

I

I

30.0

40.0

50.0

60.0

70.0

80.0

Fig. 7. XRPD pattern of a red sample (PM-9361A) (*Cu° reflections).

aggregates (Brun et al. 1991). The XRD pattern, shown in Figure 7, confirmed this hypothesis, showing, along with a strong background caused by the glass matrix, two very weak diffraction peaks attributable to the strongest reflections ((111) and (200)) of Cu °. To clearly determine the oxidation state of Cu atoms, further spectroscopic studies were performed by our group (Arletti 2005), based on Cu K-edge absorption spectroscopy (XAFS). The results obtained from these investigations, and in particular from the fitting procedure of the EXAFS spectral region, indicate the presence of metallic clusters, along with a minor presence of Cu 1+, incorporated in the glass matrix. The presence of Cu ~+ in the glass network is not relevant for the colour of the sample, which is determined by the number and size of the metallic clusters (Nakai et al. 1999). This result represents an interesting affinity with the situation found by Padovani et al. (2003) for Renaissance lustre decorations, suggesting a temporal continuity in the basic technological operations during the glass production cycles for this type of red artefact. Further spectroscopic studies are in progress on our red glass sample. According to Nakai et aL (1999), the glass composition and melting conditions are the crucial parameters to control the Cu oxidation state in these artefacts. Tin, iron and lead, in appropriate oxidation states, could be used as reducing agents. Several Cu-rich Celtic enamels analysed by Brunet al. (1991) contained high levels of lead, introduced to avoid the oxidation of copper and to allow the precipitation of cuprite crystals. In our sample the lead content is rather low and the reducing role could be played by iron which is present in higher amounts (see Table 1).

R. ARLE'UI'I ETAL.

182

~ 10.0

I 20.0

I 30.0

I 40.0

I 50.0

I 60

0

[ 70.0

2-Theta, deg

Fig. 8. XRPD pattern for the opaque vessels, showing peaks attributable to Ca2Sb207 (labelled with *) and Pb2Sb207 (labelled +).

Opaque artefacts As discussed in the introduction, opaque artefacts were not very widespread in finds from Pompeii. Therefore, among the samples analysed, there is only one opaque blue-green vessel fragment (PM-35117), which, notwithstanding its very small dimensions, was classified as a fragment of an Isings 42 cup. Its chemical composition, reported in Table 1, shows a high amount of Sb205 (2.67%), PbO (3.11%) and Cu20 (3.22%). The XRPD analysis proved the presence of both Ca2Sb207 and Pb2Sb207 crystalline phases (Fig. 8), the former being the most abundant phase. The S E M - B S E images clearly confirmed the presence of a phase with mean atomic number higher than that of the glass matrix. The EDS analysis of these particles suggests that most of them are calcium antimonate particles: the conclusion is that the high amount of PbO found in the chemical analysis is mainly due to the matrix composition and that the dominant opacifying agent in this sample is calcium antimonate, accompanied by a subordinate presence of Pb2Sb207. The intense blue-green colour is otherwise due to the substantial amount of CuO present in this glass.

Transparent samples All the transparent game counters analysed show the typical compositions of Roman transparent glass as regards major elements. The main differences were detected in the minor elements and are related to the colour of the samples. In particular, the purple colour of the transparent portion of the Millefiori sample (PM- 11313-8) is probably due to the high amount of Mn 4+ in the glass matrix. As can be seen in Table 2, the

contents of Pb and Sb (which are, in general, the elements responsible for the opacity in our glass) are <0.5% in the transparent samples, with the exception of the colourless sample PM- 11313-9, which contains > 1% of Sb205. It is reasonable to suppose that, in this sample, Sb was added as decolorant. It is well known that antimony oxide was used as the main decolorant up to the first millennium BE, before being replaced in Roman times by manganese oxide. Nevertheless, it was also a common practice in Roman times to add Sb to a Mn-rich glass or to the batch to produce a colourless and more brilliant glass. Several colourless glasses containing both Sb and Mn have been found from between the first and the fourth centuries AD (Henderson 1985). High levels of Co (see Table 2) are present in the blue samples PM-52659 and PM 35094, always associated with Cu, whereas a high level of Cu, responsible for the dark green colour, is found in sample PM-35050. Some other samples (PM- 11313-7, PM-52527, PM-52658) show a light blue colour typical of common Roman glass, which is mainly due to Fe, present as an impurity in the initial batch (see Henderson 1985; Quartieri et al. 2002). The black and brown colours of samples PM 11313-10 and PM 11313-4, respectively, can again be ascribed to the presence of Fe 2+ (Arletti 2005). The deliberate use of high concentration of Fe has been confirmed starting from the second to first centuries BC to obtain very dark or black glass (Henderson 1985).

Discussion

Calcium antimonate-bearing samples The stoichiometry of the two calcium antimonates found in opaque white and blue glass can be expressed by the following oxide percentages: 14.77% CaO and 85.23% Sb205 (CaO/ Sb205 = 0.17); and 25.74% CaO and 74.26% Sb205 (CaO/Sb205 = 0.35) for CaSb206 and CazSb207, respectively. If calcium antimonate was added as an external component, then the lime content of these opaque samples should be higher than that of the translucent ones (Shortland 2002a). Let us assume that CazSb207 was added to sample PM-11313-3a (which contains 10.82% of Sb205) as a crystalline phase. In this case we should expect an increase of 3.6% in the CaO content with respect to the typical value found for silicasoda-lime transparent glass. A similar value (2.9%) should be expected for sample PM-I1313-8a, whereas the increase of CaO in

GAME COUNTERS FROM POMPEII sample PM- 11313- l, containing CaSb206 along with a lower level of Sb205 (5.42%), should be rather low (0.89%). In contrast, comparing the amount of CaO of these opaque samples with that found for the other translucent glass, no differences can be found. This implies that, as discussed above, Sb was added (probably as oxide) to the glass batch or to the raw glass to obtain an opacifying effect. In this case, calcium antimonate should be a 'neo-formation' phase, grown during the cooling of the glass after adding an Sb source and using the Ca content of the matrix. As previously discussed, blue and blue-green opaque glasses, contain lower levels of Sb than the white glass. This led to the formation of fewer well-crystallized crystals (Fig. 3a inset). The different colour of these two samples (PM-11313-5 and PM-11313-6) is related to the colouring agents employed: Co in the blue glass and Cu in the blue-green one. Co is the most powerful colorant used in ancient times. Because of its linear absorption coefficient, its colouring power is five times greater than that of other transition metals: to produce a deep blue colour only a few hundred ppm are needed. Figure 9 shows the values of Sb205 v. CaO of white and blue-green opaque samples, analysed in this study, compared with literature data for glass from two Egyptian localities (Malkada and Lisht; Mass et al. 2002) and from Sicily

•

12

B l u e - g r e e n G a m e C o u n t e r s - P o m p e i i (this w o r k ) Lisht ( M a s s e t a l . , 2 0 0 2 )

o 1;I.

M a l k a t a (Massetal., 2002) Blue M o s a i c T e s s e r a e - Sicily ( T r i s c a r i e t a l . , 2 0 0 5 )

,

10

W h i t e G a m e C o u n t e r s - P o m p e i i (this w o r k )

il -I-

,

,

,

i

,

,

,

,

,

,

,

i

,

,

,

,

,

,

,

,

,

,

i

....................................................................................................................................................

ill

8

......................

6

, ........................

~ ........................

,.........................

, .........................

, ......................

......................i............................................................................................................................ • a

i 4

oi

..................... - ................................................

i

r ........................ : ...............................................

~i

! i

o

%

i

oi

o

o

2 ...................... i.+.-,ri...o...~...~..O.m ~!-~. ~o -+

+

; O

t •

L i i 5

i

i

o

o o

~ , ~ i ...................... o

°

"o~

+

I i i | i 111 i i i , i I i ~,2/~~)~/).,'>lI'/>/;IA 6

7

8

9

I0

I I

i | II

CaO

Fig. 9. Sb205 v. CaO values for the calcium antimonate-bearing samples analysed in this study, and for others reported in the literature.

183

(Triscari et al. 2005). White and blue-green opaque samples are plotted with the same symbol for each of the Egyptian localities, because of the similarities found in Sb205 and CaO contents. This procedure was not possible for the Pompeii samples, as the Sb205 content is considerably higher for the white opaque glass than for the blue ones. The range reported at the bottom of the plot roughly corresponds to the CaO content generally found for translucent silica-soda-lime glass: almost all the samples analysed fit inside this range, so this clearly demonstrates that the opacifying phases were not added to the batch as calcium antimonate but probably grew inside the glass after the addition of an Sb source. Regarding the Sb205 contents, the plot clearly shows that the highest values are always related to the white Pompeian samples; this is consistent with the high number of very small crystals dispersed in the matrix of these samples, as is usual in a crystallization process in oversaturation conditions. Lead antimonate-bearing samples The PbO/Sb205 ratio required by the stoichiometry of lead antimonate Pb2Sb207 is 1.38. The PbO/Sb205 ratio found by the bulk chemical analyses of the yellow opaque samples is more than three times higher: 4.98 for PM-11313-2, 3.68 for PM-12412A1 and 10.83 for PM11313-3. This is consistent with the EDS chemical analysis, which clearly indicates the presence of Pb (and the absence of Sb) in the glass matrix. Figure 10 is a plot of PbO v. Sb205 values for the yellow opaque samples analysed in this study and for others reported in literature. The slope of the continuous line corresponds to the PbO/ Sb205 ratio in Pb2Sb2OT. In the overall sample set, Sb205 contents vary from 0.83 to 1.71%, whereas PbO shows a wider range of variation: from about 5.23% to 18.52%. Pompeii samples are the most spread out and show the highest content of Sb and rather high contents of Pb. Moreover, all the analysed glass deviate from the 'stoichiometric' line in showing a higher content of Pb. We can definitely affirm that the ragged edges seen on the PbzSb207 crystals in all the samples are the consequence of a partial dissolution, indicating that P b z S b 2 0 7 w a s not a neo-formation phase, but was added to the batch as a further component. Moreover, in the chemical analysis the presence of quantities of PbO higher than those required for the formation of PbzSb207, as well as the presence of Pb in the glass matrix of all the yellow opaque glasses, suggests that PbzSb207 was added to the glass under conditions of Pb excess.

184

R. ARLETFI ETAL. l • ol

GameC,ounte~'s - ~

(this~ork)

(Sho~and, 2002a) Malka~ (?¢~sset al., 2002) Lish (Massetal., 2002) Sicily(Triscadeta/., 2005) Malta

i

i

1

A 3:

15

g. ,

10

o g o

lap

=1.38

1

I

I

I

2

3

4

Sb20 s

Fig. 10. PbO v. Sb205 values for the yellow opaque samples analysedin this study, and for others reported in the literature.

Many hypotheses have been put forward on the origin of the yellow colour in opaque glass, but the issue is still unresolved. Mass et al. (2002) stated that high-Sb-litharge (Sb:PbO), deriving from the cupellation of argentiferous Egyptian galena and containing crystals of Pb2Sb207, might have been used as an opacifier for the yellow opaque glass both in the Bronze Age and in the Roman world. Although Pb2SbzO7 has certainly been found in some litharges, the hypothesis of Mass et al. raises some doubts. (1) The level of Sb in the Egyptian galena is not high enough for the formation of Pb2Sb207; moreover, the content of Zn found in the Pb2Sb207 glasses is far higher than Zn levels in Pb metals in Egypt (Rehren 2002). (2) The Pb isotope composition shows that the majority of Pb in Egyptian Pb2Sb207 glasses probably originated from mines in the Red Sea region, which have no significant amount of silver; moreover, there is no evidence that they were used for silver production (Shortland 2002b). Even if we cannot exclude that argentiferous galena could have been used for a few Egyptian glass varieties, the hypothesis proposed by Mass et al. seems to be inappropriate to Pompeian game counters. To justify the high amount of Pb, always found in the antimonate-containing yellow glass, another hypothesis has been proposed. As Pb was sometimes used to lower the softening

point of glass and to improve the working properties of the melt, it was proposed that natural Pb2Sb2OT, known as the mineral bindheimite, was added to a Pb-rich batch (Galli et al. 2004). The Pb excess would have led to a more fluid melt and consequently to more homogeneous glass. However, this does not explain why Pb is present only in the matrix of yellow opaque glass and not in other opaque glass. Moreover, several BSE images (Shortland 2002a), collected on opaque yellow samples, showed clumps of Pb2Sb207 dispersed in glass matrix strips richer in Pb, suggesting that both Pb and Pb2Sb207 crystals were added later to a solidifying glass. Furthermore, the natural phase bindheimite is, in general, not widely diffused in nature, hence we suggest that PbzSb207 might have been produced artificially and then added to a raw glass. In this sense, we are in agreement with the hypothesis formulated by Shortland (2002a), who, on the basis of the PbO/Sb~O5 ratio, proposed that Pb2Sb207 was produced by the combination of Pb and Sb ore minerals, with a Pb excess. The minerals most probably used would have been galena (PbS) for Pb, and stibnite (Sb2S3) for Sb. The PbO/ Sb205 ratio found in the yellow opaque glass closely corresponds to the eutectic composition; this means that a mixture of PbO and Sb205 in this proportion would be fully molten below 850 ~C and on cooling would produce PbO and Pb2Sb207. With a lower PbO/Sb205 ratio the reaction would have required higher temperatures and this could explain the excess of Pb in the yellow glass (Shortland 2002a). The yellow-green opaque glass (PM-3191A) could have been produced by mixing two glasses (yellow opaque and blue translucent), by adding a source of Cu to a yellow opaque glass, or by adding Pb2Sb207 to blue glass. The chemical analysis of green opaque glass shows a lower level of Pb and Sb compared with the yellow samples, and an amount of Cu very similar to that of the blue transparent samples. Hence, according to Shortland (2002a), it could be assumed that green opaque glass was produced by adding lead antimonate to a normal translucent blue glass. P r o d u c t i o n cycle

After the chemical and mineralogical characterization of several glassy game counters and artefacts, it is possible to formulate some hypotheses regarding the origin and the techniques employed for the game counter production. As seen before, the chemical composition of transparent game counters is very close (almost

GAME COUNTERS FROM POMPEII identical, for the major elements) to that found for the common transparent Roman glass. This led us to suppose that, as stated by Plinius (Historia Naturalis), the transparent game counters derive directly from the remelting of recycled glass, after the possible addition of colouring elements. As regards the opaque game counters the situation is more complex. As stated in the previous sections, it is difficult to hypothesize the recycling of opaque artefacts, as they are so rare in Pompeii finds. The other, much more realistic hypothesis is that the opaque game counters were produced by recycling the material used for the widely diffused mosaic tesserae. This could be a convincing hypothesis, as usually the type of colouring and opacifying agents found for mosaic tesserae (Galli et al. 2003, 2004) and in the game counters analysed in this study are almost the same. A detailed characterization of several glassy mosaic tesserae of different colours from the Pompeii excavation is in progress to provide a comparison between materials of the same age and provenance.

Conclusions Summarizing what has been presented above we can observe that the opacifying agents used in game counter production were: (1) calcium antimonates (CaeSb207 and/or CaSbeO6) for white, blue, and b l u e - g r e e n samples; (2) Pb2SbeO7 for yellow and y e l l o w - g r e e n ones; (3) metallic Cu for red ones. Co and Cu-bearing phases were added to glass opacified with antimonates to obtain colour hues from blue to green. Financial support was provided by Italian MIUR (COFIN 2004 'Scienza dei materiali antichi derivati da geomateriali: trasferire le conoscenze di base delle geoscienze allo studio di vetri e metalli'). The Centro Interdipartimentale Grandi Strumenti (CIGS) of the University of Modena and Reggio Emilia is acknowledged for the use of the SEM. The paper was greatly improved by the comments of two anonymous referees.

References ARLETTI, R. 2005. The ancient Roman glass: an archaeometrical investigation. PhD thesis, Universith degli Studi di Modena e Reggio Emilia. ARLETTI, R., GIORDANI, N., TARPINI, R. & VEZZALINI, G. 2005. Archaeometrical analysis of ancient glass from western Emilia Romagna (Italy) belonging to the Imperial Age. Annales du 162 Congrks de l'Association lnternationale pour l'Histoire du Verre, London 2003, 80-84. BRILL, R. H. & CAmLL, N. D. 1988. A red opaque glass from Sardis and some thoughts on red

185

opaques in general, Journal of Glass Study, 30, 16-27. BRUN, N., MAZEROLLES, L. & PERNOT, M. 1991. Microstructure of opaque red glass containing copper. Journal of Materials Science Letters, 10, 1418-1420. CALVI, M. C., TORNATI,M. & SCANDELLARI.,M. L. 1963. Ricerche tecnologiche. In: I Vetri Romani del museo di Aquileia. Associazione Nationale per Aquileia Ed, Aquileia. DONOVAN, J. J. & Rivers, M. L. 1990. PRSUPR--a PC-based automation and analysis software package for wavelength-dispersive electron-beam microanalysis. Microbeam Analysis, 66-68. FRANZINI, M. & LEONI, L. 1972. A full matrix correction in X-ray fluorescence analysis of rock samples. Atti della Societa Toscana di Scienze Naturali, Memorie, Serie A, 79, 7-22. FREESTONE, I. C. 1987. Composition and microstructure of early opaque red glass. In: BIMSON,M. & FREESTONE, I. C. (eds) Early Vitreous Materials. British Museum, London, 173-191. GALLI, A., MARTINI,M., MONTANARI,C. & SIBIL1A,E. 2003. The use of antimony and its implication for the luminescence properties of ancient mosaic tesserae. Journal of Non-Crystalline Solids, 323, 72-77. GALLI, S., MASTELLONI, M., PONTERIO, R., SABATINO, G. ~¢ TRISCARI, M. 2004. Raman and SEM + EDX techniques for characterization of colouting and opaquening agents in Roman mosaics glass tesserae. Journal of Raman Spectroscopy, 35, 622-627. GIALANELLA, C. 1999. Una fornace per il vetro a Puteoli. In: PICCIOLI, C. & SOGLIANO,F. (eds)// vetro in Italia meridionale e insulate. De Frede, Napoli, 151 - 160. HENDERSON,J. 1985. The raw materials of early glass production. Oxford Journal of Archaeology, 4, 267-291. LEONi, L. & SAITTA, M. 1976. X-ray fluorescence analysis of 29 trace elements in rock and mineral standards. Rendiconti della Societa ltaliana di Mineralogia e Petrologia, 32(2), 497-510. MASS, J. L., WYPYSKY,M. T. & STONE, R. E. 2002. Malkata and Lisht glassmaking technologies: toward a specific link between second millennium BE metallurgist and glassmaker. Archaeometry, 44(1), 67-82. MIRT1, P., DAVIT,P. & GULMINI,M. 2002. Colourants and opacifiers in seventh and eight century glass investigated by spectroscopic techniques. Analytical and Bioanalytical Chemistry, 372, 221-229. NAKAI, I., NUMAKO,C., HOSONO,H. & YAMASAKY,K. 1999. Origin of red color of Satsuma copper-ruby glass as determined by EXAFS and optical absorption spectroscopy. Journal of the American Ceramic Socie~, 82(3), 689-695. NEWTON, R. & DAVIDSON, S. 1989. Conservation of Glass. Butterworth, London. PADOVANI, S., SADA, C., MAZZOLDI, P. et al. 2003. Copper in glazes of Renaissance lustre pottery: nanoparticles, ions, and local environment. Journal of Applied Physics, 93, 158.

186

R. ARLETTI ET AL.

PLINIUS, Nan~ralis Historia, XXXVI(199). Einaudi, Turin. QUARTIERI, S., TRISCARI, M., SABATINO, G., BOSCHERIN1, F. & SANI, A. 2002. Fe and Mn K-edge XANES study of ancient Roman glasses. European Journal of Mineralogy, 14, 749-756. REHREN, T. 2002. Comment I on J. L. MASS, M. T. WYPYSKY, R. E. Stone, Malkata and Lisht glassmaking technologies: toward a specific link between second millennium BC metallurgist and glassmaker. Archaeometry, 44(1), 67-82. RINALDI, R. 1981. La microsonda elettronica. In: ARMIGLIATO, A. & VALDRE, U.(eds) Microscopia elettronica a scansione e microanalisi parte H microanalisi. Lo Scarabeo, Bologna. SAYRE, E. V., SMITH, R. V. 1961. Compositional categories of ancient glass. Science, 133, 18241826.

SHORTLAND, A.J. 2002a. The use of antimonate colorants in Early Egyptian glass. Archaeometry, 44(4), 517-530. SHORTLAND, A.J. 2002b. Comment II on J. L. Mass, M. T. Wypysky, R. E. Stone, Malkata and Lisht glassmaking technologies: toward a specific link between second millennium BC metallurgist and glassmaker. Archaeometry, 44(1), 67-82 (Reply, Archaeometr3,, 45(1), 185-198). TRISCARI, M., QUARTIERI, S., SABATINO, G., VEZZALINI, G., ARLETTI, R. & MASTELLONI, M. A. 2005. Analisi archeometrica di tessere musive in pasta vitrea da un pavimento di Lipari. Quaderni del Museo Regionale di Messina (in press). TURNER, W. E. S. 1956. Studies in ancient glasses and glassmaking processes. Part IV: The chemical composition of ancient glasses. Journal of the Society, of Glass Technology, 40, 162-186.

Pre-industrial glassmaking in the Swiss Jura: the refractory earth for the glassworks of Derriere Sairoche (ct. Bern, 1699-1714) G. E R A M O Department of Geosciences, Mineralogy and Petrography, University of Fribourg, Chemin du Mus~e, 6, CH-1700, Fribourg, Switzerland (e-mail: giacomo.eramo @unifr.ch) Abstract: Fragments of the melting furnace and several crucibles of the glassworks of

Derriere Sairoche are compared with local raw materials. Principal component analysis (PCA) based on the chemical composition and on the grain-size distribution of the archaeological and natural materials demonstrates that the analysed samples were made from the same raw material and that local clayey sands (Hupper, Sidrrolithique) were exploited. Availability in situ of good raw materials made tempering unnecessary. Their high melting point (c. 1600 °C) allowed good performance in service conditions at temperatures up to 1500 °C. Moreover, because of low Fe203tot concentrations, batch-glass contamination was avoided.

In the Middle Ages the spread of wood-ash glass production ('Waldglas') in northern Europe implied substantial changes from the natron glass produced in antiquity. The high CaO ( 1 0 - 2 0 w t % ) content of the former required higher melting temperatures (up to 1400 °C) than the 1000-1100 °C range sufficient for the ancient N a - C a glasses (Turner 1956; Cable & Smedley 1987; Cable 1998; Brill 1999; Henderson 2000; Stern & Gerber 2004). To reach temperatures up to 1400 °C, a more efficient pyrotechnology and better performing refractory materials were necessary (Charleston 1978; Cable 1998; Eramo 2005a). Recent studies on the pre-industrial glassworks of Derriere Sairoche (1699-1714) discussed some aspects of glass technology in the Bernese Jura (Gerber 2003; Stern & Gerber 2004; Eramo 2005a). In this area, dozens of glassworks were active during the second half of the 17th and the first half of the 18th century (Sveva Gai 1991; Sternini 1995). Here, glassmakers found pure quartz sand (see below) as well as extensive forests and streams to transport wood (Amweg 1941; Michel 1989; Gerber et al. 2002). Stern (1991) reported a C a - K composition for glasses from this area and made glass replicas using local quartz sand and wood ash (Stern & Gerber 2004). Several outcrops of pure quartz sand and refractory earth are historically known near Derrirre Sairoche (Schlaich 1934; Amweg 1941); however, to obtain a better understanding of the role of the raw

materials and their influence on local glass technology, an archaeometric characterization appears necessary. In a recent paper Eramo (2005b) showed that the crucible samples of Derriere Sairoche were not tempered with recycled crucibles and refractory fragments ('grog') as suggested in old glassmaking treatises. Nevertheless, some processing of the raw materials cannot be excluded (e.g. sand tempering). This paper attempts to prove, by multivariate analysis of the grain-size and chemical data, whether or not the unprocessed local raw materials (Hupper, Sidrrolithique) could be technologically suitable to produce the crucibles and the refractory samples, and whether or not there are compositional differences between crucibles and refractory samples for technological reasons.

The Sid6rolithique The name Sid6rolithique was introduced in the geological literature by Thurmann (1836) and is still used to indicate a complex geological unit deposited during the Eocene (Early Oligocene?) on the karstified surface of the Mesozoic limestones in the Jura region. The accumulation of different lithologies (i.e. kaolinitic clays, iron pisoliths, quartz sand, etc.) occurring in karstic pockets, or rarely as continuous beds (e.g. valley of Del6mont) (Fleury 1909; Schlaich 1934; yon Moos 1941; Aubert 1975; Pfirter 1997), marks the stratigraphic limit between Mesozoic limestones and Molasse sediments

From: MAGGETTI,M. & MESSIGA, B. (eds) 2006. Geomaterials in Cultural Heritage. Geological Society, London, Special Publications, 257, 187-199. 0305-8719/06/$15.00 @) The Geological Society of London 2006.

188

G. ERAMO

Alluvial deposits (Quaternary) Molasse (Cenozoic) Jura limestones (Mesozoic) Sid&olithique pockets (Eocene) ~t

Townor village Derriere Sairoche glassworks

Fig. 1. Schematic geological map of the Court area (shown by the black rectangle in the outline map of Switzerland), in the northern part of canton Bern (shaded area in the outline map). Location of the Hupper sand samples: A, Sur Fr~te; B, Lac Vert; C, Champoz-P. Mont Girod; D, For~t de B6role; E, Chfitelat; F, Monible-C6te; G, La Fuet; H, Souboz-Montaigu.

throughout the Jura region (Fig. 1). Generally, the outcrops are distributed along the flanks of the valleys of the Jura belt. As these terrains do not have lateral continuity, they may occur in several associations or some lithologies may be lacking. The Sid6rolithique is composed of red or yellow clays (rarely white, green or violet) called Bolus; levels rich in iron pisoliths (Bohnerz); quartz sands that can sometimes be clayey (Hupper) and calcareous conglomerate (Gompholithe) containing iron pisoliths. The chemical composition of Hupper varies as a function of clay content. A1203 concentrations up to 15 wt% for clay-rich samples were reported by De Quervain (1969). Hoffmann & Peters (1969) reported kaolinite as the major clay mineral ( 7 0 - 9 0 wt%) and illite and montmorillonite as minor components ( 0 - 1 0 wt%) in the grain-size fraction < 2 ~zm. Hupper layers rich

in clay were quoted as suitable for refractory materials by Fleury (1909). Their melting point exceeds 1500 C (von Moos 1941 ; Hoffmann & Peters 1969). On the other hand, the quartz sand (up to 99 wt% SiO2) was exploited since the Middle Ages as a raw material for glass (Fleury 1909; Amweg 1941; Kfindig et al. 1997).

Sampling strategy The glassworks of Derri6re Sairoche is located in the valley of Chaluet (Bernese Jura) with several outcrops of Hupper nearby (Fig. 1). The exploitation of some of these outcrops for glass, refractory and ceramic production has been described in the literature (Fleury 1909; Schlaich 1934; von Moos 1941; De Quervain 1969). To prove the utilization of natural raw materials suitable to produce the crucibles and the refractory

PRE-INDUSTRIAL GLASSMAKING, SWISS JURA

189

Table 1. Analysed samples of Hupper sand (Sid~rolithique) Sample

Coordinates

Locality

Mineral content

ER125 ER126 ER127 ER131 ER136 ER 137 ER138 ER139 ER140 ER141 ER248

591.510/231.200 591.510/231.200 591.510/231.200 591.075/233.150 591.075/233.150 591.075/233.150 591.075/233.150 591.075/233.150 590.300/233.625 590.300/233.625 579.620/235.230

Sur Fr~te Sur Fr6te Sur Fr~te Lac Vert Lac Vert Lac Vert Lac Vert Lac Vert Champoz-P. Mont Girod Champoz-P. Mont Girod For~t de B~role

ER249 ER250 ER251 ER252 ER253 ER254 ER255 ER256

581.370/235.510 581.370/235.510 582.140/235.490 582.140/235.490 580.100/232.700 580.100/232.700 586.750/235.875 586.700/236.000

Ch~telat Ch~telat Monible-C6te Monible-C6te La Fuet La Fuet Souboz-Montaigu Souboz-Montaigu

Qtz + Kin Qtz + Kin + Goe Qtz + Kin Qtz + Kin Qtz + K l n + (Cal) Qtz + Kin Qtz + Kin Qtz + Kin + Ill/Mus Qtz + Kin Qtz + Kin Qtz + Kin + Ill/Mus + Chl + Kf + P1 Qtz +Kln Qtz + Kin Qtz + Kin Qtz + Kin Qtz + Kin + Cal Qtz +Kln Qtz +Kln Qtz + Kin + (Cal)

Mineral abbreviationsas in Kretz (1983): Qtz, quartz; Kin,kaolinite;Cal, calcite;Goe, goethite;Chl, chlorite; Ill, illite;Mus, muscovite; Kf, potassium feldspar; PI, plagioclase.

material, only those outcrops having somewhat plastic materials were chosen for sampling (Fig. 1). In some cases it was possible to sample sediments of different grain-size distributions (Lac Vert, Souboz-Montaigu and Sur Fr~te). Nineteen samples of Hupper were collected (Table 1). The analytical methods used in this study and a discussion of the precision of the grain-size analyses are given in the Appendix.

Results

Hupper sand Petrography and mineralogy.

The clastic portion of the samples consists of mono- and, rarely, polycrystalline quartz. The grains are angular to subrounded (Fig. 2a). The argillaceous matrix consists of kaolinite (determined by X-ray diffraction; XRD) and is sometimes brown because of the presence of iron hydroxides and

Fig. 2. Optical microphotographs (5 x, plane-polarized light) of a Hupper sand sample (a, ER250) and of a crucible fragment (b, ER52).

190

G. ERAMO

oxides. Some calcite is present in ER136, 253 and 256, and ferruginous aggregates were detected in ER126 and 248. This latter sample contains small amounts of K-feldspar, plagioclase, muscovite-illite and chlorite (Table 1).

Chemistry. Bulk chemical compositions of the samples are characterized by high percentages of SiO2. Except for A1203, the other oxides are generally <1 wt%. On the whole, the concentrations of the trace elements are low (< 100 ppm), except for Zr (Table 2). Grain-size analysis (sieving). Table 3 shows grain-size data for Hupper sand. The cumulative frequency curves (Fig. 3) show the grain-size variability in the sampling area. Almost all samples are poor in coarse sand ( < - 1 4 ) and -1-04)) and show a large dispersion of fine sand (3-44)) and ' s i l t + c l a y ' (>44)) percentages. ER139 and 248 consist of >95 wt% of silt and clay. The samples from Lac Vert and Souboz-Montaigu show the widest grain-size variability in the same outcrop. Crucibles and refractory fragments As reported by Eramo (2005a,b) both the refractory and the crucible fragments are composed almost completely of SiO2 and A1203. Contents of Fe203tot and TiO2 are generally <1 wt% (Table 2). Monocrystalline and, rarely, polycrystalline quartz grains were detected as non-plastic inclusions originally present in the raw material (Fig. 2b). Quartz grains, partially thermally transformed to tridymite and cristobalite, are surrounded by a low-birefringent matrix composed of cristobalite and mullite. Although the XRD spectra show very low background, the presence of a glassy phase cannot be excluded.

Grain-size analysis (thin section). The results of the grain-size analysis in thin section of the crucible and furnace fragments are shown in Table 4. Both types of samples are characterized by low percentages of coarse sand and by an increase of standard deviation values for finer size classes. The cumulative frequency curves of crucible fragments are similar to one another (Fig. 4a) and show 2-34) and 3-44) values more dispersed than those of the refractory samples (Fig. 4b). The crucible samples ER23, 48 and 52 are richer in silt and clay than the others.

Data processing In general, the petrographical and chemical features of the analysed Hupper sand samples are consistent with those of the refractory and crucible samples reported by Eramo (2005a,b). In both natural and archaeological materials, monocrystalline quartz forms the non-plastic portion, whereas A1203 percentages are related to mullite or original kaolinite. Furthermore, the grain-size analyses carried out on the archaeological and natural materials show that most of the samples have similar grain-size distribution curves (Figs 3 and 4). A multivariate statistical analysis using both chemical and grain-size variables appeared useful to compare the Hupper sand with the crucibles and refractory samples. Principal component analysis (PCA) was carried out on the complete dataset (84 samples: 43 crucibles, 22 refractories, 19 Hupper sand) using chemical and grain-size variables, which have few missing values and higher variance (Table 5). The '<0.01' in the dataset were approximated to 0.01. As the 14 variables (SiO2, TiO2, A1203, Fe203tot, MgO, CaO, K20, Cr, Sr, Zr, 1-24), 2-34), 3-44) and >44)) used for the PCA are expressed in different units, standardization was necessary to ensure a similar order of magnitude and variance. The contributions to the total variance and the loadings of the first three principal components (PCs) are shown in Table 6. PC1 is very dominant and accounts for 41.78% of total variance. This component is characterized by negative loadings of SiO2 and sand fractions and by the association of A1203, K20, Cr and pan fraction. Although some associations occurring in PC1 are still present, PC2 features higher variances of 1-24) and 3-44). The 2-34) contribution to the total variance of PC2 and PC3 is very low compared with PC1, giving different information about the data structure. Whereas PC2 is characterized by high loadings of 1-24) and 3-44), PC3 features high loadings of Zr, MgO, A1203 and Sr. The component plot PC1 v. PC2 (Fig. 5a) shows strong positive correlations between K20 and the pan fraction, TiO2 and Zr, Fe203tot and MgO, and between SiO2 and 2-34). This last association of variables is negatively correlated with K20 and pan fraction and, to a lesser extent, with the other variables located in the positive quadrants of the diagram. The PC I v. PC3 diagram shows less obvious correlation, such as Z r - M g O or Sr-AI203 and C r - > 4 4 ) (Fig. 5b). These relations between variables may be interpreted as the chemical and grain-size signature of the studied materials. PC2 provides a more grain-size-sensitive

8.89 1.25

8.81 1.21

Crucibles* mean 90.08 0.81 cr 1.39 0.23

Refractory fragments mean 87.98 0.79 cr 1.49 0.24

0.60 0.26

0.43 0.13

0.33 6.52 0.93 0.53 1.83 0.23 0.32 1.24 0.08 0.10 3 0.63 0.42 0.30 0.43 0.73 0.17 0.45 0.13

0.01 0.00

0.01 0.00

0.10 0.35 0.17 0.05 1.52 0.05 0.07 0.32 0.02 0.03 0.56 0.15 0.13 0.09 0.07 3.72 0.08 0.29 0.51

0.14 0.36 0.11 0.12

0.12 0.21 0.05 0.07

<0.01 0.04 0.03 0.26 <0.01 0.14 <0.01 0.06 <0.01 0.88 <0.01 0.09 <0.01 0.19 <0.01 0.28 <0.01 0.10 <0.01 0.10 0.05 0.85 <0.01 0.19 <0.01 0.14 <0.01 <0.01 <0.01 0.10 0.03 0.24 <0.01 0.08 <0.01 0.07 <0.01 0.12

*Chemical data for crucibles from Eramo (2005a). t Chemical data for refractoryfragmentsfrom Eramo (2005b). LOI, loss on ignition.

9.78 10.4 3.95 6.95 12.47 3.40 3.60 26.11 2.85 3.80 9.49 8.35 8.59 13.73 5.31 4.22 3.17 14.82 3.17

85.33 79.5 92.49 87.71 78.93 91.46 93.94 65.31 95.43 94.54 80.23 88.65 85.83 81.48 90.54 89.55 92.66 80.16 94.65

1.03 0.67 0.45 0.47 0.77 0.25 0.26 1.25 0.24 0.91 1.07 0.47 0.47 0.78 0.34 0.46 0.32 0.75 0.22

ER125 ERI26 ER127 ER131 ER136 ERI37 ER138 ER139 ER140 ERI41 ER248 ER249 ER250 ER251 ER252 ER253 ER254 ER255 ER256

0.13 0.19

0.19 0.19 0.21 0.11

0.29 0.16

96.68 98.26 98.29 95.86 98.34 95.64 98.68 97.28 98.77 99.54 98.28 98.89 96.19 96.50 96.96 99.38 96.58 96.94 98.85

3.64 5.14 2.36 2.65 4.70 1.24 1.14 7.32 0.94 1.25 2.68 2.81 2.88 5.05 2.19 4.61 1.78 5.30 1.68

0.03 0.01

99.07 0.15 0.58 0.07

0.02 101.08 0.14 0.03 0.39 0.06

<0.01 0.02 <0.01 <0.01 0.37 0.08 <0.01 0.08 0.03 <0.01 0.05 <0.01 <0.01 1 . 7 7 0.07 <0.01 0.14 <0.01 <0.01 0.26 0.01 0.07 2.29 0.26 <0.01 <0.01 0.01 <0.01 <0.01 0.01 0.84 1 . 7 5 0.29 <0.01 0.36 0.04 0.22 0.31 0.02 <0.01 0.04 0.03 0.04 0.08 0.02 <0.01 0.16 0.2 <0.01 0.05 0.02 <0.01 0.32 0.03 <0.01 <0.01 0.02

14.45 7.60

59.21 15.09

<12 64 <12 14 111 <12 34 396 <12 <12 256 47 36 <12 <12 <12 <12 51 <12

62.77 11.34

70.93 8.89

61 97 34 47 91 11 16 121 10 20 112 65 57 71 26 28 <5 73 13

24 14 9 11 16 7 7 21 9 26 19 9 11 18 10 13 9 13 7

15 40 19 20 68 8 12 28 8 9 39 17 12 16 13 14 9 5 5

7.36 15.45 25.64 9.59 4.04 14.72

2.86 17.58 30.07 2.66 4.22 12.86

<2 7 <2 <2 13 <2 3 77 7 19 27 <2 <2 <2 <2 <2 <2 <2 <2

6.91 6.23

6.86 2.13

14 27 14 10 20 12 10 114 13 15 26 18 9 27 13 16 8 20 11

26 107 35 23 58 26 23 328 16 17 85 35 32 38 24 320 52 61 12

22 22 17 17 15 15 15 27 15 16 28 23 15 22 17 18 14 19 14

3 49 8 12 96 17 14 35 3 <2 84 13 9 4 7 16 3 5 <2

304 263 267 222 241 185 133 255 275 329 596 180 193 333 199 226 161 163 177

10.05 10.34

29.77 17.32 14.41 232.82 7.08 1.49 25.41 32.83

12.91 37.79 12.84 27.56 307.88 6.69 44.16 1 . 7 3 88.57 40.49

<3 29 4 <3 118 <3 17 119 <3 <3 89 23 12 <3 <3 16 <3 11 <3

Sample SiO2 TiO2 A1203 Fe203tot MnO MgO CaO Na20 K20 P205 Sum LOI Ba Cr Cu Nb Ni Pb Rb Sr Y Zn Zr (wt%) (wt%) (wt%) (wt%) (wt%) (wt%) (wt%) (wt%) (wt%) (wt%) (wt%) (wt%) (ppm) (ppm) (ppm) (ppm) (ppm) (ppm) (ppm) (ppm) (ppm) (ppm) (ppm)

Table 2. Chemical composition of the Hupper sand samples and the means and standard deviations for the crucibles (n = 43) and the refractory fragments (n = 22)

192

G. ERAMO

Table 3. Grain-size data for the Hupper sand samples by sieving (wt%) 4): mm:

<- 1 >2

- 1-0 1-2

0-1 0.5-1

1-2 0.25-0.5

2-3 0.125-0.25

3-4 0.063-0.125

>4 <0.063

ER 125 ER 126 ER 127 ER131 ER 136 ERI37 ER138 ER139 ER140 ER 141 ER248 ER249 ER250 ER251 ER252 ER253 ER254 ER255 ER256

0.00 7.47 0.54 0.14 0.78 0.02 0.00 0.00 0.02 0.00 0.04 0.34 0.30 0.90 0.08 3.76 0.00 0.30 1.68

0.23 0.46 0.05 0.32 0.02 0.27 0.18 0.04 0.22 0.02 0.04 0.80 0.86 0.78 0.82 0.32 0.02 0.32 0.84

1.65 1.45 2.80 1.38 0.23 3.97 2.34 0.22 3.24 i .22 0.06 4.28 3.46 1.06 4.20 1.96 1.00 0.82 4.60

9.35 9.30 16.79 6.48 8.07 26.09 25.61 0.85 35.81 23.39 0.30 16.76 15.54 7.46 16.38 7.30 5.48 2.62 20.70

22.19 19.09 28.97 24.76 14.72 38.81 39.18 1.50 21.84 34.85 0.98 25.76 28.96 33.14 28.74 16.86 15.32 21.06 35. ! 0

21.33 9.40 21.70 1.22 9.01 15.77 15.59 2.56 16.50 23.37 1.96 14.40 14.58 19.58 19.38 22.72 16.20 8.14 17.88

45.24 52.82 29.16 65.71 67.18 15.07 17.10 94.84 22.37 17.16 96.62 37.66 36.30 37.08 30.40 47.08 61.98 66.74 19.20

fingerprint of the dataset, whereas PC3 is rather chemistry-sensitive. The A1203, K20 and pan fraction positive correlation reflects the presence of these two oxides in clay minerals, whereas those of CaO, MgO and Sr may be explained as a co-occurrence in minor amounts of calcareous clasts that are present. Although PC1, PC2 and PC3 account only for 66.07% of the total variance, their bivariate plots show a close distribution and a complete overlap of the refractory and crucibles data points (Fig. 5c and d). In contrast, Hupper sand data points are more dispersed and only ERI25 and 251 fall in the region of crucible and refractory fragments if PC I and PC2 are considered (Fig. 5c), whereas in the PC1 v. PC3 plot, ER131, 249, 250 and 254 also fit the archaeological materials (Fig. 5d). According to the position of the data points, it can be inferred that the archaeological materials are characterized by a high content of fine quartz sand and a low content of clay, Fe-oxides and calcareous fragments. Summarizing, a large number of natural samples fit the archaeological materials from a chemical point of view, whereas the grain-size distribution is a more stringent factor to fit the archaeological materials.

Discussion and conclusions Grain-size analysis by sieving and by point counting on natural and archaeological samples, respectively, added the grain-size distribution variable as a tool not commonly used in provenance studies of ceramic materials.

Multivariate statistics involving chemical and grain-size variables revealed that the refractory and crucible samples analysed were made with the same raw materials. Chemical, petrographical and grain-size characteristics of the Hupper sand samples collected in the proximity of Derriere Sairoche show their compatibility with the crucibles and refractory fragments. As stated in the introduction there is historical evidence for exploitation of refractory earth and pure quartz sand in the Berner Jura. Although it was not possible to locate exactly the old pits, owing to their lack of accessibility, it could be shown that the analysed samples cover a wide grain-size and chemical range. Only ER125 and ER251 are really consistent with the archaeological materials if the most significant PCs are taken into account. Sur Fr&e (ERI25) is the Hupper deposit closest to the glassworks, whereas Monible-Crte lies about 15 km away. Hence it is reasonable to consider Sur Fr&e as the most probable source of clayey sand. However, the occurrence of other suitable raw materials at a greater distance indicates that these features may be found in several places in the area. It must be kept in mind that the heterogeneity of Hupper sand and its stratigraphical position made possible the occurrence of good raw material almost everywhere in the Swiss Jura. Furthermore, availability in situ of good clayey sand made any further treatment unnecessary. The absence of recycled refractory and crucible fragments in refractory and crucible

P R E - I N D U S T R I A L G L A S S M A K I N G , SWISS J U R A Sur Fr~te (A)

Ch&telat (E)

100

100

ER126

80

"

o~" 8O

ER127

/

'

~" >~ 40

193

.................

60

..... / ' / >~ 40

.

.

.

.

.

3

.

2o

L) i

-1

0

1 2 3 Grain size (~)

4

-1

PAN

gt.-

8O

60

•

2

3

4

PAN

Monible-COte (F) , -

•

1

Grain size ((~)

Lac Vert (B)

•

0

100

_,,41 . , ~

ER131

........ : : . ~".I~; ,t" •.: .... : ........ : ~ I ~ ' ~"

ER2511

t r

/

8o

g,

~ 60 (1) ~ 40

4o

~ ~o

2 -1

0

1 2 3 Grain size (~)

4

Z

2o

-1

PAN

0

1 2 3 Grain size (~)

4

PAN

La Fuet (G)

Champoz - P. Mont Girod (C) 100 o~

-

/7"

ER140

8o

-

o~ ~. 80

-

. -

ER253

ER254

.... / I

/

I

60

~ 4o

>= 40

"5

20

8 -1

0

1 2 3 Grain size (~)

4

-1

PAN

ForSt de Berole (D)

0

1 2 3 Grain size (~)

4

PAN

Souboz-Montaigu(H)

100

100 ER248

ER255 4 ER256// / /

8o

8o

~ ~o g

~

g

60

.1= ~ 4o

~ 4o

~ ~°

~ O ....

0 Grain size (~)

/

20 0

-1

0

1 2 3 Grain size (~)

Fig. 3. Cumulative grain-size frequency curves of the Hupper sand samples (sieve analysis).

,

,

4

PAN

194

G. ERAMO

Table 4. Grain-size data for the refractor)." and crucible samples by thin-section analysis (vol% ) 05: mm:

< - 1 >2

- 1-0 1-2

0-1 0.5 - 1

1-2 0.25 -0.5

2-3 0.125 -0.25

3-4 0.063 -0.125

>4 < 0.063

0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00

0.00 0.09 0.00 0.77 0.00 0.19 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.20 0.00 0.00 0.25 0.00 0.00 0.00 0.27 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.72 1.71 0.00 0.00 0.00 0.00 0.00 0.00 0.10 0.00 0.00 0.00 0.10 0.30

3.92 1.04 1.71 1.93 2.25 4.08 2.96 4.64 3.50 1.38 2.64 1.88 1.09 0.39 0.39 2.15 0.86 2.75 3.96 1.99 1.43 0.98 1.43 0.98 1.28 3.57 0.51 2.51 1.62 2.66 2.91 0.72 0.57 0.91 1.48 1.69 1.47 1.98 1.13 0.93 0.56 1.19 0.97 1.84 1.11

8.81 12.16 7.62 10.89 13.30 6.21 11.50 12.96 8.47 10.80 7.91 4.33 3.83 7.56 5.10 10.67 6.39 9.17 9.66 7.40 6.51 5.69 9.17 3.54 4.95 5.35 8.81 5.20 7.94 6.45 4.73 4.15 7.60 6.92 3.87 4.14 7.56 5.49 7.16 8.55 6.40 5.25 4.65 7.32 2.58

16.15 22.51 9.14 27.12 24.75 21.17 23.98 28.43 26.34 33.45 21.66 29.94 25.32 26.16 23.92 27.25 21.59 21.83 28.15 22.56 21.75 26.67 33.05 25.20 27.11 23.53 22.37 16.67 21.48 22.77 20.18 14.26 25.67 26.05 29.52 21.09 29.46 28.51 26.74 31.12 19.59 21.02 25.78 24.21 4.80

26.10 26.50 12.95 19.32 14.85 31.26 17.48 23.98 29.10 23.43 23.16 23.35 23.68 26.16 41.18 19.86 42.31 29.54 18.14 34.84 31.11 32.55 22.79 36.61 33.88 27.81 32.88 20.79 25.27 32.26 24.00 19.49 33.08 32.24 33.76 25.05 24.79 39.02 32.39 25.11 39.36 37.63 38.57 28.08 7.29

45.02 37.70 68.57 39.98 44.85 37.09 44.08 29.98 32.60 30.93 44.63 40.49 46.08 39.73 29.41 39.87 28.84 36.70 39.84 33.21 39.21 34.12 33.29 33.66 32.78 39.75 35.42 54.84 43.68 35.86 48.18 60.65 31.37 33.88 31.37 48.02 36.72 25.00 32.58 34.20 34.09 34.92 30.04 38.45 8.53

0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00

0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00

2.23 2.11 1.47 2.64 4.85 1.60 1.20 1.52 1.51 0.53

4.28 6.01 7.89 6.78 5.04 8.53 5.60 7.77 8.47 5.49

19.74 22.24 30.09 28.44 30.22 21.31 22.40 21.79 22.22 23.72

31.47 30.68 28.81 37.66 32.09 22.91 21.80 24.83 24.11 30.27

42.27 38.96 31.74 24.48 27.80 45.65 49.00 44.09 43.69 40.00

Crucible fragments ER21 ER22 ER23 ER24 ER25 ER26 ER27 ER28 ER29 ER30 ER31 ER32 ER33 ER34 ER35 ER36 ER37 ER38 ER39 ER40 ER41 ER42 ER43 ER44 ER45 ER46 ER47 ER48 ER49 ER50 ER51 ER52 ER53 ER54 ER55 ER56 ER57 ER58 ER59 ER60 ER61 ER62 ER65 mean cr

Refractory fragments ER63 ER64 ER66 ER67 ER68 ER69 ER85 ER86 ER87 ER88

(Continued)

PRE-INDUSTRIAL GLASSMAKING, SWISS JURA

195

Table 4. Continued

4': mm: ER89 ER90 ER91 ERI02 ER103 ER267 ER276 ER277 ER278 ER279 ER280 ER281 mean O"

< - 1 >2

- 1-0 1-2

0-1 0.5-1

1-2 0.25-0.5

0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00

0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00

1.49

0.00 0.00 0.59 1.91 0.95 0.36 0.54 0.70 3.02 0.55 3.91 1.53 1.24

100

Crucibles (n = 43) ~

80

¢-

~ 60 ~ 4o E 20 O

a

-'1

6

i

½

3

4

PAN

2-3 0.125-0.25

3-4 0.063-0.125

>4 <0.063

5.65

21.99

25.65 20.43 27.50 27.48 33.21 20.91 20.75 22.26 25.66 26.06 26.82 24.59 3.74

25.71 41.14 38.72 43.42 42.37 34.92 31.64 31.13 36.35 40.00 41.77 35.75 33.07 6.66

45.17

3.14 1.17 2.55 5.53 7.44 4.55 4.83 5.04 4.34 4.62 4.84 5.43 1.87

30.07 39.69 25.93 22.71 23.47 42.55 42.75 35.65 26.98 26.99 28.68 35.38 8.54

samples (Eramo 2005b), as reported by old glassmaking treatises, suggests that this practice was not economically and technologically relevant because of the abundance of suitable refractory earth deposits. The differences between crucibles and refractory samples are minimal, indicating that the natural raw material was considered technologically valid for both. They are characterized by fine monocrystalline quartz sand, which minimized the thermal expansion problems and gave more stability to the artefacts (Htibner 1991). Normative calculation of the original mineralogical compositions for the crucible fragments indicates about 2 0 w t % of kaolinite (Eramo 2005b). This clay mineral supplied enough plasticity to the raw material to form the crucibles and bricks, and to be applied as a

Grain size (¢) 100

i/

Refractory (n = 22) ~

Table 5. PCA of the archaeological and natural samples: eigenvalues of the correlation matrix

~ /

Component

8o

Eigenvalue

% of variance

Cumulative %

5.85 2.00 1.40 1.33 0.91 0.68 0.64 0.36 0.33 0.25 0.15 0.07 0.03 0.00

41.78 14.31 9.97 9.53 6.49 4.84 4.61 2.56 2.32 1.77 1.10 0.51 0.18 0.02

41.78 56.09 66.07 75.59 82.09 86.92 91.53 94.09 96.42 98.18 99.29 99.79 99.98 100.00

O f..

® "I

,/

60

ao 4 0

E 20 o

b

0

-~

o

1 2 g Grain size (¢)

,i

P~,N

Fig. 4. Cumulative grain-size frequency curves of crucible (a) and refractory samples (b) obtained by thin-section analysis.

1 2 3 4 5 6 7 8 9 10 11 12 13 14

196

G. ERAMO

Table 6. PCA of the archaeological and natural

samples: loadings of the first three PC PCI SiO2 TiO2 A1203 Fe203tot MgO CaO K20 Cr Sr Zr 205 305 405 pan

1.00

PC2

-0.834 0.596 0.759 0.535 0.592 0.234 0.785 0.788 0.545 0.448 - 0.547 - 0.790 - 0.374 0.856

-

-

PC3

0.113 0.552 0.245 0.266 0.273 0.456 0.188 0.421 0.361 0.409 0.536 0.066 0.654 0.206

0.276 0.201 -0.384 0.269 0.517 - 0.297 0.147 0.005 - 0.368 0.608 0.353 0.057 - 0.222 - 0.017

plaster in the m e l t i n g c h a m b e r . M o r e o v e r , its l o w s h r i n k a g e r e d u c e d the f o r m a t i o n o f c r a c k s in the artefact, i n c r e a s i n g the m e c h a n i c a l resistance. S u c h a c o m p o s i t i o n (i.e. Qtz 8 0 w t % + K i n 20 w t % ) has a s o f t e n i n g point o f a b o u t 1600 °C ( A r a m a k y & R o y 1962) and h e n c e g u a r a n t e e s g o o d r e f r a c t o r y b e h a v i o u r in s e r v i c e c o n d i t i o n s (up to 1500 :C). A low Fe203tot c o n t e n t is v e r y i m p o r t a n t for r e f r a c t o r y m a t e r i a l s u s e d in glassm a k i n g . E v e n a f e w p e r c e n t o f Fe203tot m a y c o m p r o m i s e the c o l o u r o f glass and, o f course, l o w e r the e u t e c t i c point o f the r e f r a c t o r y materials.

This paper is part of a PhD thesis in Archaeometry at the University of Fribourg (Switzerland). I am grateful to M. Maggetti and G. Thierrin-Michael for their continuous support and guidance. C. Gerber and N. Stork

1.00

"_1

bl

4,

•

0.50-

.go r Zr

Ti02

0.50' ~Zr~

1~Cr

• 20 Si~..~ ~

FezO3tot

T'Oz K,O

e',,I

t,,J o.oo. 0.

,

a,. 000'

/ -0.50"

\

-/2,

-1.00 -1.00

"--;',0o

l

-0.50

Sr

-050'

Lo |

O00

0.50

100

-100 -100

i

|

-050

i

000

PC1

AIzO3

050

1.00

PC1

cl A A ER249~

~ER251

•

(~/

•

ER254~ER125 ER250~'.~

~ER251

A

4

-2

2

PC1

•

ER;a~J • •

,

.

-2

0

2

4

PC1

Fig. 5. The component plots of PC 1 v. PC2 (a) and PC 1 v. PC3 (b) showing the contributions to the PC variance and the correlations between variables (see text for details). Scatter plots of the PC scores using PC 1 v. PC2 and PC 1 v. PC3 are shown in (c) and (d), respectively. In (c) only two Hupper sand samples (ER125 and 251) fit the crucible and refractory samples, whereas in (d) four other samples (ERI31, 249, 250 and 254) are compatible with the archaeological materials. White area, crucibles; grey area, refractory samples; &, Hupper sand.

PRE-INDUSTRIAL GLASSMAKING, SWISS JURA

197

are gratefully acknowledged for their help in surveying and sampling. I am indebted to V. Serneels (University of Fribourg) for XRF analyses.

d i f f r a c t o m e t e r with Cu-K~ radiation at 40 k V and 40 m A (step angle o f 0.02 °, 20 f r o m 2 ° to 65 °, m e a s u r i n g time 1 s per step).

Appendix

Loss on ignition (LOI)

Sample weight

Dry p o w d e r e d s a m p l e (3 g) was c a l c i n e d at 1000 °C for 1 h and w e i g h e d to d e t e r m i n e the LOI.

The Hupper sand samples were quartered until a 75 g portion was obtained. This portion was

used for sieving (50 g), for thin-section preparation (c. 5 g) and for X R D and X R F (c. 20 g, powdered).

X-ray fluorescence (XRF) Analyses were carried out on glassy tablets, prepared by melting 0.700 g of calcined samples,

Petrographic analysis Five s a m p l e s o f H u p p e r sand (ER125, 136, 248, 250 and 255) w e r e i m p r e g n a t e d with e p o x y resin to obtain thin sections, w h i c h were analysed u n d e r a Carl Zeiss Standard polarizing m i c r o s c o p e .

X-ray diffraction (XRD)

0.350 g of Li fluoride and 6.650 g of Li tetraborate at 1150 °C in a Pt crucible. Bulk chemical analyses for major and trace elements were performed using a Philips PW 2400 XRF spectrometer equipped with a rhodium X-ray tube. As the standards used do not cover the very

high percentages of SiO2 in the samples,

T h e mineral c o m p o s i t i o n was resolved by X R D analyses carried out on a Philips P W 1 8 0 0

d e v i a t i o n s up to 4 wt% f r o m 100 w t % o c c u r (Table 2).

Table 7. Precision of the grain-size analyses by point counting (pc) and sieving (s) Oh: mm:

2-3 0.125 -0.25

3-4 0.063 -0.125

>4 < 0.063

7.00 1.03

15.16 0.39

19.48 0.93

57.49 1.25

1.65 0.52

9.35 0.35

22.19 2.67

21.33 1.43

45.24 3.02

0.00 0.00

0.06 0.10

2.98 0.57

16.59 0.23

10.98 0.21

69.39 0.81

0.78 0.08

0.02 0.01

0.23 0.01

8.07 2.45

14.72 0.89

9.01 1.33

67.18 1.93

0.00 0.00

0.00 0.00

0.00 0.00

1.56 0.47

3.19 0.91

1.72 0.32

93.53 0.71

0.04 0.01

0.04 0.02

0.06 0.01

0.3 0.65

0.98 0.80

1.96 0.97

96.62 1.23

0.00 0.00

0.00 0.00

3.78 0.71

12.91 2.01

30.11 1.55

19.15 2.55

34.05 2.93

0.27 0.15

0.85 0.20

4.00 0.62

16.44 1.24

27.58 1.71

12.99 1.82

37.88 2.24

0.00 0.00

0.00 0.00

1.46 0.12

2.61 0.57

18.54 0.43

1 1.87 0.67

65.52 0.36

0.23 0.08

0.64 0.28

1.03 0.18

2.98 0.37

17.87 2.79

7.99 0.46

69.25 2.44

<- 1 >2

- 1-0 1-2

0-1 0.5-1

0.00 0.00

0.00 0.00

0.87 0.58

0.00 0.00

0.23 0.15

0.00 0.00

1-2 0.25-0.5

ER125pc mean o-

ER125s mean o-

ER136pc mean

ER136s mean er

ER248pc mean cr

ER248s mean er

ER250pc mean cr

ER250s mean o-

ER255pc mean cr

ER255s mean cr

Means of the percentagesof three-time repeated analyses and the standard deviations are shown.

198

G. ERAMO

100 • O • ~7 •

=_

- -

~' 60

in & values (Tables 3 and 4) and are represented by cumulative frequency curves (Figs 3 and 4).

//"~/ // / // // / /-'" // / //"

ER125 ER136 ER248 ER250 ER255

./ / .// .. -" ~ , ~ ' . . -" "

regr~..~o. ,o~

- - - - - 95°/*confidence

// / *"//./" // // /

~/

Precision o f the grain-size analyses

~ 4o o

g

20 ..~ :-. ~

y = 0.2241 + 0.98,43x r2=0.98

'IIcl

20

40

60

80

100

Frequency (vol. %) thin section

Fig. 6. Linear regression of the grain-size frequencies determined on the five test samples by point counting in thin sections and sieve analysis. The standard deviation bars based on the three repeat measurements are shown. Dotted lines represent the 95% confidence limits around the regression line.

The two methods provide grain-size frequencies from weight percentages (sieving) and from number of counts (thin section). The grain-size frequencies obtained by the two methods on the five test samples are plotted against each other in Figure 6. The correlation coefficient (r 2 - 0.98) is highly significant and only one out of 35 observations lies outside the 95% confidence limits. Such a result shows that the grain-size analysis data in thin section have a precision comparable at the 95% level of significance with that of sieving. Underestimation of the particle size in thin section as a result of a sectioning effect (Krumbein 1935) was not relevant in PCA because of standardization of variables.

Grain-size analysis by sieving Grain-size distribution data were obtained by wet sieving. Fifty grams of each Hupper sand sample were analysed. The samples were dispersed in water and exposed to ultrasound waves to separate sand grains from clay. Six sieves with different aperture sizes (63, 125, 250, 500, 1000 and 2000 Ixm) and a terminal pan to retain the < 6 3 Ixm fraction were used. The analyses were performed with the aid of a Fritsch shaker for 3 0 m i n per sample. The size fractions (pan content included) were dried and weighed, and their percentages were normalized to 100 wt%. The precision of the sieving method was assessed by repeating three times the analysis on samples ERI25, 136, 248, 250 and 255 (Table 7).

Grain-size analysis in thin section Twenty-two thin sections of refractory materials (reported as unit ~"by Eramo 2005a) and 43 of the melting crucibles (Eramo 2005b) were analysed. Moreover, five thin sections of Hupper samples (ER125, 136, 248, 250 and 255), with different grain-size distribution, were analysed to estimate the precision of this method (Table 7). A Swift & Sons point counter, mounted on the petrographic microscope, was used (with 1/3 mm as both line distance and lateral step). The maximum apparent diameter of grains was measured with the aid of a micrometer eyepiece at 10× magnification. The same size classes as in sieving were distinguished. Between 500 and 600 points per thin section were counted as the minimum number of counts necessary for routine analyses (Friedman 1958). Grain-size data were reported

References

AMWEG, G. 1941. Verrerie. In: Les Arts dans le Jura bernois et h Bienne, Arts appliques, Tome II. Porrentruy, 403-446. ARAMAKY, S. & RoY, R. 1962. Revised phase diagram for the system A1203-SIO2. Journal of the American Ceramic Society, 45, 229-242. AUBERT, D. 1975. L'~volution du relief jurassien. Eclogae Geologicae Helvetiae, 68(1), 1-64. BRILL, R. H. 1999. Chemical Analysis Of Early Glasses. Vol. 2, Tables of Analyses. Coming Museum of Glass, New York. CABLE, M. 1998. The operation of wood fired glass melting furnaces. In: MCCRAY, P. (ed.) & KINGERY, W. D. (series ed.) The Prehisto~ and Histora' of Glassmaking Technology, Ceramics and Ci'vilisation Series, Vol. 8. American Ceramic Society, Westerville, OH, 315-329. CABLE, M. & SMEDLEY, J. W. 1987. Liquidus temperatures and melting characteristics of some early container glasses. Glass Technology, 28, 94-98. CHARLESTON, R. J. 1978. Glass furnaces through the ages. Journal of Glass Studies, 20, 9-33. DE QUERVAIN, F. 1969. Die nutzbaren Gesteine der Schweiz. Ktimmerly and Frey, Bern, 248-250. ERAMO, G. (2005a). The melting furnace of the Derribre Sairoche glassworks (Court, Swiss Jura): heat-induced mineralogical transformations and their technological signification. Archaeometry, 47, 571-592. ERAMO, G. (2005b). The glass-melting crucibles of Derriere Sairoche (1699-1714 AD, Ct. Bern, Switzerland): a petrological approach. Journal of Archaeological Sciences, doi: 10.1016/j.jas. 2005.09.002. FLEURY, E. 1909. Le Sid~rolithique suisse. M~moires de la Soci&6 fribourgeoise de Sciences naturelles, 6, 1-260.

PRE-INDUSTRIAL GLASSMAKING, SWISS JURA FRIEDMAN, G. M. 1958. Determination of sieve-size distribution from thin-section data for sedimentary petrological studies. Journal of Geology, 66, 394-416. GERBER, C. 2003. Court-Chaluet bei Moutier (Berner Jura, Schweiz): eine Schwarzw/ilder GlashiJtte. In: STEr'PUHN, P. (ed.) Glashiitten im Gesprgich. Berichte und Materialien vom 2 Internationalen Symposium zur archiiologischen Erforschung mittelalterlicher und friihneuzeitlicher Glashiitten Europas. Schmidt-Rrmhild, Lfibeck, 64-69. GERBER, C., PORTMANN, M. & KUNDING, C. 2002. Fours ~ chaux, four ~ fer et charbonnikres dans le Jura bernois. Direction de l'instruction publique du canton de Berne, Berne. HENDERSON, J. 2000. The Science and Archaeology of Materials. An Investigation of Inorganic Materials. Routledge, London. HOFFMANN, F. • PETERS, T. 1969. Untersuchungen fiber die Verwendbarkeit schweizerischer Rohstoffe als Bindetone ftir Giessereiformsande. Beitriige zur Geologie der Schweiz, Geotechnische Serie, Lieferung, 47, 27. HOBNER, G. 1991. Natural and synthetic raw materials for technical ceramics. European Journal of Mineralogy, 3, 651-665. KRETZ, R. 1983. Symbols for rock-forming minerals. American Mineralogist, 68, 277-279. KRUMBmN,W. C. 1935. Thin-section mechanical analysis of indurated sediments. Journal of Geology, 43, 482-496. K~3NDIG, R., MUMENTHALER,T., ECKARDT, P., et al. 1997. Die Mineralischen Rohstoffe der Schweiz. Schweizerische Geotechnische Kommission, Bern.

199

MICHEL, G. J. 1989. Verriers et verreries en FrancheComt~ au X V I I f me sikcle (2 volumes). Collection Recherches, ERTI, Paris. PFIRTER, U. 1997. Feuille 1106, Moutier. Atlas g~ologique Suisse, 1:25 000. Notice explicatives, 96. SCHLAICH, E. 1934. Geologische Beschreibung der Gegend yon Court im Berner Jura: mit Berficksichtigung der Molassebildungen. Beitriige zur geologischen Karte der Schweiz, 26, 1-41. STERN, W. B. 1991. Zur chemischen Analyse der Bieler G1/iser. In: GLATZ, R. (ed.) Hohlglasfunde der Region Biel, Zur Glasproduktion im Jura. Paul Haupt, Bern, 83-86. STERN, W. B. 8z GERBER, Y. 2004. Potassiumcalcium glass: new data and experiments. A rchaeometry, 30( 1), 137-156. STERNINI, M. 1995. La fenice di sabbia. Storia e tecnologia del vetro antico. Edipuglia, Bad. SVEVA GAI, A. 1991. La produzione del vetro preindustriale in Germania sudoccidentale. Stato delle ricerche e prospettive. In: MENDERA, M. (ed.) Archeologia e storia della produzione del vetro preindustriale. All'insegna del gigtio, Firenze, 375-410. THURMANN, J. 1836. Essai sur les soulkvements jurassiques. 2 .... cahier, avec la carte du Jura bernoise. TURNER, W. E. S. 1956. Studies in ancient glasses and glassmaking processes. Part V. Raw materials and melting processes. Journal of the Society of Glass Technology, 40, 277-300. YON Moos, A. 1941. Uber Vorkommen und Abbau von Giessereiformstoffen in der Schweiz. Eclogae Geologicae Helvetiae, 34(2), 229-240.

Glass production in Late Antiquity and the Early Islamic period: a geochemical perspective IAN C. F R E E S T O N E

Cardiff School of History and Archaeology, Cardiff University, Humanities Building, Colum Drive, Cardiff CFIO 3EU, UK (e-mail: freestonei@ cardiff.ac, uk) Abstract: First millennium AD glass production was divided between a relatively small

number of workshops that made raw glass and a large number of secondary workshops that fabricated vessels. Glass compositions reflect the primary glassmaking source. For most of the period, Egyptian mineral soda was fused with lime-bearing siliceous sand to produce soda-lime-silica glass. The location of the Belus glassmaking sand, which is known from the classical literature, is located on that part of the Levantine coast where iron contents are lowest. 87Sr/S6Sr of primary glass from workshops in the Levantine region is close to that of modern seawater, and confirms the use of beach sand, which contained shell. Heavy mineral assemblages of Levantine beach sands are dominated by hornblende, hence the primary glasses are characterized by very similar trace element signatures. Glasses believed on archaeological grounds to have been made in other regions, for example in inland Egypt, may have higher 87Sr/Srsr, reflecting terrigenous sources of lime, and have different trace element signatures. Compositional data for glasses from as far away as Britain suggest origins of the glass material in the Eastern Mediterranean. Recycling of old glass may be recognized by the presence of elevated transition metals. The use of plant ash as a flux became dominant practice in the ninth century and preliminary data for plant ash glasses from the early Islamic world indicate that primary production centres may be separated using strontium and oxygen isotopes as well as by major and trace elements.

Small glass objects, such as beads and pendants, are reported from the archaeological record as far back as the late third millennium Be, but production of glass vessels on any significant scale seems to have begun in the Late Bronze Age, in the late sixteenth century to mid-fifteenth century BC, in northern Syria and Mesopotamia (Moorey 1994) and at about the same time in Egypt (Lilyquist & Brill 1993). Glass at this time was strongly coloured and a rare material of high status, equated with semi-precious stones such as lapis lazuli (Oppenheim et al. 1970). Glass is relatively rare in the so-called 'Dark Age'at the end of the second millennium BC, but by the ninth to eighth centuries vessels were again being made and are known from Egypt and Mesopotamia. Glass production expanded towards the end of the first millennium BC, when it was widely traded in the form of beads and hemispherical bowls, in general still strongly coloured. The adoption of blowing in the first century BC and its rapid spread from its origins in Syria-Palestine throughout the

Roman Empire accompanied a major expansion in production. Transparent glass vessels became relatively common and inexpensive commodities and by the mid- to late first century AD strong deliberate colours had largely been replaced by weak 'natural' transparent blues and greens, along with glass that was essentially colourless. From this time, the use of blown, transparent glass for drinking and eating, for storage and for windows, continued throughout the Near East and much of Europe. Whereas clay-based ceramics have been routinely subjected to elemental analysis to determine provenance for several decades, archaeological glass has proved far less tractable. Substantial databases of major element analyses of glass exist (Brill 1999), but beyond broad technological affiliations, meaningful compositional groupings have been difficult to establish. Where trace element or isotopic data were generated, on the whole, it proved possible to interpret them only in very general terms (e.g. Hunter & Heyworth 1998; Brill et al. 1999).

& MESSIGA,B. (eds) 2006. Geomaterialsin CulturalHeritage. Geological Society, London, Special Publications, 257, 201-216. 0305-8719/06/$15.00 @ The Geological Society of London 2006. From: MAGGETTI,M.

202

I.C. FREESTONE

Recent advances in glass compositional studies originate partly in improvements in analytical capability, particularly in the ability to analyse small samples, or even to conduct essentially non-destructive analyses. In addition, archaeological discoveries have radically changed our understanding of the organization of glass production in the past. Whereas at one time it was assumed that the glass material itself was made from silica and alkali in the workshops where the vessels, window panes and beads were fabricated, it is now recognized that raw glass was widely traded, for example as lumps of colourless or naturally coloured glass in the Roman and early medieval periods (Foy et al. 2000), and as ingots of coloured glass in the Late Bronze Age (Nicholson et al. 1997; Rehren & Putsch 1997). The primary workshops that made the raw glass were thus distinct from the secondary workshops that shaped the glass into vessels, and a single primary workshop could supply many secondary workshops, dispersed over a very large area (Gorin-Rosen 2000; Nenna et al. 2000). In the primary workshops glass appears to have been made on a scale of many tonnes in a single firing, as illustrated by the 8 tonne glass slab at Beth She'arim, Israel, probably dated to the early ninth century AD (Fig. 1; Brill 1967; Freestone & Gorin-Rosen 1999) and the excavations of 17 tank furnaces of similar capacity at Bet Eli'ezer, Hadera, Israel, probably dated to the seventh to eighth centuries AD (Gorin-Rosen 1995, 2000). Three similar furnaces of sixth to seventh century AD date have been excavated at Apollonia-Arsuf, Israel (Tal et al. 2004) and four 10th to 1 l th century AD furnaces at Tyre, Lebanon, one of which

Fig. 1. Large glass slab in a cave at Beth She'arim, Israel. The glass appears to have been melted in situ but was unusable because of excess lime in the batch. Dimensions 3.40 m × 1.95 m x 0.45 m. Photograph: Freestone/Milton.

has an estimated capacity in excess of 30 t (Aldsworth et al. 2002). The concept of a division of production leads to a very different interpretation of analytical data, so that glass compositions reflect predominantly the primary glassmaking sources, rather than the secondary workshops in which the objects were made (e.g. Nenna et al. 1997; Foy et al. 2000; Freestone et al. 2000, 2002a). This new interpretation is leading to an improved understanding of glass production and how it may be investigated using physico-chemical analytical techniques. Informed by an appreciation of the geochemical processes that control the compositions of the raw materials, rapid and significant progress is being made. This is particularly the case for the blown, transparent glass of the Roman period and later. This paper reviews some of the recent work and its implications, focusing in particular upon case studies on glass carried out by the author and collaborators. A summary of the sites and production groups referred to in the text is provided as Table 1.

Categories of glass and sources of soda The great majority of ancient glass was based upon silica, fluxed with either soda or potash. Before the medieval period, lead-rich glasses were rare, excepting some strongly coloured opaque glasses and certain glass in the Far East (Brill & Martin 1991). K20-rich glasses are confined largely to medieval Europe, after the tenth century, and to SE Asia, where potash-rich glass beads, probably made in India, were widely traded (Glover & Henderson 1995). This paper focuses upon soda-lime-silica glasses, which generally lie in the low melting temperature region of the system N a 2 0 - C a O - S i O 2 (e.g. Wedepohl 2003). Soda-lime-silica glass is the earliest known from late Bronze Age Mesopotamia and Egypt, in the second millennium Be, continued as the dominant compositional type across Western Asia and the Mediterranean through to the modern period, and was also important in the glass industries of Renaissance and modern Europe. In a seminal paper, Sayre & Smith (1961) observed that ancient soda-lime-silica glasses fell into two main categories: high-K20, highMgO glass, with greater than about 1.5% of each of these oxides, and low-K20, low-MgO glass, typically with less than 1.5% of each oxide. Although exceptions and intermediate compositions are now recognized, this general subdivision is still regarded as a useful one in that it corresponds to the use of the two principal

203

FIRST MILLENNIUM AD GLASS PRODUCTION Table 1. Summary information on sites and assemblages mentioned in the figures and text Find site/or group

Probable primary source areas/groups

Leicester, UK Bet Eli'ezer, Israel

?Levantine Bet Eli'ezer, Israel Syria-Palestine

lst-3rd 7th-8th

Apollonia (Arsuf), Israel

Apollonia, Israel

6th -8th

North Sinai

4th-5th

Maroni Petrera, Cyprus Tel el-Ashmunein, Egypt Jarrow, UK

HIMT (?Egypt) and Levantine HIMT (?Egypt) and Levantine Egypt II (?Middle Egypt) Syria-Palestine

Consumer site Primary furnaces Secondary workshop Primary and secondary workshops Field survey

6th-7th

Church

Natron

8th-9th

Secondary workshop Monastery

Natron

Beth She'arim, Israel Banias, Israel

Bet She'arim, Israel ?Syria-Palestine

?9th

Nineveh, Iraq Tyre, Lebanon

?Mesopotamia Tyre, Lebanon

4th-7th 10th/11 th

Raqqa, Syria

Raqqa, Syria

8th-9th

Ras al Hadd, Oman

Unknown

1lth-14th

Bet She'an, Israel

Period (century Ao)

6th-7th

7th

10th/11 th- 13th

Context

Primary furnace Secondary workshop Consumer site Primary furnaces Workshop complex Consumer site

Glass form

Glass type

Natron Natron

Vessel fragments Raw glass chunk

Natron

Raw glass chunk

Natron

Raw glass chunk and vessel fragments Vessel fragments

Natron

Natron

Vessel and window Vessels and waste

Plant ash

Window fragments Large slab

Plant ash

Raw glass chunk

Plant ash Plant ash

Vessel fragments Raw glass chunk

Plant ash

Chunk and vessel

Plant ash

Vessel fragments

Referencesare givenin the text.

soda-rich fluxes available in the ancient world. Table 2 provides some examples of these glass types and data are plotted in Figure 2 to illustrate how glass from Late Antiquity and the early Islamic period in the Middle East may be divided into the two groups. Since the original

work, it has become generally accepted that these two glass groups correspond to the use of two principal sources of soda: evaporitic minerals and plant ash. Evaporites rich in sodium bicarbonate were available from Egypt and possibly other

Table 2. Major element compositions of selected soda-lime-silica glasses Blue-green Leicester1 Cultural period: Roman Date (century AO): lst-3rd n: 75 SiO2 A1203 FeO MnO MgO CaO Na20 K20

n.a. 2.33 0.60 0.26 0.55 6.43 18.4 0.69

HIMT North Sinai 2 Late Roman 4th-5th 18

Levantine Bet She'an 3 Byzantine 6th-7th 17

Levantine Bet Eli'ezer4 Early Islamic 7th-8th 27

Plant ash Banias 4 Early Islamic 10th-13th 17

Plant ash Nineveh5 Sassanian 4th-7th 19

65.8 2.69 2.18 1.51 0.94 5.99 18.0 0.46

69.3 3.03 0.45 <0.1 0.59 9.17 15.6 0.63

74.9 3.32 0.52 <0.1 0.63 7.16 12.1 0.46

70.5 1.06 0.40 1.00 2.72 8.55 12.5 1.89

63.2 2.70 0.93 0.08 4.49 7.57 16.5 3.23

Sourcesof data: IJacksonet al. 199l; 2Freestoneet al. 2002b;3Freestone& Gorin-Rosen,unpubl.;4Freestoneet al. 2000; 5Freestone& Leslie, unpubt. n.a., not analysed.

204

I.C. FREESTONE

6.0

o o oo

,

4.0

o

~

o °8 o e o % o ~' ~~' " ", 0 % o° t,~o6~Oo o o o o ~ ~-~g~o oOO°O°,~

o

.. " ° °

£0

x x

3.0

~-:lm~-x~'-~"

,,,

2.0

a

~-~,~

oo-':'

o

o

o

%

o

o

o

.

x xK

~

1.o

~

x x"~t~i',~

o

Xo • Byzantine

",

o Sassanian

~

x Syrian Islamic

x

0.0

0.0

1.0

2.0

30

4.0

50

6.0

70

8.0

Wt % MgO

Fig. 2. Compositions of Middle Eastern glasses, showing subdivision between low-magnesia and highmagnesia glasses. Also shown is the proposed subdivision of high-magnesia glasses into the Sassanian and Syrian-Islamic traditions. Data of Brill (1999), Freestone et al. (2000), Freestone (2002) and Freestone & Leslie (unpubl.).

locations (Shortland 2004; Shortland et al. 2005), where they are precipitated as saline lakes begin to evaporate in the spring. The Egyptian deposits, known mainly today from the Wadi Natrun, about 100km NW of Cairo, but also from al-Barnuj in the Western Delta (Fig. 3), comprise predominantly one or more of the minerals trona, thenardite, burkeite and halite. References to the extraction of soda from Wadi Natrun are found in the ancient and medieval literature, and al-Barnuj appears to have been referred to as a source of natron in the fourth century AD (Shortland et al. 2006). Furthermore, several primary glassmaking sites are known in the Wadi Natrun (Nenna et al. 2005). The tronarich varieties are considered to have been preferred for glassmaking, because sulphates and chlorides do not readily decompose and

Fig. 3. Trona precipitating at the edge of a lake at al-Barnuj, Nile Delta, April 2004. Photograph: I. Freestone.

react with silica at the temperatures of 10001200 :C that are likely to have been attainable in traditional glassmaking furnaces. Traditionally, in archaeology and Egyptology, these salts are commonly known as natron; however, to avoid confusion with the mineral of that name, in the present paper they will be referred to as mineral soda. It is unlikely that glassmakers restricted themselves to the use of pure trona. Virtually all glass of this period contains 0.5-1.2% C1 and 0.20.5% SO4, which are likely to have been close to the solubilities of these components in the glass when it was made (e.g. Bateson & Turner 1939). The sulphur and chlorine in the glass are likely to have been derived from the alkali. It is probable that the excess sodium chloride and sulphate separated as an immiscible 'scum' or 'gall' on the surface of the freshly melted glass, which could be skimmed off or discarded after the glass had cooled. Evidence of an immiscible sodium sulphate phase in ancient glass has been reported by Stapleton & Swanson (2002), and Barber & Freestone (1990) reported the sub-liquidus (metastable) phase separation of sodium chloride, suggesting that the glass was near saturation at the working temperature. Egyptian mineral soda is typically low in cations other than Na + (Brill 1999; Shortland 2004) and added few impurities to the glass batch. Its use gave rise to the low-MgO, low-KzO compositional category of glass identified by Sayre & Smith (1961). These glasses are frequently termed 'Roman-type' or 'natron' glasses. In the following, they will be referred to as low-magnesia glasses. The second sodic flux available to early glassmakers was the ash produced by burning halophytic plants, for example Salicornia and Salsola sp., from semi-arid and coastal environments. The harvesting of plants to produce ashes has a long history in all parts of the world, and they were used for medicinal purposes and in the production of detergent, as well as for glassmaking (Ashtor & Cevidalli 1983). In addition to soda, such plant ashes contain high levels of lime, as well as magnesia, potash, phosphate and some silica and other components (Brill 1970; Verit/t 1985). The glasses formed by mixing soda ash with a source of silica are richer in impurities than those produced from mineral soda, and form the high-MgO, high-K20 compositional category. In the period of interest, low-magnesia glass was generally used west of the Euphrates and plant ash glass was used to the east for most of the first millennium AD (Smith 1963). As

FIRST MILLENNIUM AD GLASS PRODUCTION shown in Figure 2, there was a switch from the use of mineral soda to plant ash flux in western regions from the middle of the ninth century AD (Gratuze & Barrandon 1990), probably because of political upheavals in the Delta region (Whitehouse 2002; Shortland et al. 2006). From this time, the majority of sodalime-silica glasses were of the high-MgO, high-K20 type. An additional subdivision of the high-magnesia (plant ash) group may usefully be recognized. The plant ash glasses from the Syria-Palestine region have compositions with K20 and MgO of 2-3.5% (e.g. from the Serce Limani wreck, Brill 1999; Banias, Freestone et al. 2000; Tyre, Freestone 2002). However, plant ash glasses made up to the seventh century AD to the east of the Euphrates, in the region of the Persian Empire under the Parthians and Sassanians, tend to have MgO and K20 contents greater than 3.5% (Fig. 2). This appears to indicate a difference in raw materials and/or technology, possibly in the species of plants that were ashed for glassmaking. It may speculatively be suggested that the Partho-Sassanians were exploiting plants from a particular environment, for example the salt marshes in the south of present-day Iraq, which conferred a different composition from Syrian plant ashes. Alternatively, the high MgO in the Sassanian glass may reflect the MgO-rich alluvium of the Euphrates and the Tigris, which, from the analysis of ceramic bodies made from it, appears to have a relatively high MgO/CaO ratio (e.g. Freestone 1991). A difficulty in resolving this issue is that our understanding of the manufacture of plant ash glass is very limited. Plant ash is a very variable raw material, its composition varying according to species, environment, and even the part of the plant that is harvested. However, as has been emphasized by Rehren (2000), this variable and apparently unpredictable material was used to produce glass of a relatively restricted compositional range at any one place and time. How compositional control was maintained is not clear, and neither is the relationship between the compositions of the plants, the ashes and the final glasses. Careful selection of plant material, discarding of poor quality glass (Freestone 2002) and the melting process itself (Shugar & Rehren 2002) are all likely to have played a part in the process of compositional control. It may be that our understanding of the production of plant ash glass is hindered by the very scale on which it occurred. From the perspective of the modern laboratory analyst, it is

205

very difficult to obtain a sample of plant ash that is representative of the hundreds of kilograms that would have been added to a single batch of raw glass. Processing plant material on this scale may have resulted in a homogenization of this variable raw material towards an average composition that cannot be recognized in modern samples of individual plants or parts of plants. Silica s o u r c e s

Potential sources of silica include quarried siliceous minerals and rocks such as vein quartz, chert and quartzite, as well as pebbles of these materials and sand. It seems likely that, over much of the period and region under consideration here, sand was the preferred source of silica, as to mine and crush silica on a scale of tonnes for each firing would have been very costly. Some direct evidence for this is observed from the waste products of glass furnaces where the reaction did not proceed to completion. For example, a large block of glassy material from one of the Islamic period furnaces at Tyre, Lebanon contains patches of silica 5 - 1 0 mm across and readily visible with the naked eye (Aldsworth et al. 2002). In the microscope, these patches are seen to be aggregates of finegrained quartz sand particles, which represent the siliceous raw material (Fig. 4). The glass made at Tyre has an A1203 content of around 1.8% and Fe203 of 0.5%; even assuming that all of these components derived from the sand, it appears that the sand was mature and composed of around 95% quartz. On the other hand, Henderson et al. (2005) have presented evidence for the use of crushed quartz pebbles as a silica source in the production of Islamic period plant ash glass at el-Raqqa, Syria.

Fig. 4. Back-scattered electron image showing quartz sand particles in partially fused, vesicular glass from a primary tank furnace at Tyre, Lebanon.

206

I.C. FREESTONE

The el-Raqqa glasses are characterized by lower A1203 contents than those of Tyre (Freestone 2002), as a result of the relatively pure nature of quartz pebbles relative to a feldspar-bearing sand. A group of Islamic glasses from Nishapur were considered by Brill (1995) to have been made using pebbles, for similar reasons. Further evidence for the use of sand as a starting material comes from Pliny the Elder, who wrote his Natural Histoo' in around AD 70. Pliny's is the only surviving written account of Roman glass production and indicates that, for many years, glassmaking had depended upon a mixture of Egyptian soda with the sand from the beach near the mouth of the River Belus, on the coast of Syria: 'This is supposed to be the source of the River Belus, which after traversing a distance of 5 miles flows into the sea near the colony of P t o l e m a i s . . . The beach stretches for not more than half a mile, and yet for many centuries the production of glass depended on this area alone' (NH XXXVI, 190; Eicholz 1962). The Belus is the present-day River Naaman, which flows into the Bay of Haifa, and it is from the beach near the mouth of the fiver that Pliny indicates the glassmaking sands were obtained. The sediments of the coast of Israel are derived from the Nile, and are moved up the coast by the prevailing Mediterranean current and longshore drift. A general sedimentary model has been given by Goldsmith & Golik (1980) and Stanley et al. (1997). These sands contain calcium carbonate in the form of marine shell and limestone eroded from the kurkar cliffs. Emery & Neev (1960) determined calcium carbonate contents and colours for sands for a series of beach stations spanning the coast of modern Israel. To understand the reasons for the selection of the glassmaking sand, their data are plotted in Figure 5. The two curves show weight percent CaCO3 and Munsell colour saturation v. location (beach station). The lower the colour saturation, the paler is the sand. The figure shows that the sand from beach stations 28 and 29, situated in the Bay of Haifa, are among the palest on the coast. This implies relatively low iron oxide contents for these sands, so that they would make relatively pale or weakly coloured glasses. In addition, the calcium carbonate content of sand in the Bay of Haifa is around 15%, which would produce a s o d a - l i m e - s i l i c a glass with 8 - 9 % CaO. This is around the level required to produce a stable and workable glass. Thus it appears that sand close to the mouth of the River Belus was the most suitable glassmaking sand on this part of the Mediterranean coast, and was deliberately selected by Roman

•-,o-- Colou~-

Saturation

[

Bay of H a f f a

I

-o-- Carbonate

80

@ ~ eo

60.

~.

40,

20-

q 4

,I 7

I 10

I 13

I 16

I 19

I 22

/ 25

8

I 31

34

Beach Station

Fig. 5. Colour saturation (arbitrary scale) and carbonate content of beach sands from the coast of present-day Israel. Data from Emery & Neev (1960).

glassmakers for this reason. A detailed comparison of the chemistry of sand from the mouth of the River Belus and glass from a fourth century AD glass workshop in Israel was conducted by Brill (1988). It is unlikely that the beach in the vicinity of the Belus was the only source of glassmaking sand. The presence of primary glassmaking installations further down the Levantine coast, at Apollonia-Arsuf (Tal et al. 2004) and Bet Eli'ezer, Hadera (Gorin-Rosen 2000), as well as in Egypt (Nenna et al. 1997, 2000), suggests that other sands in the eastern Mediterranean region were suitable for this purpose. Intriguingly, much of the evidence in the Levant relates to the Islamic period, and the factory (or factories) that produced the bulk of Roman glass is yet to be discovered. Pliny also remarks that sand between Cumae and Literno, on the coast of Italy (about 50 km north of Naples) was used to make glass, and that 'sand is similarly blended in the Gallic and Spanish provinces' (Eicholz 1962). However, direct archaeological evidence for primary glass production in these regions has not been found to date.

S t r o n t i u m and sources of lime Lime, typically present at levels of 5 - 1 0 % (Table 2), is a desirable component of glass as it reduces its solubility in water, so that it is environmentally stable. Plant ash contains both alkali and lime, and it is not necessary to invoke extra additions of lime to explain the compositions of plant ash glasses (Freestone & Gorin-Rosen 1999). However, mineral soda contains essentially no lime, and the source of lime

FIRST MILLENNIUM AD GLASS PRODUCTION in low-potash, low-magnesia glasses must be sought elsewhere. There are essentially two models for the origin of lime in low-magnesia glass. Lime may have been added deliberately to the batch by the glassmakers, to a fixed recipe, to produce a glass of consistent properties. This model is supported by a comment of Pliny, who mentions shell in a list of materials that could be added to glass. This explanation has the clear advantage of explaining why so much blue-green Roman glass of the first to third centuries AD is of a remarkably constant composition. The second model explains the lime content of low-magnesia glass as deriving from panicles of shell or limestone in the sand used as the source of silica, the sands of the Levantine coast (Fig. 5) being the classic example. Once again, this model is supported by Pliny, who specified the lime-bearing Belus sand for glassmaking. The two models for the origin of lime are not mutually exclusive, as glassmakers may have exploited lime-bearing sands where available, and added crushed shell or limestone to pure quartz sand where they were not. A solution to the question of the origin of lime was offered by Wedepohl & Baumann (2000), who pointed out that most strontium in Roman glass was likely to have entered the batch in calcium-rich minerals, in particular calcite and aragonite. Late Roman (fourth century) glass from the Rhineland analysed by them had high Zr and Ba, consistent with local river sediments, as well as lead isotopes consistent with the ores of the region. This led them to suggest that the glasses were made locally, using local river sand as a source of silica. To determine the form in which the lime was added to the glass they measured the concentrations of strontium and its isotopes. The glasses have relatively high strontium/calcium ratios, appropriate to marine biogenic carbonate rather than limestone that has undergone diagenesis. Furthermore, 87Sr/86Sr ratios of around 0.7090, close to the value for modern seawater (S7Sr/ 86Sr ~ 0.7092), also suggest that the lime was derived from shell. Thus Wedepohl & Baumann concluded that their glass was made using local sand to which shell was deliberately added to give the appropriate amounts of lime. Freestone et al. (2003) measured strontium isotopes in low-magnesia glasses from the primary tank furnaces at Bet Eli'ezer, near Hadera on the Levantine coast. They are shown with a number of other glass groups, including the Rhenish glasses of Wedepohl & Baumann (2000), in Figure 6. These glasses are believed to have been made utilizing the nearby coastal

207

600

Coasta/ Sand

iI r~J

500

~-r

P/ant Ash 4OO

~

go

+

~ Io

oo Im '~ i=

300

0

Bet Eliezer Bet Shean

200 100 0 0.7075

A

Limestone-bearing sand I

I

I

0~7080

0.7085

0.7090

i i i i i r i i

z~Ashmunein 0 + _

Banias Germany

[ 0.7095

SrSr/~6Sr

Fig. 6. Strontium isotopes and total strontium in glasses from the Near East and Germany (Freestone et al. 2003; German data from Wedepohl & Baumann 2000).

sands, which naturally contain calcium carbonate in the appropriate concentrations to yield good soda-lime-silica glass (Fig. 5). The data show that the Levantine glasses have 87Sr/86Sr and total Sr similar to the Rhenish glasses. As might be expected, glass made from sand that contains intrinsic shell has similar strontium characteristics to glass to which shell has been deliberately added. Thus the strontium approach, although indicating the mineralogy of the added lime, cannot determine whether it was deliberately added. The case for deliberate addition of shell to the Rhenish glasses rests upon the inference that their trace element and lead isotope compositions imply a local origin. However, Freestone e t aL (2005) have shown that a category of fourth century glass termed HIMT (see below), which occurs widely in Europe and the Mediterranean, has similar isotopic and trace element compositions to the Rhenish glass analysed by Wedepohl & Baumann (2000). Given the fourth century dates of all of these glasses, it seems probable that they have a single origin and that the Rhenish glasses are unlikely to have been locally made. The preferred interpretation of the author follows that expressed by Brill (1987) in that lime was rarely deliberately added as a separate component in ancient glassmaking and that glass was widely made as a mixture of only two components, silica and flux. Where the flux was ash, a lime-free silica source was used. Where the flux was mineral soda, a calcareous quartz sand was used. In each case, the appropriate mixture of silica and flux gave glass with similar Na20:CaO:SiOe ratios. Rarely, soda ash and calcareous sand were combined to give a glass with lime from two sources. For example, the glass slab at Beth She'arim (Fig. 1) seems

208

I.C. FREESTONE

to have been produced in this way and, because of the resultant 15.9% CaO, devitrified to wollastonite (Brill 1967; Freestone & Gorin-Rosen 1999). It should be noted that glasses from tank furnaces with 'normal' CaO contents, < 10%, are not normally devitrified in this way (Freestone et al. 2000; Freestone 2002; Tal et aL 2004), so that the archaeologically significant variable here is composition, not cooling rate; although a slow cooling rate is required to cause the glass to devitrify, the cooling rates of these installations are likely to have been very similar. The selection of an appropriate glassmaking sand can have a critical effect on the preservation of glass in the archaeological record. On the basis of the limited evidence from recovered glasses, it appears that, when Egyptian mineral soda began to be used on a large scale in glassmaking around the tenth to eighth centuries Be, glasses were frequently made using vein quartz or sand with insufficient lime to render them stable over archaeological time. Most of these are likely to have dissolved away, leaving limited evidence of this important technological change (Reade et al. 2005). Assuming the foregoing model for the origin of the lime, then the remaining data in Figure 6 are readily explained. The low levels of strontium and low 87Sr/86Sr in low-magnesia glass from an eighth to ninth century AO workshop at Tell el-Ashmunein, in middle Egypt, represents the use of an inland sand containing clasts of limestone, rather than shell. The strontium in the glass from Banias was derived from plant ash, and reflects the bioavailable material in the soil in which the ashed plants grew. It therefore has a terrigenous isotopic signature and a relatively high strontium content, which reflects the strontium capacity of the minerals in the plant structure, which are likely to have included calcium oxalates (Freestone et al. 2003).

five commonly occurring glass groups, dating from the fourth to ninth centuries AD. Egypt II and Wadi Natrun appear to reflect Egyptian manufacture between the seventh and ninth centuries (data of Gratuze & Barrandon 1990), Bet Eli'ezer and Levantine I represent different productions on the Levantine coast (Freestone et al. 2000), and HIMT has been tentatively attributed to a late Roman primary production in Egypt (Freestone et al. 2005). These groupings are largely a reflection of calcium carbonate and feldspar minerals in the sand used to make the glass. It appears that at any one time glass was being made in a limited number of localities. Multivariate statistical analysis of compositional data for lumps of raw glass from shipwrecks and workshops along with typologically wellcharacterized vessels has allowed Foy et al. (2000, 2003a,b) to identify 10 distinctive lowmagnesia glass groups in use in the Mediterranean region in the first millennium AD, on the basis of their major element composition. Each group is assumed to relate to a primary production centre, although such a centre may consist of more than one factory in a single region. For example, the products of the Levantine coast (Belus-type sand) are grouped together, although individual workshops may have slightly different chemical characteristics (Fig. 7). Mean concentrations of selected trace elements in low-magnesia glasses from ByzantineIslamic (sixth to ninth centuries) workshops at Bet Eli'ezer (near Hadera) and Apollonia (Arsuf) on the coast of Israel are shown in Figure 8, normalized to crustal values (Freestone et al. 2000). These samples are mainly of primary glass and show a consistent pattern. This is to be expected, as the glasses were made from Levantine coastal sand of which the 14 O0 [3 Wadi Natrun 1200

A Egypt II

×

Provenancing low-magnesia glass As low-magnesia glass is essentially a mixture of soda and sand, and the evaporites of the Wadi Natrun and the Delta are relatively pure soda sources, the question of the provenance of glass is reduced in essence to that of slightty diluted sand. In this paper, by the term provenance is understood the place where the raw glass was made, not the place where it was shaped, which could have been very different. Lime and alumina concentrations reflect those in the raw material and may be used to provide an initial impression of glass groups. Figure 7 shows

<>Bet Eli'ezer

~ ~××

10.00

X Levantine I +HIMT

800 6.00

A -t4-

++

4.00 2 O0 0 O0 1.00

I 2.00

I 3.00

I 4.00

5.00

AIzO~

Fig. 7. Lime and alumina contents of a number of glass groups.

FIRST MILLENNIUM AD GLASS PRODUCTION 2.00

209

2.00 - £}--, Bet Eli 'ezer

[+

-- -X-- Apollonia

1.60

Apollonia

I -- -o -- Bet Eli 'ezer l - -El-- Maroni (n=12)

1.60

.X

120

1.20 O

~

0.80

0mS0 "

\\

•

" " x,

0.40

0.40 0.00 0.00

Ga

I

I

I

Rb

Sr

Y

I Zr

I Ba

I La

I Ce

I Pr

I Nd

Ga Th

Fig. 8. Trace elements, normalized to mean crustal composition, for Levantine primary glass from Bet Eli'ezer (n = 5) and Apollonia (n = 5) compared with a workshop at el-Ashmunein, middle Egypt (Freestone et al. 2000). major component was ultimately derived from the Nile, as discussed above. The heavy mineral assemblages of the beach and dune sands from this region are similar and are dominated by hornblende (Emery & Neev 1960; Pomerancblum 1966). The trace elements in the Levantine glasses reflect this relatively homogeneous raw material. This characteristic trace element distribution may be contrasted with that of the glass from the workshop at Tel el-Ashmunein, Egypt, which was shown by strontium isotope analysis to have been made using inland sand (see above). The el-Ashmunein glass (Fig. 8) differs from the Levantine glasses in terms of the concentration of elements such as Zr and Ba, in addition to the lower Sr, which is a reflection of the presence of limestone, rather than shell, in the sand (Freestone et al. 2000, 2003). These and other differences in trace element composition (see also Freestone et al. 2000) suggest that it should be possible to source glass by elemental analysis. This possibility was investigated using vessel glass from Maroni Petrera, a sixth to seventh century AD Byzantine church on the southern coast of Cyprus, which faces the Levantine coast and therefore could be expected to have obtained raw glass material from this region. Two compositional groups were identified on the site (Freestone et al. 2002a). A good match was found between the major and trace elements of the glass from the larger of these groups and glass from the Levantine workshops (Fig. 9; Freestone et al. 2002a), consistent with the inferred origin in Syria-Palestine.

I

I

I

I

I

Rb

Sr

Y

Zr

Ba

La

I

[

I

Ce

Pr

Nd

Th

Fig. 9. Mean trace element concentrations for 12 glasses from Maroni Petrera compared with those from Levantine tank furnaces. It should be noted that the comparison between the primary furnaces and the Cypriot glass neglected the transition metals associated with glass coloration, which are substantially higher in the vessels (Fig. 10). This is the result of recycling old glass, some of which was coloured and incorporated into the remelted batch. At these levels, the transition metals have little effect on the colour of the glass, but are clearly present at concentrations above those encountered in freshly made glass sampled from tank furnaces, or those expected in mature sands. Elements such as Cu, Co, Pb, Mn and Sb, which are associated with coloration, decoloration and opacification processes, along with associated elements such as Ag and Zn, can be misleading as their concentrations are related to the particular mix of recycled glass used in a specific batch. They can therefore give rise to compositional subgroups that are not related to the origin of the primary glass. Even where evidence of other colorants is absent, manganese oxide may be elevated, as pyrolusite (MnO2) was used as an oxidant in Roman and Islamic glassmaking to minimize the blue colour caused by divalent iron (Sayre 1963). Many Roman glasses contain low levels of MnO, of the order of 0.1-0.2%, in the presence of around 0.4% FeO t. Whether additions of manganese at such low levels would have effectively decolorized the glass is doubtful. However, as MnO/FeO averages 0.015 in the Earth's crust (Wedepohl 1995) the probability is that these levels do represent deliberate addition at some point during the manufacturing and recycling history of the glass. Technological factors, such as coloration and decoloration of the glass, and the initial sand:soda ratio chosen by the glassmaker, need to be taken into

210

I.C. FREESTONE

1000

1000 r

/

.

100

100

10

10

1

1

0.1

0.1

Maroni

I

Co

Cu

Zn

Pb

Ag

Co

Cu

Zn

Pb

Ag

Fig. 10. Colorant elements in vessels from Maroni Petrera, compared with concentrations in raw glass from Apollonia, on the Levantine coast.

consideration when analysing glass compositional data. There are literary references to hawkers on the streets of Rome who collected glass for recycling (Whitehouse 1999; Keller 2005), and it is becoming increasingly clear from compositional studies that the recycling of glass was widely practised. However, in general, the recycling process does not appear to have resulted in the blurring of the compositional differences between primary production groups. The same glass compositional groups may coexist and show clear compositional separation in localities as diverse as Cyprus, Egypt and Italy (Freestone et al. 2002a,b; F o y e t al. 2003a,b). This apparent paradox may be attributed to several factors. First, for significant periods of time, glass working in a region tended to depend on a single production centre for the supply of raw glass. Thus recycling, which is likely to have been carded out on an intra-regional basis, will have involved mixing glass of the same basic type. The apparent coexistence of separate compositional groups may in some cases reflect poor chronological resolution; their use may have been separated by a decade or more. Furthermore, it is likely that glassmakers tried to avoid mixing glasses of different colour. For example, colourless glass would have been spoiled by mixing it with blue-green glass. The traces of colorant that have been interpreted as evidence of recycling in this paper may reflect carelessness on the part of the craftsman, unavoidable contamination through the inclusion of coloured handles, decorative spots, etc., and confidence that the inclusion of a small amount of

coloured glass would not affect the colour of the batch. It would appear that Levantine glass producers supplied not only the eastern Mediterranean but also much of the early medieval world. Levantine glass appears to be present in assemblages in Italy, dating to around the fifth century (Freestone et al. 2002b) and in France (Foy et al. 2003a). Figure 11 compares trace element distributions in window glass from Jarrow, Northumbria, UK, dating to the seventh century AD, with primary glasses from the Levant. The Saxon glasses show the characteristic trace element profile of Levantine production and it seems very likely that the origin of the glass was in the Levant. It is not at this stage clear

Jarrow - -~-

• Bet Eli 'ezer

" D-- Apollonia

- -O-

Apollonia

chunk vessel

~ 20

080

0 O0 Ga

I Rb

~ Sr

I Y

I Zr

I ~

I La

1 Ce

I Pr

/ Nd

Fig. 11. Trace elements in window glass from the monastery at Jarrow, Northumbria, compared with raw glass and vessel glass from the Levant,

FIRST MILLENNIUM AD GLASS PRODUCTION whether raw lumps of glass were arriving into Britain from the Near East; the glass may have arrived as cullet from continental Europe, to be remelted by glass workers in England (Freestone 2003). Other primary production centres that were operative in the period under consideration include that which produced so-called HIMT glass, labelled on account of its high iron manganese and titanium contents (Table 2) and discussed by Freestone et al. (2005). Glasses of this group frequently show a strong linear relationship between the constituent oxides, and also between strontium and lead isotope and oxide compositions (Fig. 12). They appear to represent the mixing of two end-members, one based upon typical coastal sand with high SVSr/86Sr, the other with a higher terrigenous component, lower in radiogenic Sr. Freestone et al. (2005) suggested that these mixed compositions may reflect a source on the coast closer to the mouth of the Nile, where sands with variable amounts of terrigenous components may be juxtaposed. HIMT glass is found extensively in the North Sinai, and occurs as far away as fourth century Roman Britain (Freestone et al. 2005). It seems increasingly likely that much of the glass of the first millennium AD originated in the primary production centres of SyriaPalestine and Egypt. In this region were found the major deposits of soda and also glassmaking sands, which combined the desirable properties of low iron contents and moderate lime, needed to make good quality soda-lime-silica glass. Furthermore, large deposits of the widely used decolorant, pyrolusite, occur in the Sinai. The location of the primary glass industry, as currently understood, is a clear reflection of the

7.00

6.50

O

6,00 o

.~

/

5.50

o RZ = 0.64

5.00

4.50 0.7080

I 07082

I 0.7084

I 0.7066

I 0.7088

8~Sr/~Sr

Fig. 12. Correlationb e t w e e n 87Sr/86Srand CaOin HIMT glass.

0.7~0

211

overriding influence of the location of raw materials. Fuel would have been required in large quantities to feed the furnaces that produced the primary glass. At present, there are no published data on the nature of the fuel used in the primary glassmaking furnaces of the Levantine coast and Egypt but, given the regional climate, the requirements of glassmaking might be expected to have placed considerable demands on the fuel supply. The existence of 17 furnaces, each producing around eight tons of glass in a single firing, at Bet Eli'ezer in Israel (Gorin-Rosen 1995, 2000) indicates that fuel supply issues were resolved, but the mechanism is not yet understood. It is possible that glass furnaces were located according to locally available fuel sources, and moved within the region when fuel was exhausted. This would be consistent with the known locations of primary workshops. An outstanding issue is the origin of the bluegreen transparent glass so common on Roman sites of the first to third centuries AD (Table 2). Picon & Vichy (2003; also Nenna et al. 1997) argued cogently that this glass is compositionally of the Levantine type, and indeed in terms of major elements it is very close. However, the match is not exact; the raw glass recovered so far from Levantine workshops has higher alumina than Roman glass from western Europe (Freestone et al. 2000). Variations in major element composition of this magnitude are found in primary glasses made in the Levant (e.g. Fig. 7). However, recent work using oxygen isotopes also suggests a mismatch, although this involves comparison of data from different laboratories and must be regarded as provisional (Leslie et al. 2006). One problem is that glasses of different date are being compared. The primary furnaces discovered so far appear to be late Byzantine-early Islamic, several centuries later than the Roman glass. These relatively minor differences might be accounted for by the variations in sand composition along the eastern Mediterranean coast (Fig. 5), and it is also noted that the sedimentary load of the Nile has varied significantly over the past 7000 years because of climatic change (Krom et al. 2002), which may have resulted in some variation in the sediment supply over the several centuries under consideration here. However, given the indication by Pliny (Eicholz 1962) that glass was also made in Italy and the western provinces, some prudence is needed. It is clear that glass production in the Roman period was massive: a single building, the Baths of Caracalla in Rome, contained 16 900 m 2 of glass mosaic (Stern 1999), which would represent around

212

I.C. FREESTONE

200 t of raw glass. That the installations that produced glass on such a scale have not yet been discovered is surprising, and is an indication of the limitations of our current understanding of Roman glass production.

Towards the provenance of plant ash glasses Recent work suggests that it will be possible to group the compositions of plant ash (highmagnesia) glasses according to source, even if it will not be possible to identify the locations of the sources in all cases. In particular, production assemblages of plant ash glass from the eastern Mediterranean region dating from the ninth to eleventh centuries have been analysed, for example, glass melted at el-Raqqa in Syria (Henderson et al. 2004), glass from the large primary furnaces at Tyre (Freestone 2002), lumps of raw glass and cullet at Banias (Freestone et al. 2000) and raw lumps and cullet from the Serqe Limani wreck (Brill 1999). These demonstrate that plant ash glasses fall into coherent compositional production groups in terms of their major element compositions (Freestone 2002; Henderson 2003). Furthermore, as shown in Figure 2, it is possible that the blanket term 'plant ash glass' conflates different technological traditions that, when fully understood, will help to refine approaches to this type of glass (see also Henderson et al. 2004). Plant ash glass is more complex than glass produced from mineral soda. Whereas mineral soda typically contributes less than 25% of a glass composition and is essentially a diluent of the silica-bearing component, plant ash is a complex material, carrying many minor and trace elements at similar levels of abundance to the source of silica. Elementary considerations suggest that it makes up at least 30% of a glass batch and, given that plant ashes frequently contain combined silica, alumina and iron in excess of 10% (Brill 1970, 1999), it is probable that in some cases 40% of the material of a plant ash glass was derived from the plant ash itself. It is not inconceivable therefore that the same silica source may have been fused with different plant ashes at different times, and it is certain that the same types of plant ash were fused with different silica sources, as there was an extensive trade in plant ash in the medieval period, when ashes from Syria were imported into Europe to make glass (Ashtor & Cevidalli 1983; Jacoby 1993). In addition, as indicated above, it is still not entirely clear how the

composition of the final glass plant ash corresponds to that of the raw material batch. In spite of the complex nature of plant ash glass production, provisional investigations into the trace element and isotopic compositions of plant ash glasses indicate that they can be very variable and suggest that there is considerable potential for the discrimination of different production centres. For example, trace element distributions in three Islamic-period glasses from Ras al Hadd, Oman, are shown in Figure 13, from a pilot study carried out by the author. They show clear differences in composition, which are large compared with those observed in low-magnesia glasses, and are likely to correspond to different origins of the glass types. The isotopes of strontium and oxygen also show some potential to distinguish plant ash glass groups. Strontium can be expected to have been derived from the plant ash used to make the glass, and in turn this will reflect the bioavailable strontium of the soil on which the plants grew. Oxygen, on the other hand, will be more closely related to the silica source, as the majority of oxygen enters the glass with the silica. 87Sr/86Sr and B180 are thus complementary sources of information, and offer considerable potential for the investigation of plant ash glass when employed together. Figure 14 presents 87Sr/86Sr v. ~180 for three groups of plant ash glass from Syria-Palestine, dated to between the ninth and the 13th centuries. The material from Tyre, Lebanon, is from a primary production site (Freestone 2002), that from Bahias, Israel, is raw glass from a secondary production site (Freestone et al. 2000) and that from el-Raqqa, Syria, from a workshop that was apparently involved in primary and secondary

2.50

-. ~ . -

35456U

-x-- 35448V

2.00

354445

' • 1.50 ~.oo 0.50

0.00

Ga

t

I

t

I

I

t

I

I

I

Rb

Sr

Y

Zr

Ba

La

Ce

Pr

Nd

Fig. 13. Trace element distributionsin three glass vessels from Ras ai Hadd, Oman.

Th

FIRST M I L L E N N I U M AD GLASS P R O D U C T I O N

x

x

x x

o

13

o o A

o

k 11 0.7076

I O 7078

I 07080

I 0.7082

I

07084

07086

87SrlSSSr Fig. 14. Strontium and oxygen isotope data for three groups of plant ash glass.

production (Henderson et al. 2004). Isotopic data are from several sources (Freestone et al. 2003; Henderson et al. 2005; Leslie et al. 2006). It can be seen that Tyre and Raqqa glasses are well separated by 87Sr/S6Sr, suggesting that the plant ashes were derived from different regions. Coupled with a suggestion of a difference in oxygen isotope ratios, which needs confirmation with more data, it appears that these glasses were produced in different areas from two different sets of raw materials. Although the Raqqa glasses are separated from those of Tyre in terms of ~180, they overlap in terms of 87Sr/86Sr (Fig. 14). This is likely to be coincidental, but it could suggest the exploitation of plant ash from the same region of Syria. This would not be surprising, given the widespread trade in plant ash. The difference in ~180 is to be expected, as Henderson et al. (2005) interpreted the Raqqa glasses to have been made using quartz pebbles, whereas sand was the starting material for the Tyre glasses (Fig. 4). The interpretation that these two groups of glasses may have been made using the same alkali, but different silica sources, whether or not it is proved ultimately to be correct, would not have been possible without the isotopic data, and indicates the potential of the technique.

Conclusions Glass provides an excellent example of the importance of the production model in the interpretation of archaeometric data. The recognition that production was divided into primary and secondary workshops is at last allowing compositional data on glass from the period of interest to be interpreted in terms of origin.

213

Furthermore, it defines new questions and allows the construction of appropriate sampling strategies for problem resolution. Within the framework of a realistic production model, an understanding of the geochemistry of the raw materials is invaluable; it explains the choices made by the glassmakers and offers approaches to address problems of technology and origin. Furthermore, by providing an explanation of compositional variation, it strengthens confidence in interpretation and allows extrapolation. In addition, the chemistry of glass is in some respects more complex than that of claybased ceramics, as many different components can be added by the glassmaker in the form of colorants and opacifiers. Glass may also be recycled. These practices can have a significant influence on composition and must be allowed for if compositional groupings are to be meaningfully interpreted. By taking into account the environmental and cultural factors that determine the composition of glass, it is possible to develop and extend approaches to the determination of glass provenance. Significant progress in this area has been achieved and more may be expected over the next few years. I thank an anonymous referee for his helpful comments on this paper. I am grateful to the many archaeologists who have advised on and provided access to material for analysis, especially Y. Gorin-Rosen for her hospitality and generosity. M. J. Hughes, K. A. Leslie, K. Matthews and C. P. Stapleton are thanked for their collaboration in the laboratory. The late T. L. Martin generously funded many analyses through the Renaissance Trust, and S. G. E. Bowman, Keeper of Scientific Research at the British Museum, gave her unstinting support.

References ALDSWORTH, F., HAGGARTY, G., JENNINGS, S.

WHITEHOUSE, D. 2002. Medieval glassmaking at Tyre, Lebanon. Journal of Glass Studies, 44, 49-66. ASHTOR, E. & CEVIDALLI,G. 1983. Levantine alkali ashes and European industries. European Journal of Economic History, 12, 475 -522. BARBER, D. J. & FREESTONE,I. C. 1990. An investigation of the origins of the colour of the Lycurgus Cup by analytical transmission electron microscopy. Archaeometry, 32, 33-45. BATESON, H. M. and TURNER, W. E. S. 1939. A note on the solubility of sodium chloride in a sodalime-silica glass. Journal of the Society of Glass Technology, 23, 265-267T. BRmL, R. H. 1967. A great glass slab from ancient Galilee. Archaeology, 20, 89-95. BRILL, R. H. 1970. The chemical interpretation of the texts. In: OPPENHEIM, A. L., BRILL, R. H.,

214

I.C. FREESTONE

BARAG, D. & VON SALDERN, A. (eds) Glass and Glassmaking in Ancient Mesopotamia. Coming Museum of Glass, New York, 105-128. BRILL, R. H. 1987. Chemical analysis of some early Indian glasses. In: BHARDWAJ, H. C. (ed.) Archaeometry of Glass. Proceedings )(IV International Congress on Glass, New Delhi, 1986. Indian Ceramic Society, Calcutta, 1-25. BRILL, R. n. 1988. Scientific investigations. In: WEINBERG, G. D. (ed.) Excavations at Jalame: Site of a Glass Factor), in Late Roman Palestine. University of Missouri, Columbia, 257-294. BRILL, R. H. 1995. Chemical analysis of some glass fragments from Nishapur in the Coming Museum of Glass. In: KROGER, J. (ed.) Nishapur--Glass of the Islamic Period. Metropolitan Museum of Art, New York, 211-233. BRILL, R. H. 1999. Chemical Analyses of Early Glasses. Coming Museum of Glass, New York. BRILL, R. H. & MARTIN, J. H. (eds) 1991. Scientific Research in Early Chinese Glass. Coming Museum of Glass, New York. BRILL, R. H., CLAYTON, R. N., MAYEDA, T. K. & STAPLETON, C. P. 1999. Oxygen isotope analyses of early glasses. In: BRILL, R. n. (ed.) Chemical Analyses of Earl), Glass. Coming Museum of Glass, New York, 303-322. EICHOLZ, D. E. 1962. Pliny--Naturat Histor3, Books 36-37. Loeb Classical Library, Harvard, MA. EMERY, K. O. & NEEV, D. 1960. Mediterranean beaches of Israel. Geological Survey Israel Bulletin, 26, 1-23. FoY, D., VIEHY, M. & PICON, M. 2000. Lingots de verre en M6diterran6 occidentale. Annales du 14e Congrbs de I'Association pour l'Histoire du Verre. AIHV, Amsterdam, 51-57. FoY, D., PICON, M., VIEHY, M. & THIRION-MERLE,V. 2003a. Charact6risation des verres de la fin de l'Antiquit6en M6diterran6e occidentale: l'emergence de nouveaux courants commerciaux. In: FoY, D & NENNA, M.-D. (eds) Echanges et Commerce du Verre dans le Monde Antique, Monique Mergoil, Montagnac, 41-85. FoY, D., PICON, M. & VICHY, M. 2003b. Verres Omeyyades et Abbasides d'origine Egyptienne: les temoignages de l'arch6ologie et de l'arch6ometile. Annales du 15e Congrks de l'Association Internationale pour l'Histoire du Verre. AIHV, Nottingham, 138-143. FREESTONE, I. C. 1991. Technical examination of NeD-Assyrian glazed wall plaques, lraq, 53, 55-58. FREESTONE, I. C. 2002. Composition and affinities of glass from the furnaces on the island site, Tyre. Journal of Glass Studies, 44, 67-77. FREESTONE, I. C. 2003. Primary glass sources in the mid-first millennium AD. Annales du 15e congrks de l'Association Internationale pour l'Histoire du Verre. AIHV, Nottingham, 111 - 115. FREESTONE, I. C. & GORIN-ROSEN, Y. 1999. The great glass slab at Beth She'arim: an early Islamic glass-making experiment'?. Journal of Glass Studies, 41, 105- l 16.

FREESTONE, I. C., GORIN-ROSEN, Y. & HUGHES, M. J. 2000. Composition of primary glass from Israel. In: NENNA, M.-D. (ed.) Ateliers primaires et secondaires de verriers du second millinaire av. J.-C. au Moyen-Age. Travaux de la Maison de l'Orient Mgditerran6en, 33, 65-84. FREESTONE, I. C., PONTING, M. & HUGHES, M. J. 2002a. Origins of Byzantine glass from Maroni Petrera, Cyprus. Archaeomett)', 44, 257-272. FREESTONE, I. C., GREENWOOD, R. • GORIN-ROSEN, Y. 2002b. Byzantine and early Islamic glassmaking in the Eastern Mediterranean: production and distilbution of primary glass. In: KORDAS, G. (ed.) Hvalos- Vitrum- Glass. Histor3' Technology and Conservation of Glass and Vitreous Materials in the Hellenic WoHd. 1st International Conference, Rhodes, Greece, 1 - 4 April 2001. Glasnet, Athens, 167-174. FREESTONE, I. C., LESLIE, K. A., THIRLWALL, M. & GORIN-ROSEN, Y. 2003. Strontium isotopes in the investigation of early glass production: Byzantine and early Islamic glass from the Near East. Archaeometo', 45, 19-32. FREESTONE, I. C., WOLF, S. & THIRLWALL, M. 2005. The production of HIMT glass: elemental and isotopic evidence. Annales du 16e Congr~s de I'Association lnternationale pour l'Histoire du Verre. AIHV, Nottingham, 153-157. GLOVER, I. & HENDERSON,J. 1995. Early glass in South and South East Asia and China. In: SCOTT, R. & GuY, J. (eds) South East Asia and China: Art, Interaction and Commerce. Percival David Foundation of Chinese Art, Colloquies on Art and Archaeology in Asia, 17, 141 - 169. GOLDSMITH, V. & GOLIK, A. 1980. Sediment transport model of the southeastern Mediterranean coast. Marine Geology, 37, 147-175. GORIN-ROSEN, Y. 1995. Hadera, Bet Eli'ezer. Excavations and Surveys in Israel, 13, 42-43. GORIN-ROSEN, Y. 2000. The ancient glass industry in Israel: summary of new finds and new discoveries. In: NENNA, M.-D. (ed.) Ateliers primaires et secondaires de verriers du second millinaire av. J.-C. au Moyen-Age. Travaux de la Maison de l'Orient Mgditerran6en, 33, 49-63. GRATUZE, B. 8z BARRANDON, J.-N. 1990. Islamic glass weights and stamps: analysis using nuclear techniques. Archaeometr3', 32, 155-162. HENDERSON, J. 2003. Glass trade and chemical analysis: a possible model for Islamic glass,production. ln: FoY, D. & NENNA, M.-D. (eds) Echanges et Commerce du Verre dans le Monde Antique. Monique Mergoil, Montagnac, 109-124. HENDERSON, J., MCLOUGHLIN,S. D. & MCPHAIL, D. S. 2004. Radical changes in Islamic glass technology: evidence for conservatism and experimentation with new glass recipes from early and middle Islamic Raqqa, Syria. Archaeometry, 46, 439-468. HENDERSON, J., EVANS, J. A., SLOANE, H. J., LENG, M. J. & DOHERTY, C. 2005. The use of strontium, oxygen and lead isotopes to provenance ancient glasses in the Middle East. Journal of Archaeological Science, 32, 665-674.

FIRST MILLENNIUM AD GLASS PRODUCTION HUNTER, J. & HEYWORTH, M. 1998. The Hamwic Glass. Council for British Archaeology, York, Research Reports, 116. JACKSON, C. M., HUNTER, J. R., WARREN, S. E. & COOL, H. E. M. 1991. The analysis of bluegreen glass and glassy waste from two RomanoBritish glass-working sites. In: PERNICKA, E. & WAGNER, G. A. (eds) Archaeometry '90. Birkhauser, Basel, 295-305. JACOBY, D. 1993. Raw materials for the glass productions of Venice and the Terraferma about 1370-about 1460. Journal of Glass Studies, 35, 65 -90. KELLER, D. 2005. Social and economic aspects of glass recycling In: BRUNN, J., CROXFORD, B. & GRIGOROPOULOS, D. (eds) TRAC 2004: Proceedings of the 14th Annual Theoretical Roman Archaeology Conference. Oxbow, Oxford, 65-78. KROM, M. D., STANLEY, J. D., CLIFF, R. A. & WOODWARD, J. C. 2002. Nile river sediment fluctuations over the past 7000 yr and their key role in sapropel development. Geology, 30, 71-74. LESLIE, K. A., FREESTONE, I. C., LOWRY, D. & THIRLWALL, M. 2006. Provenance and technology of Near Eastern glass: oxygen isotopes by laser fluorination as a complement to strontium. Archaeometry (in press). LILYQUIST, C. & BRILL, R. H. 1993. Studies in Early Egyptian Glass. The Metropolitan Museum of Art, New York. MOOREY, P. R. S. 1994. Ancient Mesopotamian Materials and Industries. Clarendon Press, Oxford. NENNA, M.-D., VICHY, M. & PICON, M. 1997. L'atelier de verrier de Lyon, du Ier sibcle aprbs J.-C., et l'origine des verres 'Romains'. Revue d'Archbombtrie, 21, 81-87. NENNA, M. -D., l~CON, M. & V1CHY, M. 2000. Ateliers primaires et secondaires en l~gypte l'rpoque grrco-romaine. In: NENNA, M.-D. (ed.) La Route du Verre: Ateliers primaires et secondaires de verriers du second milldnaire av. J.-C. au Moyen-Age, Travaux de la Maison de l'Orient Mdditerranden, 33, 97-112. NENNA, M.-D., PICON, M., THIR1ON-MERLE, V. & VICHY, M. 2005. Ateliers primaires du Wadi Natrun: nouvelles drcouvertes. Annales du 16e Congrbs de l'Association Internationale pour l'Histoire du Verre. AIHV, Nottingham, 59-63. NICHOLSON, P. T., JACKSON, C. M. & TROTT, K. M. 1997. The Ulu Burun glass ingots, cylindrical vessels and Egyptian glass. Journal of Egyptian Archaeology, 83, 143-153. OPPENHEIM, A. L., BRILL, R. H., BARAG, D. & yon SALDERN, A. 1970. Glass and Glassmaking in Ancient Mesopotamia. Coming Museum of Glass, New York. PICON, M. & VICHY, M. 2003. D'Orient en occident: l'origine du verre ~ l'~poque romaine et durant le haut Moyen Age. In: FoY, D. & NENNA, M.-D. (eds) Echanges et Commerce du Verre dans le Monde Antique. Monique Mergoil, Montagnac, 17- 31. POMERANCBLUM, M. 1966. The distribution of heavy minerals and their hydraulic equivalents in sediments of the Mediterranean continental shelf of

215

Israel. Journal of Sedimenta~ Petrology, 36, 162-174. READE, W., FREESTONE, I. C. & SIMPSON, ST. J. 2005. Innovation or continuity? Early first millennium BCE glass in the Near East: the cobalt blue glass from Assyrian Nimrud. Annales du 16e Congrks de l'Association pour l'Histoire du Verre. AIHV, Nottingham, 23-27. REHREN, TH. 2000. Rationales in Old World base glass compositions. Journal of Archaeological Science, 27, 1225-1234. REHREN, TH. & PUTSCH, E. B. 1997. New Kingdom glass-melting crucibles from Qantir-Piramesses. Journal of Egyptian Archaeology, 83, 127-141. SAYRE, E. V. 1963. The intentional use of antimony and manganese in ancient glasses, ln: MATSON, F. R. & RINDONE, G. E. (eds) Advances in Glass Technology, Part 2. Plenum, New York, 263-282. SAYRE, E. V. & SMITH, R. W. 1961. Compositional categories of ancient glass. Science, 133, 1824-1826. SHORTLAND, A. J. 2004. Evaporites of the Wadi Natrun: seasonal and annual variation and its implication for ancient exploitation. Archaeometry, 46, 497-516. SHORTLAND, A. J., SCHACHNER,t . , FREESTONE, I. C. & TITE, M. 2006. Natron as a flux in the early vitreous materials industry--sources, beginnings and reasons for decline. Journal of Archaeological Science (in press). SHUGAR, A. & REHREN, TH. 2002. Formation and composition of glass as a function of firing temperature. Glass Technology 43C, Proceedings XIX International Congress on Glass, Edinburgh, 1-6 July 2001, Society of Glass Technology, Sheffield, 145-150. SMITH, R. W. 1963. Archaeological evaluation of analyses of ancient glass. In: MATSON, F. R. & RINDONE, G. E. (eds)Advances in Glass Technology, Part 2. Plenum, New York, 283-290. STANLEY, D. J., MART, Y. & NIR, Y. 1997. Clay mineral distributions to interpret Nile cell provenance and dispersal: II. Coastal plain from Nile delta to northern Israel. Journal of Coastal Research, 13, 506-533. STAPLETON, C. P. & SWANSON, S. E. 2002. Chemical analysis of glass artefacts from Iron Age levels at Hasanlu, northwestern Iran. Glass Technology 43C, Proceedings of the XIX International Congress on Glass, 151-157. STERN, E. M. 1999. Roman glassblowing in a cultural context. American Journal of Archaeology, 103, 441-484. TAL, O., JACKSON-TAL, R. E. & FREESTONE, I. C. 2004. New evidence of the production of raw glass at Late Byzantine Apollonia-Arsuf (Israel). Journal of Glass Studies, 46, 51-66. VERITA, M. 1985. L'invenzione del cristallo Muranese: una verfica analitica delle fonti storiche. Rivista delle Stazione Sperimentale del Vetro, 1, 17-36. WEDEPOHL, K. H. 1995. The composition of the continental crust. Geochimica et Cosmochimica Acta, 59, 1217-1232. WEDEPOHL, K. H. 2003. Glas in Antike und Mittelalter. Schweizbart, Stuttgart.

216

I.C. FREESTONE

WEDEPOHL, K. H. & BAUMANN, A. 2000. The use of marine molluskan shells for Roman glass and local raw glass production in the Eifel area (Western Germany). Naturwissenschaften, 87, 129-132.

WH[TEHOUSE. D. 1999. Glass in the Epigrams of Martial. Jountal of Glass Studies, 41, 73-81. WHITEHOUSE, D. 2002. The transition from natron to plant ash in the Levant. Journal of Glass Studies, 44, 193-196.

Medieval stained glass windows from Pavia Carthusian monastery (northern Italy) V A L E R I A M A R C H E S I l, E L I S A B E T T A NEGRI 1, B R U N O M E S S I G A 1'2 & M A R I A PIA R I C C A R D I 1'2

lDipartimento di Scienze della Terra, Universitb degli Studi di Pavia, via Ferrata 1, 27100 Pavia, Italy (e-mail: [email protected]) 2Centro Interdipartimentale di Studi e di Ricerche per la conservazione dei Beni Culturali, Universith degli Studi di Pavia, via Ferrata 1, 27100 Pavia, Italy Abstract: Stained glass windows of the Carthusian monastery (Certosa) of Pavia (15th century) were investigated during a recent restoration campaign, to define chemical compositions and to understand manufacture techniques. Analyses were performed on six stained glass windows, representing S. Caterina (St. Catherine), S. Gregorio (St. Gregory), S. Agostino (St. Augustine), la Nativith (the Nativity), S. Bernardo (St. Bernard) and S. Girolamo (St. Jerome), and on two undecorated windows. The methodological approach couples micro-sampling with micro-analyses. The detailed chemical analysis reveals four compositionally distinct glass-types: mixed alkali, soda and two potash glass types. These last two types display a rather constant MgO content but variable CaO content and define two main compositional clusters, CaO rich (Type II) and CaO poor (Type III), that have been ascribed to the pristine glass. The compositional boundary of Type II and Type III glass fits those already defined in recent literature. The mixed alkali and soda glass are considered to be related to later repairs. The narrow compositional cluster represents a good statistical result defining the composition of the pristine glass panes and allows compositional comparisons with stained glass from coeval cathedrals of Northern Europe (Germany, Switzerland and France). Close similarities are found to stained glass from Rouen and St. Maur des Fosses.

The history of flat glass, from its origins to the beginning of the use of stained glass windows, has been recently wonderfully outlined by Dell'Acqua (2003). In late antiquity flat glass was used to close windows and as a valuable complement to architectonic decorations. In Byzantine Italy flat glass had light colours and the addition of colouring oxides was avoided. In Europe, some examples of stained windows made during the Byzantine period are situated in Mtistair in Switzerland (ninth century), St. Maur des Fosses in France (ninth century), Psalmodi in France (ninth century) and Sion in Switzerland (fifth to sixth century). In Italy the first example of a stained glass window seems to be the findings unearthed in the St. Vitale Church in Ravenna (sixth century) (PallotFrossard 1998). Until the l lth century, flat glass was used only as a coloured screen for functional and religious purpose (Dell'Acqua 2003). The use of stained glass windows not only as decorative elements started during the Carolingian

age, and techniques of construction were described by the monk Theophilus (about AD 1100) (Caffaro 2000). During Carolingian times the continuous demand for glass caused a technological change in glass-making, with the conversion of production from Na-Ca-glass to K-Ca-glass, using ash from trees instead of ash from halophytic plants or the import of trona- soda or soda raw glass from Eastern Mediterranean countries (Wedepohl 1997). The origin of this new tradition was not related to a sudden change, but resulted from the continuity of technological and decorative traditions, taking advantage of the glass composition and technical experience of the Roman imperial period (Krueger & Wedepohl 2003). In the Romanesque period (12th century) as the demand for churches increased so did the production of decorative stained glass windows. Early Romanesque style stained glass was influenced by the linear patterning and severe frontality found in Byzantine times.

From: MAGGETTI,M. & MESSIGA,B. (eds) 2006. Geomaterials in CulturalHeritage. Geological Society, London, Special Publications, 257, 217-227. 0305-8719/06/$15.00 © The Geological Society of London 2006.

218

V. MARCHESI E T A L .

The flourishing of stained glass was certainly related to the building of European gothic cathedrals. Wedepohl (1997) estimated that a minimum amount of 10000 tons of glass were used in 200 years to built churches and monasteries throughout Germany, starting from the 13th century. The history of the Pavia Certosa, dedicated to the Blessed Virgin Mary of Graces, starts in 1396, by the will of Gian Galeazzo Visconti, Duke of Milan, to fulfil a vow of his wife Caterina. The stained glass windows represent an important historical record of glass-making art between AD 1428 and 1480. In 1472 the monastery with its cloisters and the church were almost finished, and the faqade was begun. From 1428, during an early building stage (Pirina 1995) Ambrosio de Lacte and Johanninus de Balestriis 'dictus Malandrinus' were master glassworkers. In a second building stage, 14751480, in the Certosa courtyard, were employed the most celebrated master glass-makers working on the Milan Cathedral windows: Antonio da Pandino, Cristoforo de Mottis and Nicol6 da Varallo. Work went on with indoor decorations and at the end of 15th century the monastery was finished. With a solemn ceremony it was consecrated on 3 May 1497, a hundred years after its foundation. From the 16th century, wars and shortage of food led to the deterioration and neglect of the stained glass windows, and during the following centuries replacements and repairs were made. The 13 stained glass windows, with hagiographical images, represent one of the most decorative attractions of the church. The composition of the Certosa glass (Messiga & Riccardi 2001a) is similar to that of medieval glass from Europe. The stained glass windows of Pavia Certosa belong to the medieval technological and historical context and, thanks to their high quality, may be used to define glassmaking and -working technology. Moreover, such a study may help to unravel aspects of the production and trade of flat glass in Northern Italy. A companion paper (Messiga & Riccardi 2006) deals with the alteration processes of the stained glass surfaces. This paper aims to give a complete overview of the compositional variability of glass panes from stained glass windows, based on both a new analytical dataset and some already published data (Messiga & Riccardi 2001a). The availability of an analytical self-consistent database leads to chemical classification of medieval flat glass production, providing information that helps to reconstruct archaeology, to define arthistorical attributes, to enhance our knowledge

on the evolution of glass technology, and to trace how glass panes and/or glass technology were traded through the possible so-called 'glass ways'. The study was undertaken during a recent restoration campaign. Analyses were performed on six stained glass windows, representing St. Catherine, St. Gregory, St. Augustine, the Nativity, St. Bernard and St. Jerome, and on two undecorated windows. The analyses were performed to: (1) obtain a wide dataset for glass analyses; (2) define chemical parameters that characterize Certosa glass; (3) compare them with the available literature data. The classification of Brill & Pongracz (2004) has been adopted in presenting and discussing the glass composition. Methods Chemical analyses have been performed to gain information about when and where the glass was made, about the materials from which it was made and about how it was worked and traded in antiquity. The understanding of preindustrial processes involving glass-making and -melting, together with unravelling of the expedients used in the control of colour attributes of glass allows the reconstruction of ancient glass technology. In scientific studies of ancient glass, analyses are commonly carried out using 'bulk analysis techniques' whereas microprobe spot-analyses are performed on small pieces (Brill 1999). The bulk analysis technique implies the assumption of the homogeneous nature of glass. Although the first electron probe microanalysis (EPMA) of glass was performed in Brill & Moll (1961), only in recent years have the relationships between microtextures and chemical composition, determined using SEM (scanning electron microscope) and EPMA, have been developed as potential tools to define technological expedients used in a glass productive cycle (Messiga & Riccardi 2001b). The use of inductively compled plasma-mass spectrometry (ICP-MS) coupled with laser ablation was proposed and applied to glass studies (Gratuze 1994). According to Verith et al. (1994) (Spitzer-Aronson 1976; Barrera & Velde 1989; Mtiller et al. 1994), quantitative chemical analyses of ancient glass artefacts reveal important information on the classification and history of glass. EPMA coupled with SEM and ICP-MS has advantages in terms of smaller sample size and the ability to distinguish and recognize inclusions, opacifiers and weathered areas. Moreover, samples are not destroyed, but remain available for further investigations.

MEDIEVAL STAINED GLASS

In this study an already tested analytical procedure (Messiga & Riccardi 2001a,b) is applied to obtain textural and compositional data through investigations at different observational scales. The optical microscope was e m p l o y e d to obtain textural characterization of the samples (i.e. the texture o f flashed glass), and to select and take microsamples ( < 100 Ixm in size). Forty-eight samples from six stained glass w i n d o w s were collected, chiefly under lead binders, during the restoration campaigns. Fragments were incorporated within D Y R A C T ® (a dental curative composer), then polished with diamond paste (to 0.25 txm) and carbon-coated. Some of the microanalyses were performed on a JEOL JXA 840A electron analyser equipped with three wavelength-dispersive spectrometers (WDS) (TAP, PET, LIF analysing crystals) and one Si (Li) energy-dispersive

219

spectrometer (Be-window). Analytical conditions were 20 kV accelerating voltage and 20 nA; spot size was 5 Ixm. Counting time was 20 s for all elements except Mn and Fe (40s). Data collected by the WDS were processed with the TASK correction program. Other samples were analysed by EPMA in the wavelength-dispersive mode, on an ARL-SEMQ instrument located at Dipartimento di Scienze della Terra, University of Modena, operated at 20kV, 2 0 n A beam current and with a defocused beam (spot size 20 Ixm). Natural standards were used: microcline (Si, AI, K), diopside (Ca), ilmenite (Fe, Ti), chromite (Cr), albite (Na), spessartine (Mn) and olivine (Mg). Analytical data have been collected and processed by the 'Probe' program. The results of the chemical analyses are reported in Table 1. The expected error for major and minor elements is 3% and 10%, respectively.

Table 1. Representative analyses of glass panes Sample St. Catherinel

Colour

Colourlesst Red St. Catherine 5 Colourless St. Catherine 7a Red St. Catherine 7b Blue St. Catherine 8 Colourless St. Catherinel2 Yellow St. Gregory8 Colourless St. Gregory 1 Yellow St. Gregoryla Colourlesst Redt St. Gregory 2 Greent Colourlesst St. Gregory5 Green St. Bernard4 Colourless St. Bernard5 Yellow St. Bernard Bluet Violett Nativity4 Yellow Nativity6 Yellow St. Augustine9 Colourless St. Jerome3 Greent Colourlesst St. Jerome6 Red St. Jerome2* Pink St. Jerome 8* Yellow St. Jerome l* Colourlesst Redt St. Jerome7* Colourlesst Redt St. Jerome5* White St. Jerome6* White St. Jerome4* Colourlesst Light pinkt

SiO2

TiO2 AleO3 MnO MgO

CaO

Na20

K20

P205 Fe203 CuO

51.37 51.39 55.35 52.15 51.50 56.72 51.69 55.36 52.98 51.85 50.80 49.73 52.87 56.78 56.30 52.68 52.64 47.59 53.37 51.97 57.12 52.10 52.96 55.48 62.63 58.07 53.14 53.94 55.70 52.53 53.48 55.42 55.38 52.52

0.00 0.00 0.39 0.00 0.75 0.24 0.90 0.00 0.42 0.55 0.57 0.82 0.00 0.00 0.14 0.13 0.59 0.78 0.32 0.39 0.25 0.13 0.10 0.00 0.10 0.00 0.16 0.21 0.22 0.15 0.00 0.24 0.10 0.20

17.09 16.49 12.71 13.83 15.80 15.89 18.04 13.06 16.92 13.71 14.83 13.63 12.91 18.56 12.26 13.36 18.86 16.72 11.52 12.29 t2.05 12.93 13.38 15.39 8.70 24.18 12.74 12.91 12.26 12.83 18.21 14.58 11.02 17.38

0.42 0.34 2.42 2.61 1.02 1.41 0.78 2.25 1.75 2.37 2.40 1.39 1.46 0.70 3.48 4.10 1.21 1.17 2.84 2.27 3.19 3.53 3.41 0.84 19.27 2.70 4.09 3.51 3.82 5.11 0.76 0.92 0.35 2.66

15.19 14.01 16.36 16.88 15.29 12.33 14.23 14.98 13.72 15.95 17.69 16.75 15.89 12.50 15.08 14.93 12.66 13.16 17.80 17.63 14.81 15.10 15.67 14.74 1.86 4.67 15.29 15.53 15.29 14.66 12.49 13.66 16.29 15.55

4.35 5.06 4.18 5.33 3.89 3.38 4.15 0.00 3.74 4.98 5.34 3.76 4.19 3.79 4.25 5.39 3.90 3.95 4.67 4.94 4.18 5.03 5.35 3.72 0.10 3.04 5.61 4.92 4.83 5.20 4.19 5.08 5.03 2.48

2.21 2.52 1.80 1.49 2.59 2.48 3.03 1.57 2.42 1.45 1.38 2.28 2.42 2.77 0.83 0.56 2.54 2.52 1.89 2.57 0.78 0.52 0.72 2.01 0.40 3.09 0.88 1.07 0.51 1.22 1.75 1.08 1.49 2.08

0.66 0.72 0.86 0.93 0.93 1.28 1.29 1.15 1.26 0.78 0.88 0.73 0.73 0.59 1.19 1.27 0.62 6.74 1.00 1.00 1.27 1.23 1.25 0.82 1.76 0.37 0.81 1.03 0.84 0.80 0.64 0.57 0.89 1.15

Electron microprobeanalyses abundance are given as wt%. *Analyses from the lower panel of St. Jerome (numbers are as in Fig. 1). t Flashed glass, reported in the table in adjacent rows.

7.02 7.45 4.53 5.25 4.56 5.02 5.57 5.58 5.68 5.51 5.19 4.33 4.89 6.12 5.49 6.34 6.38 5.70 5.38 5.60 5.59 5.30 6.18 5.89 2.96 2.77 5.58 5.15 5.01 5.48 7.47 6.99 4.87 4.54

0.81 0.82 0.42 0.00 0.29 0.67 0.00 0.00 0.48 0.12 0.10 0.30 0.32 0.29 0.23 0.32 0.23 0.16 0.41 0.38 0.24 0.64 0.22 0.00 0.74 0.00 0.28 0.11 0.39 0.39 0.00 0.45 0.40 0.72

0.12 0.73 0.00 0.41 0.31 0.00 0.14 0.00 0.00 0.00 0.00 4.22 0.06 0.00 0.00 0.00 0.10 0.00 0.24 0.00 0.00 1.38 0.00 0.05 0.00 0.00 0.27 1.15 0.37 1.05 0.00 0.00 0.16 0.00

220

V. MARCHESI ET AL.

On a stained glass window the intersection pattern of lead binders and their alteration rate allow us to identify the timing of replacements of glass panes and/or of lead binders themselves. This allows the reconstruction of a relative age-sequence of the substitutions made during restorations. On the basis of this relative chronology, sampling can be used to decipher glass composition changes and relate them to the replacements that occurred during the history of the window. During restoration procedures, the history of manufacture of the lower panel of the St. Jerome stained window (Fig. l a and b) was reconstructed considering: (1) the relationships between lead binders; (2) the thickness of the panes together with their cutting characters; (3) the surface degradation pattern; (4) the glass chemical compositions; (5) the microstratigraphy (i.e. flashed glass); (6) decorative style and drawings. This task was carried out with the irreplaceable co-operation of the restorer Arch. Laura Morandotti. The method gives evidence to distinguish different productions, even for glass of the same colour, allowing the reconstruction of time-related compositions of glass panes. The use of combined analytical data with the recognition of different ages of the glass panes helps us to distinguish the compositions of original glass (the pristine glass). Finally, these data can be used to make comparisons with other glass from stained glass windows in different contexts. In the following section we will deal with a few selected examples to highlight the fact that stratigraphy, microtextures and compositions represent important tools in deciphering stages in the development of manufacturing technology of a particular glass (e.g. flashed glass), and the recognition of repairs and substitutions.

Results Microtextures

The microtextures of glass panes are related to particular operations that occurred during glassmaking and -melting. Flashed glass, a particular coloured glass composed of a thin coloured layer on a colourless base glass, started to be produced in the 12th and 13th centuries by multi-layering (Newton & Davison 1997) because the glass making up some colours was so dense in tone that it did not allow sufficient light to pass through. Flashing consists of producing a multi-layered glass pane in which the thickness of coloured layers controls the colour tone. Flashed glasses are not coloured throughout their mass, they attain their coloration through

the 'flash' technique of glass blowing. One or more coloured glasses are applied to a colourless or coloured base glass referred to as the 'cartier glass'. The base colour glass is then gathered over the carrier glass and blown out into cylinder shape. The cylinder is then cut, opened out and flattened. Because coloured glass can be produced also by controlling the kiln atmosphere, the expedient of flashing was used to insulate the coloured glass by covering it with a new gather. The precise control of form and colour of the flashed glass thus requires considerable experience, which only the skilled glassblower has. The most common microstructures are mainly the defects: stones, which represent relicts of the batch materials (such inhomogeneities are due to insufficient mixing of the glass during the melting), and bubbles, which are gas inclusions (generated either during melting and/or working and retained by the glass when not sufficiently refined). The flashed glass is composed of sequences of differently coloured glass layers. The detailed structure of multi-layered glass has already been investigated by SEM and EPMA (Messiga & Riccardi 2001a); Figure 2 shows the main characteristics of zoned glass from the stained glass windows of Pavia Certosa. Red-flashed glass has the highest number of textures: sections of flashed glass panes reveal the presence of the 'carrier glass' and one or up to four levels of coloured glass. The red layer varies considerably in thickness (from 90 to 180 Ixm) as well as in its position. In Figure 2, sample 5 (St. Catherine) and sample 7 (St. Catherine) show a red flashed glass in which the red layer is the surface layer. In contrast, in other window panes the red layer is embedded between two colourless glass. Commonly, the red layer never occurs in the exact middle of the glass pane. Figure 2, sample 8, shows a complex layering in which, starting from the left, a colourless layer (extensively altered by pitting) is followed by a red layer (100 Ixm thick) and then by a very thin layering, in which red and colourless glass are intermixed and occur before the colourless cartier glass. The microtextures of the red layers are variable and dominated by the alternation of dark red and orange ribbons, in almost all the red-flashed glass. The observed microstructures alone cannot help in recognizing when and where the glass was produced. However, two tesserae of the St. Catherine stained window (Fig. 2, samples 3 and 6), have identical microstructures (except for the different width of the layers) and could reasonably come from the same pane. Also, some green, pink and blue glass are flashed. Green glass from the St. Gregory

MEDIEVAL STAINED GLASS

221

30 cm F-I M

I II

~ II

Illl

ill

V'

I I III

|m

i[lllllllllllllll l l ~ i i i i IIIIIIIIIIIIIIll, L L I I I l i l I I I ~ ~ i i i i i ii i i-l-l-i-t llllllllllll

m,~ilIIILI/ ill[ll

IIIIII1TI'

.

. . . . " .

u .

. . . • . . . . . . • •

.

\1

"

°

.

.

m .

m/

"'~"Y

• . . .

.

/

. . .

. . . . ,,(""" • • • 'B • f ~ , . . . . /. . . . . . . . /" . . . . • • • •

•

.

~r I . II . / I ~/ I 4 I /

# # l

I I

~

: " : " : "~ : : : : ;f . . . . .

7 . . . .

I

I

I

.'r-c't-.

K~i) ~

.

"

z < .~'.

.

.

30 cm

.

.

.

.

°

.

.

.

.

.

.

.

.

.

.

.

.

~//~/

~1111111 [ 1 1 1 1 1 1 1 t I l l [ l l l I I I l J l l l l l l | l l l

A

B

C

D

E

Fig. 1. (a) Sample positions (numbers) on the lower panel of the St. Jerome window, photograph by courtesy of Arch. Laura Morandotti. The colours of the glass are: 1, red flashed; 2, fuchsia; 4, light pink flashed; 5, 6, white; 7, red flashed; 8, yellow. (b) Colours of glass from the lower panel of the St. Jerome window. A, red tesserae of tunic; B, shape of yellow lion; C, fuchsia tessera; D, light pink flashed; E, colourless.

222

V. MARCHESI ETAL.

~

!:

100 pm

4/ !

l"

Fig. 2. Microtexture of flashed red glass under optical microscope. The white thin lines outline the red glass layer. Samples 1-7 are from the St. Catherine window. Sample 8 is from the St. Jerome window. Because the layers are not perpendicular to the polished surface, the thickness of the coloured bands could be overestimated. White and black dots represent the colourless and the red glass, respectively.

MEDIEVAL STAINED GLASS

windows is flashed by alternating several layers of colourless and green glass. The blue glass of the St. Bernard window, in the image of the Devil lying at the Saint's feet, is flashed with a layer (230 p~m thick) of violet glass, containing up to 7 wt% MnO, as already described by Messiga & Riccardi (2001a). This paper also reports compositional details of red layers, where the red colour is given by small rounded areas (less than 20 p~m across) in which higher concentrations of small copper particles occur.

Chemistry All the data from stained windows and unstained glass are plotted on a K20 v. CaO plot, as proposed by Brill & Pongracz (2004); 85% of analyses plot in a single area, defined as the main compositional cluster (Fig. 3), which can be interpreted once the relative ages of the groups of glass have been defined. The remaining 15% of analyses consist of scattered compositions sometimes forming secondary clusters. The soda-lime (Soda) glass plots in the left part of the diagram. The very high lime glass and particularly the K20 poor glass (Type I) plot in the upper left corner. Towards the fight side of diagram two other groups occur (Types II and III). Differences occur in the lime contents. 25

CaO ~) E 8 D Type I

20

Type II

223

In detail these are: (A) pink Ca-Na glass from the tunic of St. Jerome; (B) and (C) unstained, lozenge-shaped panes from the St. Gregory window; (D) yellow glass from the lion's muzzle in the St. Jerome window; (E) yellow tesserae from the St. Jerome window; (F) opaque white glass from different windows. To define the composition of earlier glass of Certosa stained glass windows we have considered the age relationships between seven glass fragments from the lower panel of the St. Jerome stained glass window. Samples have been collected according to the method previously outlined (Fig. 1 and Table 1). The colours were: white (5, 6), yellow (8), red (1, 7), rask (fuchsia) (2) and light pink (4). Red glass panes are flashed with similar microstratigraphies but with different microtexture and thickness of the coloured layer; the perfect condition of lead binders and the alteration rate of glass panes surfaces suggest their early origin; moreover, their textures show the flashed red layer between two colourless layers. Most probably, the light pink glass (4) is older than the fuchsia pink glass (2), which, in turn, shows its Na character (Fig. 4). This composition is peculiar among medieval glass, and in Northern Italy became ubiquitous after about the 16th century (Brill & Pongracz 2004). In contrast, the light pink glass (4) (Fig. 4) is flashed glass with a K - C a chemical character. It shows a rather wide compositional zoning in which CaO ranges from about 11 wt% in the colourless layer to 17 wt% in the pink layers. The yellow

'Wc pe

25

CaO ]Typel

20

I," ........

Soda

o

1'o

1'5

!i~.'~~i"'~ ! JType,,I

15

K20

JType tl

2s

Fig. 3. CaO v. K20 plot showing the main cluster and the six compositional groups (A-F). Soda: group A, pink Ca-Na glass from tunic of St. Jerome; group B, unstained lozenge-shaped panes from St. Gregory; Type I: group C, unstained lozenge-shaped panes from St. Gregory; group D, yellow glass from lion's muzzle in St. Jerome window; group E, yellow tesserae from St. Jerome window; Type II and Type III: group F, main compositional group of the six stained windows; the opaque-white glass panes from different windows are in the upper part of the main cluster. Compositional fields (Type I, Type II, Type III and Soda) are as defined by Brill & Pongracz (2004).., unstained lozenged-shaped panes; O, stained glass.

10 Soda

K20 0

5

10

15

20

25

Fig. 4. CaO v. K20 plot showing glass compositional fields. Symbols correspond to the sample points shown in Figure 1. A, yellow, sample 8; *, red, sample 1; [-1,red, sample 7; A, white, sample 5; O, white, sample 6 ; . , fuchsia, sample 2; O, light pink, sample 4. Dashed line outlines the composition of the pristine glass.

224

V. MARCHESI E T A L

glass from the lion's muzzle was considered to be a later insertion, on the basis of the decorative style and drawing features. It is a Ca-rich glass and plots close to Type I field. It belongs to a glass typology produced until the 16th century, according to Brill & Pongracz (2004). White opaque glass were sampled from the Saint's tunic (5) and from the base of the column (6). Both are K - C a glasses and have different Ca contents, 18 wt% and 14 wt%, respectively (Fig. 4). Consequently, only red, white and light pink glass represent the compositional trace of the pristine K - C a glass. This validates the methodological approach and allows us to define the pristine compositions. In Figure 4 a dashed line encloses the composition of investigated glass except samples A, B, C, D and E in Figure 3, and defines the compositional field of the early glass. The compositions of the pristine glass from investigated stained glass windows form two main subgroups, Type II (Ca-rich) and Type III (Ca-poor) glass, which partially overlap (Fig. 4). In the early panes (Fig. 5) and (Table 1) the SiO2 content of glass varies between 46.94 and 59.80 wt% and A1203 is always lower than 3.50 wt%. Na20 never exceeds 5.00 wt%, confirming the K-character of glass (K20 up to 18.51 wt%). CaO varies widely from 11.02 to 19.43 wt%, and MgO never exceeds 7.60 wt%. All glass samples are fairly rich in P205, up to 5.57 wt%. CuO and MnO abundances are closely related to colour attributes, and in some samples reach 4.4 and 6.7 wt%, respectively (Fig. 5). The compositional variability defined by other major elements is shown in Figure 6: Figure 6a shows a rather constant MgO content in the pristine glass accompanied by variable content of CaO, and indicates a compositional gap around

the value of 16 wt% CaO. This composition closely matches Type II (Ca-rich) and Type III (Ca-poor) glass. MgO values range between 4 and 7 wt%, although opaque-white panes (St. Catherine, St. Jerome) exceed 7 w t % . Opaque-white panes from the St. Augustine window plot in the Ca-poor composition field and have about 5 wt% MgO. Ca-poor glass shows a broad negative correlation in the N a 2 0 - K 2 0 plot of Figure 6b. All the ancient glass is in three compositional groups with

9 Mgo 8

(a)

7 6 5 4 3 CaO 2 10

11

12

Na~O

Fe20 3 P20~

19

20

o

O oCOn

o

2

o

o

Oo

1 K20

010 7

11

12

13

14

15

16

17

18

19

P~o0

20

(e)

O0 :O:o0 o%

o

11.07-18.51 wt% 0.33-5.11 wt%

m

MgO

CaO

MnO

m

AI203

ii

2

0.42-6.74 wt % 0.51-3.06 wt % 46.94-59.80 wt %

SJO 2

0

1'0

:~0

3'0

4'0

i

10

4.53-7.60 wt %

m

[]

[]

11.02-19.43 wt %

CaO

go

eb

Chemical variability of pristine glass; numbers are wt% oxides.

Fig. 5.

18

0 0

2.48-6.34 wt %

K20 Na20

17

(b)

0

0--5.08 wt %

m

16

O

3

0-1.91wt %

m

15

o°1.o

4

0-1.99 wt %

CuO

14

5

O

PbO

13

11

i

12

i

13

i

14

i

15

i

16

i

17

i

18

i

19

20

Fig. 6. Composition of pristine glass in terms of: (a) stabilizing components, MgO v. CaO plot; (b) flushing components, Na~O v. K20 (group A, St. Bernard, St. Jerome and St. Augustine; group B, The Nativity, St. Jerome and St. Gregory; group C, Ca-rich glass from St. Bernard, The Nativity, St. Jerome and St. Gregory); (c) P205 v. CaO plot.

MEDIEVAL STAINED GLASS some scattered compositions. Group A comprises Ca-poor glass from the St. Bernard, St. Jerome and St. Augustine windows, and has the highest Na20 contents; group B mainly contains Ca-poor glass from the Nativity, St. Jerome and St. Gregory windows; group C contains Ca-rich glass from the St. Bernard, Nativity, St. Jerome and St. Gregory windows, and has the lowest Na20 and K20 contents. Yellow glass from the Nativity window has a compositional zoning ranging from group C to group B which occurs even in a single sample, indicating chemical inhomogeneity of the glass pane. The Ca-poor glass (both red and opaquewhite glass) from group C belongs to the St. Jerome window. The P205 content displays a rather high variability but Ca-rich glass has P2Os contents lower than Ca-poor glass (Fig. 6c). Once each window has been considered it is possible to say that all windows contain both Type II and Type III glass compositions. An example of this is the stained window of St. Catherine, the most analysed window, in which

the composition of glass panes cover all the above defined compositional fields (Fig. 7); this indicates a wide compositional variability of glass panes even within a given stained window. If we consider glass windows where the number of analysed samples is smaller, it is not possible to ascertain if their lower compositional variability is real or is merely a consequence of the poor sampling.

Discussion and conclusions Microsampling of glass under lead binders is a powerful method that does not destroy the aesthetic and artistic appearance of the glass windows and allows us to obtain substantial information on (1) composition of glass panes, (2) microstructures of ancient glass and (3) compositional differences from or similarities to coeval stained glass, when microtextural and microchemical analyses are combined. The glass from Pavia Certosa reveals four compositionally distinct types: Soda, Type I,

25-

(a)

CaO

225

(b)

CaO

Type II

20-

o 15-

10-

Type III OStained and Unstained

O Stained and Unstained

• St Catherine

K20

K20

• St Gregory 0

10

12

14

16

18

20

10

12

14

16

18

20

25-

25

(c)

CaO

(d)

CaO 20-

20

15"

10'

O Stained and Unstained • Nativity l i S t Bernard • St Augustine

O Stained and Unstained • St Jerome

K20

0 10

12

14

16

18

20

lO

12

1'4

K20 16 -

1'8

20

Fig. 7. The chemical composition of glass from each of the Certosa stained windows plotted against the total analyses. Four CaO v. K20 diagrams highlight the chemical trend of a glass from a given stained window. (a) St. Catherine; (b) St. Gregory; (c) St. Jerome; (d) Nativity, St. Bernard and St. Augustine.

226

V. MARCHESI ETAL.

Type II and Type III glass. All potash glass are inferred to be medieval, whereas soda and Carich, K20-poor glass (Type I) are inferred to be produced later, as suggested by Brill & Pongracz (2004) for Soissons medieval glass. Potash glass defines a very tight compositional cluster representing a good statistical result and defining the composition of the pristine glass panes, which belong to the Type II and Type III glass. Both compositions are represented in almost all the investigated windows. A recent paper (Riccardi et al. 2005) show that this tight compositional cluster plots onto the liquidus surface of the K20-CaO-SiO2 phase diagram, in a rather narrow region at temperatures of about 1250 °C. These conditions slightly exceed the peritectic melting temperature, and this implies a glass-making process based on a good technological knowledge, which prevents secondary (peritectic) crystallization of mineral phases (wollastonite) and the consequent presence of defects. Pirina (1995) indicated two main periods in which stained glass windows were built. However, is it not possible to distinguish them on the basis of differences in composition because all glass windows have panes belonging to Type II and Type III glass. It is reasonable to suppose that master glass-makers used traded glass panes, instead of producing the glass themselves within the Certosa building yard as asserted by Pirina (1995). Compositionally, unstained lozenge-shaped glass match as stained glass, as shown by Figure 3. As a consequence, unstained glass cannot be considered as lower quality glass from a local production. Looking at the huge database proposed by Brill (1999), we can perceive a wide variability of the glass compositions even considering glass windows from a given cathedral. The tight clustering of Type II and Type III glass from Pavia Certosa indicates that they are surely the pristine glass. Moreover, the tight cluster enables us to obtain insight into the glass age and provenance. Figure 8 reports compositional comparisons among the investigated glass and stained glass coeval cathedrals of Northern Europe (Germany, Switzerland and France). A wide compositional scatter is evident for analyses from European cathedrals, and this could be accounted for by differences in composition related to repairs or analyses of altered fragments. Exceptions are Rouen and St. Maur des Fosses. The CaO-K20 diagram (Fig. 8) highlights a good compositional match between the investigated Ca-poor (Type III) and Ca-rich (Type II) glass from Pavia Certosa and glass

(a)30 II

CaO

GERMANY

•

°)

[]

[] 0 Ca-poor [] Ca-rich rlAUGSBURG X-XII O FREIBURG XlV ALAUTENSACH XV-XVI iOPPENHEIM XIV • ULM MUNSTER XV

5

0

10

[] []

15

K20

25

2O

(b) 25 SWITZERLAND

CaO • 20

om}

~

o

o Ca-p~or D Ca-rich

K20

• BERNE XV-XVl

5

0

10

15

20

25

(c) 30

o Ca-poor [] Ca-rich O AVIGNON XIV • ROUEN XlV II PSALMOD IXlll xST.MAUR DES FOSSES X l l l - X l ~

25

FRANCE O 0

"

CaO

X X

mol

,4 O 0

5

10

15

K20 20

25

Fig. 8. Comparison between chemical composition of Certosa pristine glass and stained glass from Germany, Switzerland and France. ©, Ca-poor (Type III); E3, Ca-rich (Type II) glass.

from Rouen and St. Maur des Fosses (13th and 14th century). Acknowledgements are due to the 'Soprintendenza per i Ben• Ambientali e Architettonici' of Lombardy for permission to study the glass samples. The authors gratefully acknowledge the help and suggestions supplied by the

MEDIEVAL STAINED GLASS restorer Arch. L. Morandotti during the sampling of the stained glass windows; she has also given permission to print the colour photograph. We thank S. Trfimpler, G. Rebay and an anonymous referee for the careful review of the manuscript.

References BARRERA, J. & VELDE, B. 1989. A study of French medieval glass composition. Journal of Glass Studies, 31, 48-54. BRILL, R. H. 1999. Chemical Analyses of Early Glasses. Coming Museum of Glass, New York. BRILL, R. H. & MOLL, S. 1961. The electron beam probe microanalysis of ancient glass. Recent Advances in Conservation, IIC Rome Conference, 145-151. BRILL, R. H. & PONGRACZ, P. 2004. Stained glass from Saint-Jean-des-Vignes (Soisson) and comparisons with glass from other medieval sites. Journal of Glass Studies, 46, 115-144. CAFEARO, A. 2000. Teofilo monaco Le varie arti (De diversis artibus) Manuale di tecnica artistica medievale. Palladio, Salerno. DELL'ACQUA, F. 2003. 'Illuminando Colorat' la vetrata tra etb tardo imperiale e I'alto medioevo: le fonti, l'archeologia. Fondazione Centro Italiano di Studi sull'Alto Medioevo, Spoleto. GRATUZE, B. 1994. Le verre: les 616ments de r6ponses que peuvent proposer les m&hodes de caract6risation physico-chimiques aux problematiques arch6ologiques pos6es par ce mat6riaux. Revue d'Archdomdtrie, 18, 75-87. KRUEGER, I. & WEDEPOHL, H. K. 2003. Composition and shapes of glass of the early medieval period (8th to 10th century AD) in Central Europe. Echanges et commerce du verre dans le monde antique. Ed. M. Mergoil, Montagnac, 93-100. MESSIGA, B. & RICCARDI, M. P. 2001a. A new assessment on the chemical composition of stained glass from The Certosa di Pavia, Italy. Rivista della Stazione Sperimentale del Vetro, 6, 177-185.

227

MESSIGA, B. & RICCARDI,M. P. 200lb. A petrological approach to the study of ancient glass. Periodico di Mineralogia, 70( 1), 57-70. MESSIGA, B. & RICCARDI, M. P. 2006. Alteration behaviour of glass panes from Medieval Pavia Charterhouse. Journal of Cultural Heritage (in press). MOLLER, W., TORGE, M. & ADAM, K. 1994. Ratio of C a O / K 2 0 > 2 as evidence of a special Rhenish type of medieval stained glass. Glass Science Technology, 67(2), 45-48. NEWTON, R. & DAVISON, S. 1997. Conservation of Glass. Butterworth Heinemann, London. PALLOT-FROSSARD, I. 1998. La place du vitrail dans l'architecture. Le Matdriau vitreux: verre et vitraux. Edipuglia, Ravello. PIRINA, C. 1995. Le due fasi delle vetrate nella Certosa di Pavia. In: PIRINA, C. (ed.) Le Due Fasi della Certosa di Pavia. Rendiconti dell'Istituto Lombardo Accademia di Scienza e Lettere, 127, 129-169. RICCARDI, M. P., MARCHES1, V. & MESSIGA, B. 2005. Melting path-ways of medieval glass from Certosa di Pavia (Italy). Thermochimica Acta,

425, 127-130. SPITZER-ARONSON, M. 1976. Contribution h la connaissance des vitraux du Moyen Age. Insuffisance de la diffusion pour expliquer la non-concordance stricte entre la pr6sence de cuivre et la couleur l'int6rieur des verres des vitraux rouges. Verres et Refractaires, 30( 1), 56-61. VERIT.~,, M., BASSO, R., WYPYSKI, M. T. & KOESTLER, R. J. 1994. X-ray microanalysis of ancient glassy materials: a comparative study of wavelength dispersive and energy dispersive techniques. Archaeometo,, 36(2), 241-251. WEDEPOHL, H. K. 1997. Chemical composition of medieval glass from excavations in West Germany. Glass Science Technology, 70(8), 246-255.

Petrographic features and thermal behaviour of the historically known 'pietra oilare' from the Italian Central Alps (Valchiavenna and Valmalenco) F A B R I Z I O A N T O N E L L I l, P A T R I Z I A S A N T I 2, A L B E R T O R E N Z U L L I 2 & ALESSANDRA BONAZZA 3

1Laboratorio di Analisi dei Materiali Antichi, University luav of Venice, S. Polo 2468, 30125 Venice, Italy (e-mail: [email protected]) 2Istituto di Vulcanologia e Geochimica, University of Urbino 'Carlo Bo', Campus Scientifico, 61029 Urbino, Italy 31SAC-CNR, Istituto di Scienze dell'Atmosfera e del Clima, Via Gobetti 101, 40129 Bologna, Italy Abstract: Thermal and porosimetric properties of different lithotypes of 'pietra ollare'

(magnesite-bearing talc-shists, chlorite-schists, tremolite-bearing chlorite-schists and serpentine-schists) from the Italian Central Alps (Valchiavenna and Valmalenco) have been investigated. Some cross-correlations are established among the main mineralpetrographic and textural features, thermal behaviour and historical utilization of these lithotypes for the production of stoves and cooking pots during the Middle Ages. All the analysed samples show (1) low total open porosity (0.73-2.85%) with meso- and micropores prevailing over macropores; (2) regular linear expansion up to c. 700 °C, good thermal stability up to 1200°C and negligible weight loss (<1%) to c. 500 °C; (3) high thermal expansion (5.57 x 10 6 , c - l < Ot25-100 C "~ 8.89 x 10 -6 °C-I). The results indicate that, under the thermal conditions typical of the traditional medieval 'open fire system' (T < 600 °C), the Italian 'pietra ollare' from the Central Alps was an excellent fire-resistant geomaterial, which did not undergo any significant transformation as a result of thermal shocks.

The aim of the present study is to characterize the thermal behaviour of 'pietra ollare' from the Italian Central Alps, which was widely used in the Middle Ages as a raw material for the production of stoves and pots for cooking and preserving food. According to Mannoni & Messiga (1980) and Mannoni et al. (1987), the term 'pietra ollare' groups basic and ultrabasic metamorphic rocks belonging to the greenschist facies whose outcrops are generally linked to the presence of ophiolite units (Dietrich 1980). On the basis of the different mineralogy, grain size and colour, all the Alpine metamorphic rocks known as 'pietra ollare' have been classified by Mannoni et al. (1987) into 11 groups: coarse- to fine-grained grey to pale green carbonate talc-schists (Groups C and D); grey to pale green amphibole-bearing talc-schists _ carbonates (Groups B and E); fine- to coarsegrained, green chlorite-schists (Groups F and G); green to whitish serpentine-schists (Group A); less common lithotypes represented by metagabbros, amphibole-schists, olivine-schists and

prasinites, with colours varying from dark green to grey-white (respectively Groups H, I, K and L). Groups C and D proposed by Mannoni et al. (1987) mainly comprise magnesite-bearing talc-schists, and therefore can be considered soapstones stricto sensu (according to Bucher & Frey 2002). Some physical features of the 'pietra ollare' have already been reported in the literature. In particular, it is well known that they are soft and easily workable materials, having low values on the Mohs hardness scale: c. 1 for soapstones (Robertson 1982), and between 1 and 4 for nearly all the Alpine 'pietra ollare'-forming minerals (Mannoni et al. 1987). Very low porosity, relatively good thermal capacity and high refractory properties were hypothesized by Mannoni et al. (1987), although they did not provide any measurements on the physical properties of these rocks. Mineralogy and chemical compositions of a wide range of 'pietra ollare' samples (from both outcrops and archaeological finds) from the Valmalenco and Valchiavenna (Central Alps, Sondrio

From: MAGGETTI, M. & MESSIGA, B. (eds) 2006. Geomaterialsin CulturalHeritage. Geological Society, London, Special Publications, 257, 229-239. 0305-8719/06/$15.00 ~, The Geological Society of London 2006.

230

F. ANTONELLI ET AL.

province, Lombardy) have been provided in recent years by Bonazza et al. (1999) and Santi et al. (2005). It is worth noting that during the Middle Ages, 'pietra ollare' artefacts from the Valchiavenna and Valmalenco were widely used throughout northern and central Italy (Alberti 1997; Malaguti & Zane 1999 and references therein; Santi et al. 2005). The present paper will focus, for the first time, on the physical-thermal features of samples representative of these two renowned Alpine quarrying and production centres. Porosity, differential and gravimetric thermal data, thermal linear expansion coefficients (~), thermodilatometric curves and fusion flow test results will be discussed, providing some insights into the thermal behaviour of the Valchiavenna and Valmalenco fire-resistant rocks. Relationships between the physical-thermal features of the 'pietra ollare' and their modal mineralogy, texture and whole-rock geochemistry will also be highlighted.

The 'pietra oilare' in antiquity Alpine 'pietra ollare' (from the Latin olla, i.e. container) was already known in the Neolithic Age and became a geomaterial widely employed in the production of fire-resistant artefacts between the Late Roman Empire period and the Middle Ages (Mannoni & Messiga 1980). The easy workability and relative toughness of these rocks allowed the production of food containers that were comparable with pottery. A standardization of 'pietra ollare' artefacts, characterized by the progressive thinning of the walls, was developed from the fourth century AD as the result of using heavy lathes (Alberti 1997). During the Middle Ages, when the trade network swiftly expanded, the extra-regional distribution of 'pietra ollare' artefacts became of paramount importance to the economy of the Alpine populations (Alberti 1997). Among the numerous Alpine 'pietra ollare' outcrops, the quarries of Valchiavenna and Valmalenco were two of the most exploited production centres in the Medieval period. In particular, Valchiavenna attained prominence as the crossroads of the trading routes, becoming an important centre for the production and exchange of 'pietra ollare' artefacts, within and beyond the confines of northern Italy. In this context, the proximity of 'pietra ollare' quarries to the artisan production centres, located close the rivers and main commercial routes, greatly favoured the local trade of these artefacts (Gaggi & Leoni 1984). The distribution of the goods outside the region was effected along the main waterways (rivers and lakes; Alberti

1997): from Lake Como, following the course of the Adda river, the Valchiavenna 'pietra ollare' artefacts were able to reach directly the region of the Po Plain, from whence they could be transported along the River Po to the Adriatic coast, and finally shipped southwards (Santi et al. 2005). The 'pietra ollare' artefacts from the Valmalenco production centre were mainly distributed in north-eastern Italy (Bonazza et al. 1999). The distribution of these rock artefacts began later (13th century AD) compared with those of Valchiavenna, probably because most of the quarries were located slightly farther from the main trading routes and waterways (Gaggi & Leoni 1984).

Sampling and analytical methods In the present study we analysed representative samples of different lithologies of 'pietra ollare' from the most exploited outcrops of the Italian Central Alps (Table 1). The Valchiavenna *pietra ollare' lithology (fine-grained and grey to pale green coloured) comes from ancient and modern quarries located at Piuro, very close to the village of Chiavenna. Valmalenco samples are from different historical quarries located in the neighbourhood of Chiesa in Valmalenco; they belong to different lithologies although they show a common green colour and a very fine grain size. Bulk density and porosimetric characteristics (total open porosity and pore-size distribution) were measured by a mercury porosimetry (Pascal 140-240) following the Italian recommendations (Normal 4/80; CNR-ICR 1980). The X-ray diffraction analyses (XRD) were performed on powdered portions of selected samples using a Philips PW1840 diffractometer characterized by Cu Ko~/Ni radiation, at 40 kV and 20 mA. The thermal behaviour was investigated by means of differential thermal analysis (DTA) coupled with thermogravimetric analysis (TGA) on c. 50 mg of powdered sample, placed in aluminium crucibles of 150 ill capacity. The DTA and TGA analyses were carried out using a Mettler Toledo TGA/SDTA 851 system, equipped with a TSO 800GCI programmable gas switch at a constant air flow, at temperatures ranging between 25 and 1100 ~C, with a thermal gradient of 10 9C min-l. The fusion flow test was performed by a Misura hot stage microscope, with an increasing thermal gradient of 10 °Cmin -~ up to the complete melting of a previously pressed powder sample on a 3 m m x 2 m m cylinder (Dondi et al. 2001). The thermal expansion values were determined using a Netzsch 402

231

CHARACTERIZATION OF 'PIETRA OLLARE' Table 1. Analysed samples of 'pietra ollare' from Valchiavenna and Valmalenco Sample

Lithotypes

Petrography XRD Porosity T G A + D T A

Fusion Thermal test expansion

Valchiavenna VCO 1 VCO2 VCO3 VCO4 VCO5 VCO6 VCO7

Magnesite-bearing talc-schist Magnesite-bearing talc-schist Magnesite-bearing talc-schist Magnesite-bearing talc-schist Magnesite-bearing talc-schist Magnesite-bearing talc-schist Magnesite-bearing talc-schist

X

X

x

x

x

x

x

x

X

X

X

X

X

X

x

*

x

x

x

*

x

x

x

*

x

x

*

x

x

*

x

x

*

x

x

*

x

x x x x x x

X

X

X

X

X

X

X

X

X

X

X

X

Valmalenco PO 1 PO2 PO3 PO5 POA PO6

Chlorite-schist Chlorite-schist Chlorite-schist Tremolite-bearing chlorite-schist Tremolite-bearing chlorite-schist Serpentine-schist

*XRD data from Santi

et al.

X X

X X

X

(2005).

mechanical dilatometer with a thermal gradient of 10 ° C m i n -1 up to the final temperature of 1000 °C (following ASTM C 372, ASTM 1994). The thermal linear expansion coefficient oL is calculated as the ratio between the length change of the sample At and the original sample length expressed as 10 . 6 °C -~ (at a constant pressure and a precise temperature interval). Since thermal expansion is an anisotropic physical rock property, to obtain an average value of oL, the adopted sample size corresponded to a parallelepiped of 40 mm x 4 m m x 4 mm, cut at c. 45 ° with respect to the foliation planes.

Results Mineralogy, petrography and geochemistry According to the petrographic data reported by Santi et al. (2005), the Valchiavenna samples (labelled VCO) are magnesite-bearing talcschists (i.e. soapstones; Group D of Mannoni et al. 1987) showing a heteroblastic texture. Their mineralogy consists of talc 4- magnesite 4chlorite 4- opaque minerals, in decreasing order of abundance. The carbonate xenoblasts are magnesite with minor or occasional dolomite and/or calcite (Table 2); chlorite is usually

Table 2. XRD analyses of 'pietra ollare' Sample

Tic

Chl

Mgs

Dol

VCO1 VC02 VC03 VC04 VCO5 VCO6 VCO7

+++ +++ +++ +++ +++ +++ +++

+++ +++ + ++ ++ ++ +

+ + ++ + +++ ++ ++

+

Sample

Chl

Tr

Atg

Ilm

Mag

Cal

+

+

Mag

Cal

Valchiavenna samples

Valmalenco samples PO1 PO2 PO3 POA PO5 PO6

+++ +++ +++ +++ +++ +++

+

Tic, talc; Chl, chlorite; Mgs, magnesite; Dol, dolomite; Mag, magnetite; Cal, calcite; Tr, tremolite; Atg, antigorite; llm, ilmenite. + + + , very abundant; +, present; +, trace. Valchiavenna samples from Santi et al. (2005).

F. ANTONELLI ETAL.

232

abundant and represented by clinochlore (Mg-rich term; Santi et al. 2005). Although diablastic texture may locally occur, talc and chlorite give to the rock a more or less developed schistosity. Opaque minerals such as Cr-magnerite, ilmenite and pyrite are generally present as trails along the foliation. These lithologies are generally fine-grained ( 6 < l mm), with a very variable range in grain size from c. 0.1 to 1 mm. However, some samples are characterized by xenoblasts of talc from 1-2 mm to 4 mm in size (VCO1), chlorite (VCO4) and magnesite (VCO6) up to 1.5 mm in size. The Valmalenco samples (labelled PO) are represented by three lithologies: chloriteschists, tremolite-bearing chlorite-schists and serpentine-schists. The chlorite-schists (Group F of Mannoni et al. 1987) are heteroblastic, fine-grained, with an average grain size of 0.5 ram. Their mineralogy is dominated by chlorite, showing a well-developed schistosity and rare diablastic texture. Opaque minerals (magnetite and ilmenite in PO3; Table 2) and rare tabular epidotes are also present in grains, locally aligned along the schistosity. Tremolite-bearing chlorite-schists (Group I of Mannoni et al. 1987) show weak schistosity and heteroblastic texture with an average grain size of 0.5 mm; nevertheless, tremolite idioblasts range between 1.5 and 3.5 mm. Chlorite is decidedly more abundant than diablastic tremolite. Rare small epidote crystals and opaque minerals (magnetite; Table 2) are also found.

Serpentine-schists (Group A of Mannoni et al. 1987) are dominated by foliated antigorite (Table 2), and are generally fine-grained, with an average grain size of c. 0.3 mm (although opaque grains may reach 0.5 mm). Tiny tabular tremolite and opaque minerals are also present. Secondary micrite may locally occur along microfractures. Major element data for selected 'pietra ollare' samples from Valchiavenna and Valmalenco, published respectively by Santi et al. (2005) and Bonazza et al. (1999), are summarized in Table 3 and plotted in Figure 1. Although at the scale of the outcrops these metamorphic rocks from the Central Alps are described as generally inhomogeneous, in terms of mesoscopic appearance, mineralogy and chemistry (e.g. Trommsdorff & Evans 1977), some differences in chemical composition can be emphasized (Table 3). These distinctions appear to be linked to the source areas. In fact, on the basis of A1203, MgO and SiO2 values, few compositional variations, strictly associated with the different mineralogical assemblages (Table 2), can be seen in Figure l. In particular, because of the mineralogy, dominated by talc and magnesite, the soapstones of Valchiavenna are characterized by lower contents of A1203 and higher values of MgO with respect to the other investigated lithologies of 'pietra ollare' (Fig. 1). Furthermore, among the Valchiavenna samples, the variable contents of SiO2 could reflect the different talc/magnesite ratios. The Valmalenco

Table 3. Major element data for selected 'pietra ollare' samples from Valchiavenna and Valmalenco SiO2

A1203

Fe203

MnO

MgO

CaO

Na20

K20

3.17

8.83

0.16

40.29

0.51

<0.01

<0.01

3.25 2.42

8.46 7.50

0 . 1 3 39.85 0 . 7 7 <0.01 0 . 1 2 33.65 1.61 0.02

2.79 4.44 2.35 2.40

6.64 14.21 7.37 7.24

0.13 0.15 0.11 0.14

33.58 32.40 34.90 33.94

1.55 0.24 0.11 1.50

24.61 25.81 24.31

0.18 0.21 0.2 !

20.00 17.75 22.95

TiO2

P205

LOI

Total

Modern quarries VCO1

40.63

VCO2 40.98 VCO3 33.16

0.06

0.01

5.89

99.54

0.09 0.02

0.12 0.05

0.01 0.01

5.05 20.73

98.69 99.31

0.01 <0.01 <0.01 < 0.01

0.02 0.07 <0.01 0.03

0.04 0.13 0.08 0.03

0.01 0.02 0.01 0.01

18.43 18.16 19.61 19.27

99.08 98.53 99.67 99.43

0.79 1.49 0.68

<0.01 <0.01 <0.01

<0.01 <0.01 < 0.01

2.22 1.90 2.64

0.42 0.16 0.37

8.80 8.45 7.90

100.00 100.00 100.00

Ancient quarries VCO4 VCO5 VCO6 VCO7

35.87 28.72 35.17 34.91

Chlorite-schists PO1 PO2 PO3

23.85 23.84 24.07

19.13 20.39 16.88

Tremolite-bearing chlorite-schists POA PO5

34.75 36.77

14.92 10.19

6.76 9.01

0.20 0.09

28.54 30.80

5.19 3.44

<0.01 <0.01

0.01 <0.01

0.55 0.22

0.03 <0.01

9.05 9.47

100.00 99.99

2.41

7.88

0.1 I

35.49

1.57

<0.01

<0.01

0.02

<0.01

10.70

100.00

Serpentine-schist PO6

41.82

Valchiavenna (VCO) data from Santi et al. (2005); Valmalenco (PO) data from Bonazza et al. (1999).

CHARACTERIZATION OF 'PIETRA OLLARE' 25 20 A

[]

lO <

SiO2

VCO5 VCOla-lb-2 • VCO3-4-6-7

22 24 26 28 30 32 34 36 38 40 42 44 48 VCOla-lb-2

43 38

VCO3-4-6-7

33 28 23

*#*

vco5

[]

©

[]

A ....

I ....

SiO2 I ....

I ....

I ....

i ....

I~'"I

....

I ....

I ....

I,,,~

22 24 26 28 30 32 34 36 38 40 42 44 • A [] O

open porosity shown by VCO3 (2.85%) is probably linked to the marked schistosity. In all the lithologies, medium and small pores (7.5/xm > ~b < 0.0037 lxm) always predominate (generally >> 65 vol %) over the macropores (7.5-100 Ixm). The Valmalenco samples generally show a polymodal pore-size distribution, and are characterized by major peaks around 0.03-0.1 Ixm, 2 - 5 Ixm, 7 - 2 0 txm and 3 0 - 5 0 Ixm (Fig. 2). Conversely, those from Valchiavenna display a roughly bimodal distribution, with two main porous domains around 0.02-0.2 Ixm and 2 - 5 Ixm, and a less frequent domain between 25 and 60 Ixm (Fig. 2). However, all the samples are characterized by a main peak between 2 and 5 Ixm, which often represents a volume > 20% of the whole distribution (20-27% and 15-31% in the Valmalenco and Valchiavenna 'pietra ollare', respectively).

Thermal gravimetric analysis (TGA ) and differential thermal analysis (DTA)

A A

18 13

.

233

Valchiavenna magnesite-bearing talc-schist Valmalenco chlorite-schist Valmalenco tremolite-bearing chlorite-schist Valmalenco serpentine-schist

Fig. 1. A1203 and MgO v. SiO2 binary diagrams of the 'pietra ollare' representative samples. 'pietra ollare' (with the exception of the serpentine-schists) shows a higher percentage of chlorite and hence higher A1203 contents (Fig. 1). Finally, the lower MgO content of the Valmalenco serpentine-schists, compared with some of the Valchiavenna magnesitebearing talc-schists, is due to the presence of tremolite in the former.

Porosity In the different 'pietra ollare' lithologies, total open porosity is always low (Table 4, Fig. 2), varying from 0.73% (VCO2) to 2.85% (VCO3), with an average value of 1.63%. Samples from Valchiavenna show a total open porosity and an average pore radius generally lower than those from Valmalenco, as well as relatively more variable values of bulk density (Table 4). However, the relatively high total

In all samples investigated, the DTA and TGA results indicate a stable thermal behaviour, with negligible weight loss (< 1%) up to c. 500 °C. Depending on the mineralogy, endothermic peaks associated with significant weight loss in the various samples are revealed within a temperature range between 500 and 990 °C (Table 5). The DTA curves of the magnesite-bearing talc-schists of Valchiavenna are characterized by an endothermic peak in the temperature range between 595 and 626 °C, which may be associated with a first stage of chlorite dehydroxylation and magnesite thermal dissociation into periclase and carbon dioxide (Mackenzie 1970; Chang et al. 1996). Both reactions contribute to a decrease in mass varying from 2.8% (VCO1) to 16.0% (VCO5), as a result of the increasing percentage volume of chlorite and, predominantly, magnesite (Fig. 3a). A second minor endothermic peak is present in the temperature range between 690 and 820 °C, associated with the dehydroxylation of chlorite and talc. Another endothermic peak is also present for samples VCO4 and VCO7, at 817°C and 791 °C, respectively, as a result of the thermal decomposition of dolomite. In the temperature range between 820 and 860 °C, an exothermic peak, owing to chlorite recrystallization into forsteritic olivine (Mackenzie 1970; Deer et al. 1971), is revealed. Finally, an endothermic peak, occurring between 900 and 990 °C, is associated with a negligible weight loss (0.72.2%) and is probably caused by a second stage of talc dehydroxylation.

234

F. ANTONELLI ETAL.

Table 4. Results of the porosimetric analyses Sample

Total open porosity (%)

VCOI VCO2 VCO3 VCO4 PO1 PO2 PO3 PO5 PO6

Bulk Apparent Average density_ density_ pore (g cm -3) (g cm -3) radius (Ixm)

0.99 0.73 2.85 1.35 1.89 1.35 2.25 1.55 1.75

2.02 1.75 0.06 2.05 2.53 0.24 2.78 2.72 2.05

Pore radius distribution (% relative volume) 100-7.5 (Fm)

2.95 2.97 2.89 2.94 2.96 2.96 3.00 2.77 2.67

2.98 2.99 2.98 2.97 3.02 3.00 3.07 2.81 2.72

22.39 25.58 11.18 13.58 22.82 26.09 21.80 32.00 17.35

Max.

Min.

Average

S.D.

Total open porosity (%) VCO group 2.85 PO group 2.25

0.73 1.35

1.48 1.76

0.96 0.34

7.5-1 1-0.1 0.1-0.01 ( I x m ) (#.m) (Ixm) 37.31 39.53 18.63 23.46 28.86 19.13 26.32 34.67 35.65

13.43 18.60 23.60 25.93 22.82 26.96 24.06 17.33 18.06

0.01-0.0037 (I.tm)

17.91 16.28 42.86 28.40 23.49 23.48 25.56 14.67 22.00

8.96 3.73 8.64 2.01 4.45 2.26 1.33 6.94

Main statistical porosimetric data

Bulk density VCO group PO group

2.97 3.00

Average pore radius VCO group PO group

I

)

f

"-,,.

IIj/fllll Ogl

0,001

I "~

~

I

2.05 2.78

0.01

0.06 0.24

I

i

i

I "~11 ~

1.47 2.06

15

I

-~1

I l j l [ l l l III I l l l l l l l l l l I I I I I II11111 I II 11"tl0 0.1 1 10 100 Pore radius(p.m) f i Sample V C 0 4

o~

I

-

I 0.001

2.94 2.87

~ ~ Sample P 0 1

I

o

2.89 2.67

0.1

30

---% I li'--N 1 Pore radius(prn)

~ 10

100

Fig. 2. Representative pore-size distribution of the samples belonging to the PO (polymodai range) and VCO (roughly bimodal range) groups.

Average VCO + PO

S.D. VCO ÷ PO

1.63

0.64

2.90

O. 11

1.80

0.99

0.03 0.14 0.95 1.06

As regards the Valmalenco 'pietra ollare', chlorite-schists show two endothermic peaks in the ranges 6 2 7 - 6 6 7 ~C and 7 5 0 - 7 6 6 °C, as a result of two stages of chlorite dehydroxylation (Fig. 3b). Such transformations give rise to a loss of mass that varies from 6.4% (PO3) to 7.6% (PO2), and are followed by an exothen•ic peak (850-859 °C) related to chlorite recrystallization into forsteritic olivine (Mackenzie 1970; Deer et al. 1971). Tremolite-bearing chlorite-schists are characterized by two main endothermic peaks ( 5 8 9 596 ~C and 8 0 7 - 8 5 0 :C) with weight loss of 4 . 1 - 7 . 1 % and 2.1-5.5%, respectively (Table 5 and Fig. 3c). The former is mainly due to chlorite dehydroxylation, although in sample POA a small contribution from the first stage of amphibole decomposition has to be taken into account. The second peak is related only to the amphibole decomposition. A clear exothermic peak caused by Mg-chlorite recrystallization into forsteritic olivine is detected at 850 °C (according to Mackenzie 1970). Finally, the DTA curve of the serpentine-schist is characterized by two

CHARACTERIZATION OF 'PIETRA OLLARE'

235

Table 5. Results of the DTA-TGA analyses Magnesite-bearing talc-schists Reaction

Temperature range:

Weight loss (%) VCO1 VCO2 VCO3 VCO4 VCO5 VCO6 VCO7

Chlorite recrystallization into olivine

2nd talc dehydroxylation

500-690 °C

2nd chlorite and 1st talc dehydroxylation+ dolomite decomposition 690-820 °C

820-860 °C

900-990 °C

2.8 2.9 4.3 12.9 16.0 15.7 13.1

1.3 0.9 0.5 3.7 1.0 1.1 3.5

0.2 0.2 1.8 <0.1 1.0 < 0.1 < 0.1

0.8 0.7 <0.1 2.0 1.0 2.2 2.0

1st chlorite dehydroxylation + magnesite decomposition

Chlorite-schists Reaction: Temperature range:

Weight loss (%) PO1 PO2 PO3

1st chlorite dehydroxylation

2nd chlorite dehydroxylation

620-670 °C

750-770 °C

3.9 7.6 3.6

3.0 2.8

Chlorite recrystallization into olivine 850-860 °C 1.0 0.5 1.0

Tremolite-bearing chlorite-schists Reaction:

Temperature range:

Weight loss (%) POA PO5

Chlorite dehydroxylation + I st amphibole decomposition 580-600 °C

2nd Amphibole decomposition + chlorite recrystallization into olivine 800-850 °C

4.1 7.1

5.5 2.1 Serpentine-schists (PO6)

Reaction: Temperature range:

Weight loss (%)

Antigorite dehydroxylation 730-770 °C

Olivine formation

9.0

0.5

830-870 °C

endothermic peaks at 7 3 4 ° C and 768 °C, corresponding to a total mass loss of 9.0%, related the dehydroxylation of antigorite. The subsequent exothermic peak at 836 °C (Fig. 3d) may be attributed to forsterite formation (according to Mackenzie 1970; Deer et al. 1971).

Fusion flow tests, thermal linear expansion and thermo-dilatometric curves The fusion flow tests (Table 6) show that all the samples maintain a solid state up to a temperature of 1200 °C. Sintering takes place between

1200 °C (serpentine-schist, PO6) and 1265 °C (tremolite-bearing chlorite-schist, PO5), whereas total melting is attained between 1405 °C (PO5) and 1545 °C (magnesite-bearing talc-schist, VCO1). All the samples can be therefore considered good fire-resistant materials. On a scale of increasing refractory nature, they can be arranged as follows: tremolite-bearing chlorite-schist (PO5), chlorite-schist (PO2), serpentine-schist (PO6), magnesite-bearing talc-schist (VCO4 < VCO3 < VCO1). The Valchiavenna samples of 'pietra ollare' (fine-grained soapstones) reveal the greatest fire-resistance, as a

236

F. ANTONELLI E T A L .

(a)

I

5mg

~

(b)

12mg !

100

200

300

400

500

600

700

800

llnlll.,l .... I .... l . , , = l l . l a l a l , l J . , l l l , , l , . , , l . | U , l , , , . l , . , . l , , , , l , , . n l . . ,

0

10

O0 Ill I..

20

30

200

300

I Ill I l I I l l . . . .

0

40

50

400

ll]l

tl

500

I ~I

. . I

20

30

40

700

I I flirt

II

60

800

200

300

..

, . . . .

10

1000 *C

I l ]I

II

[ Ill

400

ill, .

.

.

.

30

I

50(] .

40

.

.

.

!

600

700

100 ...I

....

400 I ....

500 I ....

600 I ....

700 I . . . .

'60' '" ' ; 0 '

800 I ....

dO"

900 I ....

~'

1000 *C llllll

i ~ ' ~n

500 600. . . . .700 1000 . , , 100 ;. ;, ;200 {. ,.,.,3~ ~;,. ,,.,400 ~.; ....,.. j ,[......,.. . , .,...,.. 800 . , '. . . . .900 ., .,,,,..,,, ..,..,,*CI I0 20 30 40 50 60 70 80 90 100 rain

.

.

.

50

,

.

60

.

.

.

.

.

.

.

.

70

o~ 900

80

100 200 300 400 500 600 700 800 900 1000 *C | ; ; .. L ; ; ; .. L, ; - - , ; ..; .. ,..... ;.. ,..,.. • .. ' .,...... '..,...... '..,...... '..,......'..,...... ~ u . . . 10 20 30 40 50 60 70 80 90 100 rain 100 rain 0

1000 °c

I .... I .... , . . . . . . .

I ....

.

90

.

2°C 200

300

I ....

[iii.iiiiii,,i.

10

....

100 mill 0

90

1 °C

0

300

.

800

.... .

200

lllll..l.,.ll

(c)

tllllllJl.llil=lll

I . . . . u

20

I

. I . . . . . . . . n . . .

80

100 [

;00' ~n 0 . . . . ; # ' ' " £ . . . . 30''" '~ . . . . ~ ' ' ' f ~

~

l,l*ll,

. . . .

llllnl

~

f.~ 100

1,,

0

1000 *C

....

900

Ill I l l I l l

70

]2~o t.

900

'dO"' ' ~ ' "

. . . . . . . . . . .

50

...... I

i,.11

70

600

I I l II

. . I . . . .

10

60

400

I ....

I ....

20

30

I ....

500

600

I ....

I ....

700

800

l . , l l l

....

900

1000 *C

l , . t . l , l l . l

l l , l l l l l . . l l . . . l . , , , l l l l l l l l , , i , l

40

50

60

70

80

90

I

100

200

300

400

500

600

700

800

900

,..I .... | . . . , l , , , , l l . , , l .... I .... I .... l .... I .... | l = . , l , , . . l l . . . | . l l l | , l , , l l e e | l , , l , l l P l l l l l . l l , P , e l l , !

I00 m~ 0

I0

20

30

40

50

60

70

80

90

1000 °C l.,..l

100 min

Fig. 3. Representative D T A (lower) and T G A (upper) curves for the "pietra ollare" lithologies showing the endoand exothermic reactions. Chl, chlorite: Mgs, magnesite; O!, olivine; Tic, talc: Tr, tremolite; Atg, antigorite. (a) Magnesite-bearing talc-schists (VCO5); (b) chlorite-schists (PO1): (c) tremolite-bearing chlorite-schists (PO5); (d) serpentine-schists (PO6).

Table 6. Fusion flow tests Sample

TI (~C)

72 (:C)

73 (~C)

T4 ( C )

VCO1 VCO3 VCO4 PO2 PO5 PO6

1250 1250 1225 1240 1265 1200

1440 1355 1405 1295 1340 1410

1525 1495 1460 1410 1395 1470

1545 1510 1475 1425 1405 1475

TI, sintering; T2. softening; T3. starting fusion. T4. ending fusion.

result of the presence of a Mg-silicate such as talc (Calvino 1967) and the highest values of MgO (Table 3) compared with the other lithologies. Considering the thermal linear expansion coefficients ((x; Table 7), some magnesite-bearing talc-schists (VCO3 and VCO4), the chlorite-schist (PO2) and the tremolite-bearing chlorite-schist (PO5) have ¢x between (7.5-7.9)× 10-6 ~C-1 (0/-25-100 C ) and (9.6-11.8) × 10 -6 :C- 1 (¢x25-5oo~c). One magnesite-bearing talcschist (VCO1) shows higher coefficient values, between 8.9 X 10 -6 °C -1 (~25-1oo c) and

12.6 x 10 -6 :C -I (oL25_5ooc), whereas the serpentine-schist PO6 is characterized by the lowest thermal linear expansion coefficients, between 5.6 × 10 -6 :C -1 (0¢25_100 c) and 8.1 × 1 0 - 6 ~ C - I (0L25-500 c)- Different expansion or shrinkage (positive v. negative slopes) on the thermo-dilatometric curves (Fig. 4) was also considered. With the exception of the VCO1 soapstone, which does not show any thermal shrinkage, all the other investigated samples are characterized by shrinkage at c. 740 °C (PO6 serpentine-schist) or 800 °C (soapstone VCO3 and chlorite-bearing lithologies sensu lato). According to the D T A - T G A data, these two temperatures correspond to the breakdown of antigorite and chlorites, respectively (Mackenzie 1970; Deer et al. 1971). Shrinkage starts after a linear expansion ranging from 0.58% (PO6) to 1.25% (PO5), with an average value around 0.9% (Fig. 4). However, at temperatures corresponding to the rudimentary domestic open fires (500-600~C) typically used in historical times, shrinkage does not occur in all the investigated samples.

CHARACTERIZATION OF "PIETRA OLLARE' Table 7. Thermal linear expansion coefficient (a) (expressed as 10 -6 Sample

VCO1 VCO3 VCO4 PO2 PO5 PO6

2 5 - 1 0 0 °C

2 5 - 2 0 0 °C

2 5 - 3 0 0 °C

2 5 - 4 0 0 °C

2 5 - 5 0 0 °C

2 5 - 6 0 0 °C

2 5 - 7 0 0 °C

8.89 7.60 7.51 7.86 7.57 5.57

10.08 8.79 8.32 9.03 8.51 6.23

11.31 10.06 8.86 9.57 9.39 7.35

12.41 10.99 9.43 9.73 10.16 7.90

12.61 11.77 9.90 9.65 10.56 8.08

11.89 12.08 10.21 9.40 10.95 8.09

11.36 11.51 9.77 9.36 11.80 7.91

1.0~ 0.9 ~,°

VCO1- Magnesite-beannglalc-schist ~ /

/

thermo-dilatometric curves are similar) up to c. 700 °C, and preserve a good thermal stability up to 1200 °C (sintering temperature), with the beginning of softening between 1295 °C and 1440 °C (Table 6). The thermal behaviour of Valchiavenna and Valmalenco 'pietra ollare' clearly explains their importance in antiquity as highly prized, easily

-

~

"

'

07

Temperature °C

0.9_ 0.8- ~ VCO3-Magnesite-bearingtalc-sc~st j 0.70,60.50.40.30.2-

"

0,80.605. 0.4~ 0,30.2. 0.

~

Temperature °C

0

I()O°C 21)0°C 300°C 400°C 500°C 600°C 700°C 800°C 900°C1000°C l12 :lJ

,.oI

0.94

160°c 2~0°c360oc460°c560oc660°c 760°c860oc960°clobooc PO5-Tremolite-beanngchlorit .... hist ~ ' ~

Z

I \

~

I

\

0.7 0.6 0.5 0.4 0,3 0.2

0.100.9-

°c-l)

ot

Conclusions In the Italian Alpine 'pietra ollare' the dominant mineral phases are talc, magnesite, chlorite __+ serpentine + amphibole, all characterized by a well-known thermal stability range (temperature breakdown of these minerals occurs at > 5 3 0 540 °C). In particular, all the analysed samples show a regular linear expansion (slopes of the

1.00.90.80.706 0.5 0.40.302. 0.1. 0

237

o 100oc 200°C 300°C 400oC 500°C 600oC 700°C 800oC 900°C 1000oC

Tempe lO0OC 200°C 300°C 400°C 500°C 600°C 700°C 800oC 900°C1000°C

0.80.70.60.5. 0.40.3. 0.20.1. 0 /

Temperature °C

-0.1]

I()0°C 2()0°C 3()0°C 400°C 500°C 6()0°C 700°C 8()0oc 9()0°C 1000°C

,

Temperature °C /

100oC 200°C 3()0°C 400°C 500°C 6()0°C 7()0°C 800oC 900°C 1000°C

Fig. 4. Thermo-dilatometric curves of the analysed samples representing all the investigated lithologies of 'pietra ollare'.

238

F. ANTONELL1 ET AL.

workable and fire-resistant rocks. It is noteworthy that, although the maximum burning temperature of wood ranges from 560 to 920 ~C in a system without strong heat dispersions (Gosselain 1992; NunnaUuni 2001), we can assume that in a traditional 'open-fire system' the surfaces of the ancient stoves and pots made of 'pietra ollare' were heated to a maximum temperature always below 600 °C. This hypothesis is supported by the modal mineralogy of many archaeological finds of soapstone artefacts from Central Italy (Santi et al. 2005). In fact, in these analysed items, the absence of periclase in the external layers ( 1 - 4 ram) indicates that a temperature of c. 5 2 0 - 5 4 0 °C, which marks the beginning of the endothermic reaction transforming magnesite into periclase 4- CO2, was never attained. The generally low values of the total open porosity and the small average pore radius (between 0.06 and 2.78 ~m) characterizing both the Valchiavenna and Valmalenco "pietra ollare' constitute an additional reason for the widespread domestic use of this geomaterial in medieval times. To produce cooking pots, the stones were cut, then progressively abraded and probably roughly polished, yielding a marked further reduction in the superficial porosity of the internal surface, and therefore a very low absorption of liquids and foods. Normally, thermal conductivity of the rocks is inversely proportional to their total open porosity (Calvino 1967; Robertson 1982) and in low porosity geomaterials, like the investigated samples, such a thermal feature could be enhanced by the fine grain size. Therefore, 'pietra ollare' of Valchiavenna and Valmalenco has a good capability to transfer and distribute heat from the external (in direct contact with fire) to internal layers. Clearly, the thermal conductivity and capacity of foliated rocks are anisotropic, with the highest value being recorded along the direction of the cleavage and the lowest one perpendicular to it. Literature data on the thermal conductivity of soapstones from Finland (Nunna deposit; currently used to produce fireplaces), mainly consisting of talc and magnesite, give, at 50 °C, a value of 2 - 4 W m K - 1 perpendicular to the cleavage plane and 4 - 5 . 5 W mK-~ along the schistosity (NunnaUuni 2001). This apparent thermal anisotropy of the 'pietra ollare', together with its easy workability, could explain why most of the archaeological finds show a 'cuttingworking' direction developed parallel to the schistosity of the rock. It is noteworthy that, despite its excellent refractoriness, the investigated samples of 'pietra ollare' have a relatively high linear expansion coefficient oL (Table 7). Therefore, it

could be expected that cooking containers made of this stone had a low resistance to thermal shock. Nevertheless it is inferred that the physical properties of the abundant Mg-phyllosilicates contributed to mitigating the effects of possible thermal shock in the 'pietra ollare'. In other words, thermal expansion at high temperatures had to be accompanied by greater flexibility of the stone. Indeed, although showing the highest thermal expansion, the Valchiavenna sample VCO1, containing large talc xenoblasts (up to 4 mm in size), is characterized by a steady increase in linear expansion (low slope; Fig. 4), with a generally good thermal behaviour (no shrinkage at all). The abundant content of phyllosilicates such as talc and chlorite also contributes to give excellent toughness to the 'pietra ollare' artefacts, as required by any domestic object for every day use. The first draft of the manuscript benefited from a critical reading of L. Lazzarini, which is acknoweledged by the authors. We are also grateful to D. Malferrari for the stimulating discussion during the processing of the DTATGA data. C. D'Amico and an anonymous referee are also sincerely thanked for helpful reviews.

References ALBERTI, A. 1997. Produzione e commercializzazione della pietra ollare in Italia settentrionale tra tardoantico e altomedioevo. In: Proceedings, 1 ° Congresso Nazionale di Archeologia Medievale, Firenze, 335-339. ASTM 1994. Linear thermal expansion of porcelain enamel and glaze frits and fired ceramic whiteware products br the dilatomer method. ASTM C 372. American Society for Testing and Materials, Philadehphia, PA. BONAZZA, A., LAZZARINI,L. & VACCARO, C. 1999. Nuovo contributo archeometrico allo studio della 'pietra ollare' padana, 6~' Giomata 'Le Scienze della Terra e l'Archeometria ', Proceedings, 171 - 180. BUCttER, K. & FREY, M. 2002. Petrogenesis of Metamorphic Rocks. Springer, Berlin. CALVINO, F. 1967. Lezioni di Litologia Applicata. Cedam, Padova, 23-35. CHANG, L. L. Y., HOWIE,R. A. & ZUSSMAN,J. 1996. Rock-forming Minerals. Non-silicates, Vol. 5b, 2nd edn. Longman, Harlow, 136-149. CNR-ICR. 1980. Distribu:ione del volume dei pori in fimcione del Ioro diametro. Raccomandazioni Normal, 4/80. CNR-ICR, Roma. DEER, W. A.. HOWlE, R. A. & ZUSSMAN, J. 1971. Rock-forming Minerals. Sheet Silicates, Vol. 3, Longman, Harlow. DIETRICH, V. J. 1980. The distribution of ophiolites in the Alps. Ofioliti, Special Issue on Tethyan Ophiolites, 1 (Western Area), 7-51. DONDI, M., GUARINI, G. & VENTURI, I. 2001. Assessing the fusibility of feldspathic fluxes for

CHARACTERIZATION OF 'PIETRA OLLARE' ceramic tiles by hot stage microscope. Industrial Ceramics, 21(2), 67-73. GAGGI, S. & LEONI, B. 1984. La pietra ollare. Estratto da: Quaderni della Provincia. Amministrazione Provinciale di Sondrio, Sondrio, Italy. GOSSELAIN, O. P. 1992. Bonfire of the enquiries. Pottery firing temperatures in archaeology: what for? Journal of Archaeological Sciences, 19, 243-259. MACKENZm, R. C. 1970. Differential Thermal Analysis 1. Academic Press, London. MALAGUTI, C. & ZANE, A. 1999. La pietra ollare nell'Italia nord-orientale. Archeologia Medievale, 26, 463-479. MANNONI, Y. & MESSIGA, B. 1980. La produzione e la diffusione dei recipienti di pietra ollare hell'Alto Medioevo. 6 ° Congresso Internazionale di studi sull'Alto Medioevo, Proceedings, 501-522. MANNONI, T., PFEIFER, H. R. & SERNEELS, V. 1987. Giacimenti e cave di pietra ollare helle Alpi.

239

In: La Pietra Ollare dalla Preistoria all'Etgt Moderna, Proceedings, 7-45. NUNNAUUNI 2001. The magic of gold fire with Mammutti Stone http://www.nunnauuni.com/ ROBERTSON, E. G. 1982. Physical properties of building stone. In: Conservation of Historic Stone Buildings and Monuments. Report of the Committee on Conservation of Historic Stone Buildings and Monuments. National Academy Press, Washington, DC, 62-86. SANTI, P., ANTONELLI, F. & RENZULLI, A. 2005. Provenance of Medieval pietra ollare artefacts found in archaeological sites of Central-Eastern Italy: insights into the Alpine soapstone trade. Archaeometry, 47, 253-265. TROMMSDORFF, V. & EVANS, E. W. 1977. Antigoriteophicarbonates: contact metamorphism in Valmalenco, Italy. Contributions to Mineralogy and Petrology, 62, 301-312.

Obsidian localization and circulation in northwestern Patagonia (Argentina): sources and archaeological record C. B E L L E L L I 1, F. X. P E R E Y R A 2 & M. C A R B A L L I D O 3

1Consejo Nacional de Investigaciones Cienffficas y Tdcnicas (CONICET), Instituto Nacional de Antropolog[a y Pensamiento Latinoamericano (INAPL), 3 de Febrero 1370, (1426) Buenos Aires, Argentina (e-mail: [email protected]) 2Instituto de Geologfa, Servicio Geol6gico Minero Argentino (SEGEMAR), Avda. Roca 651, Piso 8, Sector 8, (1322) Buenos Aires, Argentina 3Instituto Nacional de Antropolog{a y Pensamiento Latinoamericano (INAPL), 3 de Febrero 1370, (1426) Buenos Aires, Argentina Abstract: The objectives of the current research are to characterize obsidian sources and

lithic archaeological remains in two areas of NW Patagonia, so as to establish spatial distribution and network circulation patterns. Obsidian tools and remains were found in archaeological contexts dated back 3200 years and up to the 16th-18th centuries. Geological and artefact sample rocks were analysed by instrumental neutron activation analysis and inductively coupled plasma emission mass spectrometry. Four potential sources were detected (Sacanana, Angostura Blanca, Portada Covunco and Laguna La Larga). Obsidians in the region appear to be related to two petrotectonic associations: (1) an arc source, located in the western part of Argentina and Chile, of Plio-Pleistocene age; (2) an intraplate and back-arc source, located in the North Patagonian Massif, east of the studied area, of Miocene age. The latter, because of the higher frequency of tools and remains that appear in archaeological context, seems to be more intensively used. Some hypotheses about obsidian circulation patterns, the mobility of northern Patagonia hunter-gatherers, and landscape occupation throughout the last 3200 years are outlined based on the results obtained in this work.

Before the arrival of Europeans in Patagonia, hunter-gatherer populations intensively used obsidian, as they valued its excellent qualities for tool making, especially projectile points. This rock appears in few places in North Patagonia, therefore it was a valuable good that implied long distance exchanges and an intense circulation among hunter-gatherers throughout the last 3200 years. Geochemical studies of archaeological tools and sources (potential or actually used) are an open door that allows access to the knowledge of pre-Hispanic occupation dynamics of different Patagonian spaces. These help us to become aware of aspects related to social groups' mobility and connected with the supply, circulation and use of this lithic raw material in different environments separated by long distances. The aim of this work is to identify and locate sources within the region, and to characterize the obsidian varieties present in the area, so as to deepen the knowledge of its spatial distribution, its circulation on a wide regional scale

and the chronological trends. These tasks were accomplished on the basis of two datasets: samples obtained by excavating archaeological sites and recently published data from nearby areas (Stern et al. 2000; Bellelli & Pereyra 2002; G r m e z Otero & Stern 2005). The information comes from two archaeological research areas, both located in NW Patagonia: (1) Piedra Parada, in the middle course of the Chubut River; (2) the northern sector of the Futaleuf6 River basin, specifically the Cholila locality and Los Alerces National Park (see Fig. 1). These two sectors are located in different environments. Piedra Parada is in a herbaceous and shrub-like steppe, which is typical of the Extra-Andean Patagonia Ecoregion, whereas Cholila is located in the forest-steppe ecotonal area, which is typical of the North Patagonia Mountain Range Ecoregion. The whole area is located between 41 ° and 44°S in the Argentine Patagonia and occupies about 100 000 km 2. The oldest hunter-gatherer occupations in these regions are recorded in the area of Piedra

From: MAGGETTI,M. & MESSIGA,B. (eds) 2006. Geomaterialsin CulturalHeritage. Geological Society, London, Special Publications, 257, 241-255. 0305-8719/06l$15.00 © The Geological Society of London 2006.

242

C. BELLELLI ET AL. %

t N Argentina { /

Atlantic Ocean

=

O

420

•

South A m e r i c a

o Pacific Ocean

,.-~ \

//

ia

, I.

~j

0

, 100

, 200 k,m

Fig. 1. Map of north Patagonia. Geological sources (circles) and archaeological localities (triangles) mentioned in the text.

Parada (5000 years old, Bellelli 1987). The archaeological use of obsidian occurred from 3200 years ago (Bellelli 1988; P~rez de Micou et al. 1992). In the Cholila area, obsidian began to be used later, after the first occupations known in the area (1900 years old) (Bellelli et al. 2003). O b s i d i a n r o c k s in north P a t a g o n i a Obsidian rocks usually appear as small domelike bodies associated with volcanoes, generally in relation to lateral fractures or fissures. Sometimes they can also appear as small domes or flows, with little longitudinal development. They can also be found as a vitreous shell of rhyolitic flows in boiler environments and fissure eruptions. These obsidian outcrops occupy small areas, so their identification in the field is usually difficult, as it implies a detailed study of a region. In the case of Patagonia, the areas where there are possibilities to find these rocks can reach tens of thousands of square kilometres, hence, there may be numerous nonidentified outcrops. Usually domes and siliceous glass lava domes are associated with deposits of pumices and pyroclastic breaches, and it is possible to differentiate various sectors where

obsidian rocks show different hydration stages (perlitization) and devitrification (with spherulites and microlites). Obsidian can be found in interplate and backarc environments, generally associated with extensional dynamics. It is quite frequent, and at the geological sites that have been studied in this paper, the obsidian rocks used as sources are secondary, occurring as boulders or fragments of boulders, generally associated with short-distance transport in alluvial fans or ejecta cones, or, when transported a longer distance, as structural plains of boulders and lava (typical Extra-Andean Patagonia plateaux). Finally, the physical weathering of obsidian outcrops, because of the diversity they show, produces variable proportions of boulders, potentially usable as almost a primary source. Obsidian sources taken into account for this paper are those of Telsen and Sacanana (Stern et al. 2000), Chait~n Volcano (Chile) (Stern & Curry 1995), Rfo Villegas, Portada Covunco, Laguna La Larga and Angostura Blanca (see Fig. 1). There are other sources in the Patagonian Andes and Precordillera: in the area of Cerro Leones close to Bariloche city and in Lolog lake, near San Martfn de los Andes city (Albornoz, pers. comm.; Prrez & Ldpez 2004). As

ARCHAEOLOGICAL OBSIDIAN SOURCES, PATAGONIA

243

geochemical analyses are not available for these sources, they are not included in this paper. Telsen and Sacanana samples can be related to an acid vulcanite environment, restricted to large effusive centres located in the southern area of the Somuncura Massif related to great basaltic lava plains. Volcanism in this region began in the Eocene and continued until the late Pliocene. In general, the pyroclastic manifestations, probably associated with acid effusions, took place essentially in the Miocene and are included in the Complejo Eruptivo Quifielaf and maybe also in the Somuncura Formation. This magmatism corresponds, according to Ardolino & Franchi (1993) and Ramos & Kay (1992), to the presence of a hotspot, in the intraplate or backarc area, with which extensional structures are associated. Samples from Portada Covunco and Angostura Blanca indicate a similar petrotectonic association. Conversely, the Rio Villegas and Laguna La Larga samples are related to basic lava effusions, which are arc meso-siliceous and back-arc related to extensional events resulting from the collision and subduction of an oceanic ridge (Ramos & Kay 1992). Finally, the source of Chait6n Volcano is associated with arc volcanism with oceanic participation.

heading of 'volcanic-pyroclastic complex of middle Chubut River' to account for the intense volcanism phenomena, which began with magmatic activity in the late Paleocene and continued until the mid-Eocene. These volcanic phenomena left signs of an old caldera in the landscape in a collapsed epi-continental environment, the products of which cover more than 900 km 2. Aragdn & Mazzoni (1997) also described the effusive rocks and acid domes that include obsidian with different degrees of perlitization (Buitrera Vitrophere and Cretton Rhyolites). Buitrera Vitrophere is of great interest, as its manifestations appear in the homonymous wash, where the archaeological site Campo Moncada 2 is located. This is a set of domes and shallow intrusive rhyolites with a greenish-blackish colour. These have a strong perlitic character and show a speckled aspect; the size of the darkest glass spheroids ranges from 1 mm to several centimetres, on the altered greenish glass. For this reason, it has a 'conglomerate' appearance. Recently, it has been described as a vitrophere dome that could include obsidians within the same complex that was called 'Domo de Escuela Piedra Parada' by Aragdn et al. (2004).

P i e d r a P a r a d a r e s e a r c h area

Cholila r e s e a r c h area

The research area includes a sector of the middle Chubut River basin, with an east-west extent of c. 70 kin. It is a transitional environment between the Pre-mountainous and the Extra-Andean Patagonian areas. The latter is characterized by great structural plains crowned by boulders cemented by calcium carbonate ('Patagonian Boulders') of different origins (outwash plains and fluvial terraces) of Plio-Pleistocene ages. In the Chubut River valley, fluvial terraces are well developed. The volcanic morphology is widely represented by calderas or volcanic domes, or, especially, by large lava and ignimbrite structural plains. The predominant geological formations in this area are the Huitrera and E1 Mirador Formations, both Tertiary and of volcanic origin (Lage 1982). To a lesser extent, there are also sedimentary formations: the Colldn Curfi and Norquinco (Paleocene), Lefipfin (at the boundary between the Paleocene and the Cretaceous) and Paso del Sapo (Cretaceous) Formations. In addition, most of the research area includes fluvial and alluvial piedmont deposits. After the publication of the Geological Sheet that shows the above information, Aragdn & Mazzoni (1997) grouped the rocks that makeup the Huitrera and E1 Mirador formations (Paleocene) under the

This area includes the northern part of the upper Futaleufti basin on the eastern slopes of the Andes. Forests border the Patagonian steppe and this transition defines an ecotonal strip with species from these two diverse environments (Bellelli et al. 2000; Podestfi et al. 2000). The landscape originated by the interaction between tectonic dynamics and glacial processes. Valleys are aligned almost in a northsouth direction, controlled by major inverse faults within a folded strip and back-arc environment. The area is characterized by significant relief, which implies considerable geomorphological dynamics, not only now but also in the past. Glaciers flowed through longitudinal valleys and, in some sectors, they crossed the dominant structure integrating the river basins by means of transverse valleys. It is important to mention vulcanite outcrops and Tertiary pyroclastic deposits (PaleoceneEocene) of the Ventana Formation. This volcanic association corresponds to an arc magmatism related to the first pulses of the Andean orogeny. To a smaller extent we can find marine sedimentites, Mesozoic continental formations (Lago La Plata, Lago Puelo and Divisadero Formations), and Tertiary sediments and pyroclastic deposits of the Cholila, lqorquinco

244

C. BELLELLI ET AL.

and Coll6n Cura Formations (Lizuafn 2005). The noticeable presence of Tertiary vulcanites would involve the presence of vitreous rhyolites and especially obsidian rocks. Nevertheless, it is worth mentioning that this volcanism is essentially meso-siliceous (andesitic) and the obsidians are generally more frequent in relation to a basic and trachytic volcanism.

Materials and methods The origins and basic characteristics of the 32 samples analysed are described in Table 1. These samples are limited to some archaeological cases because there are few obsidian tools from the excavated or merely prospected sites in the two studied areas. Source samples were obtained directly or kindly supplied by other researchers. In Piedra Parada area we analysed samples from three stratified sites (Campo Moncada 2, Campo Cerda 1 and Campo Nassif 1), seven surface sites (Campo Moncada 3, Piedra Parada 5, San Ram6n 2, Tranquera Colorada, Bajada del Tigre, Barda Blanca Pasarela and Barda Blanca 4) and two boulders recovered in the valley (Angostura Blanca). There is a very small proportion of obsidian tools at all of the sites, not over 10%. In Campo Moncada 2, which is the stratified site that showed the oldest and best-known stratigraphic sequence in the region at present, obsidian tools occur in the upper levels (surface and layers 1 and 2), dated after 3210 + 50 years BP (UGA 7621) (Pfrez de Micou 2002, p. 59). Both samples of the lower layers (5080 ___ 100 years BP (AC 666) and 4770 + 90 years BP (AC 671), which we include here, are not tools but small and angular fragments that are not suitable for knapping (Samples 4 and 5; Table 1). Campo Nassif 1 and Campo Cerda 1 are rockshelters with rock art. The first is dated at 4 8 0 _ 75 years BP (AC 665) (Onetto 19861987), and at the second site occupations range from 2850 + 50 to 5 8 0 _ 50 years Be (UGA 7454 and LP 415). The oldest date corresponds to the stratigraphic level from which the obsidian sample was extracted, and it also agrees with the beginning of the basket maker technology on the site (Bellelli 1994, 2002). This shortage of obsidian at the stratified sites is found again at surface sites located in different landforms and environments (valley and high fields). Analysed samples in the Cholila area come from the surface sites Juncal de Calderfn 1 and 2. These are located in the western margin of the homonymous 'mallin' (typical Patagonian

wetland) and Los Guanacos 3, on the edge of another 'mallfn'. Two samples from Cerro Pintado were analysed. This rock-shelter with rock art is to the SW of Cholila town. Excavations undertaken at this site gave three radiocarbon dates: a modem one, 680 _ 60 years Br, and 1870 + 80 years BP (LP 1319, LP 1333 and LP 1313) (Bellelli et al. 2003). Finally, a large flake with cortex, a sample from an isolated find, was analysed ('Hallazgo Aislado'). Coming from Los Alerces National Park, another of the analysed samples is an isolated find in the forest (Laguna La Larga I), near an outcrop. The sample called Laguna La Larga 2 belongs to this outcrop, and the sample Laguna La Larga 3 is an isolated core fragment found close to the outcrop. Samples of well-known sources have also been analysed. One of them belongs to the X Region in Chile. It is a boulder recovered on the slopes of Chaitfn Volcano. Two samples from the Somuncura massif (Telsen and Sacanana) were also analysed. These are two boulders of excellent knapping quality from the localities of Cafiad6n Salamanca and Cerro Guacho, which belong to the sources sampled and published by Stern et al. (2000). Finally, we took samples from the Rio Villegas and Portada Covunco areas (river-bed boulders). All 32 samples described were analysed at Activation Laboratories Ltd (ACTLABS) using instrumental neutron activation analysis (INAA), and inductively coupled plasma emission mass spectrometry (ICP-MS). INAA is capable of measuring up to 35 elements at the ppb to percent level in most materials. INAA is dependent on measuring primary gamma radiation, which is emitted by the radioactive isotopes produced by irradiating samples in a nuclear reactor. Each element that is activated will emit a 'fingerprint' of gamma radiation that can be measured and quantified. INAA is exceptionally sensitive to a number of trace elements. This technique does not require the slow ashing procedure of other chemical methods, thus preventing potential loss of certain elements and improving the reliability of data because it minimizes sample handling and potential human error (Hoffman 1992, in Activation Laboratories Ltd. 2005). ICP is a lithium metaborate-tetraborate fusion, which is unique for the scope of elements and detection limits. The trace element in the fusion solution provides research quality data using research detection limits (Activation Laboratories Ltd. 2001, p.15). The sample preparation includes, as a routine practice, the crushing of entire

19 21 22 24

Cholila

10 15 16 17

Piedra Parada 1 2 3 6 7 8 9

Juncal de Calder6n 2, surface Cerro Pintado, Level 2 Los Guanacos 3, sample 1, surface Hallazgo Aislado

Campo Moncada 2, surface Campo Moncada 2, Level 1 Campo Moncada 2, Level 2c Campo Cerda 1, Level 5 (base) Campo Moncada 3, surface Piedra Parada 5, surface San Ram6n 2, surface Tranquera Colorada, surface Barda Blanca Pasarela sample 1, surface Barda Blanca Pasarela sample 2, surface Barda Blanca 4, surface

site, site, site, site, site, site, site, site, site, site, site,

archaeological artefact achaeological artefact archaeological artefact archaeological artefact archaeological artefact archaeological artefact archaeological artefact archaeological artefact archaeological artefact archaeological artefact archaeological artefact

Archaeological site, archaeological artefact Archaeological site, archaeological artefact Archaeological site, archaeological artefact Archaeological isolated find, archaeological artefact

Archaeological Archaeological Archaeological Archaeological Archaeological Archaeological Archaeological Archaeological Archaeological Archaeological Archaeological

Black; Black; Black; Black;

Black; Black; Black; Black; Black; Black; Black; Black; Black; Black; Black;

conchoidal conchoidal conchoidal conchoidal

conchoidal conchoidal conchoidal conchoidal conchoidal conchoidal conchoidal conchoidal conchoidal conchoidal conchoidal

fracture fracture fracture fracture

fracture fracture fracture fracture fracture fracture fracture fracture fracture fracture fracture

(Continued)

Black; conchoidal fracture

Geological source, boulder

Portada Covunco

31 Sacanana/Group C

Portada Covunco

Reddish with black spots; conchoidal fracture Brown with black bands; conchoidal fracture

Black with red bands; brittle Black with red bands; brittle Black; brittle

Black with grey bands; conchoidal fracture

Appearance

Archaeological site, archaeological artefact Archaeological site, archaeological artefact

Isolated surface find, fragment from the outcrop Geological source, fragment from the outcrop Isolated surface find, core fragment

Archaeological site, archaeological artefact

Origin, evidence

Los Guanacos 3, sample 2, surface Juncal de Calder6n 1, surface

Laguna La Larga sample 1 Laguna La Larga sample 2 Laguna La Larga sample 3

Cerro Pintado, Level 4

Site, level

23 18

Cholila

25 26 27 Portada Covunco/ Group B

Los Alerces National Park

20

Cholila

Laguna La Larga/ Group A

Sample no.

Table 1. Analysed and discussed samples

Telsen 28 R{o Villegas 3O Chile 32 Unknown 1 Piedra Parada 4 Unknown 2 Piedra Parada 5

Without correspondance

Piedra Parada 11 12 13 14

Angostura Blanca/ Group D

Sacanana 29

Sample no.

Table 1. Continued

Black; brittle Black with red bands: brittle Black; brittle Black; brittle

Geological source, boulder Geological source, boulder Archaeological site, fragment of a boulder Archaeological site, fragment of a boulder

El Rinc6n

Volczin Chait6n, X Regi6n

Campo Moncada 2, Level 3b

Campo Moncada 2, Level 4a

conchoidal fracture conchoidal fracture brittle brittle

Black; conchoidal fracture

Black; Black; Black; Black;

Black; conchoidal fracture

Appearance

Geological source, boulder

Archaeological site, archaeological artefact Archaeological site, archaeological artefact Isolated surface find, boulder Isolated surface find, boulder

Geological source, boulder

Origin, evidence

Cafiad6n Salamanca

Campo Nasif l, Level 1 Bajada del Tigre, surface Angostura Blanca sample ! Angostura Blanca sample 2

Cerro Guacho

Site, level

ARCHAEOLOGICAL OBSIDIAN SOURCES, PATAGONIA samples to a nominal minus 10 mesh (1.7 mm), followed by mechanical splitting (riffle) to obtain a representative sample and then pulverizing to at least 95% minus 150 mesh (106 I,zm). The laboratory automatically uses cleaner sand between each sample (Activation Laboratories Ltd. 2001, p.7). Samples are weighed into a small custom-made polyethylene vial to totally fill the vial. For every 11 samples, a CANMET WMS-1 standard is co-irradiated with flux wires at a thermal neutron flux of 7 x 1012 neutrons c m - 2 s -1 for 15 min in the RIFLS site of the McMaster Nuclear Reactor. After a decay of 7 days the samples are counted on a high-purity co-axial GE Detector with a resolution of better than 1.7 keV for the 1332 k e V 6° Co photopeak. Using the flux wire monitors the data are corrected for decay and compared with a calibration developed from multiple international reference materials (c. 50). The CANMET standard WMS-1 is used solely as a check on the procedure and not for calibration proposes. Selected samples are re-counted and compared with the original as part of the Q A / Q C procedure (Hoffman 1992, in Activation Laboratories Ltd. 2005). The values given by the laboratory were processed with the NewPet program developed by the Centre for Earth Resources Research (1987-1993) of the Memorial University of Newfoundland, Canada, especially designed for the management of geochemical data (see Tables 2 and 3). Results for major elements were expressed in oxide percentage and hydration percentage (LOI), whereas trace elements and rare earth elements (REE) were expressed in txg g-~. These analytical methods have proven to be highly effective for the chemical fingerprinting of geological and archaeological samples (Glascock et al. 1998). Trace elements and mainly REE have great resolution power, as their concentrations vary within a wide range of values. In the case of some rocks such as obsidian, their different concentrations are specific for each geological formation or part of a formation, so, this 'chemical trace' allows, within certain statistical limits, the identification of possible origin sources of obsidian tools (Escola et al. 2000, p. 17). Results Table 2 shows the chemical composition of samples expressed in oxide percentages (SiO2, A1203, N a 2 0 , K 2 0 and FexOy (total iron, ferrous plus ferric iron). These percentages are in the range expected for obsidian rocks and contribute to their characterization and

247

classification as rhyolites. The water content is also indicated as LOI in Table 2. Trace elements and REE are present in concentrations < 1% and are shown in Table 3, expressed in ~g g-1. The REE help to distinguish obsidian rocks from different sources in a very reliable way, as they differ by one or two orders of magnitude between sources, whereas the variations within each source are smaller and more difficult to determine. We are dealing mainly with obsidian rocks with a high siliceous content (in some cases over 75%, metaluminous to peraluminous, and with a large content of K). In addition to the differences detected in the siliceous and aluminium content, the alkalinity of samples from different sources also varies, which would indicate their different petrotectonic associations. Some of our samples are of very poor knapping quality, a fact that is closely related to large degrees of hydration of the sample. Water content for obsidian rocks apt for knapping should be between 0.1 and 0.5%. As these values increase, the rocks become, as noted by Glascock et al. (1998, p. 19), 'a less useful glass form, known as perlite'. Those authors have established the limit at 3.5% and, according to Table 2, Samples 4 and 5 (Campo Moncada 2), 13 and 14 (Angostura Blanca), 25, 26 and 27 (Laguna La Larga), 30 (Rfo Villegas) and 32 (Chaitrn volcano), greatly exceed that limit. Nevertheless, their values were compared with each other and with the rest of the studied samples. Highly incompatible trace element and REE values as well as the observed relationship between them were plotted on bivariate and trivariate graphs. Only some of these are presented here (Fig. 2 a - c and Tables 1-3) and it is possible to identify four groups. Group A. A survey in Los Alerces National Park, south of Cholila, showed the location of a poor-quality obsidian outcrop in the northern margin of Laguna La Larga at 45 km from Cholila (samples 26 and 27). One isolated find of the same type (sample 25) was made. Information has also been published by Arrigoni (1999) regarding the existence of obsidian outcrops in the eastern ridge of the Futalaufquen Lake, a few kilometres far from Laguna La Larga. These perlitized samples have the same geochemical values, and are similar to archaeological sample 20 obtained at Cerro Pintado (Cholila). Group B. Archaeological samples 18 and 23 are associated with a concentration of good knapping quality boulders located in the Portada Covunco area (sample 31) (430km north of Cholila).

Piedra Parada 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 Cholila 18 19 20 21 22 23 24 b~s Alerces National Park 25 26 27 Telsen 28 Sacanana 29 R{o Villegas 3O Portada Covunco 31 Chile 32

Sample no. 76.46 75.70 75.71 72.84 70.19 76.03 76.02 76.06 76,10 75.79 74.02 74.27 71.70 7 !.44 74.65 74.94 75.06 76.21 75.94 69.69 69.93 75.25 74.91 74.53 71,04 70.75 70.86 73.49 77.05 55.79 75.48 69.24

Juncal de Calder6n I Juncal de Calder6n 2 Cerro Pintado, Level 4 Cerro Pintado, Level 2 Los Guanacos 3, sample 1 Los Guanacos 3, sample 2 Hallazgo Aislado

Laguna La Larga, sample I Laguna La Larga, sample 2 Laguna La Larga, sample 3

Cafiad6n Salamanca

Cerro Guacho

E1 Rinc6n

Portada Covunco

Volcfin Chait6n, X Regi6n

SiO2

Campo Moncada 2, surface Campo Moncada 2, Level 1 Campo Moncada 2, Level 2c Campo Moncada 2, Level 3b Campo Moncada 2, Level 4a Campo Cerda 1, Level 5 base Campo Moncada 3 Piedra Parada 5 San Ram6n 2 Tranquera Colorada Campo Nassif I, Level 1 Bajada del Tigre Angostura Bianca, sample 1 Angostura Blanca, sample 2 Barda Blanca Pasarela, sample I Barda Blanca Pasarela, sample 2 Barda Blanca 4

Site, level

Table 2. Chemical composition of the samples (in % of oxides)

0.20

0.138

0.176

0.12

0,12

0.04 0.04 0.042

0.13 0.12 0.051 0.127 0.116 0.134 0.1 i 7

0.1 I 0.12 0.12 0.13 0.27 0.12 0.12 0.1 I 0.12 0. ! 2 0.26 0.26 0.23 0.23 0.117 0.112 0.117

TiO2

12.85

13.19

3.12

1 1.90

I 1.03

12.60 12.63 13

12.98 12.31 12.61 1 1.47 12.37 12.98 12.48

! 2.16 12.38 12.40 I 1.67 I1.79 12.25 12.30 12.25 12.26 12.37 13.38 13.37 12.53 12.44 12.2 12.13 12.36

A!203

2.08

1.23

3.38

1.75

3.57

1.32 1.31 1.42

I. 15 1.89 1.65 2.33 2.34 1,55 2.35

1.79 ! .92 2.02 1.76 3.04 1.92 i .9 i 1.86 1.89 1.92 2,02 2.00 1.73 ! .71 2.15 2.18 2.24

Fe203

0.10

0.05

0.04

0.04

0.13

0.07 0.07 0.07

0.06 0.05 0.07 0.07 0.05 0.06 0.05

0.04 0.05 0.05 0.04 0.06 0.05 0.05 0.05 0.05 0.05 0.04 0.04 0.03 0.03 0.05 0.05 0.05

MnO

0.25

0.09

0.37

0.02

- 0.01

0.05 0.05 0.09

0.09 0.01 0.22 0.56 0.03 0. i 0.04

0.02 0.03 0.06 0.09 0.28 0.02 0.02 0.02 0.02 0.02 0.32 0.31 0.19 0.19 0.04 0.03 0.02

MgO

1.36

0.54

0.41

0.46

0.09

0.7 ! 0.73 0.73

0.57 0.49 0.92 1.29 0.5 0.55 0.57

0.47 0.50 0.53 0.53 1.26 0.50 0.50 0.48 0.49 0.49 1.17 1.15 0.85 0.84 0.49 0.5 0.48

CaO

4.27

4.44

0.24

4.09

6.48

4.01 4.03 4.44

4.21 4.25 3.73 4.07 4.44 4.33 4.38

4.16 4.20 4.12 3.23 3.93 4.19 4.20 4.19 4.22 4.32 4.23 4.22 3.92 3.95 4.32 4.26 4.34

Na20

2.65

4.7

0.16

4.45

3.92

3.68 3.57 3.19

4.66 4.49 4.66 4.5 4.65 4.73 4,83

4.56 4.59 4.39 4.82 3.10 4.58 4.49 4.50 4.67 4,57 4.02 3.95 3.96 3.75 4.68 4.75 4.8

K20

0.07

0,02

0.01

0.02

0,03

0.04 0.10 0.03

0.03 0.02 0.18 0.73 0.01 0.02 0.03

0.02 0.02 0.02 0.03 0.09 0.02 0.03 0.02 0.02 0.02 0.08 0.07 0.13 0.04 0,03 0.02 0.01

P205

7.00

0.32

35.55

0.42

0.36

5.60 6.87 6.47

0.29 0.49 5.32 4.19 0.54 0.23 0.76

0.37 0,35 0.70 5.05 5.94 0.49 0.46 0.39 0.41 0.51 0,15 0.06 4.72 5.13 0.54 0.69 0.8

LOI

100.08

100.20

99.24

100.32

99.23

99.16 100.14 100.34

100.39 100.06 99.10 99.27 i (X).30 99.6(I 100.34

100.16 99.86 100. ! 1 100.19 99.93 100.15 i 00.10 99.93 100.25 100.18 99.70 99.70 99.98 99.77 99.27 99.66 100.28

Total

:~

m t" t't'-" t--' t~

Site, level

15

14

13

12

11

9 10

8

7

6

5

4

3

2

1

Campo Moncada 2, surface Campo Moncada 2, Level 1 Campo Moncada 2, Level 2c Campo Moncada 2, Level 3b Campo Moncada 2, Level 4a Campo Cerda 1, Level 5 base Campo Moncada 3 Piedra Parada 5 San Ram6n 2 Tranquera Colorada Campo Nassif 1, Level 1 Bajadadel Tigre Angostura Blanca, sample 1 Angostura Blanca, sample 2 BardaBlanca Pasarela, sample 1

Piedra Parada

no.

Sample

4.3

4.5

6.8

283

286

156

4.3

279

292

142

130

127

137

4.4

5.6

5.6

2.8

2.9

4.3 3.9

4.6

313

278 279

4.3

289

182 53

4.7

Cs

296

Rb

3 4

3

4

3

68

26

6

4

3

Sr

11

527

494

609

3

67

60

95

616 103

15 18

15

22

18

379

680

25

18

14

Ba

Ta

3.1

2.9

17

16

16

18

19

18

18

19

123 123

119

137

128

45

35

123

122

117

Nb

9.44 127

1.8

1.8

2.0

2.2

29 10 29 11

27 10

31 11

29 11

27

22

29 11

28 10

28 l0

Ga

0.72 31

1.0

0.9

0.6

0.8

0.9 0.8

1.0

0.9

0.9

1.3

0.6

0.5

0.8

1.0

T1

372

366

334

Zr

390 384

378

419

374

665

14.3 359

7.5 296

7.4 276

5.8 250

6.1 250

14 13

14

14

14

14

9.7 346

13

13

13

Hf

Trace elements (Ixg g - l ) Th

52 25.5

29 14

27 14

22 18

22 18

56 26 56 26

54 27

58 26

54 26

70 l l

60 13

55 26

56 26

56 26

Y

6.8

3.3

3.1

4.0

4.0

5.8 5.8

5.9

6.2

6.0

2.3

3.1

5.8

5.8

6.2

U

T a b l e 3. Chemical composition of the samples (trace elements and rare earth elements)

86.7

34

33

40

41

86 89

83

92

91

87

58

86

85

76

La

164

62

62

70

70

167 169

163

173

172

167

113

164

167

147

Ce

18.4

6.37

6.18

6.60

6.62

16.3 16.9

16.3

17.2

16.6

18.0

12.4

16.0

16.6

14.9

Pr

62

25

26

24

25

63 64

62

67

66

76

54

63

63

59

Nd

9.8

9.4

9.8

0.42 0.42

5.7

5.6

0.09 10.4

5.3 0.67

5.2 0.69

3.4

3.4

9.5 9.3

0.41 10

0.42 10

0.45

1.79 14

1 . 4 8 12

0.43

0.43

4.1 0.65

13

Gd

0.44 II

Eu

4.3 0.64

13 13

12

13

13

15

12

13

12

12

Sm

5.1

5.0

3.3

3.7

11 10

10

11

11

12

12

10

10

10

Dy

1.89 10.4

0.8

0.8

0.5

0.5

1.6 1.7

1.7

1.7

1.7

2.0

1.9

1.7

1.6

1.7

Tb

Rare earth elements (Ixg g - l )

3.1

3.0

2.0

2.1

5.1 5.1

5.1

5.2

5.0

7.1

6.4

5.0

5.0

5.2

Er

0.49

0.50

0.34

0.38

0.72 0.74

0.75

0.81

0.76

1.10

1.00

0.77

0.70

0.71

Tm

Lu

4.6 0.64

3.2 0.60

3.3 0.57

2.3 0.45

2.5 0.47

4.4 0.56 4.6 0.58

4.3 0.58

4.4 0.54

4.6 0.57

6.5 1.02

6.1 0.93

4.6 0.59

4.5 0.63

4.4 0.60

Yb

(Continued)

1.91 5.43 0.81

1.0

1.0

0.7

0.7

1.8 1.9

1.9

1.9

1.9

2.5

2.2

1.8

1.8

1.9

Ho

4~

>

,..] >" 0

t'rl

O 7~

>

> t" O

O

O

Z: >

>, 7Z

Barda Blanca Pasarela, sample 2 Barda Blanca 4

Site, level

4.4

4.2

4

4.3

8.1

4.3

292

118

274

286

169

272

6.0

8.0

4.3

4.4

Cs

165

288

293

Rb

b~s Alerces National Park 25 Laguna La 128 Larga, sample I

Cholila 18 Juncal de Calder6n I 19 Juncal de Calder6n 2 20 Cerro Pintado, Level 4 21 Cerro Pintado, Level 2 22 Los Guanacos 3, sample I 23 Los Guanacos 3, sample 2 24 Hallazgo Aislado

17

16

Sample no,

T a b l e 3. C o n t i n u e d

923

10

252

11

II

818

18

259

10

11

Ba

60

4

46

4

5

46

3

42

3

4

Sr

Ga

16

0.7

18

0.68 29

1.4

0.75 30

0.53 29

132

Nb

3.5

75.6

1.9

19

8.99 l l0

3.04

Zr

4

14 94

405

5.1 163

1 3 . 7 364

13.7 342

Hf

4.0 I07

13.9 364

5.2 148

13.6 372

8 5 . 1 13.5 344

15

9.04 136

9.01

1.62

127

26

9.21 137

9.3

Ta

29 II

15

0.44 18

0.9

1.1

0.57 32

0.81 31

TI

Trace elements (t-tg g i) Th

9.17

38

9.5

53 24.3

19 26.7

53 25.2

50 24.2

33

57 27

19 28

52 25.4

52 25.6

Y

86

34

88.3

83.1

La

2.1

6.47

7.63

6.7

6.47

25

83.5

32.5

82.8

85.2

2 . 3 1 25.1

6.1

6.9

6.71

6.8

U

51

158

56.2

158

160

50.7

164

57

166

159

Ce

5.64

17.6

5.71

17.5

17.9

6.11

16.8

5.46

18.4

17.9

Pr

24

60

18

58

59

22

64

19

62

60

Nd

0.09

7.5

0.09 10

5.7 (}.48

12

1.7

0.4

1.9

0,6

1.81

5.0

2.0

0.77

0.35

5.14 0.78

0.78

Tm

1.79 0.33

1.0

6.2

1.2

3.9

0.66

1.77 9.74 1.82 5,08 0.78

2.75 0,58

4.88 0.77

9.58 1.81 4.95 0.76 1 . 7 6 9.72 1.8

1.79

Er 1.88 5.2

Ho

5.56 1,15 3.53 0.58

10

2.9

1.77 l0

2.41 0.44

9.9

0.09 10.2

3.2 0.36

12

12

Dy 1 . 8 7 10.2

Tb

4.95 0.91

0.42 10

2.6

0.09 10.2

5.1 0.64

12

Gd

0.09 10.3

Eu

3.4 0.63

13

12

Sm

Rare earth elements (Ixg g J) Lu

4.3 0.80

4.6 0.63

2.3 0.36

4.4 0.61

4.4 0.61

3.7 0.59

4.4 0.54

2.7 0.58

4.5 0.62

4.7 0.62

Yb

t" t"

t-" t-"

Cafiad6n Salamanca

LagunaLa Larga, sample 2 LagunaLa Larga, sample 3

5.5

44

1.2

1.9

80

3.2

678 121

0.4

14

0.88 15

0.7

3.02

0.07

31 13

56 41

3

18

8

36.1

36.7

141

471

1 . 5 1 35.6

1.9

35

36

8.82

9.2

32

353 1.16

5.9 223

35 11

18 26.8

5

64 32

752 232 80

98

5.2 142

0.8

13

78

3.7

3.8 103

2.4

7.58

0.8

7.9

18

2.22

2.1

32

35

1.52

78

208

24

26

64

59.4

3.33

153

334

48.2

51

7.02

6.01

0.47

15.6

46.3

5.82

5.55

30

19

2.5

60

176

22

25

0.16

1.9

7.0

6.7 1.08

7.2

1.0

2.47 0.46

0.9

0.42 12

3.3 0.38

1.0

4.87 0.92

7.2

3.85 44

0.9 0.26

13

42

5.1 0.61

5.5 0.58

6.0

2.8

3.9

0.65

2.0

7.6

1.2

0.58

0.96

3.8

0.59

1.87 0.35

4.3 0,85

2.4 0,37

0.8 0.11

5.5 0.76

2.69

3.7 0.56

4.3 0.83

3.23 19

0.57 0.1

5.9

21

1 . 1 1 3.36 0.58

1.2

0.96 0.2

11

41

5.4

6.0

z.

> C~ 0

>

0

7~

0

©

7Z 9 3~

0

Volc~inChait6n

47

30 - 0 . 0 5

4

17

0.79 17

1.0

~

263

50

14

7 -2

1044 123

849

Chile 32

7.7

0.3

4

156

5.6

9.1

325

585

193 35

160

Portada Covunco 31 Portada Covunco

Sacanana 29 Cerro Guacho R{o Villegas 30 El Rinc6n

Telsen 28

27

26

252

C. BELLELLIETAL.

1000

1

'

u.m~o,.,,2 t

¢

r..,ou~A G ~ 8

~++

100

(a) i

(b) ,

i

....

lO

t

i

i

100

1000

p,b~)

I

,

,

,

,

,

,,1

i

i

,

i

,

,,

/-

rh 1~9)

Fig. 2. Plots of geochemical REE and trace element values.

Group C. Archaeological samples 1-3, 6 - 1 0 and 15-17 from Piedra Parada, and samples 19, 21, 22 and 24 from the Cholila region come from the source located in Sacanana (sample 29), 230 km and 160 km respectively from both archaeological areas. Group D. Isolated find recovered in the Piedra Parada valley (perlitized samples 13 and 14 from Angostura Blanca) are identical and also show the same values for some elements, (Rb, K, Nb, Ce, Eu and Ti). There is a correspondence between these samples and archaeological tool samples 11 and 12. Perlitized samples 4 and 5 from Piedra Parada valley do not have clear correspondence to any known source or archaeological sample. In addition, they each have different values. Therefore we have called them 'Unknown 1' and 'Unknown 2'; nevertheless, the former has a slight resemblance to Group D. Samples 28, 30 and 32 belonging to the sources Telsen (east of Somuncura massif), Rio Villegas and Chait6n volcano (Chile) have no geochemical relationships with any archaeological sample analysed here. Finally, samples that belong to Group C show high percentages of silica in non-perlitized samples. The Si-O2 contents exceed 75%, with

high contents of K and Na and comparatively low contents of Al and Ca. Consequently, they are vitreous rhyolites with an alkaline tendency, metaluminous, that fall in the field of high-K rhyolites (Tables 2 and 3). On the other hand, Group D (Table 1) shows a lower content of SiO2, although it is still high, and it has a slightly less alkaline tendency despite the fact that it also corresponds to high-K rhyolites (Tables 2 and 3). Group A samples have lower SiO2 contents, but looking at the LOI values it is possible to relate this fact to a higher degree of perlitization. Group B has similar features to Group C (highK rhyolites, slightly alkaline and metaluminous).

Discussion and conclusions Obsidian may be characterized as one of the best quality raw materials for the knapping of lithic tools and this is the major reason why it has been widely used in the past. In Patagonia, it is a scarce rock that appears in precise locations. Its geochemical characteristics allow a chemical fingerprint analysis. It is one of the few rocks that offers the possibility of establishing potential sources and origin of tools within statistical limits.

ARCHAEOLOGICAL OBSIDIAN SOURCES, PATAGONIA Archaeological obsidian tools from several stratified and surface sites of Piedra Parada could be identified as deriving from the Sacanana source (Group C), 160 km NE from Piedra Parada, in the Somuncura massif. Additionally, there are two samples from two sites in different environments (Campo Nassif 1 in the river valley and Bajada del Tigre in the high fields, 13 km apart) that are identical and are related to perlitized samples found in Angostura Blanca (Group D). In a previous paper, we called them 'unknown X' (Bellelli & Pereyra 2002), but after the analysis presented here, it is certain that the tool samples come from somewhere near Angostura Blanca, and therefore have a local origin. Cholila archaeological samples show the use of three sources: Groups A, B and C. The Sacanana source (Group A) is 230 km to the east and the Portada Covunco source is more than 400 km to the north (Group B). Previously (Bellelli & Pereyra 2002) the latter group was labelled 'unknown Y'. Three facts lead us to consider the existence of good knapping quality obsidian rocks in the region: the results showing similarities between the archaeological sample from Cholila and the perlitized samples from Laguna La Larga (50 km to the south) (Group A), the geological data and the references of Arrigoni (1999). At present, Laguna La Larga would be the only local source regionally used, as the archaeological obsidians in groups B and C come from very distant sources. Summarizing, in the research area, obsidian rocks from four sources were used. Sacanana is the only source represented in both archaeological areas (Piedra Parada and Cholila) and also is the most frequently used. This source supplied archaeological sites located in the centre, west and east of Chubut, such as Cerro Castillo (90 km NW of the source), Los Altares (more than 100 km south), Las Plumas (more than 180km SE) (Stern et al. 2000, p. 288) and several locations on the Atlantic coast (between 270 km and 400 km east of the source) (Gdmez Otero & Stern 2005). Other sources that were taken into account in this paper are not present in analysed archaeological assemblages. This is the case for the source called Telsen by Stern et al. (2000), in the Somuncura massif, which seems to make a restricted raw material contribution to the centre-east of Chubut (Las Plumas and Peninsula de Vald~s) (G6mez Otero & Stern 2005). Also, in Argentine northwestern Patagonia there is no archaeological evidence for the use of obsidian from Chaitrn volcano, on the western slope of the Andes.

253

Moreover, the sample from Rio Villegas shows different characteristics from the rest and therefore it could not be classified. This obsidian has very poor knapping quality and a high water content. Identified obsidians appear to be associated with two petrotectonic environments: (1) an arc environment, located in the western area of Argentina and Chile, which is younger (Chait~n volcano, Chile; Laguna La Larga, Los Alerces National Park, Chubut); (2) an intraplate environment, located in the Northpatagonic Massif, to the east of the surveyed area, which is older (Sacanana, Telsen, Portada Covunco and Angostura Blanca). The sources that belong to this latter environment, based on the frequency of stone flakes and tools taken as samples in archaeological sites, seem to have been used more intensively. Volcanism in this region began in the Eocene and continued until the Pliocene. Associated pyroclastic rocks are included in the Miocene age 'Complejo Eruptivo Quifielef' and Somuncura Formation. Regarding the chronological framework, the use of obsidian (sparsely represented in the archaeological record of both areas) has been recorded in the contexts of Piedra Parada with ages of 3200 years or younger, and was absent in the previous period (5000-3200 years ago) (Bellelli 1988; Prrez de Micou et al. 1992) (the brittle fragments recovered in these contexts were not archaeological tools: Unknown 1 and 2). One of the samples analysed here (from the Sacanana source, Group C) comes from layer 2c of Campo Moncada 2, dated 3210 + 50 years Bp (UGA7621) (Prrez de Micou 2002). This indicates for the first time that this raw material was used in the research area. Its use continued until 480 + 75 years Be (Onetto 1986-1987) in Campo Nassif 1, where the sample from Angostura Blanca was collected (Group D). Regarding the Cholila area, the oldest occupations are dated 1900 years ago. The use of raw material belonging to Groups A (local) and C (non-local) was recorded since this time at the stratified site Cerro Pintado and at surface sites. In two of the latter, obsidian from Portada Covunco (Group B) was also observed.

Concluding remarks With ages from 3200 years onwards, plant-based manufactures and remains of non-local plants (from the Andean-Patagonia mountain range, where Cholila is located) obsidians are found in the archaeological contexts of Piedra Parada (P~rez de Micou et al. 1992; Marconetto 2002).

254

C. BELLELLI ETAL.

W e believe that an exchange of ideas, information, goods a n d / o r groups and a more intensive occupation of space after that date would be taking place in northwestern Patagonia, based on long-term networks of environmental knowledge, resources and communication channels. In addition to obsidian, in this period certain goods in 'prestige technologies' (sensu H a y d e n 1998) were widely circulated in North Patagonia. These items are manufactured with special care and are highly curated. A m o n g these are engraved slate plates, necklace beads, egg shells, w o o d e n and bone tools, all of them decorated, which have been recovered in both research areas. The Andes mountain range, because of its topography and its biophysical conditions (deciduous forest and dense Valdivian forest to the west) has conditioned the circulation of goods, people, information and ideas. That the Chaitrn Volcano source is not represented in the analysed samples supports this idea, and is an important clue in determining the circulation of goods on both sides of the Andes. This research has been supported by Fundaci6n Antorchas and Consejo Nacional de Investigaciones Cientfficas y Trcnicas. Special thanks are due to the two reviewers. who offered valuable suggestions. We appreciate the editors' help and patience. We also thank G. Gur~ieb, who kindly helped us with the translation.

References ACTIVATION LABORATORIES LTD. 2001. Fee Schedule. Quality Analysis. hmovative Technologies. Activation Laboratories Ltd., Ancaster, Ont. ACTIVATION LABORATORIES LTD. 2005. Techniques: Trace Elements Analysis. http://www.actlabs. com/trace_elem_analysis.htm. ARAGON, E. & MAZZONI, M. M. 1997. Geologfa y estratigraffa del complejo pirocl~istico del rfo Chubut medio (Eoceno), Chubut, Argentina. Revista de la Asociacirn Geolrgica Argentina, 52(3), 243-256. ARAGON, E., GONZ,~LEZ, P., AGUILERA, Y., MARQUETTI, C., CAVAROZZI, C. ~:; RIBOT, A. 2004. E1 domo vitroffrico Escuela Piedra Parada del Complejo Volc~nico piroclfistico del rio Chubut Medio. Revista de la Asociaci6n Geolrgica Argentina, 59(4), 634-642. ARDOLINO, A. &; FRANCHI, M. 1993. Hoja Geolrgica 4366-1 Telsen. Direcci6n Nacional del Servicio Geol6gico, Buenos Aires, Boleffn 215. ARRIGON1, G. 1999. An~ilisis del territorio de explotaci6n de los recursos b~isicos para la subsistencia de los grupos prehistrricos que habitaron el valle del r/o Desaguadero. Parque Nacional Los Alerces. Chubut. Abstracts XIII Congreso Nacional de Arqueolog[a Argentina, C6rdoba.

BELLELLI, C. 1987. El componente de las capas 3a, 3b y 4a de Campo Moncada 2 (CM2) (Pcia. del Chubut) y sus relaciones con las industrias laminares de Patagonia Central. Comunicaciones. In: 1 Jornadas de Arqueologfa de la Patagonia, Serie Humanidades. Gobierno de la Provincia del Chubut, Rawson, 27-32. BELLELLI, C. 1988. Recursos minerales: su estrategia de aprovisionamiento en los niveles tempranos de Campo Moncada 2, Valle de Piedra Parada, Rio Chubut. In: YACOBACCIO, H. (ed.) Arqueolog{a Contemporrnea Argentina. Btisqueda, Buenos Aires, 147- 176. BELLELLI, C. 1994. Excavaciones en Campo Cerda 1 (Valle medio dei rio Chubut). Actas v Memorias del XI Congreso Nacional de Arqueolog{a Argentina (Res~hnenes). Revista del Museo de Historia Natural de San Rafael, Mendoza, 14(1-4), 285287. BELLELLI, C. 2002. Dataciones por AMS de artefactos realizados con trcnicas cesteras en Campo Cerda 1 (Valle de Piedra Parada, Chubut). Cuadernos del Instituto Nacional de Antropolog{a, 19, 660-662. BELLELLI, C. & PEREYRA, F. X. 2002. An~ilisis geoqufmicos de obsidiana: distribuciones, fuentes y artefactos arqueolrgicos en el Noroeste del Chubut (Patagonia argentina). Werken, 3, 99-118. BELLELLI, C., PEREYRA, F. X., FERN,~NDEZ, P., SCHEINSOHN, V. & CARBALLIDO,M. 2000. Aproximaci6n geoarqueol6gica del sector sur de la Comarca Andina del Paralelo 42:S (Cholila, Chubut). Cuaternario y Ciencias Ambientales. Publicaci6n Especial CADINCUA-COMINCUA, Buenos Aires, 1, 15-21. BELLELLI, C., CARBALLIDO, M., FERNANDEZ, P. & SCHEINSOHN, V. 2003. El pasado entre las hojas. Nueva informaci6n arqueol6gica del noroeste de la provincia del Chubut, Argentina. Werken, 4, 25 -42. CENTER FOR EARTH RESOURCES RESEARCH 1987-1993. NewPet (~. Memorial University of Newfoundland, St. John's. ESCOLA, P., V~.ZQUEZ,C. • MOMO, F. 2000. An~ilisis de procedencia de artefactos de obsidiana: una vfa metodolrgica de acercamiento al intercambio. Arqueolog(a Contemporrnea, 6, 11-32. GLASCOCK, M. D., BRASWELL,G. E. & COBEAN, R. H. 1998. A systematic approach to obsidian source characterization. In: SHACKLEY, M. S. (ed.) Archaeological Obsidian Studies. Plenum, New York, 15-65. GOMEZ OTERO, J. c~¢ STERN, CH. 2005. Circulaci6n, intercambio y uso de obsidianas en la costa de la provincia del Chubut (Patagonia, Argentina) durante el hoioceno tardio. Intersecciones, 6, 93-108. HAYDEN, B. 1998. Practical and prestige technologies: the evolution of material systems. Journal of Archaeological Method and Theory, 5(1), 1-55. LAGE, J. 1982. Descripci6n geolrgica de la hoja 43c, Gualjaina. provincia del Chubut. Servicio Geol6gico Nacional, Buenos Aires, Boletfn 189. LIZUAfN, A. 2005. Hoja Geoldgica Esquel. SEGEMAR-IGRM, informe inrdito.

ARCHAEOLOGICAL OBSIDIAN SOURCES, PATAGONIA MARCONETTO, M. B. 2002. Anfilisis de los vestigios de combusti6n de los sitios Alero Don Santiago y Campo Moncada. In: PEREZ DE MIcou, C. (ed.) Plantas y cazadores en Patagonia. Facultad de Filosoffa y Letras, Universidad de Buenos Aires, Buenos Aires, 33-53. ONETTO, M. 1986-1987. Nuevos resultados de las investigaciones en Campo Nassif 1. Valle de Piedra Parada, Provincia del Chubut. Relaciones de la Sociedad Argentina de Antropolog(a, 17(1), 95-121. PI~REZ,A. & LOPEZ, L. G. 2004. Obsidianas Lolog. Una cantera de obsidiana en el bosque meridional neuquino. In: TAMAGNIM, M. & MENDON~A, O. (coord.) Publicacidn de rest~menes del XV Congreso Nacional de Arqueologfa Argentina. Universidad Nacional de Rfo Cuarto, Rfo Cuarto, 415. PEREZ DE MICOU, C. 2002. Tecnologfa cestera en Patagonia. Fechando artefactos. In: PEREZ DE MICOU, C. (ed.) Plantas y cazadores en Patagonia. Facultad de Filosoffa y Letras, Universidad de Buenos Aires, Buenos Aires, 55-63. PI~REZ DE MICOU, C., BELLELLI, C. & ASCHERO, C. 1992. Vestigios minerales y vegetales en la determinacidn del temtorio de explotaci6n de un sitio. In: BORRERO, L. A. & LANATA, J. L. (eds) An6lisis

255

Espacial en la Arqueolog{a Patag6nica. Ayllu, Buenos Aires, 53-82. PODESTA, M. M., BELLELLI, C., FERN,/~NDEZ, P., CARBALLIDO, M. & PANIQUELLI, M. 2000. Arte rupestre de la Comarca Andina del Paralelo 42°: un caso de anfilisis regional para el manejo de recursos culturales. In: PODESTA, M. M. & DE HOYOS, M. (eds) Arte en Ias rocas. Arte rupestre, menhires y piedras de colores en Argentina. Sociedad Argentina de Antropologfa & Asociaci6n Amigos del Instituto Nacional de Antropologfa, Buenos Aires, 175-201. RAMOS, V. & KAY, S. 1992. Southern Patagonia basalts and deformation: back arc testimony of ridge collision. Tectonophysics, 205, 261-282. STERN, CH. & CURRY, P. 1995. Obsidiana del sitio Pose Las Conchillas, Isla Traigtien (45 ° 30'S), Archipi61ago de los Chonos, Chile. Anales del lnstituto de la Patagonia, Serie Ciencias Humanas, 23, 119-124. STERN, CH., GOMEZ OTERO, J. & BELARDI, J. B. 2000. Caracterfsticas qufmicas, fuentes potenciales y distribuci6n de diferentes tipos de obsidianas en el Norte de la Provincia del Chubut, Patagonia, Argentina. Anales del Instituto de la Patagonia, Serie Ciencias Humanas, 28, 275-290.

Prehistoric polished stone artefacts in Italy: a petrographic and archaeological assessment CLAUDIO D'AMICO 1 & ELISABETTA STARNINI 2

1Dipartimento di Scienze della Terra e Geologico-Ambientali, Universitgl di Bologna, Piazza San Donato, 1, 1-40126, Bologna, Italy (e-mail: [email protected]) 2Soprintendenza per i Beni Archeologici della Liguria, Via Balbi 10, 1-16126, Genova, Italy (e-mail: estarnini@ hotmail.com) Abstract: The paper illustrates the results of an archaeometric project on the raw material

characterization of some collections of prehistoric polished stone tools, dated from the Early Neolithic to the Bronze Age, from sites located in Northern Italy. The petrographic analyses (surface and thin-section microscopy, X-ray powder diffraction, scanning electron microscopy-energy-dispersive spectrometry, X-ray fluorescence, atomic absorption spectrometry) revealed a raw material circulation network involving the whole of Northern Italy. Here occur the outcrops of high-pressure (HP) meta-ophiolites, which were widely utilized from the Early Neolithic onwards for the manufacture of polished cutting-edged tools, which are represented by axes, adzes and chisels. Other raw materials, such as serpentinites, seem to have been preferred for the production of other types of artefacts, including stone rings used as bracelets. The analyses revealed that the prehistoric polished stone artefacts were made from uncommon lithologies such as Alpine eclogites, jades and other HP meta-ophiolites. These rocks were exploited from primary and secondary sources, mainly located in Piedmont, the Aosta Valley and Liguria. During the Neolithic these lithologies are the dominant raw material for the polished stone tools in Northern Italy and southeastern France. In the same period, in other European countries the same lithologies occur less frequently as axe or adze blades; in NW Europe they were frequently used for manufacturing long ceremonial axes, which have a typology that does not appear to belong to the Italian tradition.

Despite the fact that the Italian peninsula is rich in stone resources that can be used for polished implements, archaeometric analyses so far conducted on some of the most important prehistoric stone tool assemblages have demonstrated that high pressure (HP) meta-ophiolites were generally preferred. However, comparisons of archaeological data show that the greenstone circulation might be correlated with the exchange network of other raw materials, such as flint and, to some extent, obsidian (Williams Thorpe et al. 1979; Ammermann & Polglase 1997; Binder 1998; Barfield 2000; Blet et al. 2000). The most important flint outcrops generally exploited since the beginning of the Neolithic are located in the Gargano Promontory (Di Lernia & Galiberti 1993), in the JurassicCretaceous formations of the Lessini Hills, the Venetian Alps and the Marchigian Apennines (Cremaschi 1981; Barfield 1987, 1999; Benedetti et al. 1994, 1994-1995; Peresani 1994; Starnini 1997; Ferrari & Mazzieri 1998). Other common lithologies exploited in prehistory are

represented by other varieties of cherts, defined as radiolarites or jaspers (Del Soldato 1990), the prehistoric quarrying of which has been discovered in the Ligurian and Emilian Apennines (Maggi et al. 1994; Negrino 1998; Ghiretti et al. 2002; Campana & Maggi 2003; Negrino & Starnini 2003). Less important lithic raw material sources are known only in areas where micro-regional surveys and detailed studies have been carried out (Negrino 1999; Del Lucchese et al. 2003). Finally, the existence of four of the most important insular obsidian sources of the Mediterranean should be mentioned, namely those of Sardinia, Pantelleria, and the Pontine and Lipari Islands. Nevertheless, as mentioned above, the petroarchaeometric analyses so far conducted demonstrate that, especially in Northern Italy, only the HP meta-ophiolites were widely used and traded for the manufacture of polished stone, cutting-edged implements, especially axe and adze blades, since the beginning of the Early Neolithic (D'Amico 1998a, 2002a,b; D'Amico

From: MAGGETTI,M. & MESSIGA,B. (eds) 2006. Geomaterials in Cultural Heritage. Geological Society, London, Special Publications, 257, 257-272. 0305-8719/06/$15.00 © The Geological Society of London 2006.

258

C. D'AMICO & E. STARNINI

et al. 2002). In fact, it has been demonstrated by

experimental studies that these rocks, because of their mechanical resistance and hardness, are the best raw material for the production of woodworking implements. Their outcrops are located in the western Alpine Arc (Fig. 1) (Franchi 1900; D'Amico et al. 1987; Dal Piaz et al. 1993; Morten 1993; D'Amico 1998b). Secondary Oligocene deposits, containing greenstone cobbles and pebbles, are distributed over a wider area, which covers a great part of Piedmont and southwestern Lombardy. Secondary deposits containing pebbles of these rocks can be found in the alluvial cones along the Ligurian Alpine-Apennine fringes (Fig. 1).

The archaeological and archaeometric framework The time-span covered by this paper extends from the Early Neolithic, when the first polished tools were manufactured, to the Copper-Early Bronze Ages, during which the last polished

stone axes were produced before being replaced by metal tools (i.e. from the seventh to the beginning of the fourth millennium uncalibrated

BP). The state of research is uneven. Petroarchaeometric analyses and systematic studies of the assemblages have been carried out mainly in Northern Italy (Fig. 1), whereas very little information is available for the rest of the peninsula. Petrographic information was obtained by mean of different analytical approaches; these include simple stereomicroscopic observations and qualitative density determinations (100% of the samples), X-ray powder diffraction (XRPD) analyses (95%), thin-section (more than 50%), and scanning electron microscopy-energy-dispersive spectrometry (SEM-EDS) analyses (less than 5%) (D'Amico et al. 1991, 1995, 1997, 1999; Compagnoni et al. 1995; Chiari et al. 1996). Samples for thin-section, XRPD, chemical and S E M - E D S analyses were obtained from broken artefacts, using small flakelets or microcores, under the

Fig. 1. Map of Northern Italy with the location of the archaeological complexes and areas investigated during the archaeometric project. White numbers indicate more recent data. Black hatched areas indicate the HP meta-ophiolite primary outcrops; cross-hatched area represents the main distribution of secondary occurrences of these rocks among the Oligocene conglomerates.

PREHISTORIC

POLISHED

STONE ARTEFACTS

supervision of archaeologists; a number of X-ray diffraction (XRD) determinations were performed directly on the surface of the artefacts (Chiari et al. 1996). Recently, some researchers began the study of polished stone tools from the Neolithic sites of Sardinia (Bertorino et al. 2002), and Leighton published the finds from Southern Italy, in particular Sicily (Leighton 1989, 1992; Leighton & Dixon 1992a,b). Furthermore, a new small assemblage from a Neolithic site in Calabria is currently under study and slowly we hope to achieve a complete picture of the entire Italian territory. At present, the occurrence of HP metaophiolite polished tools in the various geographical areas seems to reflect the 'down-the-line' model of distribution, proposed by Renfrew (1975), slowly falling off as we move away from their primary sources. Their occurrence is documented in Tuscany and more occasionally in Latium, Campania, Calabria and Sicily. The few data so far available are summarized in Table 1. The data from Sardinia (Bertorino et al. 2002) indicate that, during the Neolithic, the imports of raw materials, among which are nephrites and glaucophane schist, were probably from the continent. However, at the present state of research, a provenance from Corsica for nephrite and glaucophane schists cannot be excluded (Bertorino et al. 2002). Local raw materials, such as phonolite, are also represented as artefacts, indicating the exploitation of outcrops probably located in the Pliocene volcanic complex of Montiferru, in central Sardinia. The problem of the provenance of nephrite artefacts and the presence of HP meta-ophiolites in museum collections of Southern Italy will be discussed with more detail below. Regarding the typology of the polished stone tools, we have to point out the absence of a codified type-list (Barfield 1996). This is also

IN ITALY

259

due to the intrinsic difficulty in assigning to a precise type instruments that often simply represent different stages of use, resharpening, recycling, and even exhausted and discarded tools. The most significant classes that occur in Italy are listed in Table 2. The assemblages of studied polished stone tools are rather numerous in Northern Italy. More than 20 sites of different age were analysed (Table 3), numbering more than 1000 implements. The location of these sites is shown in Figure l and the results concerning the sites will be discussed below. Archaeometric analyses were also conducted on a large sample (182 specimens) of artefacts collected from a workshop for the production of polished stone tools located near Rivanazzano (Pavia, NW Italy: Mannoni & Starnini 1994, 1996; Mannoni et al. 1996; D'Amico et al. 2004a). In our opinion this locality deserves a more detailed description because, at present, it is the only prehistoric polished stone tool workshop so far analysed in Northern Italy. The site lies on a terrace of the Staffora River, along the fringes of the northern piedmont of the Ligurian Apennines (Fig. 1, site 18). Here, more than 400 artefacts have been collected on the surface. They consist of hammer-stones, by-products, fragmentary wastes, flakes and rough-outs that result from the manufacture of adzes, axes and chisels. They demonstrate the existence of one important workshop for the production of polished, cutting-edged tools, which can be attributed to the Neolithic period. Petrographic investigations were carried out with the aim of understanding the pattern of exploitation of the stone resources and their possible provenance (D'Amico & Starnini 2001; D'Amico et al. 2003). The sample for the petrographic analyses was selected according to two methods. The first consists of a random sampling of 90 artefacts, the second of 92 samples collected according to the macroscopic

T a b l e 1. Distribution of polished stone tools made from HP meta-ophiolites and other lithologies in

Central-Southern Italy

Lithology

Tuscany

HP meta-ophiolites Nephrites

7 -

Serpentinites Metamorphic rocks Volcanic or Plutonic rocks Sedimentary rocks

5 5

Umbria

59 -

1 3(V) 3

Latium

South Italian sites in museum collections

4 1

18 2

7 28 (V) 14

5 26 130 (V + P) 81

Calabria

-

8 6 (P) 16

Siciiy

Sardinia

1

10 24

4 68 (V) 4

7 12 (V + P) 3

P, plutonic; V, volcanic lithologies. Data from D'Amico et al. (2004b), with addition of unpublished data from Latium and Calabria.

260

C. D'AMICO & E. STARNINI

Table 2. Typology of the tools from the Italian prehistoric sites

Tool type

Chronology and cultural context

Polished stone ring bracelets Small chisels with two opposite cutting edges Shoe-last adzes Cutting-edged tools such as axes and adzes Perforated hammers or axes Large ceremonial axes Hammerstones, often recycled cutting-edged tools

rock differences. The cumulative analysis shows a predominance of eclogites, followed, in order of importance, by glaucophane schists, jades and other HP meta-ophiolites. The occurrence of jades is surprisingly low, in comparison with the usual pattern observed for other Northern Italian sites (D'Amico & Starnini 2000, 2001), whereas that of glaucophane schists is, in contrast, rather high. This can be perhaps explained by the different character of this site, which is a workshop of primary production, where the finished tools are absent, in contrast to settlements,

Early Neolithic Middle Neolithic-Square Mouthed Pottery Culture Early-Middle Neolithic of NW, probably imports From Early Neolithic to Copper Age From Copper to Early Bronze Age Context often undefined; generic Neolithic period Neolithic

from which only finished and used tools have usually been analysed. All the lithotypes of this site can be considered of local provenance, collected as pebbles from the alluvial deposits, naturally enriched in these rocks, which result from the erosion of the Oligocene conglomerates (Fig. 1). It is thus possible to interpret this site as a workshop for the production of such tools, mainly because of the absence of other categories of finds, which are typical of the settlement areas. On the basis of the typology of the artefacts collected, the

Table 3. Polished stone tool assemblages from Northern Italy studied and published during the petro-archaeometric project (see also Fig. 1) Site Alba (CN-Piedmont) Brignano Frascata (AL-Piedmont) Arene Candide (SV-Liguria) Rivanazzano (PV-Lombardy) Vh6 (CR-Lombardy) Ostiano, Dugali Alti (CR-Lombardy) Ostiano, Casotte (CR-Lombardy) Volongo, La Pista (CR-Lombardy) Isorella (BS-Lombardy) Bogliaco (BS-Lombardy) Lonato, Fornasetta (BS-Lombardy) Lonato, Pr~ (BS-Lombardy) Lonato, Case Vecchie (BS-Lombardy) Monte Netto (BS-Lombardy) Remedello (BS-Lombardy) Cascina Ferramonda (BS-Lombardy) Malegno (BS-Lombardy) C~ dei Grii (BS-Lombardy) Casatico di Marcaria (MN-Lombardy) Bancole (MN-Lombardy) Porto Mantovano (MN-Lombardy) San Lazzaro di Savena (BO-Emilia) Gaione (PR-Emilia) Ponte Ghiara (PR-Emilia) Adige Valley, Trentino Valer (PN-Friuli) Sammardenchia (UD-Friuli)

Site type

Chronology

No. of artefacts

Open-air settlement Open-air settlement Cave site Open-air workshop Open-air settlement Open-air settlement Open-air settlement Surface find Open-air settlement Surface find Open-air settlement Open-air settlement Open-air settlement Open-air settlement Graveyard Open-air settlement Surface find Cave site Open-air settlement Open-air settlement Open-air settlement Open-air settlement Open-air settlement Open-air settlement Open-air settlement Open-air settlement Open-air settlement

Early and Middle Neolithic Early Neolithic Early and Middle Neolithic Neolithic Early Neolithic Early Neolithic Middle Neolithic Neolithic Early Neolithic Neolithic Neolithic Copper Age Early Bronze Age Neolithic Copper Age Early Neolithic Copper Age Middle Neolithic Middle Neolithic Middle Neolithic Middle Neolithic Copper Age Neolithic Middle Neolithic Neolithic Early Neolithic Early and Middle Neolithic

115 34 19 182 15 10 31 1 2 1 1 1 1 1 I 2 1 5 4 7 16 100 261 39 80 1 291

AL, Alessandria; BO, Bologna; BS, Brescia: CN, Cuneo; CR, Cremona; MN, Mantova; PN, Pordenone; PR, Parma; PV, Pavia; SV, Savona; UD, Udine.

PREHISTORIC POLISHED STONE ARTEFACTS IN ITALY workshop can be attributed to the Neolithic period. Most of the finds come from settlements, although from the Middle Neolithic onwards the practice of sometimes placing polished stone tools as grave goods in burials began. This ritual is known for the Square Mouthed Pottery Culture, in Northern Italy. It indicates not only the economic value of these objects, but also their symbolic significance. The use of polished stone axes or adzes as grave goods continued until the Copper Age, as documented from the finds of the Remedello cemetery (Fig. 1, no. 8). A general diachronic trend was observed, comparing the various assemblages: the number of polished stone tools is usually low during the Early Neolithic; it increases dramatically during the Middle Neolithic. This discussion has shown that at present we do not have a complete and clear picture for the Late Neolithic, as we lack a sufficient number of studied assemblages. Nevertheless, we are able to deduce that a dramatic change in the raw material exchange system took place at the onset of the Copper Age: the exploitation and procurement of raw materials from long distances dropped in favour of the use of local stone resources, even though they are not of optimal quality. This pattern has been observed not only for the polished stone tool production, but also for the chipped stone industry (D'Amico et al. 1998, 2001; Negrino & Starnini 2003). This phenomenon, however, is particularly evident in NE Italy, where eastern serpentinites replaced western HP meta-ophiolites for the production of the characteristic shaft-hole hammer axes of the Copper Age (D'Amico et at. 1996b; Montagnari Kokelj 2001). A similar pattern also occurs in the Appennine area near Bologna, where local basalts became very common as a raw material during the Copper Age (D'Amico et al. 2000b).

HP meta-ophiolites as a dominant raw material for cutting-edged tool production in Northern Italy The HP meta-ophiolites, employed for the manufacture of axes and other tools, consist of high-pressure-low-temperature (HP-LT) metabasites (eclogites with minor omphacite schists, N a - P x metabasalts, glaucophane schists, greenschists that have undergone retrograde metamorphism) and their differentiates (jades or Na-pyroxenites, including jadeitites,

261

omphacitites, Px-mixed jades), as well as ultramafic rocks (serpentinites), typical of the Alpine geology. As mentioned above, in Northern Italy, fine-grained H P - L T alpine eclogites, jades (Na-pyroxenites) and a few geologically related HP meta-ophiolitic lithologies (omphacite schists, glaucophanic rocks, retrograde-metamorphosed greenschists, serpentinites and other unusual lithologies), represent the main raw material exploited for the manufacture of Neolithic axe blades (usually 90% or more of all lithologies, at least 70% at each site). They are known as 'greenstones' or 'pietre verdi' in the archaeological terminology. More than 1000 axes, adzes and chisels, some ornaments and other less common greenstone tools, and many fragments, sampled from both important sites and museum collections, have been analysed. Optical surface observation, density, thin-section, XRPD, microprobe analyses and bulk chemistry have been employed, alone or in combination, for their petrographic identification, aimed at an archaeometric interpretation. A review of the results so far obtained has been published by D'Amico et al. (2004b), together with a discussion and reinterpretation based on the extensive petro-archaeometric literature. That study is a preliminary report of the data currently available for the HP metaophiolites of Europe. From a typological point of view, most of the Italian polished stone tools consist of axe or adze blades, used for cutting or working wood (Table 4): they are often damaged (fragmented) and/or worn-out. Unworn, complete, ritual or ceremonial axes are much more rare. These tools were mainly employed for woodworking and forest clearance from the Neolithic onwards all over Europe. Chisels and ring bracelets are less frequent, and other types of implements, among which are burnishers, polishers, rough-outs, reused instruments and pebbles, occur in varying percentages. Table 5 gives an updated, summary picture of the number and percentage of blade axe or adze blades, chisels, ornaments and other tools of Northern Italy, according to their lithology. Most of them are Neolithic, although a few belong to the Copper and Bronze Ages. The HP meta-ophiolites predominate, with percentages up to 90%. They are mainly represented by alpine eclogites and jades, although less frequent lithologies also occur. The relationship between lithology and typology (Table 4) shows the predominance of the HP meta-ophiolites for the production of axe blades and chisels (eclogites >jades>> other lithologies), whereas different lithologies

262

C. D'AMICO & E. STARNINI

Table 4. Typology v. lithology of the anah'sed Italian polished stone tools from the Neolithic to the Bronze Age Tool types

Axes Adzes Large, ceremonial axes Rough-outs Shafthole hammeraxes Chisels Hammers, burnishers, polishers Pebbles and unidentifiable Shoe-last chisels Small, miniature axes or adzes Ring bracelets

Eclogites Jades Omphacite Glaucophane Other HP Serpen- Chlorite- Paragonite Other (jadeite) schists metatinites schists schists lithologies schists ophiolites XXXX XXXX XX

XXX XXX XXX

X X X

(X) (X)

X

X

X

XX

(X)

(X) (X) (X)

X X X

X

XX X

X X

X X

X (X)

(X)

(X) XX

(X) X

X

X

X

X

X

X

X

X

XXX

XXX

X

(X)

X

X

XX

XX

XXX

X

The number of x s expresses the frequency, from higher to minimal and sporadic (X).

prevail for the o r n a m e n t s and o t h e r tools. T h i s is d u e to the p r e f e r e n t i a l use o f hard, tough, h e a v y m a t e r i a l s (eclogites, jades, o m p h a c i t e schists, etc.) for the m a n u f a c t u r e o f f u n c t i o n a l tools, and softer r o c k s (serpentinites, p a r a g o n i t e schists, etc.) for the p r o d u c t i o n o f rings and other ornaments.

The predominance of eclogites and jades and o t h e r H P m e t a - o p h i o l i t e s e x t e n d s f r o m northw e s t e r n to all o f N o r t h e r n Italy. T h i s p r e d o m i n a n c e has cultural, l i t h o - t e c h n o l o g i c a l r e a s o n s r e l a t e d to the h a r d n e s s , t o u g h n e s s and d e n s i t y o f the r a w material, and aesthetic r e a s o n s (green colour and translucency of a number of

Table 5. Distribution of the different lithologies according the tool t3"pes Lithogical groups

HP meta-ophiolites

All samples (n = 1018)

Axe or adze blades (n = 668)

No.

%

No.

%

No.

%

No.

%

No.

%

847

83.2

602

90. l

27

90.0

13

32.5

205

73.2

including Alpine eclogites Jades (Na-pyroxenites) Omphacite (jadeite) schists Glaucophane rocks Green schists Serpentinites Other lithologies t

415 240 40 26 23 103 171

40.8 23.6 3.9 2.5 2.3 10.1 16.8

including 327 183 28 15 18 31 66

48.9 27.4 4.2 2.3 2.7 4.6 9.9

Chisels (n = 30)

including 14 7 1 4 1

46.7 23.3 3.3 13.3 3.3 -

3

10.0

Ornamental objects (n = 40)

including 2

5.0 -

11 27

27.5 67.5

Other artefacts* (n = 280)

including 74 48 11 7 4 61 75

26.4 16.8 3.9 2.5 1.4 21.8 26.8

"Undeterminableobjects, hammers, pestles, burnishers, polishers, pebbles, reused fragments and flakes. Paragonite schists, chlorite schists, spotted slates, limestones for manufacturingbracelets. Nephrites, cinerites-tuffites, porphyritic volcanites or sub-volcanites, sandstones, siltites, silexites, gabbros, basalts, granites, etc. for axe or adze blades, chisels, pendants, burnishers and polishers, or as pebbles. n, total number.

35.6 29.4 52.6 54.4 35.9 48.7 63.9 36.7 45.4 53.2 60.4

58.3 20.0 36.1

34 19 182 39 261 36 30 44 47

96

24 80 291

Eclogites

115

No. o f a n a l y s e d samples

2.5 1.3

-

5.9 6.0 20.5 5.7 5.6 2.3 6.4

2.6

Omphacite schists

25.0 16.2 22.3

19.8

29.4 36.8 11.5 15.4 21.8 13.9 40.0 22.7 19.1

36.5

Jades

8.3 1.3 0.7

-

23.1 2.6 5.0 2.8 2.3 2.1

3.5

Glaucophane schist

HP meta-ophiolites

4.2 -

-

2.9 0.5 2.6 5.0 4.6 2.1

6.1

Other H P metaophiolites

4.2 32.5 :I: 7.0 §

6.3

20.6 . 5.3 1.1 20.5* 6.9* 5.6* i 3.4 11.4 8.5

7.0

Serpentinites

.

. 2.5 0.7

-

2.1

0.9 .

Nephrites

.

.

.

.

-

-

2.7

1.4

-

--

--

-

3.3

0.9

Paragonite shists

0.4 3.3

-

Chloriteschist

Other lithologies

25.0 27.8

13.5 t

11.8 5.3 3.3 2.6 6.5 8,3 3.3 1 1.4 6,3

7.0

Other/ local

Data sources: Alba, D'Amico et al. (2000a); Brignano Frascata, D'Amico & Starnini (1996); D'Amico et al. (2000c); Rivanazzano, D'Amico et al. (2003); Ponte Ghiara, Bernab6 Brea et al. (2000); Gaione, Bernab6 Brea et al. (1996), S. Lazzaro di Savena, D'Amico et al. (2000b); Vhr, Starnini et al. (2005); Ostiano, D'Amico (1995); Starnini et aI. (2005); Mantua and Brescia Provinces; Starnini et al. (2005); Province of Verona, Lunardi (2003); Fimon D'Amico & Lunardi (unpubl. data); Trentino, Sammardenchia, D'Amico et al. (1997). *Probably total or partial, Apennine provenance. ? Only preliminary examinations. Others are undefined green stones, probably some type of HP meta-ophiolite. ~.Many pebbles from Adige river, local/regional contribution. § At least partial provenance from easternmost Alps.

Alba (CN), Pigorini M u s e u m collection B r i g n a n o Frascata (AL) A r e n e C a n d i d e (SV) R i v a n a z z a n o (PV) Ponte Ghiara (PR) G a i o n e (PR) S. L a z z a r o di S a v e n a (BO) V h 6 (CR) Ostiano (CR) M a n t u a and Brescia P r o v i n c e s sites V e r o n a P r o v i n c e sites (Natural History. M u s e u m collections) F i m o n (VI) Trentino sites S a m m a r d e n c h i a (UD)

Sites/areas

T a b l e 6. D i s t r i b u t i o n o f l i t h o l o g i e s at the N o r t h I t a l i a n sites ( v a l u e s a r e p e r c e n t a g e s )

~Z

O

t~

© ~

('3

~ ©

264

C. D'AMICO & E. STARNINI

lithologies). These both most probably justify the almost complete absence of other lithologies, which are nevertheless common in other European countries (D'Amico et al. 2004b). Table 6 shows the basic data for the lithologies from each North Italian site or region so far analysed. The Alpine eclogites+jades predominate in general, and in each single assemblage, with varying percentages (eclogites from some 30 to >60% and jades from some 12 to 40%). Differences can be noted between a few assemblages in the relative numbers of eclogite and jade samples. For example, the Alba assemblage (D'Amico & Ghedini 1996; D'Amico et al. 2000a; for the location see Fig. 1) shows roughly the same number of jades and eclogites, whereas those of Rivanazzano and, less markedly, Sammardenchia (D'Amico et al. 1992, 1996a; D'Amico & Felice 1994; Pessina & D'Amico 1999; for the location see Fig. 1) and Gaione (Bernab6 Brea et al. 1996) show a higher predominance of eclogites over jades. Other differences concern the presence or absence or different percentages of single minor HP meta-ophiolite and non-meta-ophiolite lithologies. It is possible that they all have a meaning, which at present is unknown.

Nature and provenance of the HP meta-ophiolites employed for making tools As shown in Tables 5 and 6, eclogites and jades are the most representative rocks. It is well known that HP-meta-ophiolites have undergone subduction of the oceanic crust, HP metamorphism and metasomatism, and subsequent tectonic exhumation from depth. Similar H P - L T metaophiolites and related rocks crop out along the Western Italian Alpine watershed (and in part of Valais, Switzerland) as primary geological bodies within the Penninic Tectonic Units of the Western Alps (Compagnoni 1977; Messiga et al. 1993; Mottana 1993). This is a unique case in Europe and one of the few in the world, as far as the association of eclogites and jades is concerned (Compagnoni et al. 1995; D'Amico et al. 2002). Apart from the above-mentioned primary outcrops, the same rocks are also present in the Oligocene conglomerate secondary deposits of the NW Apennines, which consist of an old basin rich in blocks and pebbles of these and other lithologies, as well as boulders, cobbles and pebbles in the Quaternary moraine and alluvial deposits of Piedmont, Liguria and SW Lombardy, which are derived from both the primary outcrops and the Oligocene deposits. In all

these regions, the HP meta-ophiolites employed for the manufacture of the Neolithic stone tools, and in particular axes and chisels, represent at least 90% of the total assemblage. This demonstrates a widespread utilization of these rocks during the Neolithic. An almost identical situation is known from both Northern Italy and Southern France (Compagnoni & Ricq-deBouard 1993). The exceptions are represented by the northeasternmost Italian regions and the Alpine valleys (e.g. Friuli, Trentino), where the percentages drops to 60-70%. Here other rocks were exploited or imported for tool manufacture. West and north of Provence, south of the Apennines, and in the Danubian and Dalmatian regions, the use of HP meta-ophiolitic rocks drops gradually or abruptly (D'Amico et al. 2004b; D'Amico 2005) following a 'down-theline' model (Renfrew 1975; Barfield 1996). The HP meta-ophiolitic rocks were probably gathered mainly from secondary sources, i.e. the fiver valley fluvial and glacial pebblecobble-boulder deposits along the fringes of the Po Plain (Ricq-de-Bouard et al. 1990; Ricq-de-Bouard & Fedele 1993; D'Amico et al. 1995, 2004b; D'Amico & Starnini 2000, 2005). Recently the exploitation of highland zone primary outcrops of the Voltri Group, Monviso and Valais, has been proposed by Errera (2004) and P6trrquin et al. (2005), for the manufacture of beautiful ceremonial jade axes in France and other Western European countries. This idea is reasonable and convincing, as the production of this variety of long, thin axes from cobbles is undoubtedly difficult and improbable, but it remains to be verified. However, this source is considered improbable so far for most of the Italian tool typologies.

Petrographic characterization of the main iithologies Axe or adze blades

Axe or adze blades and chisels are woodworking implements (Fig. 2). They are mainly obtained from HP meta-ophiolites and rarely from other rocks. The lithologies are briefly described below (readers are referred to D'Amico et al. (2004b) for further details), with relative percentages shown in Tables 5 and 6. The chemical characterization of the HP meta-ophiolites is shown in Table 7. Eclogites (Figs 3 and 4) are commonly finegrained, heterogranular and variously sheared to mylonitic texture. Their components are Na-pyroxenes (omphacite _+ jadeite) + garnets (or derived chlorite) + some compatible or

PREHISTORIC POLISHED STONE ARTEFACTS IN ITALY

Fig. 2. Smallaxe blade from the Early Neolithicsite of Vh6 (Lombardy).

secondary minerals: rutile or ilmenite and/or titanite, always present and often relatively abundant; zoisite-epidote and paragonite, mostly within whitish pseudomorphs of plagioclase (or lawsonite: Compagnoni et al. 1995); common apatite, zircon and other accessory minerals; very sporadic quartz or chloritoid; rare secondary glaucophane, actinolite and albite. Fetich (dark green), Mg-rich (bright or medium green) and intermediate eclogites can be chemically distinguished (Table 7). They are typical Alpine eclogites, very different from the coarser-grained H P - H T eclogites from the Variscan basements of Central and Western Europe (or some areas in the Alps), which have rarely been used for the manufacture of stone axe blades in Europe (Hovorka et al. 2001). Omphacite (jadeite) schists are similar to the eclogites, without garnets or chlorite pseudomorphs after garnet. Zoisite-epidote, chlorite, paragonite, and sometimes albite may be important mineral phases. Jades (Na-pyroxenites) are lacking in garnet and usually contain 90% or more of Napyroxenes of varying composition, plus accessory and secondary mineral phases, among which paragonite is relatively common, as well as rutile or ilmenite or titanite; zircon is often abundant as an accessory mineral. Jades have been conventionally classified as jadeitites and Fe-jadeitites (jadeite or Fe-jadeite 90% or more), omphacitites (omphacite 90% or more) and Px-mixed jades (both jadeite and omphacite in various percentages). Both Fe-rich and Mgrich compositions, dark and light respectively, although of various shades, are chemically recognized (Table 7). Blastic ('saccharoidal') and variously sheared to mylonitic textures are

265

present (Figs 5 and 6). Blastic textures are more common in jadeitites than in other jades. Glaucophane schists (or felses) contain glaucophane (or crossite) as a dominant or very important phase, in various combinations with zoisite-epidote, chlorite, garnet, lawsonite, albite, ore minerals and often residual Napyroxenes. The greenschists, from which a few tools are made, are all eclogites that have undergone retrograde metamorphism; they are very fine grained and composed of albite, chlorite, epidote, actinolite, ore minerals, and relict omphacite and garnets. Serpentinites are usually characterized by the association antigorite + magnetite, + chlorites + residues of Ca-pyroxenes, amphiboles, and rarely olivine. The raw material provenance of the serpentinite polished tools may be from not only the NW Italian sources, but also the Central and Eastern Alps (D'Amico et al. 2004b), and, possibly, the Apennines (lizardite instead of antigorite). The definition of the different provenances is rather ambiguous and therefore attributions are often uncertain. Many other lithologies, which occur in small quantities, differ from the HP meta-ophiolites. They include nephrites, actinolite-hornblende schists, cinerites, andesite-dacites, porphyries, basalts, gabbros, granites, limestones, sandstones, silexites, vitric tufts, cherts and spotted slates. Many of them have been used to produce non-cutting-edged tools, such as burnishers, polishers and pestles, or occur as pebbles or unidentifiable tool fragments. Only a few have been used for axe blades or axe or adze rough-outs, or ornaments or ring-bracelet rough-outs. Among these latter lithologies only the following two merit mention (D'Amico et al. 2004b). Actinolite-hornblende schists, with Amph >>Ab, Ep, ore minerals, etc., in the shape of shoelast adzes of transalpine typology ('Hinkelstein' type) occur sporadically: in the Adige Valley (one specimen from the Middle Neolithic site of La Vela near Trento, dated to c. 5500 uncal, m,) and Friuli (from Sammardenchia, near Udine, probably imports from the Dalmatian coast: P. Biagi, pers. comm.) and testify, by their morphology and petrography, to occasional exchange with transalpine cultures (Barfield 1970). Nephrite (nearly 100% tremolite) has been employed in the production of only a few axes. It is characterized by a fine-grained to felty or schistose texture, occasionally diablastic. It is clearly different in mineralogy and texture from the nephrites described by Kalkowsky

266

C. D'AMICO & E. STARNINI

Table 7. Mean chemical composition (values are percentages) of selected HP meta-ophiolites Rock: n:

Fe-eclogites 32

Mg-eclogites 21

SiO2 TiO2 A1203

48.88 ± 4.29 2.45 ± 0.92 12.19 _ 1.55 15.05 ± 2.91 0.29 ± 0.11 4.68 ± 2.00 7.06 ± 2.29 7.41 ± 2.22 0.07 ± 0.08 0.82 ± 0.69 1.32 ± 1.10 14 ± 16 21 ± 20 356±469

49.81 _ 2.08 1.28 +_ 0.33 13.77 ± 1.90 8.70 _ 1.31 0.18 ± 0.08 10.20 ± 2.06 8.59 ± 1.10 5.65 ± 1.00 0.09 ± 0.08 0.14 ± 0.15 1.78 ± 0.85 237 + 108 113 ± 48 1 4 6 ± 121

Fe203tot MnO MgO CaO Na20 K20 P2Os LOI Cr Ni Zr Rock: n:

Jadeitites 16

Fe-jadeitites 15

Mixed jades 12

SiO2 TiO2 A1203

58.37___1.04 0.61 __+0.34 19.34 ± 1.05 2.61 ± 0.87 0.06 ± 0.02 2.74 ___ 1.09 2.79 ± 0.86 11.90 ___ 1.48 0.24 ± 0.50 0.22 ± 0.26 1.19 ± 0.67 15 ± 13 41 ± 32 735 ± 467

55.99 ___ 1.12 1.07 ± 0.26 15.69 ± 1.37 9.13 ± 1.88 0.16 ± 0.03 2.16 _ 1.16 3.19 ± 0.94 11.45 ± 0.70 0.07 ± 0.07 0.32 ± 0.21 0.83 ± 0.57 18 ± 14 37 ± 48 t045 _ 779

56.76 0.69 15.19 5.03 0.11 5.35 5.88 9.79 0.08 0.29 0.89 60 69 326

Fe203tot MnO MgO CaO Na20 K20 P205 LOI Cr Ni Zr

Intermediate eclogites 6 48.76 2.37 12.07 12.94 0.28 8.18 9.23 5.42 0.05 0.08 1.16 115 53 291

4- 1.37 + 1.09 + 1.41 ± 0.98 + 0.09 ± 0.44 ± 1.09 ± 0.76 ± 0.03 ± 0.10 +_ 1.02 ± 108 ± 23 ± 149

Fe-mixedjades 7

+ 1.12 + 0.39 + 1.63 ± 1.76 ± 0.03 ± 1.79 _ t.71 ± 1.23 ± 0.09 ± 0.85 ___ 0.62 _ 74 ± 64 ± 390

54.82 1.50 13.88 9.38 019 3.42 5.58 9.54 0.05 0.77 1.03 13 25 902

Omphacite schists 9

+ 2.06 + 0.66 ± 1.51 + 2.30 ± 0.04 ± 0.78 __+ 1.91 _ 0.74 ± 0.03 _ 0.69 _ 0.53 ± 8 ± 18 _ 548

48.38 1.24 15.27 8.23 0.15 10.08 8.37 5.27 0.12 0.12 2.82 266 143 115

Omphacitites 5 54.63 0.93 10.10 7.84 0.13 8.29 9.96 6.89 0.13 0.09 1.21 203 147 546

+ + ± ± + + ± ± ± _ ± ± + ±

0.53 0.58 1.83 0.96 0.03 1.62 1.40 0.95 0.14 0.18 0.60 267 174 703

± 1.44 ± 0.40 4- 0.50 ± 1.12 ± 0.02 + 2.53 ± 0.94 ± 1.08 ± 0.12 ± 0.08 ___0.87 ± 58 + 46 ±51

Fe-omphacitites 2 54.86 1.18 14.30 13.09 0.15 1.96 5.92 6.47 0.07 0.85 1.17 17 15 1706

X-ray fluorescencedeterminationsfor all chemicalvaluesexceptNa20 and MgO (atomicabsorption spectrometrydeterminations).Mean values (%), ppm, standard deviation (___)and numberof analysed samples. Samples selected from Tables 2 and 3 (see D'Amico et al. (2004b) for furtherdetails), n, total numberof samples; LOI, loss on ignition. (1906) for Liguria and cannot c o m e f r o m these outcrops. A p r o v e n a n c e f r o m Switzerland (Grisons) can be proposed. Here both nephrite outcrops and their local use for the production

o f Neolithic tools have been well d o c u m e n t e d ( M u t s c h l e c h n e r 1948; Giess 1994). Nephrite axes are m e n t i o n e d here because they are mistakenly considered, by some

Fig. 3. Petrographic thin section of very fine eclogite (sample from Sammardenchia, cross-polarized light (NX), 51 x).

Fig. 4. Petrographic thin section of sheared eclogite (sample from Sammardenchia, NX, 51 ×).

PREHISTORIC POLISHED STONE ARTEFACTS IN ITALY

267

Fig. 5. Petrographic thin section of jade with blastic texture (sample from Sammardenchia, NX, 51 × ).

Fig. 7. Ring bracelet from the Early Neolithic site of Vhb (Lombardy).

Fig. 6. Petrographic thin section of jade with mylonitic texture (sample from Sammardenchia, NX, 51 x ).

workers, to have the same origin as the eclogite and jade axes (Bertorino et al. 2002; P~tr~quin et al. 2002, 2005; Errera 2004). This suggestion does not seem to have any basis. From a geological point of view, nephrites are not high-pressure rocks as the jades and eclogites are. Therefore their provenance must be from different formations. Archaeologically speaking, nephrite axes or adzes are very few or absent in Northern Italy (Table 6).

potentially derived from many areas (Table 7) (D'Amico et al. 1995, 2004b; D'Amico & Starnini 2000), and were exploited for making ornaments or working implements other than cutting-edged tools (Table 5). Felty and pure chlorite-schists and felses are present in a reasonable quantity exclusively at Sammardenchia, in Friuli (D'Amico et al. 1997). Occasionally. unusual lithologies, among which are spotted slates and limestones, may occur. Hard-stone ring bracelets, such as jades, are very rarely represented (Alba in Piedmont). 'Others' are represented by shapeless finds, flakes from manufacture and maintenance of implements, hammers, pestles (Fig. 8),

O r n a m e n t s a n d o t h e r tools

Ornaments are mainly represented by ring bracelets (Fig. 7) and a few pendants, made from soft rocks, which may be classified as follows. Paragonite schists (about 100% paragonite) occur at many sites (Venturino Gambari 1996; D'Amico et al. 1997, 2000a) and most probably are from Piedmont HP metamorphic deposits, probably the 'Sesia-Lanzo' zone (Traversone 1996). Serpentinites are present in all the collections, usually in small quantities (Table 6). They are

Fig. 8. Recycled cutting edged-tools reused as pestles (from the Neolithic sites of Vhb and Ostiano, Lombardy).

268

C. D'AMICO & E. STARNINI

burnishers, polishers, pebbles and reused fragments, of all lithologies, as reported in Table 6. Rather frequent are HP meta-ophiolite, reused, worn-out axe blades; splintered and shapeless fragments of the same and/or different material; and occasional, locally collected lithologies, as mentioned by D'Amico et al. (2004b).

Discussion and conclusions To conclude, the data described here show a complicated and articulated picture. It is interesting to mention briefly the situation outside Italy. Thicker and shorter axe blades, obtained from the same lithologies, similar to those of the Italian tradition, occur in other areas of Europe together with ceremonial axes. These latter are often polished all over the surface and are not worn. This indicates that they were not used for working wood. The noticeable occurrence of jade and alpine eclogites, especially amongst the Western, and more rarely, Central European axe blades (mainly in form of symbolic or ceremonial axes), demonstrates the distribution of the HP meta-ophiolites from NW Italy to France, Germany, Benelux, Britain, etc. (Campbell Smith 1965; Wolley et al. 1979; Compagnoni & Ricq-de-Bouard 1993; D'Amico 1993, 2000, 2005; Ricq-deBouard 1996; Ricq-de-Bouard et al. 1996; D'Amico et al. 2004b), up to a distance of 1000-1500 km from their source. The presentday unsystematic and still incomplete petrographic knowledge of the raw materials, including, possibly, HP meta-ophiolitic rocks, of the polished stone tools in Europe, including the Italian peninsula, allows only a preliminary comparison of these areas with the situation described here for Northern Italy. The currently available picture of the widespread Neolithic greenstones diffusion and circulation, all over Europe, will need further research and more petrographic identifications, if we are to achieve a correct interpretation of the manufacture and variability of use of the same lithological material, from a generally functional use in Northern Italy and its closely related regions, to its very important exploitation for the production of 'symbols of social prestige' in N W Europe.

References AMMERMANN,A. & POLGLASE,C. 1997. Analyses and descriptions of the obsidian collections from Arene Candide. In: MAGGL R., STARNINI, E. & VOYTEK, B. (eds) Arene Candide: a functional and environmental assessment of the Holocene sequence

excavated

by

L.

Bernabb

Brea

(1940-50).

Memorie deil'Istituto Italiano di Paleontologia Umana, Nuova Serie, 5, 573-592. BARHELD, L.-H. 1970. La stazione neolitica de 'La Vela" presso Trento. Considerazioni sulle tombe a cista nel Trentino Alto Adige. Studi Trentini di Scienze Naturali, sezione B, 47(1), 35-55. BARF[ELD, L.-H. 1987. Recent work on sources of Italian flint, h~: SIEVEKING, G. DE G. & NEWCOMER, M.-H. (eds) The Human Uses of Flint and Chert. Proceedings of the 4th International Flint Symposium, Cambridge, 231-239.

BARFIELD, L.-H. 1996. Le asce in pietra levigata nel Neolitico d'Europa e d'Italia. In: VENTURINO GAMBARI, M. (ed.) Le vie della pietra verde. L'industria litica levigata nella preistoria dell'Italia settentrionale. Omega, Torino, 57-66.

BARFIELD. L.-H. 1999. Neolithic and Copper Age flint exploitation in Northern Italy. Universitdtsforschungen zur priihistorishen Archiiologie, 55,

245-252. BARFIELD, L.-H. 2000. Commercio e scambio nel Neolitico dell'Italia settentrionale. In: PESSINA, A. & Musclo. G. (eds) La Neoliti~azione tra Oriente e Occidente, Atti del Convegno, Udine, 2 3 - 2 4 aprile 1999, Museo Friulano di Storia Naturale, Udine, 55-66. BENEDETTI, R., CAMPANA, R., D'AMtco, C. & NANNETTI, M. C. 1994. Petroarchaeometry of Epigravettian and Mesolithic flints in the Val Cismbn-Lagorai area (NE Italy). The flint supply question. Preistoria Alpina, 28(1), 33-49. BENEDETTI, M. R., D'AMICO, C. & NANNETTI, M. C. 1994-1995. Studio petroarcheometrico preliminare di selci neolitiche alpino-padane a scopo di confronto tra siti e ricerca di provenienza. Ani della Societh per la Preistoria e Protostoria della regione Friuli- Venezia Giulia, 9, 171 - 179. BERNABO BREA, M., D'AMICO, C., GHEDINI, M., GmRETTI, A. & OCCHL S. 1996. Gaione, loc. Case Catena. h~: VENTURINO GAMBAR1, M. (ed.) Le vie della pietra verde. L 'industria litica levigata nella preistoria dell'Italia settentrionale. Omega, Torino, 122-136. BERNABO BREA, M., D'AMICO, C., GHED1NI, M. & MAZZ1ERI, P. 2000. La pietra verde di Ponte Ghiara, Fidenza (PR). Mineralogica et Petrographica Acta, 43, 233-243. BERTORINO, G., FRANCESCHELLI, M., LUGLII~, C., MARCHI, M. & COLUMBU, S. 2002. Petrographic characterization of polished stone axes from Neolithic Sardinia. Archaeological implications. Periodico di Mineralogia, 71(special issue), 87-100. BINDER, D. 1998. Silex blond et complexit6 des assemblages lithiques dans le Nrolithique liguroprovenqal. Rencontres mdridionales de Prghistoire r~cente, Deuxibme session, Arles, 1996, APDCA, Antibes, 11 i - 128. BLET, M., BINDER, D. & GRATUZE, B. 2000. Essais de caractrrisation des silex Brdouliens provenqaux par analyse chimique 616mentaire. Revue d'Archgomgtrie, 24, 149-167.

PREHISTORIC POLISHED STONE ARTEFACTS IN ITALY CAMPANA, N. & MAGGI, R. (eds) 2003. Archeologia in Valle Lagorara. Diecimila anni di storia intorno a una cava di diaspro. Origines, Studi e Materiali pubblicati a cura dell'Istituto Italiano di Preistoria e Protostoria, Firenze. CAMPBELL SMITH, W. 1965. The distribution of jade axes in Europe with a supplement of the catalogue of those from the British Isles. Proceedings of the Prehistoric Society, New Series, 31, 25-33. CHIARI, G., COMPAGNONI, R., GIUSTETTO, R. & RICQ-DE-BOUARD, M. 1996. Metodi archeometrici per lo studio dei manufatti in pietra levigata. In: VENTURINO GAMBARI, M. (ed.) Le vie della pietra verde. L'industria litica levigata nella preistoria dell'Italia settentrionale. Omega, Torino, 35-53. COMPAGNONI, R. 1977. The Sesia-Lanzo Zone: high pressure-low temperature metamorphism in the Austro-alpine continental margin. Rendiconti della Societd Italiana di Mineralogia e Petrologia, 33(1), 335-374. COMPAGNONI, R. & RICQ-de-BOUARD, M. 1993. French neolithic stone axes in 'jade': petrological variability and origin. Terra Nova Abstracts, 5(Supplement 4), 5-6. COMPAGNONI,R., R1CQ-de-BOUARD,M., GIUSTETTO,R. & COLOMBO, F. 1995. Eclogite and Na-pyroxenite stone axes of southwestern Europe: a preliminary petrologic survey. Bollettino del Museo Regionale di Scienze Naturali, 13(Supplement 2), 329-359. CREMASCHI, M. 1981. The source of the flint artefacts for the Central Po Plain and Apennine sites, between the 7th and the 2nd millennium he. Staringia, 6, 139-142. DAL PIAZ, G. V., Gosso, G., PENNACCHIONI, G. & SPALLA, M. I. 1993. Geology of eclogites and related rocks in the Alps. In: MORTEN, L. (ed.) Italian Eclogites and Related Rocks. Accademia Nazionale delle Scienze, Scritti e documenti, 13, 17-58. D'AMIcO, C. 1993. Eclogites and jades as prehistoric implements in Europe. Terra Nova Abstracts, 5(Supplement 4), 7-8. D'AMIcO, C. 1995. Petroarcheometria dei reperti in pietra levigata. In: BIAGI, P. (ed.) L'insediamento neolitico di Ostiano-Dugali Alti (Cremona) nel suo contesto ambientale ed economico. Monografie di Natura Bresciana, 22, 92-96. D'AMIco, C. 1998a. La circolazione delle materie prime nel Neolitico: le pietre verdi. In: PESSINA A. & MuscIo, G. (eds) Setternila anni fa. I1 primo pane, Comune di Udine and Museo Friulano di Storia Naturale, 177-183. D'AMIcO, C. 1998b. Introduzione alia provenienza geologica dei materiali culturali. In: Atti della 4 a Giornata Le Scienze della Terra e l'Archeometria, February 1998. CUEN, Napoli, 133-137. D'AMIcO, C. 2000. La pietra levigata neolitica in Italia settentrionale e in Europa. Litologia, produzione e circolazione. In: PESSINA, A. & MuscIo, G. (eds) La Neolitizzazione tra Oriente e Occidente, Atti del Convegno, Udine, 23-24 aprile 1999. Museo Friulano di Storia Naturale, Udine, 67-80.

269

D'AM~cO, C. 2002a. La pietra levigata tra Neolitico e Bronzo Hell'area alpino-padana. Atti della XXXIII Riunione Scientifica dell'Istituto ltaliano di Preistoria e Protostoria (Trento, 1997), 417-431. D'AMtCO, C. 2002b. Polished stone between the Neolithic and Bronze Ages in Northern Italy. Preistoria Alpina, 34, 297-305. D'AMICO, C. 2005. Neolithic 'greenstone' axe blades from Northwestern Italy across Europe: a first petrographic comparison. Archaeomet~, 47(2), 235-252. D'AMICO, C. & FELICE, G. 1994. Petroarcheometria di reperti neolitici di Sammardenchia (Udine). Notizie preliminari. Atti della XXIX Riunione Scientifica dell'Istituto Italiano di Preistoria e Protostoria. Firenze, 159-165. D'AMICO, C. & GHEDINI, M. 1996. La pietra levigata della collezione 'Traverso' di Alba, nel Museo etnografico 'L. Pigorini' di Roma. Museologia Scientifica, 13(Supplement 1-2), 293-312. D'AMICO, C. & STARNINI, E. 1996. Brignano Frascata (Scavi Tin6 1984). In: VENTURtNO GAMBARI, M. (ed.) Le vie della pietra verde, L'industria litica levigata nella preistoria dell'Italia settentrionale. Omega, Torino, 74-79. D'AMIco, C. & STARNINI, E. 2000. Eclogites, jades and other HP metaophiolites of the Neolithic polished stone tools from northern Italy. Krystalinikum, 26, 9-20. D'AMICO, C. & STARNINI,E. 2001. The polished stone axe workshop of Rivanazzano PV-Northern Italy): analyses of the lithological diversities in comparison with other Italian Neolithic sites. Slovak Geological Magazine, 7(4), 420. D'AMIco, C. & STARNINI, E. 2006. Approvvigionamento e circolazione di manufatti neolifici in eclogite, giada e altre litologie connesse in Italia settentrionale: lo stato della ricerca. In: Archeologie, Studi in Onore di Tiziano Mannoni, Edipuglia (in press). D'AM~cO, C., INNOCENTI, F. & SASSI, F. P. 1987. Magmatismo e metamorfismo. (Appendice 6). UTET, Torino. D'AMICO, C., BARGOSSI, G. M., FELICE, G. & MAZZEO, M. 1991. Giade ed eclogiti in pietra levigata. Studio petroarcheometrico. Mineralogica et Petrographica Acta, 34, 254-283. D'AMICO, C., FEL1CE, G. & GHEDINI, M., 1992. Lithic supplies in the Early Neolithic to Sammardenchia (Friuli), Northern Italy. Science and Technology for Cultural Heritage, l, 159-176. D'AMICO, C., CAMPANA,R., FELICE, G. • GHEDINI,M. 1995. Eclogites and Jades as prehistoric implements in Europe. A case of petrology applied to cultural heritage. European Journal of Mineralogy, 7, 29-41. D'AMICO, C., FERRARI, A., GHEDINI, M. & PESSINA, A. 1996a. Sammardenchia (Pozzuolo del Friuli). In: VENTURINO GAMBARI, M. (ed.) Le vie della pietra verde. L'industria litica levigata nella preistoria dell'Italia settentrionate, Omega, Torino, 84-91. D'AMtcO, C., GHEDINI, M., MICHELI, R. & MONTAGNAR1 KOKELJ, E. 1996b. Le asce rotate del Friuli-Venezia Giulia. In: VENTURINO

270

C. D'AMICO & E. STARNINI

GAMBARI, M. (ed.) Le vie della pietra verde. L'industria litica levigata nella preistoria dell 'Italia senentrionale. Omega, Torino, 229-238. D'AMIcO, C., FELICE, G., GASPAROTTO,G., GHEDINI, M., NANNETTI, M. C. 8£ TRENTINI, P. 1997. La pietra levigata neolitica di Sammardenchia (Friuli). Catalogo petrografico. Mineralogica et Petrographica Acta, 40, 385-426. D'AMIco, C., FELICE, G. & GHEDINI, M. 1998. Neolithic-to-Bronze polished stone in Northern Italy. Proceedings of the XIII Congress UISSP, Forli, 1996, Volume 3, Section 9, 389-399. D'AMIcO, C., NANNETTI, M. C., TRENTINI, P., De SALVO, F., FELICE, G. & GHEDIN1, M. 1999. Informazioni petrologiche dall'archeometria. Eclogiti e giade del Neolitico Italiano. Atti 6" Giornata 'Le Scienze della Terra e l'Archeometria ', Este, 151 - 158. D'AMICO, C., GHEDIN1, M., NANNETTI, M. C. & TRENTINI, P. 2000a. La pietra levigata neolitica di Alba (CN). Catalogo petrografico e interpretazione archeometrica. Mineralogica et Petrographica Acta, 43, 176-206. D'AMICO, C., LENZI,F., MARGUTTI,S. & NENZIONI,G. 2000b. Temoignage de la pierre polie du chalcolithique a l'est de Bologna: analyses quantitatives, qualitatives et typologiques. 2nd International Congress 'Science and Technology for the Safeguard of Cultural Heritage in the Mediterranean BasbF. 5 - 9 J u l y 1999, CNRS-CNR, Paris, 271- 278. D'AMICO, C., STARNINI, E. & VOYTEK, B. 2000c. L'industria litica di Brignano Frascata (AL): dati paleoeconomici di un insediamento del Neolitico antico padano attraverso l'analisi tipologica funzionale e lo studio della provenienza delle materie prime. Preistoria Alpina, 31, 91-124. D'AMICO, C., GASPAROTTO, G., GHEDINI, M. & SABETTA, T. 2001. Serpentiniti e metaultramafiti ad anfiboli e cloriti in asce-martello eneolitiche del NE Italiano. Geohalia 3 ~ Forum, sessione 21, Chieti, 5 - 8 settembre 2001, Riassunti, 631-632. D'AMIcO, C., BERNABO BREA, M., BIAGI, P., PEDROTTI, A. L., PESSINA, A. & STARNINI, E. 2002. Archaeometrical analyses of polished stone tools from the Neolithic to the Bronze Age in Northern Italy. In: JEREM, E. & BIR0, T. (eds) Proceedings of the 31st International Symposium of Archaeometry 98, Budapest, 1998. British Archaeological Report International Series, 104301), 691- 696. D'AMIco, C., MINALE, M., STARNINI,E. ~ TRENTINI, P. 2003. L'officina di produzione di asce in pietra levigata di Rivanazzano (PV). Dati archeometrici e catena operativa. Atti della XXXV Riunione Scientifica IIPP, Le comunitb della Preistoria italiana. Studi e ricerche sul Neolitico e le Eth dei metalli, Firenze, 981-986. D'AMICO, C., SIMONE ZOPFI, L. & STARNINI, E. 2004a. Rivanazzano, loc. La Cascinetta. hz: VENTURINO GAMBARI, M. (ed.)Alia conquista dell'Appennino. Le prime comunith delle valli Curone, Grue e Ossola. Omega Edizioni, Torino, 203 -205.

D'AMICO, C., STARNINI, E., GASPAROTTO, G. & GHEDINI, M. 2004b. Eclogites, jades and other HP-metaophiolites employed for prehistoric polished stone implements in Italy and Europe. Periodico di Mineralogia, 73(Special Issue 3), 17-42. DEL LUCCHESE, A., MARTINI, S., NEGRINO, F. & OrTOMANO, C. 2003. 'I Ciotti' (Mortola Superiore, Ventimiglia, Imperia). Una localit~ di approvvigionamento della materia prima per la scheggiatura durante il Paleolitico. Bullettino di Paletnologia ltaliana, 91-92, 1-26. DEL SOLOATO, M. 1990. Analisi petrografiche deli'industria litica scheggiata, bz: MAGGI, R. (ed.) Archeologia dell'Appennino Ligure. Collez. Di Monografie preistoriche ed Archeologiche, VIII, 209-218. D! LERNIA, S. & GAL1BERTI, A. 1993. Archeologia mineraria della selce nella Preistoria. Definizioni, potenzialit~ e prospettive di ricerca. Quaderni del Dipartimento di Archeologia e Storia delle Arti, sez. Archeologica, Universith di Siena, 36, 1-83. ERRERA, M. 2004. D6couverte du premier gisement du jade-jad6ite dans les Alpes (6t6 2004). Implications concernant plusieurs lames de hache n6olitique trouv6es en Belgique et dans les r6gions limitrophes. Notae Prehistoricae, 24, 191-202. FERRARI, S. • MAZZIERI, P. 1998. Fonti e processi di scambio di rocce silicee scheggiabili. In: PESS1NA, A. & MuscIo, G. (eds) Settemila annifa, llprimo pane, Comune di Udine and Museo Friulano di Storia Naturale, 165-169. FRANCHI, S. 1900. Sopra alcuni giacimenti di rocce giadeitiche helle Alpi occidentali e Appennino ligure. Bollettino della Reale Commissione Geologica Italiana, 31, 119-158. GHIRETTI, A., NEGRINO, F. & TOZZI, C. 2002. Estrazione del diaspro e produzione di strumenti a ritocco bifacciale in localit/t Ronco del Gatto (Bardi, Parma): modificazioni economiche e tecnologiche tra la fine del Neolitico e l'eth del Rame nell' Appennino ligure-emiliano. In: FERRARI, A. & VISENTINI, P. (eds) 1l declino del mondo neolitico. Ricerche in Italia centrosettentrionale fra aspetti peninsulari, occidentali e nord alpini. Atti del Convegno, Pordenone, 403 -408. GIESS, H. 1994. Jade in Switzerland. The Bulletin of Friends of Jade, 8, 1-46. HOVORKA, D., SPISIAK, J., MERES, S., llla~ova, L. & Dc Bl KOVA, K. 200 I. Petrological and geochemical evidences of the eclogite hammer-axe from Nitriansky Hradok site (Neolithic, Lengyel culture, Slovakia). Slovak Geological Magazine, 7, 355-363. KALKOWSKY, E. 1906. Geologie de Nephrites im siidlichen Ligurien. Zeitschrift der Deutshen. Geologischen Gesellschaft, 58, 307-378. LEIGHTON, R. 1989. Ground stone tools from Serra Orlando (Morgantina) and stone axes studies in Sicily and Southern Italy. Proceedings of the Prehistotqc Society', 55, 135 - 159. LEIGHTON, R. 1992. Stone axes and exchange in south Italian prehistory: new evidence from old collections. Accordia Research Papers, 3, 7-36.

PREHISTORIC POLISHED STONE ARTEFACTS IN ITALY LEIGHTON, R. & DIXON, J. E. 1992a. Jade and greenstone in the prehistory of Sicily and Southern Italy. Oxford Journal of Archaeology, 11(2), 179-200. LEIGHTON, R. & DIXON, J. E. 1992b. Alcune considerazioni sulle asce levigate in Italia Meridionale ed in Sicilia. In: Papers of the Fourth Conference of Italian Archaeology, Volume 3, New Developments, Part 1, London, 9-28. LUNARDI, A. 2003. Le lame d' ascia in pietra verde del territorio veronese dal Neolitico all'et~ del Bronzo: petrografia, tipologia e funzione. Atti della Societb per la Preistoria e Protostostoria della regione Friuli- Venezia Giulia, 13, 57-110. MAGGI, R., CAMPANA, N., NEGRINO, F. & OTTOMANO, C. 1994. The quarrying and workshop site of Valle Lagorara (Liguria, Italy). The Accordia Research Papers, 5, 73-96. MANNONI, T. & STARNINI,E. 1994. I1 contributo delle analisi petrografiche nello studio dell'officina litica di Rivanazzano (PV). In: D'AMICO, C. & CAMPANA, R. (eds) Le scienze della Terra e l'Archeometria. Giornata di studio, Bologna, 21 Aprile 1994, 21. MANNONI, T. & STARNINI, E. 1996. Rivanazzano-Caratteristiche tecniche della lavorazione--La petrografia. In: VENTURINO GAMBARI, M. (ed.) Le vie della pietra verde. L'industria litica levigata nella preistoria dell'Italia settentrionale. Omega, Torino, 119-120. MANNONI, T., STARNINI,E. & SIMONE ZOPFI, L. 1996. Rivanazzano. In: VENTURINO GAMBARI, M. (ed.) Le vie della pietra verde, L'industria litica levigata nella preistoria dell'Italia settentrionale. Omega, Torino, 119-122. MESSIGA, B., SCAMBELLURI,M. & TR1BUZIO, R. 1993. Petrology of eclogitized mafic ophiolites from Western Alps. In: MORTEN, L. (ed.) Italian Eclogites and Related Rocks. Accademia Nazionale delle Scienze, Roma, Scritti e Documenti, XIII, 79-96. MONTAGNARI KOKELJ, E., 2001. Pietra verde, Neolitico e Post-Neolitico, Carso e Friuli (Italia Nord-orientale): lo stato della questione. A t t i e Memorie della Commissione Grotte 'E. Boegan', 38, 71-86. MORTEN, L. (ed.) 1993. Italian Eclogites and Related Rocks. Accademia Nazionale delle Scienze, Roma, Scritti e Documenti, XIII. MOTTANA, A. 1993. Italian eclogite-facies minerals: phase relationships, chemical compositions and crystal structures. In: MORTEN, L. (ed.) Italian Eclogites and Related Rocks. Accademia Nazionale delle Scienze, Roma, Scritti e documenti, XIII, 121 - 141. MUTSCHLECHNER, G. 1948. Vorkommen der Steinzeitlichen Werkstoffe Nephrit und Jadeit in unsere Alpen. Der Schlern, 107-109. NEGRINO, F. 1998. The quarrying and workshop site of Valle Lagorara (Maissana, La Spezia). In: Lithic technology: from raw material procurement to tool production. Workshop 12, Workshops-Volume II, Proceedings of XIII International Congress of Prehistoric Sciences (Forli, 1996), Number 6, 831-836.

271

NEGRINO, F. 1999. Palaeolithic evidences of quarrying activity at 'I Ciotti' (Mortola Superiore, Ventimiglia, Imperia, Italy). In: Abstracts, VIII International Flint Symposium, Bochum, 55-56. NEGRINO, F. & STARN1NI, E. 2003. Patterns of lithic raw material exploitation in Liguria from the Palaeolithic to the Copper Age. In: Actes de la Table ronde internationale: Les Matibres Premikres Lithiques en Prdhistoire. Inventaire, caractdrisation et circulation des matikres premikres lithiques durant la pr~histoire (paldolithique, m~solithique et n~olithique). Aspects rdgionaux et inter-r~gionaux. Aurillac (France), 20-22 Juin 2002. Prehistoire du Sud-ouest, 2003, Supplrment, 5, 235-243. PERESANI, M. 1994. Flint exploitation at epigravettian and mesolithic sites on the Asiago Plateau (Veneto Prealps). In: BIAGI, P. (ed.) Highland Zone Exploitation in Southern Europe. Monografie di Natura Bresciana, 20, 221-243. PESSINA, A. & D'AMICO, C. 1999. L'industria in pietra levigata del sito neolitico di Sammardenchia (Pozzuolo del Friuli, Udine). Aspetti archeologici e petroarcheometrici. In: FERRARI, A. & PESSINA, A. (eds) Sammardenchia--Cueis. Museo Friulano di Storia Naturale pubblicazione, 41, 23-92. PI~TRI~QUIN,P., CASSEN, S., CROUTSCH, C. & ERRERA, M. 2002. La valorisation sociale des longues haches dans l'Europe Nrolithique. In: GIULAINE, J. (ed.) Matdriaux, production, circulations du Ngolithique ~ l'Age du Bronze. Errance, Paris, 67 -98. PI~TRI~QUIN,P., PI~TRI~QU1N,A. M., ERRERA, M., et al. 2006. Voltri, Viso et Valais. A l'origine des grandes haches plies alpines au V ° millrnaire en Europe occidentale. In: Materie prime e scambi nella Preistoria Italiana. Atti della XXXIX Riunione Scientifica dell'Istituto Italiano di Preistoria e Protostoria, Firenze, 25-27 november 2004 (in press). RENFREW, C. 1975. Trade as action at a distance. In: SABLOFF, J. & LAMBERG-KARLOVSKY, C. C. (eds) Ancient Civilization and Trade. Omega Edizioni, Albuquerque, NM, 1-59. RICQ-DE-BOUARD, M. 1996. Petrographie et socidtds neolithiques en France mdditerran~enne. L'outillage en pierre polie. Centre de Recherches Archrologiques, Monographie, 16. RICQ-DE-BOUARD, M. & FEDELE,G. F. 1993. Neolithic rock resources across the Western Alps: circulation data and models. Geoarcheology, 8(1), 1-22. RICQ-DE-BOUARD, M., COMPAGNONI,R., DESMONS,J. & FEDELE, F. 1990. Les roches alpines dans l'outillage poli nrolithique de la France Mrditerranrenne. Classification, origine, circulation. Gallia Prdhistoire, 32, 125-149. RICQ-de-BOUARD, M., COMPAGNON1, R., COLOMBO, F. & DEISS, W. 1996. Le materiel des Musres de Provence et du Dauphin~. Contribue 7t la connaissance des courants de circulation transalpins, au Neolithique. Museologia Scientifica, 13(Supplement 1- 2), 313- 328. STARNINI, E. 1997. Raw material procurement and use strategies. In: BARONI, C. & BIAGI, P. (eds)

272

C. D'AMICO & E. STARNINI

Excavations at the high altitude mesolithic site of Laghetti del Crestoso (Bovegno, Brescia--Northern Italy). Ateneo di Brescia, Brescia, 21-24. STARNINi, E., D'AMICO, C., BIAGI, P., GHEDIN1,M. & PITTI, G. 2005. Strumenti in pietra levigata dalla Lombardia orientale: aspetti archeometrici e culturali. Bullettino di Paletnologia Italiana, 95, 1-62. TRAVERSONE, B. 1996. Oggetti ornamentali. In: VENTURINO GAMBARI, M. (ed.) Le vie della pietra verde. L'industria litica levigata nella preistoria dell'Italia settentrionale. Omega, Torino, 197-202.

VENTURINO GAMBARI, M. (ed.) 1996. Le vie della pietra verde. L'industria litica levigata nella preistoria dell'Italia settentrionale. Omega, Torino. WtLLIAMS THORPE, O., WARREN, S. E. & BARFIELD, L.-H. 1979. The sources and distribution of archaeological obsidian in Northern Italy. Preistoria Alpina, 15, 73-92. WOLLEY, A. R., BISHOP, A. C., HARRISON, R. J. KINNES, I. A. 1979. European Neolithic jade implements: a preliminary mineralogical and typological study. Stone Axes Studies, 23, 90-96.

The stone of the inscribed Etruscan stelae from the Valdelsa area (Siena, Italy) A. G A N D I N l, E. C A P E Z Z U O L I 1 & A. C I A C C I 2

lDipartimento di Scienze della Terra, Universitg~ di Siena, Via Laterina 8, 53100 Siena, Italy (e-mail: gandin @unisi, it) 2Dipartimento di Archeologia e Storia delle Arti, Universitgt di Siena, Via Roma 56, 53100 Siena, Italy Abstract: The results of an archaeometric study on five inscribed Etruscan funerary stelae,

produced during the second half of the sixth century ac, in the Valdelsa area (NW of Siena, Italy), are reported. The style of the inscriptions suggests the presence of a local culture and a common language. The stone used to make the stelae is a laminated limestone consisting of the alternation of more or less compact bands whose petrographic features can be compared to those of travertines (continental carbonates deposited from hot springs). The fabric of the stone of these stelae implies provenance from a thermal deposit, in the Valdelsa area, where calcareous tufa dominates; however, hot water deposits are probably to be found in the Gracciano Val d'Elsa area, not far from the localities where the stelae were found.

Five inscribed funerary stelae of Etruscan age have been found in the Valdelsa valley, NW of Siena in southern Tuscany. They were made from a concretionary limestone referred to as 'travertine', a term that, in Italian, encompasses a wide variety of continental carbonates deposited by lacustrine, palustrine or fluvial waters, or by hot or cold spring waters (i.e. a very diverse range of physico-chemical, biological and climatic conditions). Historically, 'travertine' has been used in central Italy for the production of stone objects and as a building stone because of its good stratification, availability on the surface and ease of quarrying. Recent studies have shown that these concretionary carbonates (Pentecost 1995; Ford & Pedley 1996) can be classified on the basis of their depositional environment. Thus, laminated accretionary deposits related to hydrothermal waters are called travertine. Travertines are relatively unaffected by climatic factors. They are characterized mainly by macrocrystalline facies, consisting of large ray-crystals, associated with coated bubble travertine, paperthin rafts and shrub travertine (Guo & Riding 1998). Travertine exhibits regular plane parallel, centimetre-scale lamination, high growth rates and low organic content, and is slightly enriched in 13C (813C from - 2 to 10%0 and total dissolved inorganic carbon (TDIC; 1 5 - 6 0 mM 1-1), magnesium, strontium and sulphur (Pentecost 1995; Ford & Pedley 1996).

The term calcareous tufa has been coined in the recent English language literature (Viles & Goudie 1990; Pedley 1990) to indicate nonthermal continental carbonates (i.e. those derived from fluvial or spring waters in balance with subaerial surface temperatures). They form as encrustations of microcrystalline calcite on supports consisting mainly of higher plants (bryophytes, reeds, etc.), usually at a low deposition rate. They normally have irregular or massive stratification, are affected by climatic conditions, and have low contents of TDIC ( 1- 7 -(10) mM 1-1), strontium and magnesium, with strong depletion in 13C (~13C from - 2 to - 12%0 (Pentecost 1995; Ford & Pedley 1996). Although the distinctive features of the various types of carbonates are still under discussion, some workers believe that a reliable classification can be made on the basis of the petrographical and geochemical characters (Pedley 1990; Pentecost 1995; Ford & Pedley 1996; Capezzuoli & Gandin 2004). There are extensive outcrops of travertine and calcareous tufa in southern Tuscany. This paper reports a study of the lithological and genetic characteristics of the stone from which the five stelae were cut, aimed at establishing the stone's genesis and possible provenance. The latter is important information for the archaeology of the production of that period, as it permits the reconstruction of historical human-environment relationships and of the

From: MAGGETTI,M. & MESSIGA,B. (eds) 2006. Geomaterials in Cultural Heritage. Geological Society, London, Special Publications, 257, 273-282. 0305-8719/06/$15.00 © The Geological Society of London 2006.

274

A. GAND[N ET AL.

commercial links between diverse areas in Etruria (Mangani 1992). However, the identification of the origin of stone used in antiquity is often a complex archaeometric problem, particularly for continental limestones, because of the great variability in the vertical and lateral distribution of the limestone facies that have been utilized. Here we discuss the possible location of the quarry on the basis of our study of the microfacies of the stone and of the characteristics of travertine and calcareous tufa present in the area surrounding the Valdelsa valley (Fig. 1).

The Etruscan stelae The studied stelae are housed in various museums in their discovery area, as follows. Stele 184531: Museum of Santa Maria della Scala (SMS), Siena, deriving from Campassini (Monteriggioni); 120cm high, 60cm wide, about 15-22 cm thick, 412 kg weight (Becatti 1933) (Fig. 2a). Stele 13918: Museum of SMS, Siena, deriving from Toiano (Sovicille); 131 cm high, 70 cm wide, about 15 cm thick, 327 kg weight (Pauli & Danielsson 1893) (Fig. 2b).

A stele in the Museum of San Gimignano, recently discovered near Ulignano (San Gimignano); about 70cm high, 55 cm wide, about 15 cm thick (Fig. 2c). Stele 139435: Museum of Colle Val d'Elsa, deriving from Morticce di Mensanello (Colle Val d'Elsa); 93.5 cm high, 51 cm wide, 12 cm thick (Fig. 2d). A stele in the Museum of Casole d'Elsa, deriving from Le Poggiola (Casole d'Elsa); 69cm high, 101 cm wide, 17.5-20cm thick (Chigi 1877) (Fig. 2e). These Etruscan stelae are an interesting example of funerary artefacts used in northern Etruria in the second half of the sixth century Be, particularly in the areas of Volterra, Rusellae, Vetulonia and in the Valdarno (Fig. 1). In most cases, the stele is in the form of an arch (thus the name 'horseshoe stele') and follows a model common in the Padanian Etruria region at the end of the fifth century BC and particularly typical of the Volterra area, where it first appeared during the second half of the sixth century BC (Bruni 1995). In such stelae, the deceased is celebrated mainly by representations of warriors or characters holding the lituum (sceptre), but banquets (with a few people), dances and games are also

,./_

LIVORNO

Elba IStar~d /% ~

(}

0

/

I1(

[ ] Ulignanc ~, ~. [] Morticce di' : Mensanello~ [ ] Le Poggiola 0

[ ] Campassini

10 20

30Kin

[ ] Toiano ,~

Outcrop of travertine

~

Outcropof

calcareous tufa area of Figure 6

~ "'"''

....

~'

'~

""

"

' '

~

I

I

Fig. 1. Distribution of the Valdelsa stelae and location of the surrounding concretionary carbonate deposits.

STONE OF INSCRIBED E T R U S C A N STELAE

275

a - Campassini b - Toiano c- Ulignano d - Morticce Mensanello e - Le Poggiola

Fig. 2. The five inscribed Etruscan stelae studied.

di

276

A. GANDIN ET AL.

illustrated. Unlike other funerary styles (e.g. the so-called 'Chiusi cippi', characterized by more complex scenes and produced by an urbanized society; see Cristofani 1978), stelae addressed a mainly rural world and emphasized with a few essential elements (including inscriptions) the prestige and power of Etruscan princes (Zifferero 1991). The inscribed Etruscan stelae of the Valdelsa valley are part of this theme of death-celebration, although their specific function is not always clear (e.g. funerary symbol, commemorative monument or tomb door). Moreover, it has been suggested (Zifferero 1991) that stelae were used to mark and possibly organize the funerary space on the land owned by noble Etruscan families. The localities where the Valdelsa stelae were found (Fig. 1) are in the valleys of the Elsa River and Rosia Creek (a tributary of the Merse River), which are separated by the Montagnola Senese metamorphic ridge: the former valley includes the sites of Ulignano (San Gimignano), Le Poggiola, Morticce di Mensanello (Casole d'Elsa) and Campassini (Monteriggioni), whereas the latter includes Toiano (Sovicille). The Valdelsa stelae lack figurative representations and the deceased is celebrated by inscriptions (Fig. 2). The incised cornices, following the curved profile of the stele, also have a 'decorative' function in addition to their informative role. This tradition is represented by an even older example in the Siena area, the so-called 'Tomb of the Alphabet' (Bartoloni 1997) discovered near Monteriggioni. Although the tomb has been lost, archival documents testify to a wall 'decoration' consisting exclusively of inscriptions and an alphabet. Nevertheless, the primary function of the use of writing appears to be related to the social affirmation and rank of the deceased, as well as to awareness of the distinctive meaning it assumed: knowledge of writing was something to boast about. Therefore, the owners of the inscribed stelae belonged to a wealthy class, well aware of themselves and of the degree of social affirmation that writing assured (Bruni 1997, 2002). This class of stele used their inscriptions to document the inheritance of their land and the holding of power, including control of the transit routes to and from the Valdelsa valley (Ciacci 1999). Funerary writing appears to have been used in the Valdelsa area since the second half of the seventh century BE, but was most widespread around the end of the following century, with the production of specific objects such as stelae, 'cippi' and urns. The production of stelae consisted of: cutting the stone along the

horizontal stratification of the deposit; extraction, probably using wooden wedges; cutting the 'horseshoe'; smoothing the surface to receive the inscription; plus plastering and decoration (Ciacci 2004). The last two operations were revealed by analyses of superficial films, which showed the existence of a lime-finish pigmented with carbon-black overlying a grey covering plaster probably used for decoration (Guasparri et al. 2004). The operations necessary to cut and finish the stelae required unspecialized labourers, probably part of the servitus of a socially and culturally emancipated clientele. Writing was the duty of scribes, connected to the emergent families rather than to specific workshops, who were also able to create new styles of graphic writing. In fact, some recently discovered inscriptions on a stele (Martelli 1975) and a cinerary urn (Cianferoni 2002) found in the area of Colle Val d'Elsa, belonging to the Etruscan family Shekuntena (in Latin, Secundii), highlight a particular treatment of the initial sibilant of the noble name, which used a digram created ex-novo (sh) to indicate a phoneme not as yet recorded in Etruscan (Maggiani 2003; Ciacci 2004). Material and methods Material

All the studied stelae consist of a 10-15 cm thick slab of limestone formed by more or less regular alternations of centimetre-thick bands composed of flat or slightly wavy compact (Fig. 3a) and vacuolar facies, the latter characterized by the presence of calcium carbonate tubules (Fig. 3b). All the laminae tend to be lenticular and thus of variable thickness, although the more compact bands generally predominate in number or thickness. The slabs were extracted from the mother rock along detachment surfaces parallel to the stratification or lamination. Their state of preservation is generally good, although some of them show a slight alteration of pedogenetic origin, apparently attributed to late diagenetic exposure (see 'Interpretation'), or wear signs probably caused by use different from the original function (Guasparri et al. 2004). Methods o f sampling and analyses

Five cylindrical samples (2 cm diameter and 2 cm deep), one per slab, were taken with a microcorer from an unexposed area of the stelae (to preserve their aesthetic quality) to characterize the microfabric.

STONE OF INSCRIBED ETRUSCAN STELAE

277

Fig. 3. (a) Alternationof fine porous and slightly vacuolarbands. Stele 139435. (b) Regular successionof relatively compact bands. Stele from Le Poggiola. To make the sampling representative and comparable among stelae, each sample was taken from the compact facies. This allowed us to obtain a more complete sample from which was possible to acquire more information to characterize the stone. Two thin sections, parallel and orthogonal to the stratification plane of the slab, were cut from each cylinder for a detailed petrographic analysis.

Results

Fabric of the stone of the stelae The bands with a fine porous structure and compact appearance as well as those with more or less accentuated vacuolarity show a common spongy, thrombolitic fabric, consisting of a thin network of dense microcrystalline calcite (of micrite and microsparite size) containing relicts of organic matter (filaments and coccoids), and sometimes peloids with a clotted inner structure (peloidal microfacies; Pedley 1994). Internally, these peloids show a lumpy (grumous) fabric interpreted as the result of low-Mg micrite precipitation on coccoid bacterial agglomerations (Chafetz & Folk 1984). This network, referred to as microbial bindstone (bacterial mats; Folk et al. 1985), outlines small irregular pores (primary porosity) lined by cement of clear rhomboidal-scalenohedral (Fig. 4a) and/or acicular crystalline calcite (Fig. 4b), partially (Fig. 4c) or completely occluded by equant spar. Occasionally, small tubules composed of dense micrite (Fig. 4d), either unfilled or filled with equant sparry calcite, are trapped in the microbial bindstone. They represent small clusters or fragments of encrusted macrophyte stems and/or roots, usually well represented in the porous facies.

Some samples have rosettes of zoned crystals of dark calcite (Fig. 4e) caused by the presence of carbonaceous material (bacterial oncoids of Chafetz & Meredith 1983; Chafetz & Guidry 2003). They appear to have formed after cementation occurred, at the expense of the microbial fabric, probably following the circulation of meteoric or pedogenetic waters. Most samples also have relatively welldeveloped secondary porosity, associated with dissolution and/or bioturbation cavities produced by root systems, which are usually empty or partially filled with a yellowish clay (Fig. 4f). The macrofabric characteristics of the vacuolar facies can be related to the PhytoclasticPhytohermal facies (Ferreri 1985) or Reed Travertine (Guo & Riding 1998), formed by encrustation of the stems of marsh plants such as reeds. The tubules are usually massed together and oriented in the plane of the stratum (Phytoclastic facies), but are sometimes perpendicular to it (Fig. 5), arranged in small clusters of stems in a life-like position (Phytohermal facies). The stone of the Le Poggiola stele has the best preserved depositional and diagenetic fabric and the lowest total porosity (about 5%), whereas the stelae from Toiano, Campassini, Ulignano and Morticce di Mensanello have relatively high secondary porosity (up to 22%), indicating the occurrence of dissolution processes related to late-diagenetic or epidiagenetic recirculation of meteoric waters.

Interpretation The depositional and diagenetic evolution of these carbonates can be inferred from the fabric of the dominant spongy microfacies. The first phase, i.e. deposition, corresponds to formation

278

A. GANDIN ET AL.

Fig. 4. (a) Cement of clear rhomboidal-scalenohedral crystalline calcite. Stele 184531. (b) Cement of clear acicular crystalline calcite. Stele from Ulignano. (c) Thin network of dense microcrystalline calcite (spongy texture) with relicts of organic matter, which outlines irregular primary pores partially occluded by crystalline calcite. Stele from Le Poggiola. (d) Stems filled with cement and surrounded by a thin layer of micritic clots. Stele 184531. (e) Rosettes of dark calcite crystals probably related to recirculation of meteoric waters after deposition. Stele 139435. (f) Very fine clayey material deriving from the vegetated zone of the ground (soil) and deposited within small secondary cavities. Stele 13918.

of the microbial bindstone, whose genesis is believed to be related to biomediated precipitation of microcrystalline calcite on a substratum formed by mucilaginous masses of extracellular polymeric substances produced by unicellular

micro-organisms (cyanobacteria and cyanophytes) trapped inside the microcrystalline precipitate (Pedley 1994; Riding 2000). This phase alternated with ephemeral episodes of tentative colonization by reeds, fragments of

STONE OF INSCRIBED ETRUSCAN STELAE

Fig. 5. Remainsof small tufts of marsh reeds encrusted with calcium carbonate. Stele 13918. The pencil is 5 cm long.

which in the form of stems transported by water are mainly found entrapped in the spongy ground-mass. The first generation of cement, consisting of isopachous fringes of rhomboidal or acicular crystals, records the persistence during early diagenesis of active circulation of calcium carbonate-rich water within the porous network of the bindstone. The precipitation of equant spar, i.e. the second generation of cement, indicates meteoric phreatic circulation during later diagenetic stages, and the presence of incompletely occluded pores implies an interruption of water circulation. The presence of dissolution vugs and internal clayey sediments is evidence of active circulation of undersaturated or chemically aggressive waters of probable meteoric or pedogenetic origin. These observations allow a palaeoenvironmental reconstruction of these concretionary deposits. The texture characterized by mats of micro-organisms could have been preserved only in the absence of their natural enemies, i.e. grazers such as gastropods, which feed on their colonies. The concomitant scarcity or absence of macrophytes indicates environmental conditions unfavourable to higher organisms (Riding 2000). These conditions were determined by one or more physico-chemical characteristics of the water, i.e. high temperature, presence of H2SO 4 and/or HzS and low levels or lack of oxygen. Such conditions occur in palustrine systems located either near thermal sources (Folk et al. 1985) or in fluvio-lacustrine systems (Pedley 1994). However, episodes of dilution by rain or mixing with cold, more oxygenated waters could have created favourable, albeit ephemeral, conditions for the development

279

of pioneer plants such as reeds. The presence of peloids, occurring also in deposits formed in pools with stagnant non-thermal water (Pedley 1994), confirms this reconstruction. Moreover, the evidence of pedogenetic processes indicates that these deposits, formed in very shallow pools or marshes, were periodically subject, at least in the proximal areas, to wet and dry conditions. Consequently, the primary microbial deposit was affected by dissolution and/or bioturbation by roots, transport of clayey material toward the bottom and reprecipitation of calcite (recorded by rosettes of dark calcite; Fig. 4e), probably induced by oscillations of the water table. Such water-level oscillations also affected the depth of the marsh and are reflected in the carbonate sediments (Freytet & Plaziat 1982; Alonso-Zarza 2003). The presence of phytohermal structures indicates the marginal, shallower part of the marsh, with more diluted water and less energy (Toiano stele, Fig. 5; Campassini stele, Fig. 4d). The associations of phytoclastic facies indicates an input of plant remains by occasional water transport (Morticce di Mensanello stele, Fig. 3a), whereas the scarcity or absence of macrophyte remains indicates less oxygenated areas of the water basin (Le Poggiola, Figs 3b and 4c). The good preservation of the depositional and early diagenetic fabric can be related to an early interruption of water circulation, which resulted in incomplete occlusion of the residual porosity, and/or the formation of a poorly developed dissolution-vug system with internal clayey sediments. Evenly laminated carbonates characterized by similar depositional and diagenetic fabrics have also been observed in recent carbonates forming in distal pools of active hydrothermal systems at Tivoli (Folk et al. 1985), in the Mammoth Hot Springs (Yellowstone National Park, Chafetz & Guidry 2003) and in distal deposits of a well-preserved Messinian hot spring complex occurring near Volterra (Pignano) in southern Tuscany (Bossio et al. 1992), as well as in the lacustrine-palustrine environments of ambient-temperature water systems (Pedley 1994).

Discussion The stone used to make the five Valdelsa stelae is very different from the two local building stones deriving from the Mesozoic Tuscan Unit (Fig. 6) and available in the Valdelsa area, both very commonly used during the Middle Ages (Rodolico 1953): marble (a hard, not easily quarried rock, used as an ornamental stone in

280

A. GANDIN ET AL. -\

i ".~Sa ~G,mii na-~" ,~

cl

CDF

a

C • , - : 4 " . " : COF. ~ d e , F:,~ Synthem • -.-o., Ld - tn'~l~,'~: Og - co,'~lome~ ..... s - sand, a - day} • ..~.;-;...; • . • . • • . ° . • • . • . • ! CAI_ - Cal~na.a Synthem . . . . . . . . . . . . . FOC Torrente Foo Synthern • ""

•

.

•

* *"

" ""

• °"

*.'~

l ~

(:

~

o l l e Val d'Elsa •

° ""

c/'r'..." . •

[ ~

:oci r a c c l a n o j• ~

~ii~

........

\-

POG PoggibollfaSynt~em

Poggibonsi S y n t h e m (POG) Abbadia Synthem (ABB) Calcinaia Synthem (CAL) T.Foc~ S y n t h e m ( F O C ) Bellavista Synthem (BEL) tl - calcareous lithofacies t2 - detritical l i t h o f a c i e s C a m p i g l i a d e i Foci Synthern (CDF)

~ Pliocenemanne

.--..v

sediments Miocene continental sediments

i : ii i Monteriggioni

~

re-NeogeneLigurian

a n d T u s c a n Units

Le Caldane Spring

~;,

I I 1Km

Fig. 6. Geological scheme of the Valdelsa valley (modified from Capezzuoli & Sandrelli 2004). Inset: stratigraphic relationships of the Quaternary synthems recognized in the Valdelsa Basin (not to scale).

Siena Cathedral (Giamello et al. 2005)) and 'Cavernoso Limestone' (a brecciated rock, relatively soft but very vacuolar, mainly used as ashlar stonework (Rodolico 1953; Giamello et al. 2003))• The stele stone is a well-laminated, well-stratified concretionary limestone, generally rather compact, resistant to atmospheric agents, easily worked, easily extracted along the stratification surfaces and readily found on the surface without the need to open large quarries. Nevertheless, the petrographical data from the stele samples are not sufficient to determine without doubt whether the initial palustrine environment with extreme conditions was part of a hot-spring complex, and thus fed by hot water rich in poisonous gases (HzS, SO2-SO4). It may equally have been part of a fluviopalustrine system fed by cold poorly oxygenated water of karst origin. In fact, the absence of the macro-crystalline facies, exclusively developed in travertines (Guo & Riding 1998) and the scarcity of macrophyte remains in a living position and of faunas typical of calcareous tufas (e.g. ostracodes and gastropods) tend to exclude both a thermal origin s e n s u stricto and a clearly fluvio-palustrine environment• Therefore, in view of current knowledge and the high lateral variability of these deposits (Guo & Riding 1999), it is not possible to precisely define

either the general depositional context or the place of origin of the single slabs used to make the stelae. We can, however, be confident that the palaeoenvironment was marshy, and contained reed travertine. Water was possibly derived from thermal waters partially cooled after outflow from the resurgence and diluted with meteoric and/or karst-derived waters• Nevertheless, an analysis of the distribution of the fossil and currently forming laminated freshwater carbonates in central-southern Tuscany provides clues to the provenance of the material used by the Etrnscans. Travertines, and occasionally hydrothermal springs related to them, are very widespread outside southern Valdelsa, forming the deposits of Rapolano Terme (Guo & Riding 1998) and Acqua Borra (Castelnuovo Berardenga) to the SE, Pignano (Volterra, Bossio et al. 1992; Capezzuoli et al. 2004) to the west, Frosini (Chiusdino) to the SW and Iano (Gambassi Terme) to the north (Fig. 1). In these localities, proximal to the spring, well-laminated crystalline facies, associated with coated bubbles, shrub and paper-thin raft travertine, are dominant whereas distal facies, transitional to calcareous tufas, are less abundant• In contrast, both active and fossil calcareous tufas are predominant in the Valdelsa area. In fact, repeated phases of continental carbonate deposition occurred in

STONE OF INSCRIBED ETRUSCAN STELAE this area during the Quaternary, represented by five orders of terraces (Fig. 6). These are formed by thick micritic carbonates deposited in a lacustrine-palustrine environment (Campiglia dei Foci Synthem) and by generally crudely bedded, poorly cemented calcareous tufas indicative of a fluvio-palustrine environment (Abbadia Synthem, Calcinaia Synthem, T. Foci Synthem, Bellavista Synthem; Capezzuoli & Sandrelli 2004). The origin of the calcareous tufa deposits, present only in part of the valleys of some local watercourses (Elsa River, Staggia Creek and Foci Creek), is attributed to the activity of local karst and thermal springs during the Quaternary (Capezzuoli & Sandrelli 2004). However, as active springs are now rare in the area and the thermal springs are of low volume and low temperature, it can be assumed that most of them ceased to be active when local tectonic activity ceased (Capezzuoli & Sandrelli 2004). In conclusion, a potential extraction site may be sought in the Pignano area (Messinian travertines of the Volterra basin) or in southern Valdelsa, in an area with signs of thermal springs related to the deposition of Quaternary carbonates. With our current knowledge, precise location of the site is unlikely. However, the results of continuing research suggest the zone of Gracciano Val d'Elsa (in the southern sector of the area, see Fig. 6) as a possible source of the stone material used by the Etruscans. In fact, an extensive, poorly exposed tabular outcrop of concretionary limestone has recently been detected adjacent to one of the rare thermal springs still active in the area (Le Caldane Spring). In these well-bedded deposits at present under study, which show many of the lithological features of intermediate travertines or reed travertine, small-scale quarrying of limestone slabs took place until recent times. This activity occurred randomly where the small plateau morphologies allowed access to the rock stratification. For this reason, largescale quarries never existed at Gracciano Val d'Elsa and quarrying activity today is represented only by a series of holes, which are now part-filled with soil and vegetation. The possible extraction site occupies an approximately central position with respect to the localities where the stelae were found (Fig. 1). This is compatible with the view that Etruscan artisans, rather than carry materials from the farther Volterra, may have preferred to use deposits near their workshops (Petrelli et al. 2004), with the possibility of relatively short transportation of the material.

281

This paper has been financially supported by the Siena PAR (University Research Plan) grant (to A.G.). The writers thank G. C. Cianferoni (Superintendent of the Archaeology of Tuscany), M. Manganelli (President of Gruppo Archeologico Colligiano), M. Bezzini (President of Societ~t Archeologica Valdelsana of Casole d'Elsa) for kindly allowing the sampling of all the stelae, and Gruppo Archeologico of San Gimignano and A. Mennucci for facilitating access to the stele of Ulignano. The authors would like to thank P. Christie and M. Pedley for improving the English version of the text. We are grateful also to two anonymous referees for helpful discussion and comments.

References ALONSO-ZARZA, A. M. 2003. Palaeoenvironmental significance of palustrine carbonates and calcretes in the geological record. Earth-Science Reviews, 60(3-4), 261-298. BARTOLON~, G. 1997. La Tomba dell'Alfabeto di Monteriggioni. In: Etrusca et ltalica. Scritti in ricordo di Massimo Pallottino. Universith degli Studi 'La Sapienza'-Consiglio Nazionale delle Ricerche e Istituto per l'Archeologia EtruscoItalica, Istituti Editoriali e Poligrafici Internazionali, Pisa-Roma, 25-49. BECATTI, G. 1933. Monteriggioni (Siena). Scavo di una tomba a camera ed alcuni trovamenti archeologici. Notizie degli Scavi di Antichita, Accademia

Nazionale dei Lincei, Roma, 150-159. BOSSIO, A., CERRI, R., COSTANTINI,A., et al. 1992. I1 bacino di Volterra. In: Guida alle Escursioni postcongresso. 76 ° Riunione estiva SGI: l'Appennino settentrionale. Escursione B4: I bacini distensivi neogenici e quaternari della Toscana, 244-270.

BRUNI, S. 1997. La Valdera e le colline pisane inferiori: appunti per la storia del popolamento. In: Aspetti della cultura di Volterra etrusca fra l'et& del Ferro e l'etb Ellenistica e contributi della ricerca archeologica alla conoscenza del popolo etrusco, Atti del XIX Convegno di Studi Etruschi ed Italici, Volterra, 129-171.

BRUNI, S. 2002. La Valle dell'Arno: i casi di Fiesole e Pisa. In: MANGANELLI,M. ~ PACCHIANI,E. (eds) Cittb e territorio in Etruria. Per una definizione di cittb nell'Etruria settentrionale. Grupp Archaeolo-

gical Colligiane, Colle Val d'Elsa, 272-344. CAPEZZUOLI, E. 8~ GANDIN, A. 2004. I 'travertini' in Italia: proposta di una nuova nomenclatura basata sui caratteri genetici. II Quaternario, Italian Journal of Quaternaty Sciences, 17, 273 -284. CAPEZZUOLI, E. • SANDRELLI, F. 2004. I sedimenti quaternari del settore meridionale della Valdelsa (Provincia di Siena). II Quaternario, Italian Journal of Quaternary Sciences, 17(1), 33-40. CAPEZZUOLI, E., GANDIN, A. & SANDRELLI, F. 2004. Neogene-Quaternary continental carbonates: the Quaternary deposits of Valdelsa and the Miocene travertines of Pignano (Volterra). In: MOR~NI, D. & BRUM, P. (eds) The 'Regione Toscana' project of geological mapping; case histories and data acquisition. Regione Toscana, Firenze, 89-96.

282

A. GANDIN ET AL.

CHAFETZ, H. S. & FOLK, R. L. 1984. Travertines: depositional morphology and the bacterially constructed constituents. Journal of Sedimentary Petrology, 54, 289-316. CHAFETZ, H. S. t~; GUIDRY, S. A. 2003. Deposition and diagenesis of Mammoth Hot Springs travertine, Yellowstone National Park, Wyoming, USA. Canadian Journal of Earth Sciences, 40(1 l), I515-1529. CHAFETZ, H. S. & MEREDITH, J. C. 1983. Recent travertine pisoliths (pisoids) from Southeastern Idaho, USA. In: PERYT, T. (ed.) Coated Grains. Springer, Berlin, 456-455. CHIGI, B. 1877. Siena. Notizie degli Scavi di AntichitL Accademia Nazionale dei Lincei, Roma, 301-304. CIACC1, A. 1999. I1 periodo etrusco. Note sul popolamento e l'economia. In: VALENTI, M. (ed.) Carta archeologica della Provincia di Siena. lll--La Val d'Elsa (Colle Val d'Elsa e Poggibonsi), Nuova Immagine, Siena. 300-312. CIACCI, A. 2004. Le stele etrusche dell'Alta Valdelsa. Gli aspetti archeologici ed epigrafici, hi: CIACCI, A. (ed.) Monteriggioni-Campassini. Un sito etrusco nell'Alta Valdelsa. AIl'Insegna del Giglio, Firenze, 183-210. CIANFERONI, G. C. 2002. L'Alta Valdelsa in et/~ orientalizzante e arcaica, h~: MANGANELLI, M. t~¢ PACCHIANI, E. (eds) Cittgt e territorio in Etruria. Per una definizione di cittb nell 'Etruria Settentrionale. Colle di Val d'Elsa, 83-126. CRISTOFANI, M. 1978. L'arte degli Etruschi. Produzione e consumo. Giulio Einaudi, Torino, 139-147. FERRERI, V. 1985. Criteri di analisi di facies e classificazione dei travertini pleistocenici dell'Italia Meridionale. Rendiconti Accademia Scienze Fisiche e Matematiche, Napoli, 52, 1-47. FOLK, R. L., CHAFETZ, H. S. 8z TIEZZI, P. A. 1985. Bizarre forms of depositional and diagenetic calcite in hot-spring travertines, Central Italy. In: SCHNEIDERMANN,N. 8,1; HARRIS, P. (eds) The Biology of Blue- Green Algae. Blackwell Scientific, Oxford, 349-369. FORD, T. D. & PEDLEY, H. M. 1996. A review of tufa and travertine deposits of the world. Earth-Science Reviews, 41, 117-175. FREYTET, P. & PLAZIAT, J. C. 1982. Continental carbonate sedimentation and pedogenesis--Late Cretaceous and Early Tertiary of southern France. Contributions to Sedimentology, 12, 1-213. GIAMELLO, M., GUASPARRI, G., MUGNAINI, S., SABATINI, G. & SCALA, A. 2003. Lo studio dei materiali lapidei del centro storico di Siena. Arkos, 2, 22-29. GUASPARRI, G., NARDELLI,M. G. & SCALA, A. 2004. Studio delle superfici lapidee di sue stele. In: CIACCI, A. (ed.) Monteriggioni-Campassini; un sito etrusco in Alta Valdelsa. All'Insegna del Giglio, Firenze, 222-227. GIAMELLO, M., DROGHINI, F., MUGNAINI, S., GUASPARRI, G., SABATINI, G., SCALA, A. &

MORANDINI, M. 2005. I1 Pavimento marmoreo del Duomo di Siena. Caratterizzazione dei materiali e dello stato di conservazione. In: CACIORGNA, M., GUERRINI, R. & LORENZONI, M. (eds) Studi interdisciplinari sul Pavimento del Duomo di Siena. lconografia, stile, indagini scientifiche. Opera della Metropolitana di Siena, Collana di studi e ricerche. Cantagalli, Siena, 173-197. Guo, L. & RIDING, R. 1998. Hot-spring travertine facies and sequences, Late Pleistocene, Rapolano Terme, Italy. Sedimentology, 45, 163-180. Guo, L. & RIDING, R. 1999. Rapid facies changes in Holocene fissure ridge hot spring travertines, Rapolano Terme, Italy. Sedimentology, 46, 1145-1158. MAGGIANI, A. 2003. Rivista di Epigrafia Etrusca, Studi Etruschi, Istituto Nazionale di Studi Etruschi e Italici, Firenze, LXIX, 362-364. MANGANI, E. 1992. Castelnuovo Berardenga (Siena). L'Orientalizzante recente in Etruria settentrionale: tomba A della necropoli principesca del Poggione (1980). Notizie degli Scavi di Antichitb., Roma, Accademia Nazionale dei Lincei, 42/43, 5-82. MARTELLI, M. 1975. Rivista di Epigrafia Etrusca, Studi Etruschi, Istituto Nazionale di Studi Etruschi e ltalici, Firenze. XLIII, 200-201. PAOLI, C. & DANIELSSON O.A. 1893. Corpus lnscriptionum Etruscarum, I(4620). PEDLEY, H. M. 1990. Classification and environmental model of cool freshwater tufas. Sedimentary Geology, 68, 143-154. PEDLEV, H. M. 1994. Prokaryote-microphyte biofilms and tufas: a sedimentologicai perspective. Kaupia, 4, 45-60. PENTECOST, A. 1995. The Quaternary travertine deposits of Europe and Asia Minor. Quaternary Science Reviews, 14(10), 1005-1028. PETRELLI, M., PERUGINI, D., MORONI, B. & POLl, G. 2004. Travertine, a building stone extensively employed in Umbria from Etruscan to Renaissance age: provenance determination using artificial intelligence technique. Periodico di Mineralogia, 73(3), 151-169. RIDING, R. 2000. Microbial carbonates: the geological record of calcified bacterial-algal mats and biofilms. Sedimentology, 47, 179-214. RODOLICO, F. 1953. Le pietre delle cittb d'ltalia. Le Monnier, Firenze. VILES, H. A. & Goudie, A. S. 1990. Tufas, travertines and allied carbonate deposits. Progress in Physical Geography, 14, 19-41. ZIFFERERO, A. 1991. Forme di possesso della terra e tumuli orientalizzanti nell'Italia centrale tirrenica. In: HERRING, E., WHITEHOUSE, R. WILK1NS, J. (eds) The Archaeology of Power. Papers of the Fourth Conference of Italian Archaeology, 1. Accordia Research Centre, London, 107-134.

Geological tools to interpret Scottish medieval carved sculpture: combined petrological and magnetic susceptibility analysis S U Z A N N E M I L L E R 1, F I O N A M. M c G I B B O N 2, D A V I D H. C A L D W E L L 1 & N I G E L A. R U C K L E Y 3

1National Museums of Scotland, Chambers Street, Edinburgh EH1 2PB, UK (e-mail: s. mille r @nms. ac. uk) 20ffice of Lifelong Learning, The Universi~ of Edinburgh, 11 Buccleuch Place, Edinburgh EH8 9LW, UK 3The Old School House, Kirkbuddo DD8 2NQ, UK Abstract: Geological surveys of 172 early and late medieval sculptured stones from central

and western Scotland have been undertaken to determine the provenance of the materials used. Non-destructive petrological studies (including grain size, mineralogy distribution, and textural and structural characteristics) and magnetic susceptibility measurements are used to characterize the sculptured stones and potential source material. The results indicate that for the early medieval sculpture: (1) all the sculptured stones are sandstone with the exception of one siltstone and one granite; (2) the sedimentary rocks are consistent with sources in the Lower Old Red Sandstone of the area; (3) from within the Lower Old Red Sandstone, a number of different geological units have been used. For the late medieval sculpture, the results indicate that: (1) various rock types have been used including schist, slate and sandstone; (2) non-locally derived material is used extensively in some areas, suggesting a more developed network for procurement of raw materials. The analytical techniques used also provide additional information in the art historical interpretation of a number of carved stones by identifying carved fragments from the same monument.

Medieval sculpture in East Central Scotland (Angus, Tayside and Perthshire) is exemplified by some 120 eighth to 10th century (Pictish) sculptures and sculptural fragments (e.g. Fig. 1). These were published and catalogued by Cruden (1964). The Royal Commission for Ancient and Historical Monuments Scotland (RCAHMS) is in the process of drawing and photographing the collections. Medieval West Highland sculpture (e.g. Fig. 2) was defined and surveyed in an important monograph by Steer & Bannerman (1977). There are some 600 examples ranging in date from the 14th to the 16th centuries, spread over 86 sites throughout the Western Highlands of Scotland. Typically, the Pictish sculptures are standing stones with elegant designs and symbols, which often include abstract designs and pictures of objects and animals (both real animals and fantasy creatures). The West Highland sculpture is typified by grave monuments and commemorative crosses. Decoration includes effigies (generally military in nature), ships, hunting scenes, scrollwork and inscriptions. For the

purposes of this work, the early medieval sculpture (from East Central Scotland) will be referred to as 'Pictish sculpture' and the late medieval sculpture (from Iona and Oronsay) will be referred to as 'West Highland sculpture'. Many of the sculptures are in the care of Historic Scotland (e.g. the St. Vigeans collection) whereas others are part of the collections of the National Museums of Scotland, are in local authority care, or are held privately. The present project was initiated with the aim of improving previous interpretation of the sculpture by bringing geological knowledge and skills to bear on the subject, leading to a reassessment of where the sculpture was carved and a new understanding of power and patronage in medieval Scotland. The project has involved close collaboration with RCAHMS and Historic Scotland. Using the non-destructive, in situ techniques of macro-petrology and magnetic susceptibility, a survey of the most representative sculptures has been undertaken. Pictish sculpture collections surveyed include: St Vigeans, Aberlemno, Meigle, Pictavia and

From: MAGGETT1,M. & MESSIGA,B. (eds) 2006. Geomaterials in Cultural Heritage. Geological Society, London, Special Publications, 257, 283-305. 0305-8719/06/$15.00 © The Geological Society of London 2006.

284

S. MILLER ETAL.

(a)

Although previous studies have used such techniques to characterize artefacts, they have been largely restricted to objects composed of igneous rocks (e.g. Williams-Thorpe & Thorpe 1993; Peacock 1995, 1997; WilliamsThorpe et al. 1996, 1997; Markham 1997; Williams-Thorpe & Henry 2000) or low-grade metamorphic rocks (Floyd & Trench 1988). Here, the techniques are applied to sedimentary, high-grade metasedimentary and high-grade meta-igneous rock artefacts.

Geological setting Pictish sculpture

(b)

Fig. 1. Examples of carved early medieval (Pictish) symbol stones from East Central Scotland. (a) Aberlemno No. 2; (b) Meigle No. 3.

Meffan, together with individual sculptures at other sites (Fig. 3). West Highland sculpture collections surveyed are two of the largest and most representative collections, located at Iona and Oronsay (Fig. 3).

The oldest rocks underlying the sites of early medieval sculpture belong to the Dalradian Supergroup and crop out to the NW of the Highland Boundary Fault zone (Fig. 4) in the Grampian Highland terrane. They were deposited as sandstones and mudstones (principally marine), which were subsequently metamorphosed to cleaved grits, shales, phyllites and schists. They are considered to be of Cambrian age (540-490 Ma) and were subjected to polyphase deformation during the Caledonian Orogeny. During the Early Devonian ( 4 0 9 - 3 8 6 M a ) vast quantities of sediment were eroded from the Caledonian mountains and carried by large rivers to a subsiding area of low ground in the area of the present Scottish Midland Valley. Coarse detritus was deposited in large conglomerate fans whereas sandstones and siltstones were deposited on extensive alluvial plains by braided rivers, in a semi-arid environment. This dominantly arenaceous and commonly red-coloured, non-marine succession is known as the Lower Old Red Sandstone and is c. 9000 m thick. The term Old Red Sandstone has been used in the UK since 1822 (Conybeare & Phillips 1822) to denote the terrestrial sediments that are roughly equivalent in age to the Devonian marine deposits in SW England and continental Europe. These are generally recognized as being c. 4 0 0 - 3 6 0 M a old, although there is still considerable debate as to the exact age of the Scottish Old Red Sandstone (McKerrow et al. 1985). The six major lithostratigraphical divisions of the Lower Old Red Sandstone are the Stonehaven, Dunnottar, Crawton, Arbuthnott, Garvock and Strathmore groups, with various subdivisions at formation level (Armstrong & Paterson 1970). Associated with the sedimentary units are a number of contemporaneous volcanic rocks (particularly lavas) of similar age. No sculptures in this group have been carved from volcanic

GEOLOGY OF SCOTTISH MEDIEVAL SCULPTURE

(a)

285

(b)

Fig. 2. Examplesof late medieval sculpture from the West Highlandsof Scotland. (a) Slab of Angus MacDonald, Iona. (b) Slab of Murehad MacDuffieof Colonsay, 1539; Oronsay Priory.

rock, therefore the Old Red Sandstone lavas will not be considered further here. Early Devonian sedimentation and volcanism in the area was terminated by the onset of earth movements (from around 386 Ma) producing two asymmetric folds with a N E - S W trend: the Strathmore Syncline and the Sidlaw Anticline. Younger sequences were subsequently deposited and eroded, leading to the Lower Old Red Sandstone forming the principal country rock in the area of study. The sculptures that form the early medieval (Pictish) dataset are located within the

northeastern part of the Midland Valley of Scotland. This area lies almost entirely within the Midland Valley terrane and crops out as Lower Old Red Sandstone strata. One sculpture (the Dunfallandy Stone) is sited on Dalradian metasediments (Fig. 4). Although the Dunfallandy Stone is located to the north of the Highland Boundary Fault, within Dalradian metasediments, the lithology of the sculpture is of Lower Old Red Sandstone. Potential sources of rock have been identified from various outcrops of Lower Old Red Sandstone, indicating that only rocks from the

S. MILLER ETAL.

286

j I"

~c9

Oro

Fig. 3. Map of Scotland showing the distribution of Pictish sculpture (East Central Scotland (circles)) and West Highland sculpture (circles with crosses). Arbuthnott, Garvock and Strathmore Groups are likely to have been utilized for the stone sculptures under investigation.

extensively between the Moine Thrust and the Great Glen Fault (Fig. 5). The Moine is subdivided into the Morar, Glenfinnan and Loch Ell Groups. The Morar and Loch Eil Groups are shallow marine sediments whereas the Glenfinnan Group is indicative of transgressive deposition. The Dalradian Supergroup is a succession of clastic sediments and limestones (both principally marine) with minor volcanic rocks, which were subsequently metamorphosed to cleaved grits, shales, slates, phyllites, schists, metalimestones and meta-igneous rocks. Dalradian sedimentation occurred during a prolonged extensional phase, lasting c. 330 Ma (c. 800470 Ma). The rocks were subjected to polyphase deformation during the Caledonian Orogeny. The succession is at least 25 km thick and is divided into the Grampian, Appin, Argyll and Southern Highland Groups. The youngest rocks underlying the West Highland sculptures are igneous rocks of Tertiary age, associated with volcanic centres that were active between c. 60.5 and 55 Ma. The extrusive sequences range from alkali olivine basalt to trachyte, erupted from fissures. Intrusive rocks, representing the now exposed roots of major volcanic centres, display a wide range of rock types. No sculptures in this group have been carved from igneous rock, therefore the Tertiary lavas will not be considered further here. The sculptures that form the late medieval dataset reported here are located on the islands of Iona and Oronsay in the West Highlands. The geology of the two islands is dominated by Lewisian and Torridonian rocks.

Methods

West Highland sculpture The oldest rocks underlying sites of West Highland sculpture belong to the Lewisian Complex of the Hebridean terrane. The complex comprises a variety of gneissose rocks, metagreywackes, quartzite, meta-limestones and meta-igneous rocks with polyphase metamorphic and deformational histories. Locally, this ArchaeanProterozoic basement is overlain by Proterozoic sediments, the Torridonian Supergroup. Some sculpture also rests on Moine and Dalradian Supergroup suites cropping out within the Grampian Highland and Northern Highland terranes (Fig. 5). The Moine Supergroup is a thick succession of strongly deformed and metamorphosed siltstones, mudstones and sandstones that crop out

All petrological analysis was undertaken independently of the magnetic susceptibility measurements to ensure that no bias was recorded in the grouping of the rock types.

Petrology All available sculptured stones have been examined using non-destructive petrological techniques to provide a 'field identification' of the rock type. This type of petrological analysis has provided a basic identification of rock type and has been used to distinguish between general rock types. All examinations included the following measurements: (1) colour (with reference to Munsell soil colour charts); (2) grain size (with reference to standard grainsize measurements on the micrometre scale);

GEOLOGY OF SCOTTISH MEDIEVAL SCULPTURE 5

I

287

0

Upper Old Red Sandstone & younger strata Strathmore Group - sedimentary Garvock Group - sedimentary

r

] Arbuthno4t Group - sedimentary Arbuthnott Group - igneous

I

'/ "~

Lower Old Red Sandstone

Crawton Group- sedimentary & igneous

gunoC~r Group- sedimentary Sto neh=ven Group - sedimentary

(~

J

Dalradian & associated strata OUNKELE

O

Locations of early medieval

\

"~

# ..

0

5

0

15

10 10

20 3(1

20

25Mdes

40 K,IometreS

0

Fig. 4. Geological map of East Central Scotland showing the distribution of Pictish sculpture in the area under investigation. (3) macroscopic mineralogy (i.e. mineralogical content that can be ascertained by examination with 10× magnification hand lens); (4) textural and structural characteristics such as parallel bedding or lamination, cross-bedding, jointing, other planar fabric and grain-size variation;

(5) clast distribution and composition; (6) weathering characteristics. Division of the specimens into 'rock types' has been primarily based on the textural and mineralogical characteristics. Colour has been used only as a general guide to overall appearance because,

Tertiary basalts

Tertiary intrusive rocks

t

I

Dalradian metamorphic rocks

Moine metamorphic rocks

Lewisian gneiss Locations of late medieval sculpture

Fig. 5. Geological map of Western Scotland showing the location of the main late medieval sculpture centres.

288

S. MILLER ET AL.

in many cases, the sculptures have undergone varying degrees of weathering and/or cleaning, both activities that could significantly alter the colour of the surface of the specimen. All outcrop specimens (i.e. potential source rocks) have been examined using petrological techniques to classify rock type. In addition to the measurements taken for the sculptures, all examinations of outcrop specimens also include microscopic mineralogy, i.e. mineralogical content from thin-section examination. Samples of potential source (raw) materials include both historical and modern quarry outcrop and other outcrops. Obviously, some of these actual exposures are modern but by sampling at all possible localities we have aimed to assemble a comprehensive dataset, covering as many rock types as possible. All sandstones are classified according to their mineralogy and using the sandstone classification scheme of Folk, where all rocks containing less than 15% fine-grained matrix are classified in terms of the three principal components; quartz, feldspar (plus granite and gneiss clasts) and other rock fragments (Folk 1974).

Magnetic susceptibility Magnetic susceptibility was measured with an Exploranium KT-9 Kappameter, giving a measurement of the true susceptibility. The KT-9 has several operating modes but to achieve a consistency of readings the 'pin mode' of working was used throughout. The accuracy on a flat rock is estimated to be + 3% in the 'pin mode'. Care was taken to avoid magnetic contamination. For the sculptures, this can be caused by three principal sources, namely, methods of mounting (e.g. metal pins or rods), repairs to the sculptures and floor or substrate material (e.g. concrete). At all locations, stones with a thickness less than 6 c m were treated as suspect, as no tables for adjusting the readings from thin rock units are available (WilliamsThorpe et al. 2000). Whenever feasible a series of 12 readings were taken from the front and rear faces of the carved stones away from all possible sources of magnetic contamination. The average of the 12 readings represents one dataset per sculpture. Over 180 datasets for sculpture were taken. Seventy datasets taken from quarry or outcrop locations have been utilized to provide a magnetic susceptibility database of potential sources (raw materials). The complete datasets

equate to 2208 measurements for the sculptures and 840 measurements for the source rocks.

Results Lower Old Red Sandstone outcrop From the petrological examinations of the outcrop rocks, significant differences between sandstones from the Arbuthnott, Garvock and Strathmore Groups can be discerned. Within groups and formations, and even within outcrop localities, there can be significant differences in grain size and gross texture (e.g, cross-bedding, lamination, ripples, etc.). However, the overall mineralogical characteristics are generally consistent within outcrop. From the examination of thin sections, it is clear that there are very recognizable differences between outcrops on a microscopic scale (Table 1). The magnetic susceptibility data suggest that the units studied display a wide range of readings (Fig. 6). Arbuthnott group. All sandstones belong to the Dundee Fm. They are generally arkosic sandstones and litharenites. They are characteristically drab coloured (grey to pinkish grey) although the outcrop at Kingoodie is a dull red colour, as a result of very late stage weathering and development of hematite cement. Generally, the Arbuthnott sandstones contain a much more significant proportion of chlorite than either Garvock or Strathmore sandstones. The range of magnetic susceptibility measurements displayed by the rocks of the Dundee Fm is very varied (Fig. 6a). This suggests that magnetic susceptibility cannot be used as a tool for distinguishing these rocks independently. However, the readings in this group varied from sandstones to much finer silty sandstones, and micaceous flagstones. The finer material, often with an increase in mica, generally results in higher readings. It is therefore possible to characterize different lithologies within the formation. Garvock group. Sandstones cropping out to the NW of the Sidlaw Anticline lie within the Scone Fro. Those cropping out to the SE of the Sidlaw Anticline belong to one of the Red Head Fm, Arbroath Sandstone unit or Auchmithie Conglomerate unit. The Scone Fm sandstones are predominantly litharenites (containing very little or no calcite). The sandstones of the Read Head Fm and Auchmithie Conglomerate are sub-arkosic whereas the Arbroath Sandstone units are calcareous

Litharenite

Litharenite

Litharenite

Calcareous arkose

NO 5074 5312

NO 5074 5312

NO 4605 4948

Arbuthnott Gp Dundee Fm (Baldardo Quarry, Turin Hill)

Arbuthnott Gp Dundee Fm (Baldardo Quarry, Turin Hill)

Arbuthnott Gp Dundee Fm (Balmashanner E Quarry)

Arbuthnott Gp Dundee Fm (Aberlmeno Churchyard outcrop) Arbuthnott Gp Dundee Fm (Baldardo Quarry, Turin Hill)

NO 5074 5312

Petrological classification

Arkose

Grid reference

NO 5222 5557

Stratigraphic group and formation (location) Very angular, well sorted; poly- & monocrystalline qtz, weathered fsp, minor aligned bt & ms, extensive clay matrix, opq grains, rim & cement Very angular, poorly sorted; poly& monocrystalline qtz, weathered fsp, aligned, kinked ms, bt & chl, opq grains, rims & interstitial cement, clay cement, lithic (volcanic) clasts Very angular-subrounded, poorly sorted, immature; poly- & monocrystalline qtz, weathered fsp, kinked bt, ms & chl, extensive hem & clay matrix, dominant lithic (volcanic) clasts Angular- subrounded, poorly sorted, immature; poly- & monocrystalline qtz, weathered fsp, extensive aligned, kinked ms, bt & chl, extensive opq rim & cement, clay & qtz cement Very angular, mod well sorted, grain supported; poly- & monocrystalline qtz, weathered fsp, significant chl, bt & ms, matrix of clay & cc, opq grains and some cement

Petrological characteristics

Weak red

Red claystone blebs Massive and crossbedded units

Massive and silty sandstone interbedded

vcg-cg

f-mg

(Continued)

Reddish grey

Weak red Volcanic

Massive and crossbedded units

vcg-cg

Reddish brown

Colour

Weak red

Massive and crossbedded units

cg

Clast composition

Volcanic

Massive sandstone

Fabric

f-mg

Grain size

Table 1. Petrology of the potential source rocks from the Lower Old Red Sandstone (LORS) of East Central Scotland

Litharenite

Subarkosic siltstone

NO 5547 4485

NO 5547 4485

NO 4516 3955

NO 4516 3955

NO 4405 4846

Arbuthnott Gp Dundee Fm (Carmyllie NE Quarry)

Arbuthnott Gp Dundee Fm (Carmyllie NE Quarry)

Arbuthnott Gp Dundee Fm (Dodd Quarry)

Arbuthnott Gp Dundee Fm (Dodd Quarry)

Arbuthnott Gp Dundee Fm (Forfar Bypass by South Leckaway)

Calcareous arkose

Calcareous arkose

Arkose

Litharenite

Petrological classification

NO 5547 4485

Grid reference

Arbuthnott Gp Dundee Fm (Carmyllie NE Quarry)

Stratigraphic group and formation (location)

Table 1. Continued

Very angular-subrounded, poorly sorted, immature; poly- & monocrystalline qtz, weathered fsp, aligned, kinked bt, ms & chl, some opq grains & cement, clay & qtz cement, extensive lithic (volcanic) clasts Very angular-subrounded, poorly sorted, immature; poly- & monocrystalline qtz, weathered fsp, minor opq grains & cement, qtz and clay matrix, extensive lithic (volcanic & granitic) clasts Very angular, well sorted; monocrystalline qtz, weathered fsp, aligned, kinked chl, bt & ms, minor opq grains & cement, clay cement Very angular, mod well sorted; poly- & monocrystalline qtz, weathered fsp, major kinked aligned chl, ms & bt, opq grains, clay cement Very angular, mod well sorted; poly- & monocrystalline qtz, weathered fsp, major kinked aligned chl, ms & bt, opq grains, calc & clay cement Very angular, mod well sorted; poly- & monocrystalline qtz, weathered fsp, major kinked aligned chl, ms & bt, opq grains, minor opq cement, calc & clay cement

Petrological characteristics

f-rag

vfg

Cross-bedded and massive sandstone <1.5 m thick; some lamination on m m cm scale Cross-bedded and massive sandstone < 1.5 m thick; some lamination on m m cm scale Massive and crossbedded units

Andesite

Massive and crossbedded units with some flaggy sst units

silt

vfg

Weak red Volcanic, granitic

Massive and crossbedded units with some flaggy sst units

granules

Dark grey

Grey

Olive grey

Weak red

Olive

Colour

Volcanic, mud

Clast composition

Cross-bedded units (0.15-1.2 m thick); bands of pebbles and grey mud clasts; some interbedded fissile units

Fabric

vcg

Grain size

Arkose

Arkose

NO 4334 4666

NO 4318 4673

NO 3364 2929

NO 6013 4782

NO 4030 4162

NO 5898 4861

Arbuthnott Gp Dundee Fm (Kinettles Ice House Quarry)

Arbuthnott Gp Dundee Fm (Kingoodie cliff)

Arbuthnott Gp Dundee Fm (Leysmili Quarry)

Arbuthnott Gp Dundee Fm (Lumley Den Low Quarry)

Arbuthnott Gp Dundee Fm (Middleton Quarry)

Calcareous arkose

Arkose

Arkose

Calcareous arkose

Lithic arkose

NO 6037 5817

Arbuthnott Gp Dundee Fm (Hillhead of Burghill Quarry) Arbuthnott Gp Dundee Fm (Kinettles House E Quarry)

Variable colour, some fissile sst units, some more massive; some cross-bedding

Inter-bedded sst, siltstone (flaggy) and conglomeratic units; sst massive and crossbedded, some ripple marks flaggy & blocky outcrop with some shale horizons

Massive, fissile and cross-bedded units on c m - m scale

f-mg

fvg

Fissile sandstone units

Cross-bedded and channel sst with laminae on m m - c m scale; some pebbly horizons Fissile sandstone units

f-rag

m-cg

mg

Very angular, well sorted; polyfg & monocrystalline qtz, weathered fsp, aligned, kinked bt, ms & chl, predominantly clay cement with some opq cement, minor mud clasts Very angular, mod well sorted; m-cg poly- & monocrystalline qtz, weathered fsp, kinked bt & chl, opq grains & cement, calc cement, clay cement, very minor lithic clasts (volcanic)

Very angular, mod sorted; poly- & monocrystalline qtz, weathered fsp, aligned & kinked bt, ms & chl, extensive clay & qtz cement, v minor opq grains Very angular, mod well sorted; poly- & monocrystalline qtz, weathered fsp, major kinked aligned chl, ms & bt, opq grains & cement, clay cement, v minor lithic clasts Very angular, rood well sorted; poly- & monocrystalline qtz, weathered fsp, major kinked aligned chl, ms & bt, opq grains & cement, clay cement Very angular, well sorted; poly& monocrystalline qtz, weathered fsp, aligned kinked bt, ms, chl, clay, cc & opaque cement Very angular, well sorted; poly& monocrystalline qtz, weathered fsp, aligned & kinked bt, ms & chl, dominated by clay & qtz cement, minor opq cement

Lithic

Lithic clasts (up to 1 m)

Lithic

Quartzite, basalt, foliated granite, mud, jasper, felsite

(Continued)

Reddish grey

Grey

Grey

Weak red

Pinkish grey

Pale red

Grey

O

O "~

r~ O O

Grain size

vfg

vfg

vfg

m-cg

Very angular, well sorted; monocrystalline qtz, weathered fsp, aligned, kinked chi, bt & ms, minor opq grains & cement, clay & minor calc cements Very angular, well sorted; monocrystalline qtz, weathered fsp, aligned, kinked chl, bt & ms, minor opq grains & cement, clay & minor calc cements Very angular, well sorted; poly- & monocrystalline qtz, weathered fsp, aligned, kinked bt, ms & chl, calc, clay & opq cement

Very angular, mod well sorted; poly- & monocrystalline qtz, weathered fsp, kinked chl, ms & bt, clay & opq cement, minor lithic clasts

Mudstone

Calcareous siltstone

Calcareous arkose

Arkose

NO 5298 5371

NO 5298 5371

NO 4997 3725

NO 5037 4890

Arbuthnott Gp Dundee Fm (N Mains of Turin Quarry)

Arbuthnott Gp Dundee Fm (N Mains of Turin Quarry)

Arbuthnott Gp Dundee Fm (Pitairlie Quarry)

Arbuthnott Gp Dundee Fm (W Mains of Dunnichen Quarry)

Very angular-angular, mod sorted; cg poly- & monocrystalline qtz, weathered fsp, kinked bt, ms & chl, minor opq grains & cement, clay & qtz cements, lithic clasts

Petrological characteristics

Lithic arkose

Petrological classification

NO 5298 5371

Grid reference

Arbuthnott Gp Dundee Fm (N Mains of Turin Quarry)

Stratigraphic group and formation (location)

Table 1. Continued

massive, crossbedding and channels; some intercalated fissile units Thick sst units (0.11.0 m); massive, cross-bedding and channels; some intercalated fissile units Shale with more massive sst at base of quarry (30 m); thin sst units, highly micaceous and fissile; good plant fossil preservation Massive sst with some laminated units and cross-bedding

(0. I - 1,0 m ) ;

0.1 - I m thick units, massive and current cross-bedded; channels; most units show planar laminae on m m - c m scale; erosional surfaces assoc, with clay ripup clasts and pebble layers; minor silty layers Thick sst units

Fabric

Lithic

Lithic

Lithic

Clast composition

Brownish red

Grey

Dark grey

Olive grey

Weak red

Colour

t,~

Calcareous sub-arkose

NO 0238 1823

NO 5707 3448

NO 5707 3448

NO 0120 1783

Garvock Gp Scone Fm (Cairnie Cottage A9 roadside)

Garvock Gp Arbroath Sandstone (Carnoustie foreshore)

Garvock Gp Arbroath Sandstone (Carnoustie foreshore)

Garvock Gp Scone Fm (Chapelbank, A9 roadside)

Sublitharenite

Calcareous sublitharenite

Calcareous sublitharenite

Calcareous sublitharenite

NO 6628 4120

Garvock Gp Auchmithie Conglomerate (Arbroath Cliffs)

Litharenite

NO 3688 4480

Arbuthnott Gp Dundee Fm (Wester Rochelhill Quarry)

cg

Angular, poor-mod sorted, m-cg immature, polycrystalline qtz, weathered fsp, minor mica, volcanic clasts, minor clay, cc cement, opaque rim and interstitial cement. Very angular, mod well sorted; fg poly- & monocrystalline qtz, weathered fsp, aligned, kinked chl, ms & bt, opq grains & cement, clay & cc cements Very angular-angular, mod well cg-vcg sorted, immature, mono- & polycrystalline qtz, weathered fsp, kinked bt, ms & chl, clay clasts, minor lithic (volcanic) clasts, opq grains, matrixsupported cc cement Very angular-angular, mod well vcg sorted, immature, mono- & polycrystalline qtz, weathered fsp, kinked bt, ms & chl, clay clasts, minor lithic (volcanic) clasts, opq grains, matrixsupported cc cement Very angular, mod sorted; poly- & c g - v c g monocrystalline qtz, weathered fsp, aligned & kinked ms, bt & chl, opq grains, some opq cement, clay & cc cements (iron stained), minor lithic clasts

Very angular, rood sorted; poly- & monocrystalline qtz, kinked bt, ms & chl, extensive clay & qtz cement with minor cc and opq cements, volcanic clasts

Massive units (1-1.5 m) interbedded with shaly horizons (up to 0.5 m thick); crossbedding in some units; vcg, v mica rich, v. fissile & crumbly

Hard, massive units, some cross-bedding, some pebbly layers

Hard, massive units, some cross-bedding, some pebbly layers

Massive sst beds with some partings at cm scale

Hard, massive units and channel cross-bedded units; pebbly sst layers; characteristic grey colour; cc veining Some massive beds, some cross-bedding and channel deposits, erosional surfaces

Grey

(Continued)

Reddish grey

Very dark grey

Volcanic, qtz

Lithic

Light grey

Grey claystone, volcanic

Quartzite, Weak red sandstone, clay, granite in variable amounts; clay rip-up clasts Weak red

Quartzite, lithic

t,o

t"

t"

<

©

r~ © tO ¢) .¢ ©

NO 0535 1885

NO 0535 1885

NO 0581 1917

NO 2728 3917

Garvock Gp Scone Fm (Dupplin Estate Quarry)

Garvock Gp Scone Fm (Dupplin Lodge Quarry)

Garvock Gp Scone Fm (Hill of Baldowrie Quarry)

Grid reference

Garvock Gp Scone Fm (Dupplin Estate Quarry)

Stratigraphic group and formation (location)

Table 1. Continued

Sublitharenite

Litharenite

Litharenite

Litharenite

Petrological classification

Grain size

Very angular-subrounded, poorly vcg sorted, immature; poly- & monocrystalline qtz, weathered fsp, aligned & kinked ms, bt & chl, opq grains & cement, clay & qtz cement (hem-stained), lithic (volcanic & sed) clasts Very angular-angular, poorly cg sorted, v. immature, poly- & monocrystalline qtz, weathered fsp, kinked bt & ms, calc & clay cement, rim and late interstitial opq cement

Angular-subrounded; poly- & cg monocrystalline qtz, weathered fsp, aligned, kinked ms, bt & minor chl, opq grains & cement and hem-stained clay cement, lithic clasts

Very angular, poorly sorted, cg immature; poly- & monocrystalline qtz, weathered fsp, v minor bt, ms & chl, opq grains cement & rims, clay & qtz cements, lithic (volcanic) clasts

Petrological characteristics

Clast composition

Massive and laminated Claystone units; some reduction spots

Massively bedded, some Lithic pebble-rich layers, becoming very fissile towards top of quarry; base of quarry has potential for large slabs Massively bedded, some Lithic pebble-rich layers, becoming very fissile towards top of quarry; base of quarry has potential for large slabs Massive sst with finer Lithic interbedded fissile units; some pebbles (0.02-0.06 m) of qtz

Fabric

Weak red

Grey

dusky red

Olive

Colour

Litharenite (conglomerate)

Garvock Gp Scone NO 5145 5548 Fm (Nine Wells Quarry 1, Finavon Hill)

Sublitharenite

Sub-arkose

NO 3930 5460

Garvock Gp Scone Fm (Kirriemuir Quarry)

Calcareous sublitharenite

Garvock Gp Scone NO 5145 5548 Fm (Nine Wells Quarry 1, Finavon Hill)

NO 3930 5460

Garvock Gp Scone Fm (Kirriemuir Quarry)

vcg

Very angular, v. poorly sorted, immature, poly- & monocrystalline qtz, weathered fsp, major kinked bt & ms, minor chl, opq rim and cement, extensive clay cement

Angular, well sorted, poly- & fg monocrystalline qtz, weathered fsp, aligned, kinked bt, ms, chl, opq grains, rim and cement, interstitial clay cement Angular-subrounded, v. poorly vcg sorted, immature, predominant clay matrix with volcanic clasts, poly- & monocrystalline qtz, weathered fsp, very minor bt, ms & chl

cg

Very angular, v. poorly sorted, immature, poly- & monocrystalline qtz, weathered fsp, major kinked bt & ms, minor chl, opq rim and cement, calc cement

massive, crossbedding and channels; some intercalated fissile units

(0.1 - 1.0 m);

Major sandstone units; m - c g sandstone with granule-pebble clasts (to 45 x 70 mm); pebbly and conglomeratic units and pebble-filled channels; clay rip-up clasts associated with conglomeratic beds; 0 . 5 - > 1 m thick units; coarse current cross-bedding Major sandstone units; Lithic m - c g sandstone with granule-pebble clasts (to 45 x 70 mm); pebbly and conglomeratic units and pebble-filled channels; clay rip-up clasts associated with conglomeratic beds. 0.5- > 1 m thick units; coarse current cross-bedding Massive sst interbedded Quartzite, with more fissile igneous units; some ripple pebbles lamination and erosional surface Thick sst units Igneous pebbles

(Continued)

Weak red

Weak red

Weak red

Weak red; some reduction spots

Arkose

Arkose

Arkose

Calcareous sublitharenite

NO 5145 5548

NO 5 ! 20 5536

NO 5771 3494

Garvock Gp Scone Fm (Nine Wells Quarry 1, Finavon Hill)

Garvock Gp Scone Fm (Nine Wells Quarry 2, Finavon Hill)

Garvock Gp Arbroath Sandstone (West Haven foreshore)

Petrological classification

NO 5145 5548

Grid reference

Garvock Gp Scone Fm (Nine Wells Quarry I, Finavon Hill)

Stratigraphic group and formation (location)

Table 1. Continued

Angular-very angular, mod sorted, immature, matrix supported, mono- & polycrystalline qtz, weathered fsp, crude alignment of kinked bt & ms, very minor chl, extensive matrix-supported cc cement, very minor opq

Very angular, mod-poorly sorted. immature; poly- & monocrystalline qtz, weathered fsp, extensive aligned bt, ms & chi; opq grains, rim & cement, clay matrix

Very angular, poorly sorted, immature, poly- & monocrystalline qtz, dominant weathered fsp, crudely aligned bt & ms, very minor chl, extensive clay matrix, extensive opq grains, rim and cement. volcanic clasts Very angular, mod-well sorted; poly- & monocrystalline qtz, relatively fresh fsp, extensive aligned, kinked bt & ms, minor chl, minor opq grains; clay matrix

Petrological characteristics

Massive and crossbedded sst interbedded with more fissile units; some ripple lamination and erosional surface Massive and crossbedded sst interbedded with more fissile units: some ripple lamination and erosional surface Massive sandstone beds on c m - m scale; minor cross-bedding fg

mg

Massive and crossbedded sst interbedded with more fissile units; some ripple lamination and erosional surface

Fabric

cg

Grain size

Lithic

Volcanic

Clast composition

Pinkish grey

Grey

Grey

Grey

Colour

Very angular, poorly sorted, immature, polycrystalline qtz, weathered fsp, v. minor mica, minor volcanic fragments, interstitial clay matrix; red colour caused by opq rim cement

Sub-arkose (greywacke)

Sub-arkose

NO 6821 4450

NO 6821 4450

NO 3090 5010

NO 3090 5010

Garvock Group Red Head Fm (Auchmithie Cliffs)

Garvock Group Red Head Fm (Auchmithie Cliffs)

Strathmore Gp Teith Fm (Airlie Quarry)

Strathmore Gp Teith Fm (Airlie Quarry)

Sub-arkose

Litharenite

Angular, well-sorted, immature, poly- & monocrystalline qtz, minor weathered fsp, aligned mica (mainly ms), clay cement, opq rim and interstitial cement Very angular-angular, poorly sorted, v. immature, poly- & monocrystalline qtz, weathered fsp, kinked bt & ms, calc & clay cement, rim and late interstitial opq cement Very angular, poorly sorted, immature, polycrystalline qtz, weathered fsp, v. minor mica, minor volcanic clasts, interstitial clay matrix; red colour caused by opq rim cement

Arkose

NO 5276 5508

Garvock Gp Scone Fm (Woodside Quarry, Aberlemno)

Angular, poorly sorted, matrixsupported, immature, mono- & polycrystalline qtz, weathered fsp, kinked bt & chl, minor ms, angular lithic (volcanic) clasts, cc (matrix-supported) cement, minor interstitial and grain opq Angular-subangular, mod well sorted; poly- & monocrystalline qtz, weathered fsp, aligned, kinked ms, bt, chl, opq cement & grains, clay & qtz matrix

Calcareous litharenite

NO 5774 3492

Garvock Gp Arbroath Sandstone (West Haven foreshore)

Weak red

Weak red

Weak red

Quartzite, jasper, granite, claystone Quartzite

Claystone Major sandstone units; m g - c g sandstone with pebbly beds; 0.5- > 1 m thick units; coarse current cross-bedding; intercalated fissile units c. 10-12 cm thick Very variable fabric but some more massive beds with bedding on cm-1 m scale; some cross-bedding, channel crossbedding and erosional surfaces m-cg

m-cg

vcg-cg

(Continued)

Weak red

Light reddish brown

Lithic

Massive and crossbedded sst units, some coarse, pebbly layers; partings in some units on cm scale Massive, cross-bedded and trough crossbedded units; some lamination on cm scale Massive, cross-bedded and trough crossbedded units; some lamination on cm scale

m-cg

fg

Grey

White clay clasts, qtzite, volcanic

Hard, massive units and channel cross-bedded units; pebbly sst layers; characteristic grey colour; cc veining

cg

-4

Angular, well sorted; poly- & monocrystalline qtz, weathered fsp, chl with minor ms & bt, opq grains, some opq cement, mainly clay & qtz cements

Angular, well sorted; poly- & monocrystalline qtz, weathered fsp, aligned, kinked chl with minor ms & bt, opq grains, some opq cement, mainly clay & qtz cements; some banding of micachl & opq-rich layers

Sub-arkose

NN 8929 2393

Strathmore Gp Teith Fm (Cultoquhey Lower Quarry)

Petrological characteristics

Sub-arkose

Petrological classification

NN 8929 2393

Grid reference

Strathmore Gp Teith Fm (Cultoquhey Lower Quarry)

Stratigraphic group and formation (location)

Table 1. Continued

vfg

Grain size

Massive basal beds (to 2.5 m thick) with interbedded laminated sst units and ?cornstone unit; erosional surfaces between beds, channel crossbedding Massive basal beds (to 2.5 m thick) with interbedded laminated sst units and ?cornstone unit; erosional surfaces between beds; channel crossbedding

Fabric

Mud

Mud

Clast composition

Weak red

Weak red

Coiour

GEOLOGY OF SCOTTISH MEDIEVAL SCULPTURE

299

(a) 8 7

-

-

m

5

~4 ~3 l-

~2

ii

I

0.04

0,08

0.12

0.16

0.20

0.24

0.28

0.32

0.36

0.40

0.44

0.48

0.52

0.56

0,60

0.64

0.68

0.72

0.76

0.56

0.60

0.64

0.68

0,72

0.76

0,56

0,60

0.64

0.68

0.72

0.76

0.56

0.60

0,64

0.68

0.72

0.76

0.56

0.60

0.64

0.68

0.72

0.76

Magnetic susceptibility (x10-3SI)

(b) 6 5

~4 ~3 t..-

g2 1

DR 0.04

0.08

0,12

0.16

0.20

0.24

0.28

0,32

0.36

0.40

0.44

0.48

0.52

Magnetic susceptibility ( x l o 3SI)

(c) 6 5 ~4

==3 ~_2 1

CVF3 0.04

0,08

0.12

0.16

0.20

0,24

0.28

I 0

0,32

0.36

0.40

0,44

0,48

0.52

Magnetic susceptibility (x10 .3SI)

0.04

0.08

0.12

0.16

0,20

0.24

0.28

0.32

0.36

0.40

0,44

0.48

0.52

Magnetic susceptibility (x10 .3 SI)

(e) 6 5

~4 2

0

0.04

0.08

0.12

0.16

0.20

0.24

0.28

0.32

0.36

0.40

0,44

0,48

0.52

Magnetic susceptibility (x10 .3Sl)

Fig. 6, Histograms of magnetic susceptibility measurements of possible sandstone source localities in East Central Scotland. (a) Arbuthnott Gp, Dundee Fm. (b) Garvock Gp, Auchmithie Conglomerate Fm. (c) Garvock Gp, Arbroath Sandstone Fm. (d) Garvock Gp, Scone Fm. (e) Strathmore Gp, Teith Fm.

litharenites (containing significant quantities of calcite). The Scone, Red Head and Auchmithie units are a distinctive red colour and are very immature (being poorly sorted and rich in feldspar and mica). The Arbroath Sandstone

units are very characteristic matrix-supported, calcareous litharenites. They are grey in colour and contain very significant quantities of calcite. Magnetic susceptibility measurements in the Auchmithie Conglomerate and Arbroath

300

S. MILLER ET AL.

g, °1

(a) 6

3 ® m~

0

VV vlvvvvvv V V v V ~ ~

Ivlv IVviv 0 04

than granule-sized grains (clasts) also varies. In general, the sculptures can be classified into one of 12 sandstone and one siltstone types.

vivvvvlv 0.08

0.12

(b) 6

016

020

0 24

0.28

0.32

0,36

0.40

0.36

0.40

036

040

Magneticsusceptibility(x10~SI)

g4 ~2 1

MI t ~

0

0 04

0.08

0

0.IM

0.08

0.12

IMIMIMI

0.16 0.20 024 0.28 0.32 Magnetic susceptibility ( x l 0 3 S I )

I DIMnlMnI~I IMnlE IsolMnl 0.12

0.16 0.20 024 028 032 Magnetic susceptibility ( x l 0 ~SI)

Fig. 7. Histogramsof magnetic susceptibility measurements of Pictish sandstone sculpture in East Central Scotland. (a) St. Vigeans sculptures (V), Kirriemuirsculptures (K). (b) Meigle collection (M). (¢) Aberlemnosculptures (A), Fowlis Wester sculptures (FW), Menmuirsculptures (Mn), Easie sculpture (E), DunfallandySculpture (D) and St. Orlands Cross (SO).

Sandstone show a restricted range of values (Fig. 6b and c). The Scone Formation rocks, however, show a much wider range of magnetic susceptibility values (Fig. 6d). Again, it should be noted that the readings in this group varied from sandstones to much finer silty sandstones, and micaceous flagstones.

Strathmore group. These sandstones all belong to the Teith Fm and are generally sub-arkosic sandstones. They are red in colour and contain a significantly higher proportion of feldspar and significant rim and pore-fill opaque (hematite) cement. Strathmore Group rocks generally display low magnetic susceptibility readings (Fig. 6e). Pictish sculpture The geological analyses of the Pictish sculptures indicate that all are sandstones with the exception of one siltstone and one granite. The sandstones all contain quartz, feldspar, mica and opaque minerals, but differ in the relative proportions of these minerals and cement composition as well as textural features such as bedding, cross-bedding and diagenetic structures. The composition and distribution of particles larger

(1) Coarse to very coarse-grained arkose or lithic arkose with clasts of quartz and/or clay and/or siltstone and/or granite. These rocks are typically pinkish grey to red in colour. (2) Fine-grained arkose or sub-arkose. These rocks are typically grey-coloured flagstones. (3) Medium- to coarse-grained arkose or sub-arkose with clay and/or quartz clasts. They are typically dark red to pinkish red in colour and show parallel laminations. (4) Medium-grained arkose or lithic arkose with no clasts and generally pinkish grey to brownish grey in colour. (5) Fine- to medium-grained arkose. Typically pinkish red in colour. (6) Coarse- to very coarse-grained arkose with clay and/or siltstone clasts. Typically pinkish grey in colour and displaying parallel bedding. (7) Fine- to medium-grained arkose. Grey to pinkish grey in coiour. (8) Coarse-grained sub-arkose with minor quartz granules. Pinkish grey in colour. (9) Fine-grained sub-arkose. Distinctive purple colour. (10) Coarse- to very coarse-grained subarkose or lithic arkose with quartz clasts. Typically brown-grey in colour and displays parallel bedding. (11) Fine- to medium-grained arkose. Typically dark reddish brown in colour. Displays cross-bedding. (12) Very fine-grained arkose. Typically pale pinkish grey. (13) Very fine-grained siltstone. Dark olivegrey in colour and fine parallel laminations. Figure 7 is a histogram of magnetic susceptibility measurements from 83 sculptured slabs (representing 996 measurements in total). For the more comprehensive collections (e.g. St. Vigeans and Kirriemuir collections; Fig. 7a) the data show a considerable range of values with no obvious correlation with sandstone types. However, combining the magnetic susceptibility data with the petrological characteristics provides a broader characterization of the sandstone types.

West Highland outcrop The two previously postulated source rocks for many of the Iona and Oronsay sculptures are quarries at Doide, Kintyre and Loch Awe.

GEOLOGY OF SCOTTISH MEDIEVAL SCULPTURE

301

Table 2. Petrology of the potential source rocks from Doide and Loch Awe quarries, West Highlands Location (Grid ref.)

Sample number

Petrological classification

Mineralogy

Fabric

Doide Quarry (NR 7032 7672)

D001

Chlorite schist

qtz, chl, fsp, cc, epi, opq

Doide Quarry (NR 7032 7672)

D002

Chlorite schist

qtz, chl, fsp, cc, epi, opq

Doide Quarry (NR 7032 7672)

D003

Chlorite schist

qtz, chl, fsp, cc, opq

Doide Quarry (NR 7032 7672)

D004

Chlorite schist

qtz, chl, fsp, cc, epi, opq

Doide Quarry (NR 7032 7672)

D005

Chlorite schist

qtz, chl, fsp, cc, epi, opq

Doide Quarry (NR 7032 7672)

D006

Chlorite schist

qtz, chl, fsp, cc, epi, opq

Loch Awe Quarry (NN 0122 1838)

A001

Metadolerite

hbl, ab, chl, qtz, opq

Loch Awe Quarry (NN 1258 2550) Loch Awe (NM 890 039)

A002

Metadolerite

hbl, chl, ab, cc, qtz, opq

Moderately well-defined schistosity (S~ defined by chl) Very well-developed schistosity ($1 defined by chl) Moderately well-defined schistosity (S~ defined by chl); compositional banding Moderately well-defined schistosity (Sj defined by chl) Moderately well-defined schistosity (S~ defined by chl) Moderately well-defined schistosity ($1 defined by chl) Moderately developed schistosity; compositional banding Poorly developed fabric

A003

Metadolerite

ab, hbl, chl, epi, qtz, opq

Poor alignment of chl

Samples from both localities have been petrologically classified (Table 2) and magnetic susceptibility measurements taken (Fig. 8a). The rocks cropping out at Doide are chlorite calcite schist all showing well-developed fabric. Their mineralogy is dominated by feldspar (albite), quartz, calcite and accessory opaque minerals with fabric developed by chlorite. Their mineralogy and relict compositional banding indicates that they are metasedimentary. Those cropping out at Loch Awe are hornblende schist with brown-green hornblende as the principal mineral and subsidiary feldspar, quartz and opaque accessory minerals ___ chlorite + epidote. Composition suggests that the Loch Awe schist is meta-igneous. Fabric is less well developed in the Loch Awe schist than in the Doide schist. Figure 8a is a histogram of magnetic susceptibility measurements of schists from both Doide and Loch Awe quarries. Although close in average values, the two quarries have different magnetic susceptibility characteristics: the Loch Awe schist has lower values ((0.450.7) x 10-3SI) and the Doide schists has higher values ((0.75-1.25) x 10 -3 SI).

West Highland sculpture Based on the examinations carried out, there are a significant number of macroscopic differences in the stone types used for the sculptures at all of the sites examined. The rocks used for the Iona sculptures include slate, a number of varieties of schist (mica schist, chlorite schist with etched carbonate pits, chlorite schist without etched carbonate pits, garnet chlorite schist, garnet biotite schist and talc schist), metabasite, metadolerite, metagranodiorite, trachyte, gabbro and sandstone. The rock types used for the Oronsay sculptures are more restricted and include mica and chlorite schist, pelitic mica schist, metabasite and sandstone. At both these sites the most commonly used rock type is schist. The notable difference between Iona and Oronsay is that Iona has a greater variety of schist types with the exception of a distinctive pelitic mica schist that appears only on Oronsay. Figure 8b is a histogram of magnetic susceptibility measurements on 61 carved slabs from Iona (representing 732 measurements in total). There is a considerable spread of data although the majority of the schists and metabasites

302

S. MILLER ET AL

(a)

• Doide [ ] Loch

7 >,

6

group with ((0.45-0.55) carved from group with ((0.75-0.95) ite schist and

5

g ,2_ 3 2 1 0

0.1

0.2

03

04

05

o.6

07

08

09

10

1t

1.2

medium magnetic susceptibility x 10-3SI) represents sculpture pelitic mica schist whereas the higher magnetic susceptibility × 10 -3 SI) is dominated by chlormica schist lithologies.

13

Magnetic susceptibility ( x l 0 ~ S i )

(b) ~, 23 22 21 20 19 18 17 16 15

~

• Slate [][] [] [] [] [] []

Discussion

i1

Schist Metabasite Trachyte Gabbro Metadolerite Metagranodiotie Sandstone

Pictish sculpture

I

12

f..-.::.::

11 10

:::.1

9

81

:::.:

7 6

0.1

02

0.3

0.4

0.5

06

07

0.8

09

to

11

12

13

10

11

1.2

1.3

Magnetic susceptibility (x 10 ~Sl)

(c) l'J (P) Peliticmicaschist 9 8

6

(Oronsayonly) [] Micaschist [] Metabasite [] Sandstone

4 3 2 1

01

0.2

0.3

OA

0.5

06

0.7

i-lh 08

09

Magnetic susceptibility (xlO ~Sl)

Fig. 8. Histograms of magnetic susceptibility measurements of late medieval sculpture in the West Highlands of Scotland and possible source localities. (a) Doide and Loch Awe schists. (b) lona sculpture. (c) Oronsay sculpture.

show a clear grouping with higher magnetic susceptibility ( ( 0 . 7 5 - 1 . 1 5 ) × 10-3SI). There is one sandstone sculpture with very low magnetic susceptibility (<0.05 × 10-3SI). The slates tend to exhibit moderately low values ((0.250 . 5 ) × 10 - 3 SI). Figure 8c is a histogram of magnetic susceptibility measurements on 29 carved slabs from Oronsay (representing 348 measurements in total). Here the data fall into three similar groups. The low magnetic susceptibility sculpture is of sandstone (average magnetic susceptibility of 0.05 x 10 -3 $1). The other two groups are both dominated by schist but the

From the geological survey, sampling and thinsection analysis of rocks cropping out in the area (Table 1), it is clear that there are distinctive lithological differences between stratigraphical groups and even between formations. This is encouraging for potential identification of sources of carved stone. However, there is also considerable variation within groups, even on an outcrop scale. In addition, there has been considerable change in landscape since medieval times and many potential quarry sites are infilled and are no longer accessible for sampling. Nevertheless, the samples collected provide a comprehensive set of comparative material from potential sources in the Lower Old Red Sandstone of the area. This reference set is part of the permanent collections of the National Museums of Scotland (NMS.G.2003.9). Comparing the magnetic susceptibility of the sculpture sandstone types (Fig. 6) with that of the potential sources (Fig. 7), the sculpture rocks are consistent with local sources in the Lower Old Red Sandstone and would indicate that a number of different geological units from within the Lower Old Red Sandstone have been utilized for the procurement of the stones. In general it is not possible to suggest a source even at stratigraphical group level from the magnetic susceptibility values alone. For example, the St. Vigeans sculpture that has magnetic susceptibility_ values in the range of (0.040.08) x 1 0 - 3 S I corresponds to the values displayed by Garvock Gp Scone Fm, Garvock Gp Arbroath Sandstone Fm and Strathmore Gp Teith Fm. However, when combined with petrological evidence, it is possible to assign individual sculptures to a potential source from a particular stratigraphic formation (Miller & Ruckley 2005). It is also possible to discount some possible sources, e.g. the sandstone characteristic of Arbuthnott Gp Dundee Fm sandstone cropping out at Dodd quarry (Fig. 7a) has magnetic susceptibility measurements greater than any measured in the sculpture (>0.44 x 10 -3 SI). The feasibility of local production of the sculptures in the vicinity of the medieval sculpture sites is supported by local (historical)

GEOLOGY OF SCOTTISH MEDIEVAL SCULPTURE quarrying evidence. There are a number of sandstone quarries in the immediate area of each of the sites visited and a detailed study of the history of quarrying throughout the area suggests more numerous early working quarries across the region. Some quarries have also been identified as providing stone for buildings of the same age as the sculpture. In addition, the nature of the Lower Old Red Sandstone units would allow very local, non-quarry sources such as outcrops in river-cuttings, or even in drift deposits, to be utilized for the production of stone.

West Highland sculpture Combined magnetic susceptibility data and petrological classifications for Iona and Oronsay sculptures indicate that there is a good correlation between different rock types and average magnetic susceptibility measurements (Fig. 8). In particular, sandstone (both Iona and Oronsay), slate (Iona) and pelitic mica schist (Oronsay) have consistently low to medium magnetic susceptibility measurements whereas the other rock types, including all the other varieties of schist, exhibit higher average values. Again, it is interesting to note that magnetic susceptibility measurements alone cannot be used to distinguish certain rock types. For example, based on petrological examination of the sculpture on Iona there are at least four distinct schist types but they cannot be resolved in terms of their magnetic susceptibility characteristics. Indeed, the magnetic susceptibility characteristics of pelitic and some meta-igneous rocks are indistinguishable with the equipment used in this study. Based on the petrological evidence, it can be concluded that the rock types used for sculptures on Iona and Oronsay cannot be locally derived. In addition, they cannot have a single source location, as the range of rock types used indicates that a number of quarry localities must have been utilized. However, all of the rock types identified could have been sourced within the West Highlands. Steer & Bannerman (1977) proposed that West Highland sculpture developed in the first half of the 14th century (the so-called 'Iona School'). Some time after the establishment of this school of carving, others were established in Kintyre, around Loch Sween, on Loch Awe and on the island of Oronsay (Fig. 5). Steer & Bannerman (1977) suggested a possible source for many of the Iona and Oronsay sculptures as Doide 'quarry' on the Kintyre peninsula. Another source was cited as quarries at Loch Awe.

303

Combined petrological and magnetic susceptibility data indicate that Doide may be the source of schist for some of the Iona and Oronsay sculpture but is certainly not the only source. In addition, the data are consistent with some sculpture being sourced from the Loch Awe quarries. From the petrological evidence there are, however, schist types that cannot be assigned to Doide or Loch Awe, e.g. schist with characteristic pitted surface (indicative of etched carbonate minerals). In the West Highlands, it is clear that late medieval sculptures or the raw materials for that sculpture were transported significant distances. The fact that transportation from both Doide and Loch Awe, which are coastal localities, could have been effected by sea is significant, as the majority of the sculptures (or uncarved blocks) are of considerable weight.

Conclusions

Pictish sculpture Initial conclusions for early medieval sculpture in Central Scotland are that the majority of the sculpture is located very close to its stone source. It appears that little, if any, large-scale movement of raw materials or carved sculpture took place at this time.

West Highland sculpture From our work here, it is evident that there are many more sources of stone used by late medieval sculptors in the West Highlands than hitherto postulated, and that both immediately local, as well as more distant, sources were exploited. However, is it credible, as has been previously postulated, that, in late medieval times, the carvers imported blocks of stone over long distances for their sculpting? There would presumably have been a significant amount of wastage in the process of shaping and carving the slabs and crosses, best left at the quarry itself. Stone is heavy and expensive to transport, and most craftsmen would surely have wished to have been closely involved with selecting the actual slabs they were going to carve. It therefore seems much more likely that the carvers would have been based at the quarries and that long sea or land voyages transporting raw materials would have been the exception rather than the rule. Work by scholars in England on other medieval grave slabs and covers suggests that such carvers based at quarries would be the norm (Butler 1958, 1964). Further research may reveal a more complex picture of individual

304

S. MILLER ET AL.

craftsmen or workshops based at several quarries, or indeed opening up quarries from time to time to deal with local commissions.

General

As magnetic susceptibility in general can show large variation for any given lithology producing overlapping ranges between rock types (e.g. slate 0-1.2, sandstone 0.35-0.9, basalt 0.35-80, all x 10 -3 SI (Sharma 1982)), it is not possible to assign specific diagnostic values to specific rock types for provenancing purposes. However investigation of potential local sources of rock and the comparison of the magnetic susceptibility range found therein with that measured in the sculpted slabs can be very informative. This study has shown the potential benefits of combining macroscopic petrological identification and magnetic susceptibility measurements of sculptures in characterizing the rock types and comparing these with similarly characterized outcrop specimens. By using this combination of techniques a potential source can be identified with some degree of certainty. However, success in using such methods for sedimentary and metamorphic rocks may be less assured than applying similar techniques to igneous rocks. In the case of sedimentary rocks, sandstone in particular, the low levels of magnetic minerals result in low readings and relatively small ranges of data. Discerning subgroups is therefore difficult. The magnetic susceptibility data are, however, useful in refining rock-type characteristics that have been assigned by petrological analysis. In the case of metamorphic rocks, it may be that metamorphic processes alter the magnetic mineralogy. In addition, magnetic susceptibility may be significantly anisotropic, particularly in metamorphic and sedimentary rocks in which the plane of maximum susceptibility lies in the plane of foliation or bedding plane, respectively (Sharma 1982). To eradicate any resultant effects of this feature, each slab was measured for magnetic susceptibility on several orthogonal faces and the data were then averaged for each slab. This will not, however, combat any effects of metamorphic processes on the overall magnetic properties of a rock. Indeed, Williams-Thorpe & Thorpe (1993), in their provenancing study of granite columns, showed that the most dispersed group in terms of magnetic susceptibility values contains mainly foliated granites. Further work, with more detailed analysis of the magnetic properties of sedimentary and metamorphic rocks, is required.

More detailed petrological examination of the sculptures (e.g. from thin sections) would yield more detailed comparisons and potentially more robust matches between sculptures and source rocks. To date, however, it has not been possible to gain permission for destructive analysis of the sculpture. In addition, consideration of the carving properties of the different stone types relative to the type of sculpture produced may indicate a preference for particular materials depending on the desired outcome. There is clearly much more research to be done on medieval sculpture in Scotland. Future work should include chemical analyses (e.g. portable X-ray fluorescence), more petrological analyses (e.g. detailed thin-section petrology of selected sculpture from micro-cores), further identification of additional sources of raw materials, and a complete reassessment of the stylistic characteristics of the carvings. In terms of the art historical interpretation of stone sculpture, consideration of the origin of the stones is central to the thinking about economic implications for the procurement and movement of raw materials and about the place of sculpting. This can also help to assess the importance of links between sites. Petrological techniques have a place in refining models of procurement and movement of stone for sculpture during the medieval period. The authors would like to thank S. Stevenson for preparation of hand specimens and thin sections, and D. Mitchell for preparation of digitized geological maps and diagrams. The authors would also like to acknowledge the kind assistance given by staff at the various organizations and institutions whose collections were examined as part of this study. C. Graham is thanked for his valuable observations on the regional geology of the West Highlands of Scotland. The survey of Pictish sculpture was carried out in collaboration with Historic Scotland, and the survey of the West Highland sculpture was made possible by a scholarship award from the Friends of the National Museums of Scotland.

References ARMSTRONG, M. & PATERSON,I. B. 1970. The Lower Old Red Sandstone of the Strathmore Region. Institute of Geological Sciences, Natural Environment Research Council Report, 70/12. BUTLER, L. A. S. 1958. Some early Northern grave covers--a reassessment. Archaeologia Aeliana, 4th Series, 36, 207-220. BUTLER, L. A. S. 1964. Minor medieval monumental sculpture in the East Midlands. Archaeological Journal, 121, 111 - 153. CONYBEARE, W. D. & PHILLIPS,W. 1822. Outlines of the Geology of England and Wales. William Phillips, London.

GEOLOGY OF SCOTTISH MEDIEVAL SCULPTURE CRUDEN, S. 1964. The Early Christian and Pictish Monuments of Scotland. Official Guide. HMSO, Edinburgh. FLOYD, J. D. & TRENCH, A. 1988. Magnetic susceptibility contrasts in Ordovician greywackes of the Southern Uplands of Scotland. Journal of the Geological Society, London, 145, 77-83. FOLK, R. L. 1974. Petrology of Sedimentary Rocks, Hemphills, Austin, TX. MARKHAM, M. 1997. Geology and archaeology: a search for the source rock used by British Neolithic axe makers. Open University Geological Society Journal, 18(3), 48-57. MCKERROW, W. S., LAMBERT, R. St. J. & COCKS, L. R. M. 1985. The Ordovician, Silurian and Devonian periods. In: SNELL1NG, N. J. (ed.) The Chronology of the Geological Record. Memoirs of the Geological Society, London, 10, 73-83. MILLER, S. & RUCKLEY, N. A. 2005. The Role of Geological Analysis of Monuments: a Case Study from St. Vigeans and Related Sites. In: FOSTER, S. M. & CROSS, M. (eds) Able Minds and Practised Hands: Scotland's Early Medieval Sculpture in the 21st Century. Society for Medieval Archaeology Monograph 23. PEACOCK, D. P. S. 1995. The 'Passio Sanctorum Quattuor Coronatorum': a petrological approach. Antiquity, 69, 362-368. PEACOCK, D. P. S. 1997. Charlemagne's black stones: the re-use of Roman columns in early medieval Europe. Antiquity, 71, 709-715.

305

SHARMA, P.V. 1982. Geophysical Methods in Geology. Elsevier, New York. STEER, K. A. & BANNERMAN, J. W. M. 1977. Late Medieval Monumental Sculpture in the West Highlands. The Royal Commission on the Ancient and Historical Monuments of Scotland, Edinburgh. WILLIAMS-THORPE, O. & HENTY, M. M. 2000. The sources of Roman granite columns in Israel. Levant, 32, 155-170. WILLIAMS-THORPE, O. & THORPE, R. S. 1993. Magnetic susceptibility used in non-destructive provenancing of Roman granite columns. A rchaeometry, 35(2), 185-195. WILLIAMS-THORPE, O., JONES, M. C., TINDLE, A. G. & THORPE R. S. 1996. Magnetic susceptibility variations at Mons Claudianus and in Roman columns: a method of provenancing to within a single quarry. Archaeometry, 38, 15-41. WILLIAMS-THORPE, O., TINDLE, A. G. & JONES, M. C., 1997. Characterisation studies: magnetic susceptibility. In: PEACOCK, D. P. S. & MAXFIELD, V. A. (eds) Mons Claudianus, Survey and Excavations 1987-1993, Vol. 1. Topography and Quarries. Institut Francais d'Archfologie Orientale, Cairo, 287-313. WILLIAMS-THORPE, O., JONES, M. C., WEBB, P. C. & RIGBY, I. J. 2000. Magnetic susceptibility thickness corrections for small artefacts and comments on the effects of 'background' materials. Archaeometry, 42(1), 101-108.

Geochemical and petrographic approaches to chert tool provenance studies: evidence from two western USA Holocene archaeological sites M A U R Y M O R G E N S T E I N 1'2

1Archaeological Research Facility (ARF) and Near Eastern Studies Department, 250 Barrows Hall, University of California, Berkeley, CA 94720-1940, USA (e-mail: [email protected]) 2Geosciences Management Institute, Inc., 1000 Nevada Highway, Suite 106, Boulder City, NV, USA Abstract: Chert (flint, jasper and agate) from a late Holocene site in Elko, northern Nevada, USA, and a Mid-Holocene site east of Seattle, Washington, USA, represents a wide range of geological source environments for microcrystalline polymorphs of quartz. Potential chert source material and lithic artefacts from these two archaeological sites are utilized to develop laboratory approaches to provenance studies. Chert paragenesis determined by petrographic analysis and geochemical cluster fingerprinting using inductively coupled plasma (ICP) and ICP-mass spectrometry analyses accompanied by scanning electron microscope-energy dispersive X-ray fluorescence analysis provides compelling source to artefact correlations.

Chert is one of the most common raw materials used to manufacture stone tools such as projectiles, sickle blades, knife blades, choppers and scrapers. It is also, however, one of the most difficult rocks to source or characterize as it is a microcrystalline to non-crystalline composite of polymorphs of quartz that are common in the geological environment. Often these silica polymorphs, which consist of chalcedony, opalA, opal-C, opal-CT, moganite, microcrystalline quartz (tridymite or cristobalite) and megaquartz, do not have simple point source occurrences. The amorphous varieties appear to be confined to opal (Graetsch & Ibel 1997). Accurate characterization of these minerals is best accomplished by infrared (IR) absorption spectrometry, electron diffraction using a transmission electron microscope (TEM), or X-ray diffraction (XRD) (Graetsch et al. 1987). It is proposed here that mineralogical characterization by itself is not sufficient for sourcing chert and that the addition of geochemical analysis is desirable. Optical examination by polarized light microscopy is beneficial in characterizing the periodicity and crystallographic orientation of extinction bands (Runzelbanderung, which are zoned striped and rhythmic extinction banding) common to fibrous chalcedony, and for distinguishing grain boundaries between the observable quartz polymorphs (Bernauer 1927; Graetsch et al. 1987). But perhaps more important for provenance studies

than characterization of the siliceous chert minerals is the use of optical petrography to characterize the parent rock fabric that may contain diagnostic components. Such components might include the following: sedimentary mineral detritus; microfossil fragments; vapour inclusions; flow, fracture filling and metamorphic deformation structures; non-siliceous authigenic minerals; pseudomorphing minerals; entrapped plutonic, volcanic and hydrothermal wall-rock minerals (Folk & Weaver 1952; Folk 1968; Folk & McBride 1976; McBride & Folk 1977). The presence of such structures and minerals provides a means to create a fabric classification for chert, which greatly aids the process of provenance determination. Chert is used in this paper as a rock term to encompass the entire suite of sedimentary, igneous and metamorphic siliceous rock types where cryptocrystalline, microcrystalline, and non-crystalline polymorphs of quartz occur in a composite rock fabric (Folk & Weaver 1952; Folk 1968). It includes but is not limited to flint, jasper, agate, petrified wood, and lithophysae fillings. Previous studies of cryptocrystalline minerals (e.g. Frondel 1962) utilized the term chert for sedimentary material only; this is not the case here. Suites of microcrystalline siliceous mineral assemblages form as a result of numerous emplacement conditions and mechanisms, including the following siliceous fluid deposition in volcanic vesicles or vugs; fracture

From: MAGGETTI,M. & MESSlGA,B. (eds) 2006. Geomaterialsin CulturalHeritage. Geological Society, London, Special Publications, 257, 307-321. 0305-8719/06/$15.00 © The Geological Society of London 2006.

308

M. MORGENSTEIN

fillings; lithophysal cavity fillings in tuffaceous rocks; siliceous replacements of organic material in aqueous environments; mobilization of siliceous paste in sandstones, mudstones and limestones (to form nodules and bedded novaculitic sequences); devitrification of volcanic ash deposits; bacterial and nanobacterial activity. All of these microcrystalline polymorphic formational fabrics are lumped together as chert, because once the actual rock has been removed from its source area, the identification of that rock is often not attainable with only visual inspection (Folk 1968, p. 81). For example, jasper is often considered a coloured sedimentary novaculite, in addition to being considered a coloured metamorphic material (Frondel 1962). Once the chert has been worked into an artefact, or becomes debitage, visual geological classification is even more problematic. Further, an abundance of colloquial terms (sard, carnelian, moss agate, plasma, prase, chrysoprase, heliotrope, bloodstone, flint and hornstone, among others) are used to classify flint and agate on the basis of folklore or colour and texture; these have no worldwide consistency in usage. Geology Chert from a single-component late Holocene Great Basin site in Elko, northern Nevada, USA, and a multicomponent Mid-Holocene site in the western foothills of the Cascade Mountains near Seattle, Washington, USA, were used to develop laboratory approaches to provenance studies. The Elko, Nevada site, CrNV-12-9371, is located in the Humboldt River drainage system between the Adobe Mountain Range and the Ruby Mountains in a sedimentary high desert provenance associated with crustal spreading. Ostracode-bearing novaculitic chert nodules and beds occur in the Tertiary Elko Formation limestone (Elko Hills Quarry) and muddy to sandy novaculites occur in the Mississippian Joanna Limestone and the Pilot Formation in the Ruby Range. Chert tool fabrics from site CrNV- 12-9371 were confined to novaculites. The Tolt River drainage system in the State of Washington runs through the Western Cascade Mountain Range and out into the Puget lowlands. Site 45KI464 is situated on a high Pleistocene terrace above the Tolt River in the lower Cascade foothills. The Cedar River drainage system is adjacent to the south. The Tolt and Cedar Rivers flow through andesite lavas, breccia and tuffaceous marine sediment sequences and a granitic batholith. Some of these lithologies are metamorphosed, and hydrothermal dykes are scattered throughout the

foothills. The Tolt River site is dominated by a much more complex suite of chert tool fabrics than that observed from the Nevada novaculites. The potential local source materials for this more complex suite consist only of outcrop chert hydrothermal dykes, and Pleistocene drift sediments. The site appears to be a significant chert resource acquisition location as well as the location of a series of workshops. Raw material apparently was acquired during the Mid-Holocene from exposed Pleistocene outwash channel gravels and worked into a variety of tools including biface points of crystallized volcanic rock, petrified wood, hydrothermal and metamorphic chert, and quartz.

Sampling and methods Table 1 provides the sample locations and general classifications for the studied chert (42 samples). The chert samples include both potential source materials and stone tools. The dominant techniques used in the analysis are optical petrography combined with ICP (inductively coupled plasma) and ICP-MS (inductively coupled plasma-mass spectrometry). Scanning electron microscope-energy dispersive X-ray fluorescence (SEM-EDX) work also was undertaken. Petrographic analysis was accomplished on Nikon Labophot-Pol and Leitz Orthoplan petrographic microscopes with a Kodak digital megapixel camera scientific-system tied to a Macintosh computer. The geochemical analyses were performed for 41 elements; 14 of these elements had good detection values above background and are listed in Table 2. The geochemical results for the two sites were utilized in the following manner. (1) Ten elements (Ca, A1, Ba, Sr, Cs, Ce, La, Nd, Th, and U) were employed for the Nevada Elko site for statistical analysis (dendrogram produced using SPSS 10.0) to compare artefacts with potential source area outcrop samples. (2) Three elements (Cs, Th and U) were used in scattergram analyses for the Nevada Elko site. (3) Six elements (Ca, K, U, Fe, Cu and V) were used in scattergram analyses for the Washington Tolt site. (4) Two elements (Ba and S) were used together as a barite fingerprint tracer (SEMEDX) in the Washington Tolt site. (5) Two elements (Ce and Th) were used to compare the chert from the Washington Tolt site with chert from two sites in Oregon (Hess 1996). Tolt chert was analysed by ICP-MS data; the Oregon chert data were obtained by instrumental neutron activation analysis (INAA) (Hess 1996). In addition, INAA was

GEOCHEMICAL AND PETROGRAPHIC CHERT TOOL

309

Table 1. Sample location and general classification Laboratory notation

Field notation

Lithic artefacts, Nevada site CrNV-12-9371 FS4 FS-4, 102 N 116W, surface, biface FS5 FS-5, 94 N 120W, surface, biface FS7 FS-7, 88 N 126W, surface, biface FS 13 FS- 13, 105 N 130W, surface, biface FS16 FS-16, 97 N 135W, surface, biface FS19 FS-19, 60 cm at 120° from SE comer of unit 12, 124 N 115W, scraper or chopper FS21 FS-21, Unit 4, level 10-20 cm, scraper Potential source material, EIko Hills, Nevada, quarry cherts EHQM1 Elko Hills Quarry, undifferentiated EHQM2 Elko Hills Quarry, undifferentiated Elko Hills Quarry, undifferentiated EHQM3 EHQM4 Elko Hills Quarry, undifferentiated EHQM5 Elko Hills Quarry, undifferentiated EHQM6 Elko Hills Quarry, undifferentiated Lithic artefacts, Tolt River, King County, Washington, site 45K1464 12 (petrography only) Trench 1, bag 1212.3, grey brecciated chert, Fe-radiating 13 (petrography only) Block 1, bag 1181.3, 80-90 cm, chalcedony spotted chert Block 1, bag 1475.3, black micro-web chert 14 (petrography only) Square 1, bag E2034.3.3, lithophysal jasper, yellow mottled 15 16 Block 3, bag 2928.3, jasper with clear chalcedony Block 3, bag 2223.3, petrified wood, black-brown, clear 17 18 Block 2, bag 1006.3, purple banded jasper 19 Trench 3, bag 2206.3, clear chalcedony, minor banded 20 Block 4, bag 4096.3, red banded jasper 21 Block 4, bag 4096.3, red jasper, black web 22 Block 4, bag 4310.3, yellow-brown jasper 23 Block 4, bag 4259.3, red-black jasper, black web 24 Block 2, bag1203.3, petrified wood, grey-brown Block 4, bag 4134.3, red-brown jasper, quartz fracture fills 25 26 Block 4, bag 3435.3, grey chert-clear chalcedony Potential source material, Tolt River, King County, Washington, site 45K1464 S1 Pleistocene drift, T04, quartz fracture-filled red jasper $2 Pleistocene drift, TPS1, white chert-red jasper, quartz $3 Pleistocene drift, TPS2, quartz fracture-filled red jasper $4 Pleistocene drift, TPS3, quartz fracture-filled red jasper $5 Pleistocene drift, TPS4, grey chert $6 Pleistocene drift, 1534, grey petrified wood $7 Pleistocene drift, V4702, yellow-dark grey petrified wood Potential source material, Tolt Watershed, Washington Outcrop, AQ1, red jasper with quartz-filled fractures S13 S14 Outcrop, TWC-1, purple banded jasper Potential source material, Cedar Watershed, Washington $8 Pleistocene drift, CWMC-3, grey petrified wood $9 Pleistocene drift, CWMC-4, grey petrified wood S10 Pleistocene drift, CWMC-5, brown petrified wood S 11 Outcrop, CWMC- 1, dull yellow jasper with red mottling S12 Outcrop, CWMC-2, yellow-brown jasper

used for two of the Tolt samples (16 and 17), with I C P - M S data, to compare the two methods for Ce and Th. The resulting comparative analysis shows similar values that are

less than ideal because of the high INAA detection levels: (A) Sample 16, Th: I C P - M S 0.2 ± 0.1 ppm and INAA < 0 . 5 + 0.5 ppm; (B) Sample 16, Ce: I C P - M S 1.2 _ 0.1 ppm and

0.05 0.07 0.05 0.04 0.04 0.02 0,03 0.02 0.05 <0.01 0.04 <0.01 0.09 0.05 0.06 0.1 0,04 0.03 0.03 0.03 0.03 0.04 4.57 5.4 0.12 0.11 1.18 0.05 0.40 0.04 0.11 0.04 0.20 0.06

EHQM1 EHQM2 EHQM3 EHQM4 EHQM5 EHQM6 FS4 FS5 FS7 FSI3 FSI6 FSi9 FS21 SI $2 $3 $4

K ICP % 0.01

Ai 1CP % 0.01

Method: Analysis unit: Detection limit:

Mn ICP ppm 10

V ICP ppm 10

Fe ICP % 0.01

0.09 43 < 10 0.04 0.09 40 < 1 0 0.04 0.05 I 1 < 10 0.03 0.03 < 1 0 < 1 0 <0.01 0.06 20 < 1 0 0.02 0.09 I 1 < 10 0.02 11.9 352 < 1 0 0.04 4.66 136 13 <0.01 0.12 11 < 10 0.04 0.05 < 1 0 < 1 0 0.07 0.07 38 < 1 0 <0.01 0.13 26 24 0,37 0.08 12 18 0.02 0.50 199 38 3.55 1.06 288 110 1.74 0.04 134 13 0.59 0.10 154 166 7.31

Ca ICP % 0.01 15 14 17 -10 10 12 -10 I1 - 10 -10 -10 12 <10 14 16 <10 33

Sr ICP ppm 10

< 10 24 <10 23 < 10 21 14 13 <10 20 < 10 34 < 1 0 1590 <10 282 < 10 23 <10 10 <10 26 12 344 <10 15 39 n.d. 13 n.d. <10 n.d. 23 n.d.

Ni Cu ICP ICP ppm ppm 10 10

Table 2. ICP and ICP-MS geochemical analyses of the chert

4.2 13.2 0.9 0.6 1.2 0.7 0.4 0.3 0.2 0.2 0.3 37.8 0.8 n.d. n.d. n.d. n.d.

La ICP-MS ppm 0.1 2.1 4.2 1.6 0.9 0.9 0.4 0.7 0.6 0.3 0.3 0.4 67.4 1.3 9.5 82.3 1.6 1.9

Ce ICP-MS ppm 0.1 2.9 8.7 0.8 0.4 0.8 0.5 0.5 0.2 0.1 0.1 0.1 28.6 0.6 8.7 90.2 1.8 1.2

Nd ICP-MS ppm 0.1

0.4 1.2 0.2 <0.1 0.2 <0.1 0.1 <0.1 <0.1 <0.1 <0,1 3.1 <0.1 2 11.4 0.3 0.6

Dy ICP-MS ppm 0,1

I 2 0.8 0.5 0.4 0.2 0.1 0,2 0.1 0.1 0.1 12,2 0.3 1.7 0.6 0.1 0.3

Th ICP-MS ppm 0.1

5.2 2.2 I.I !.4 2.4 1.5 0,9 3.9 1.4 0.5 0.4 3.5 9.8 0.3 0.8 <0.1 0.2

U ICP-MS ppm 0.1

I 0.1 0.6 0.3 0.6 0.4 0.4 0.4 0.5 0.1 0.5 2.7 0.2 n.d. n.d. n.d. n.d.

Cs ICP-MS ppm 0.1

378 319 372 80 135 730 297 151 6.2 66.6 331 >4000 57.6 174 621 64.1 55.6

Ba ICP-MS ppm 0.1

Z

Z

0.50 0.21 2.28 0.19 0.14 0.48 0.05 0.07 1.00 0.70 0.07 0.04 0.17 0.82 0.07 0.09 0.26 0.09 0.20 0.09 0.07 0.09

n.d., no data, not analysed,

$5 $6 $7 $8 $9 SIO Sll S12 S13 S14 15 16 17 18 19 20 21 22 23 24 25 26

0.24 <0.10 186 0.04 0.04 27 0.63 0.22 710 0.03 0.03 14 0.03 0.02 12 0.07 0.03 49 0.06 0.03 74 0.05 0.04 77 0.10 0.63 70 0.32 0.27 259 0.02 0.01 17 0.02 <0.01 11 0.01 <0.01 17 0.28 0.22 71 0.04 0.02 11 0.02 0.03 82 0.09 0.04 108 0.04 0.03 85 0.05 0.03 128 0.03 0.03 35 0.04 0.03 35 0.05 0.03 < 1 0

11 <10 37 <10 <10 29 49 68 185 29 38 80 14 129 < 10 280 310 196 211 15 10 24

0.35 1.84 1.82 0.07 0.04 2.13 1.80 3.27 3.13 0.9 0.05 0.23 0.15 2.74 0.05 6.48 5.1 6.47 7.23 0.19 3.63 0.17

13 < 10 26 <10 <10 <10 26 40 27 23 15 <10 I0 <10 <10 24 16 21 19 <10 10 <10

<10 <10 24 <10 <10 <10 13 35 68 15 32 <10 428 27 <10 17 15 38 39 <10 < 10 <10

n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d.

n.d. n.d. n,d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. 2.9 12.9 23.9 0.9 0.3 1.5 0.6 2.8 8.0 4.4 1.0 1.2 0.9 6.6 0.1 1.5 0.8 0.9 2.3 0.5 0.3 1.3

1.6 14.1 16.7 0.7 0.3 1.5 0.9 2.1 4.7 4.3 0.8 0.8 0.6 1.8 <0.1 1.4 0.7 0.8 2.1 0.5 0.3 0.9

0.2 2.1 4.4 <0.1 <0.1 0.2 0.3 0.9 1.2 0.8 3.6 0.2 0.1 0.6 <0.1 0.8 0.4 0.8 1.3 <.01 <0.1 0.2

0.4 0.2 3.3 <0.1 <0.1 0.1 <0.1 <0.1 0.2 0.6 4.4 0.2 0.2 0.3 <0.1 0.2 0.6 0.5 0.3 0.1 0.2 0.2 0.3 <0.1 0.8 0.2 0.2 0.3 0.2 0.3 0.4 0.6 3.2 0.1 <0.1 <0.1 <0.1 0.3 0.2 0.2 0.2 <0.1 0.3 0.2

n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d, n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d.

62.1 40.2 781 8.4 4.3 20.9 8.8 10.4 77.5 184 7.4 12.5 30.1 220 23.8 11.3 17.0 9.0 10.1 25.3 9.4 32.6

,..q ~ C~

~ ~ ~ ~ ~ t~ ~

M. MORGENSTEIN

312

(a)

I,,,,1[~1100 2

um

,~,~,~,i 100 tJm

(b)

(c)

i,,,,i,~,, lO01Jm

, ik.

(d)

I~,,,~,,,,i lOOpm

~,. ,,,, lO01Jm (f)

(e)

-

. . . ,

-..

.

.

. t i l t . . ,,

.......... ',.,,.,,*

2001Jrn

tP,

,-~..

0

(g)

L~,,,~,,,,i 1001Jm

l 201Jm

!,,,,~,~,!lO01Jm (i)

(h)

I

I

(j) I

120pro

(k) I

J20 IJm (I) l =

..

"

,

=

.

"%.. ..~

IIIII111111100pm

(o)

.

4lb'

dl.

(n)

,20prn

i20 pm (m)

.

;, !~,i lO01Jrn

(P)

.

.

.

.

50pm

Fig. 1. Petrographic microphotographs of chert from the archaeological site CrNV-12-9371. (a-c). Elko Hills Quarry sample EHQM1, novaculite, ostracode-bearing chert. (a) and (b) were taken with plane-polarized light; (c) was taken with cross-polarized light with a quartz wedge. All three photographs are of the same area and show an opalized ostracode shell fragment in a matrix of allogenic clay and calcite ghosts with an authigenic overprint of chert composed of microcrystalline quartz and fibrous chalcedony. (d) Sample EHQM i. A short cross-section of an ostracode with micrite shell infilling and a chert overprint. There is a small vug near the bioclast. (Cross-polarized light.) (e) Sample EHQM 1. A short cross-section of an ostracode shell with a megaquartz, microquartz and fibrous chalcedony shell infilling. (Continued)

GEOCHEMICAL AND PETROGRAPHIC CHERT TOOL INAA < 3 __+ 3 ppm; (C) Sample 17, Th: I C P MS 0.2 _ 0.1 ppm and INAA 0.9 + 0.5 ppm; (D) Sample 17, Ce: I C P - M S 0.9 ___ 0.1 ppm and INAA < 3 ___ 3 ppm. One of the major problems in using geochemistry to characterize chert is that unless the chert is ore-mineralized, it commonly has very low trace element concentrations. For this reason we chose ICP and I C P - M S analyses, which have reasonably good detection sensitivities, to acquire the analytical data. Even so, many of the elements were unusable for analysis because their values fell below detection limits. From the Nevada Elko site a total of seven lithic artefacts and six potential source cherts were studied petrographically and geochemically. From the Washington Tolt site, 26 representative chert artefact and source material samples were analysed both petrographically and geochemically and are discussed here. An additional 34 samples were studied only petrographically; three of these are included in this analysis. S E M - E D X data also were obtained for spatial chemistry on selected Tolt samples analysed by I C P - M S .

Results

Petrography of the potential source materials from Nevada A total of six samples (EHQM samples in Fig. 1a f, i) from the Elko Hills Quarry, which is in the Tertiary Elko Formation and in close proximity to archaeological site CrNV-12-9371, were studied petrographically and geochemically. All six are sedimentary novaculites, dominated by ostracodes (Fig. l a - f ) and iron oxyhydroxide pseudomorphs after framboidal pyrite (Fig. l i). The ostracode valves are sometimes intact and often still composed of calcite. In some cases the calcite is not present and trains of iron oxyhydroxides delineate the carapace boundaries.

313

In other cases the carapace is totally replaced by chalcedony. When the carapace is intact (both valves present) the central portion of the animal is often filled with megaquartz. Bioclastic hash of undetermined origin is generally scattered throughout the matrix of the chert. None of the ostracodes are species identifiable in thin section.

Petrographic characterization of chert debitage and tools from archaeological site CrNV-12-9371, Elko, Nevada Chert from the single-component site in Elko, Nevada, are mostly novaculites dominated by bioclastic hash and ostracodes, chalcedony, microcrystalline quartz, detrital quartz silt, and iron oxyhydroxide pseudomorphs after framboidal pyrite. All of the Nevada chert studied are sedimentary novaculites (Fig. 1 g, h, j - p ) ; two distinctive types are present. The dominant chert is a chert-replaced marl-mudstone grading to an almost pure authigenic chert without observable allogenic components. The second is a chert-impregnated orthoquartzite (quartz arenite) siltstone (Fig. lo and p). The marl-mudstone samples are exceptionally low in iron oxyhydroxide mineralization, which, when present, appears to be a function of very local diagenesis. Many of the chert samples have iron oxyhydroxide pseudomorphs after pyrite either as individual cubes or as grouped framboidal pyrite replacements (Fig. l j - n ) . The pyrites are possibly formed during sedimentary diagenesis with bacterial action on the organic material in the original sedimentary system (Schieber 2002). Most of the chert samples have ostracode bioclastic debris scattered thoughout the fabric (Fig. lg, h, n); this is their most important characterizing feature. Occasional diatom fragments and ghosts and unidentifiable bioclastic fossiliferous debris (Fig. l j) are also present. Opal-A and chalcedony

(Continued)(Cross-polarizedlight.) (f) Sample EHQM1. A fibrous chalcedony vug with very minor amounts of microcrystalline quartz. (Cross-polarized light.) (g, h) Site CrNV-12-9371 lithic artefact FS21. Novaculite, ostracodebearing marl chert. (g) and (h) are of the same area ((g) plane-polarized light; (h) cross-polarized light). Both show an ostracode shell fragment in a matrix of microcrystalline quartz, megaquartz and chalcedony. (i) Elko Hills Quarry chert sample EHQM2. Sample shows a goethite-maghemite pseudomorph of framboidal pyrite. (Planepolarized light.) (j) Site CrNV-12-9371 lithic artefact FS7. Iron oxyhydroxide microspherulites commonly act as replacement mineralogy in fossiliferous debris in the chert. Most of these fossils are not identified. This appears to be a diatom fragment. (Plane-polarized light.) (k n) Site Cr-NV-12-9371 lithic artefact FS13. K-M are goethitemaghemite pseudomorphic pyrite framboids that have formed as bioclastic carbonate shell replacements (photograph (n)). (Plane-polarized light.) (o) Site CrNV-12-9371 lithic artefact FSI9. Quartz arenite fabric with a chert overprint containing subangular to subrounded quartz and feldspar clastics. (Cross-polarizedlight.) (p) Site CrNV-12-9371 lithic artefact FS 19. Quartz arenite fabric with subangular quartz silt clastic material with very minor amounts of clay in a chert cement. (Plane-polarized light.)

314

M. MORGENSTEIN

-

+ :..~

t

(a)

0

4001am

i l l ! i l l

(b)

(c)

50pro

0 I __,'-_~'~--

0

(d)

400pm

.

:

= .

~

0 L~_:

~__Y,~i_L L_L_J

~

~

I

=-

4001.1m i__+J

s

+

• ,

.

•

~

. ~ .

. . -

.t,

'. t -i"

(g)

•

..

(e) 0 I

"

. ~

400pm I I I I

I

(f)

0

I

400pro

...... ~÷-±-±'~-c

,+

0

4001am

tll;iii[

(h)

0

IIIIIIIIIII

lO0 m

Si

S

.it

_

,L

Ba

,A~ .

. . . .

(i) Fig. 2. Petrographic microphotographs of chert from the Tolt site 45KI464. (a) Site 45KI464, lithic artefact 15, volcanic lithophysal chert (chalcedonic agate). This is a feldspar phenocryst that is altered to clay and is situated in a groundmass of chalcedony impregnated with iron oxyhydroxides. Some vapour-phase inclusions are present. (Continued)

GEOCHEMICAL AND PETROGRAPHIC CHERT TOOL replace most of the bioclastic debris, and much of this material also contains scattered concentrations of iron oxyhydroxide spherules (pseudomorphs after pyrite). In some cases, where diagenesis is fairly complete, only the iron oxyhydroxide spherulite concentration alignments remain to identify the ostracode shell fabric (Fig. In). The clay component varies in concentration, which is probably a function of local diagenesis during chert formation and variations in sedimentary facies and stratigraphy. Large sections of the chert are dominated by an (authigenic) microcrystalline quartz and chalcedony fabric that does not contain much original (allogenic) sedimentary clastic detritus. The novaculites contain classic pelletal fabrics that overprint variable allogenic mineral concentrations and are also patterned by vapour-phase and fluid inclusions present in the opal and quartz minerals. Only one sample of the material studied (FS 19, scraper or chopper, Fig. lo and p) falls into the category of a chert-impregnated orthoquartzite siltstone. This siltstone is best classified as a well-sorted, subangular to subrounded quartz arenite. It contains very minor concentrations of K-feldspar, iron oxides and unidentified igneous rock fragments (IRFs). The cement is entirely microcrystalline quartz and chalcedony with some minor opal present. There are a variety of vugs filled with fibrous chalcedony and megaquartz. There are some bioclastic fragments of sponge spicules and what appear to be radiolarian fragments. Neither intact fossil tests nor ostracode fragments were observed. The outcrop origin of this material is at present unknown, but it is similar to chert reported by Hose & Blake (1976) from the nearby Ruby Mountains.

315

Petrography of the potential source materials from Washington Potential source materials for the Washington Tolt site are late Pleistocene glacial drift boulders and cobbles, and fracture-filling hydrothermal dykes composed of jasper. Geochemical and petrographic data were collected from 14 chert source materials. Petrographic examination of these materials indicates that they are composed of: (1) petrified wood (five samples) from Pleistocene drift sediments in the ToIt River and nearby Cedar River drainage systems (Fig. 2b); (2) hydrothermal metavolcanic replacement jasperoid chert (six samples) from Pleistocene drift sediments in the Tolt River drainage ( S E M - E D X data shown in Fig. 2i); (3) j o i n t fault and fracture-fill hydrothermal chert (three samples) from outcrops and Pleistocene drift sediments in the Tolt River drainage system: (a) Type 1, red-banded and black micro-web red and brown jasper; (b) Type 2 chalcedonic jasper and grey chert. Petrified wood pebbles are relatively uncommon in the Pleistocene drift sediments. As a consequence, not many samples could be studied. Petrographic analysis indicates that the preservation of cellular structure probably varies with plant species in addition to other undetermined factors. Two samples from the Tolt River drainage system and three from the Cedar River drainage systems were studied. The sample shown in Figure 2b, from the Tolt River drainage system, is either larch or spruce wood and shows excellent preservation of an axial resin duct. Metavolcanic chert (Fig. 2c) exhibits vesicular textures that are overprinted with chalcedonic and microcrystalline quartz. In many cases, S E M -

(Continued)(Reflected light.) (b) Site 45KI464, sample $6, petrified wood (larch or spruce wood). Tracheid cells with an axial resin duct that has been deformed by folding. The duct has a double wall. (Plane-polarized light.) (c) Site 45KI464, lithic artefact 18, metavolcanic chert. This sample shows pre-alteration vesicular volcanic rock fabric that has been replaced by iron oxyhydroxides along the vesicle walls, and chalcedonic chert with minor amounts of microcrystalline quartz and opal-CT in the interior of the vesicles and also in the vesicle walls. (Plane-polarized light.) (d) Site 45KI464, lithic artefact 22, hydrothermal joint fill chert, Type 1. Yellowish brown jasper with a flow structure that is observable because of the iron oxyhydroxide fabric. The matrix consists of radiating chalcedony grains with minor amounts of microcrystalline quartz and opal-CT. (Plane-polarized light.) (e) Site 45KI464, lithic artefact 14, hydrothermal joint fill chert, Type 1. This is a black micro-web dendritic red jasper with hematite dendrite structures in an opalCT, chalcedony and microcrystalline quartz matrix. (Reflected light.) (f) Site 45KI464, lithic artefact 13, hydrothermal joint fill chert, Type 2. This is a spotted orbicular chert with orbicular fibrous chalcedony containing concentric vapourphase and non vapour-phase, 50-100 Ixm banding. Goethite-maghemite mineralization occurs in the bands, and megaquartz vugs occur between the orbicules. (Reflected light.) (g) Site 45KI464, lithic artefact 26, hydrothermal joint fill chert, Type 2. This is a grey chert dominated by chalcedony with some megaquartz. The iron oxyhydroxide concentrations are blotchy and not present in the vug filling areas that are associated with chalcedonic mineralogy. (Cross-polarized light.) (h) Site 45KI464, lithic artefact 12, hydrothermal joint fill chert, Type 2. A grey brecciated chert is shown consisting of chalcedony, opal-CT, and microcrystalline quartz with a variety of irregular shaped vugs that are lined with fibrous chalcedony. These vugs contain radiating masses of iron oxyhydroxide mineralized chalcedonic fibres along with iron oxyhydroxide banding. (Plane-polarized light.) (i) Site 45KI464, sample S 1, metavolcanic chert, SEM-EDX. The bright microcrystailites in the image are composed of barium sulfate (EDX scan) and are classified as barite.

316

M. MORGENSTEIN

EDX has shown that barite crystals are encapsulated in the chert fabric. Very high concentrations of iron oxyhydroxides occur in this chert. Hydrothermal chert is the most varied of the petrographic fabrics. It has been divided here into two groups: Type 1 and Type 2. Type 1 hydrothermal chert usually has dendritic or banded iron and manganese concentrations that overprint the siliceous portion of the fabric. Type 2 hydrothermal chert is dominated by chalcedony, contains numerous vugs, and has iron and manganese oxyhydroxide concentrations in fairly segregated orbicular or radiating microstructures. No volcanic chert was observed in the local sources.

Petrographic characterization of chert debitage and tools from archaeological site 45K1464, Washington Chert debitage and artefacts from the Washington Tolt site are composed of sedimentary (petrified wood), metavolcanic, hydrothermal and volcanic materials. Petrographic examination of the 15 samples reported here indicates that they consist of: (1) petrified wood (two samples); (2) hydrothermal metavolcanic replacement jasperoid chert (one sample; Fig. 2c); (3) hydrothermal joint-fault, and fracture-fill chert (11 samples): (a) Type 1 red-banded and black micro-web red and brown jasper chert (Fig. 2d and e); (b) Type 2 chalcedonic jasper and grey chert (Fig. 2f-h); (4) volcanic chalcedonic chert (Fig. 2a). The sedimentary and hydrothermal fabrics are similar to the source materials described above and are dominated by opal-CT, opal-A, fibrous and non-fibrous chalcedony, moganite, microcrystalline quartz and megaquartz. The petrified wood fabrics exhibit a variety of states of cellular preservation. Most of the petrified wood is composed of chalcedony and opal-A with iron oxyhydroxides and microcrystalline quartz. The hydrothermal metavolcanic chert is dominated by siliceous overprinting of vesicular volcanic fabrics. These contain large concentrations of iron oxyhydroxide spherules and microcrystallites (jasperoid) and disseminated clays. Fracturing is common, with fracture fills dominated by microcrystalline quartz and megaquartz. The mineralogies are mostly unremarkable with respect to provenance determinations. SEMEDX analyses of the hydrothermal metavolcanic chert indicate that barite crystallites are present in the fabric, but when SEM-EDX scans are made of the sedimentary and volcanic chert no such inclusions are found. Barite is recognized from its EDX signature.

Hydrothermal joint-filling chert has a wide variety of textures within the two chert types. Type 1 chert is dominated by banded and dendritic structures composed of manganese and iron oxyhydroxides in a siliceous matrix. SEM-EDX backscatter analysis suggests that 2 - 5 ~m diameter accessory heavy mineral grains occur scattered among the iron and manganese oxyhydroxides. These microcrystallites contain a variety of transition metals. The genesis of these grains is not known. Type 2 hydrothermal chert has more segregated concentrations of iron and manganese oxyhydroxides (radial and orbicular fabrics) than does Type 1 chert. The siliceous matrix is dominated by chalcedony, with many fabrics showing evidence of brecciation and fracture-filling events dominated by microcrystalline quartz. When rugs are present they are usually filled with megaquartz. Type 2 chert contains heavy mineral microcrystallites (1-10 p~m in diameter) that are dominated by copper, nickel and tin as attested to by SEM-EDX analyses. The volcanic fabrics commonly contain inclusions of feldspar phenocrysts that have been altered to clay, microlithophysal spherules of opal-CT and iron oxyhydroxides, and iron oxyhydroxide crystallites that are not pseudomorphs after pyrite and occur in islandlike patches of dense microcrystallites and spherulites.

Geochemical results for the two sites Trace element geochemical analysis (ICP and ICP-MS) was performed on a total of 39 samples, of which 13 are from the Nevada Elko site and 26 are from the Washington Tolt site. Table 2 reports the geochemical data for both sites, presenting trace element geochemistry for source samples and for lithic artefacts and debitage. Petrographic study of the Nevada novaculites suggests that there are two sources for the material. Figure 3 shows a geochemical dendrogram with the potential source area samples from the Elko Hills Quarry and the lithic artefacts from the Elko site. This dendrogram utilizes 10 elements and shows two significant clusters that conform to the results obtained from petrographic analysis. One cluster is for the ostracode-bearing novaculites and the other (only one sample, FSI9) is for the chert-impregnated orthoquartzite. For Figure 3 data, sample FS4 is a subset of cluster 1. This is a function of the higher Ca, Mg, Mn, and Sr concentrations in sample FS4 (Table 2). These higher elemental concentrations appear to correlate with carbonate diagenesis and

GEOCHEMICAL AND PETROGRAPHIC CHERT TOOL

317

Rescaled distance cluster combine for archaeological site CrNV-12-9371 CA S E 0 Label Num + EHQMI

i

--

EHQM3

3

-

EHQM2

2

-

FS16

11

--

4

--

FS13

10

--

FS21

13

--

EHQM4

FS7

10 +

15 +

20 +

25 +

9

EHQM5

5

--

FS5

8

--

EHQM6

6

--

FS4

7

FS19

5 +--

Cluster

1

Cluster

2

12

Fig. 3. Lithic artefacts from site CrNV-12-9371, and potential source area samples from the Elko Hills Quarry;

geochemistry dendrogram using average linkage between groups for the following 10 elements: Ca, A1, Ba, Sr, Cs, Ce, La, Nd, Th and U. are apparently related to the bioclastic portion of the marl clay. Consequently, the FS4 sample appears to have been obtained from an as yet unidentified lithofacies of the standard Elko Hills Quarry. Further, sample FS4 has a very similar geochemical pattern to sample FS5, which does contain ostracodes. Figure 4 provides two scattergrams for uranium, thorium and caesium for the Elko site limestone novaculites. Similar results (two clusters) are obtained again, except that the Elko Hills Quarry chert envelope and the lithic artefacts sourced to that location show a scatter indicating that not all of the actual sources of the lithic materials are from the Elko Hills Quarry itself. The lithic artefacts derive from various facies of the Elko Formation, whose natural outcropping is more extensive than that offered by the quarry exposure alone. The geochemistry of the Washington sedimentary, volcanic, hydrothermal and metamorphic chert is provided in Figures 5 and 6. Figure 5 contains two triangular scattergrams, one for the petrified wood and one for the other fabrics. The petrified wood scattergram in Figure 5 uses the K - C a - U geochemical universe to classify both the petrified wood lithic artefacts (Fig. 2c) from Tolt site 45KI464 and their potential source materials (glacial-alluvial pebbles), which come from Pleistocene drift sediments located in the Tolt and Cedar River drainage systems. Most of the petrified wood appears to be from larch or spruce and ginkgo trees. Because of

the paucity of petrified wood artefacts and source samples, the scattergram groupings that fall into Tolt River and Cedar River drainages should not be considered conclusive. These analyses do suggest, however, that the material is of local origin. Metavolcanic, volcanic and hydrothermal chert generally classify (with scatter) into their representative morphologies by use of an F e - C u - V geochemical universe (Fig. 5). S E M EDX microcrystallite data obtained from the hydrothermal chert support the use of these transition metals for sourcing. Although the origin of these microcrystallites is not understood it appears that they are related to chert type and source location. Most (if not all) of the hydrothermal chert (Type 1 and Type 2) is derived from hydrothermal dykes from a variety of local drainages. Barite microcrystallite concentrations (SEM-EDX, Fig. 2i) appear to be dominant in the metavolcanic chert, and acts as a reasonably good fingerprint for these materials. Cummings et al. (1989) and Hess (1996) reported INAA geochemistry of chert from archaeological sites in northern Oregon. The analytical techniques used at Tolt, Washington, and in northern Oregon are not the same (see the Methods section). The Oregon sites are considerably south of the Tolt site and are used here to show how different source areas may be distinguished with the use of just a few elements in a scattergram analysis (Fig. 6). The

318

M. MORGENSTEIN A = Elko Hills Quarry Chert Envelope

~,U

B = Site CrNV-12-9371 and Elko Hills Quarry Chert Envelope C = Site CrNV-12-9371 Quartz Arenite /

0.4

T

h

. O

~

........

~

~

FS4

-----------~

•

-~,,

o

- -

_~_ o~

oCs

C FS19

U ppm

/ " FS16

B

Th ppm Fig. 4. Lithic artefacts from site CrNV-12-9371, and potentialsource area samples from the Elko Hills Quarry; geochemistry scattergrams for U, Th and Cs.

overlapping geochemical data available for the Oregon and Washington sites were very limited and thus the choice of which elements to use for comparative analysis was also limited. Even so, the results are encouraging and suggest that with a more comprehensive and comparable databank the sites would classify exceedingly well.

Discussion and conclusions Few studies have been carried out that utilize both geochemistry and petrography to acquire an understanding of chert artefact provenance.

In the previous sections of this paper the petrography and geochemical compositions of broadly varying types of chert from several widely spaced geographical areas were studied. Such a range of chert probably approximates the natural variability that one might expect to find in the archaeological record. The two main sites studied produced two markedly different chert assemblages. The Nevada Elko site, where sedimentary novaculites are present, provided a baseline to study chert that forms as nodular to bedded sequences in limestone to mudstone host rocks. The petrographic fabrics that assist in distinguishing and classifying this chert

GEOCHEMICAL AND PETROGRAPHIC CHERT TOOL

319

Fe %xlO Metavolc~

Hydrothet Hydrothe~ Volcanic (

Cu 0 ,

oo

o

o

o÷

oV

Ca

/ ~ o8 / / ~ ~

/

PeldfiedWoodToltDrainage $6and$7, Petri"fiedWoodCedarDranage " $8,$9,and$10

~x~'

/ ~ \ /

/~

\ ~.~ L~hic~facts fn~n Site 45KI464

/~,and24

0.4

0.2

K0 "O

o "0n

o "7

o 'O

U

Fig. 5. Geochemistry scattergrams for lithic artefacts from Tolt site 45KI464, and potential source material from the same site and from the Tolht and Cedar River drainage systems. relate more closely to the identification of the host sedimentary rock fabric (its fossils and clastic constituents) than to the mineralogical aspects of the overprinting chert itself. Exceptions occur when there is evidence of plastic deformation (Steinitz 1970) such as crackle breccias (McBride & Folk 1977), or when there are sufficient chert fabric data such as birds-eye structures (Folk & McBride 1976), or when the

chert mineralogy (length-slow chalcedony) can assist in classifying the environment of chert formation (Folk & Pittman 1971). The Washington Tolt site provided chert fabrics from the opposite end of the geological spectrum. In this case igneous, hydrothermal and metamorphic chert contains within its structure mineral fabrics that are unique to the individual chert material. It only rarely exhibits textural or mineralogical

320

M. MORGENSTEIN 5

VolcanicChert

45

0 []

Tolt,Washington

~

•

Troy,Oregon

,i

Flow Top Breccias 4 and Tree Molds Opal-CT 3.5 3

[ ] MackCanyon,Oregon

Th 2.5

Bedded Chert Between

~

I

Grande RondeBasalt Flows

MetavolcanicChert

OpaI-CT ~ 0.0Ol

f 0.1

ppm

lO

f

S 100o

Ce ppm

Fig. 6. Geochemistry scattergram for the Tolt site 45KI464, and for the Troy and Mack Canyon sites (data from Cummings et al. (1989) and Hess (1996)) for Th and Ce.

data from host rock or wall rock. Even so, metasedimentary fabrics (sample S l) contain trace inclusions (e.g. barite) that are observable by S E M - E D X (Fig. 2i) and that act as provenance fingerprints. The ICP and I C P - M S data provide good chemical characterization of the chert but sometimes it is necessary to utilize S E M - E D X to acquire spatial geochemical information to identify microcrystallites that are difficult to classify by optical petrography. In conclusion, our best procedure for addressing the provenance issue in chert lithic artefacts is first to acquire a comprehensive geochemical and petrographic library for potential source materials. After that is accomplished, the chert artefact study requires an understanding of the genesis of the chert, which is obtained through petrographic examination. Finally, the actual sourcing of that specific chert fabric is achieved through the use of a geochemical fingerprint that is compared with potential source materials. For the Nevada Elko archaeological site CrNV-12-9371, the analysis suggests two different sources for lithic procurement. (1) Elko Hills Quarry chert: a bedded ostracode and fossiliferous novaculite chert that exhibits lithofacies changes that are a function of the spatial variations in bioclastic debris such as diatoms, radiolarians, ostracodes, clay and biocarbonate hash. The fundamental allogenic rock is a marl-mudstone. The authigenic overprintchert variations include almost pure chert without allogenic components as well as

variations in the goethite-maghemite pseudomorphs after pyrite and pyrite framboids. The fundamental authigenic chert is composed of microcrystalline quartz, chalcedony and opal. (2) An unknown source area for lithic sample FS19: this material is a chert-cemented (overprinted) siltstone, which is a quartz arenite. Similar material has been observed (Hose &Blake 1976) coming from the Mississippian Joanna Limestone, the Pilot Formation in the Ruby Range and the Tosawihi Quarries NE of Battle Mountain. This sample was not sourced specifically to either of these latter locations because there is an insufficient geochemical library at present to do so. Future work should resolve this problem. For the Washington Tolt archaeological site 45KI464 the bulk of the chert has been sourced to the Pleistocene drift sediments at the site itself. Volcanic (agate) chert, such as sample 15, appears to be from outside the region. The petrographic and geochemical characteristics of the local chert from Tolt site 45KI464 allow chert fabrics to be distinguished not only from individual local river drainage systems but also from other archaeological sites in the Northwestern USA.

Thanks are extended to Seattle Public Utilities for supporting the archaeological efforts at site 45KI464. R. Vierra (Vierra and Associates) is acknowledged for supporting our chert studies in Nevada at site CrNV-12-9371.

GEOCHEMICAL AND PETROGRAPHIC CHERT TOOL

References BERNAUER, F. 1927. Ober Zickzackb~inderung (Runzelb/inderung) und verwandle Polarisationserscheinungen an Kristallen und Kristallaggregaten. Neues Jahrbuch fiir Mineralogie,

Geologie und Palaontologie Beil-Band, 55, 92-143. CUMMINGS, M. L., TRONE, P. M. & POLLOCK, J. M. 1989. Geochemistry of colloidal silica precipitates in altered Grande Ronde Basalt, Northeastern Oregon, USA. Chemical Geology, 75, 61-79. FOLK, R. L. 1968. Petrology of Sedimentary Rocks. Hemphill, Austin, TX. FOLK, R. L. & MCBRIDE, E. 1976. The Caballos novaculite revisited, Part l: origin of novaculite members. Journal of Sedimentary Petrology, 46(3), 659-669. FOLK, R. L. & PITTMAN, J. S. 1971. Length-slow chalcedony: a new testament for vanished evaporites. Journal of Sedimentary Petrology, 41(4), 1045-1058. FOLK, R. L. & WEAVER, C. E. 1952. A study of the texture and composition of chert. American Journal of Science, 250, 498- 510.

321

FRONDEL, C. 1962. The System of Mineralogy, Vol. III, Silica Minerals. Wiley, New York. GRAETSCH, H. & IBEL, K. 1997. Small angles neutron scattering by opals. Physics and Chemistry of Minerals, 24, 102-108. GRAETSCH, H., FLORKE, O. W. & MIEHE, G. 1987. Structural defects in microcrystalline silica. Physics and Chemistry of Minerals, 14, 249-257. HESS, S. C. 1996. Chert provenance analysis at the Mack Canyon site, Sherman County, Oregon: an evaluation study. Geoarchaeology, 11, 51-81. HOSE, R. K. & BLAKE, M. C. JR 1976. Geology and

Mineral Resources of White Pine County, Nevada, Part I: Geology. Nevada Bureau of Mines and Geology Bulletin, 85, 1-35. MCBRIDE, E. & FOLK, R. L. 1977. The Caballos novaculite revisited: Part II: chert and shale members and synthesis. Journal of Sedimentary Petrology, 47(3), 1261-1286. SCHIEBER, J. 2002. Sedimentary pyrite: a window into the microbial past. Geology, 30(6), 531-534. STEINITZ, G. 1970. Chert 'Dike' structures in Senonian chert beds, southern Negev, Israel. Journal of Sedimentary Petrology, 40(4), 1241 - 1254.

Identification, characterization and weathering of the stones used in medieval religious architecture in L'Aquila (Italy) R. Q U A R E S I M A 1, C. G I A M P A O L O 2, F. S P E R N A N Z O N I 2 & R. V O L P E l

1Department of Chemistry, Chemical Engineering and Materials, University of L'Aquila, 1-67040 Monteluco di Roio, L'AquiIa, Italy (e-mail: [email protected]) 2Department of Geologic Sciences, University of Roma Tre, Largo San Leonardo Murialdo, 1, 1-00146 Roma, Italy Abstract: In the city of L'Aquila the historical and artistic significance and the density of religious buildings is notable. From the 1 lth to the 19th century the ornamental stones quarried in the surroundings of L'Aquila have been used for religious architecture, civil artefacts and buildings of great artistic and historical relevance. Great emphasis has been placed on the recognition of ancient quarries where these ornamental stones were carved. In this paper the petrographical, physical, mechanical and technological characteristics of the limestones and marly limestones sampled in the quarries are reported. The main characteristics will be discussed and correlated to the decay forms, with the aim of achieving a general strategic approach based on a diagnostic phase and an application phase.

Sixty-five medieval religious buildings of remarkable historical and artistic significance are located in the historic city of L'Aquila (Antonini 1999a,b). Such a high density of buildings is the cause of economical and methodological difficulties of conservation. The general conservation approach based on two main phases, a diagnostic phase and an application phase (Table 1), proposed in this paper requires the full knowledge of the stone materials used. However, this cannot be fully achieved because of the destructive effects of an accurate and extensive sampling campaign. An approach for conservation and restoration of the religious faqades in L'Aquila could be defined by the guidelines proposed in Table 2. These guidelines are based on the information required and operations to be carried out during the application phase. The operational strategies list a possible series of alternative operations and materials to be adopted before, during and after working the stones in restoration. Each work phase is linked to a specific operational instruction, in which the equipment, materials and technologies necessary for the activity in progress are set. All data are to be collected in a database that should be constantly updated, and that should include new or unpredicted actions and instructions (new decay forms, stone typologies and quarries, innovative methods, etc.). The database should also contain the details of the conservation work (society, activity, materials and tech-

niques adopted, etc.). The development of the proposed conservation plan, as well as of the evaluation of the cultural heritage, could represent a tool for the conservation of the religious architectonic faqades in L'Aquila. The local ornamental stones are of great relevance to the local architecture of the city for historical, artistic and economic reasons; nevertheless, concerning the conservation of the monuments in L'Aquila, there is a lack of scientific studies, in contrast to many historical and architectonic ones. Considering the proximity of the Roman town of Aveia (today known as Fossa), some researchers have supposed that quarrying of white stone was started during the Roman Empire; also, this stone was used around the year 1000 for building the castle of Poggio (Picenze) and others nearby. The aim of this paper is to identify and to characterize the stones used in the medieval religious architecture and to evaluate the weathering process and conservation of stone faqades.

Experimental methods To identify the stones used in the architectural buildings of L'Aquila and to collect samples for their characterization and analysis, sampling was performed in the surroundings of the city, within a range of 15 km (Ponzi & Masi 1873; Jervis 1889; Rodolico 1953) as well as directly on the monuments. First, preliminary detective work was carried out by consulting geological resources agencies

From: MAGGETTI,M. & MESSIGA,B. (eds) 2006. Geomaterials in Cultural Heritage.

Geological Society, London, Special Publications, 257, 323-335. 0305-8719/06/$15.00 © The Geological Society of London 2006.

R. QUARESIMA ET AL

324

Table 1. Main actions and studies for the diagnostic and application phases Diagnostic phase

Application phase

1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11.

1. 2. 3. 4. 5. 6. 7.

Historical research Graphic documentation Photographic documentation Visual checks Characterization of decay forms Mapping of decay Analysis of construction typologies Characterization of the stones Research of the old quarries Methodologies and materials for restoration Laboratory and in situ experimental tests on products to be used

and performing historical and aerial photography studies (available at http://ww3.atlanteitaliano.it). Then, the quarries identified were inspected. Samples of shape and dimensions suitable for the methods to be adopted were obtained from blocks of stone recovered from the main quarries. The samples were characterized according to UNI NOR.Ma.L recommendations. To assess and evaluate the weathering of the lithotypes used for the stone faqades and to characterize the stones, several techniques were applied in situ (characterization by fibre-optic microscopy and by the 'pipette method' (NORMAL 44/93)) and in the laboratory (optical and electron scanning microscopy, energy-dispersive X-ray analysis, X-ray diffraction, mercury intrusion porosimetry, atomic absorption spectrometry, and mechanical tests). An approach based on the study of the state of decay of the stone and its characteristics has been designed for the classification of the stone typologies. A comparison has been carried out of the characteristics of the stones sampled in the quarries and those of the stones used in the monuments. Table 2. Main guidelines for the execution of the application phase (1) Practical instructions: Preconsolidation Preliminary cleaning tests Cleaning techniques Antifouling interventions Re-pointing and pointing of cracks, fissures and loss of material Pointing and re-pointing of mortars Consolidation Protection (2) Definition of the maintenance operations

Definition of conservation interventions Planning of conservation interventions Reuse and service of the buildings Planning of restoration program General guidelines for conservation and restoration Monitoring of the decay Maintenance actions

Results and discussion Identification and characterization of the stones Figure 1 shows the distribution of all the quarries of the 'white and reddish' stones, identified in the surroundings of L'Aquila, within a range of 15 km from the town. The quarries and the main macroscopic and microscopic characteristics of the lithotypes (ranging from limestones to marly limestones) (Accordi & Carbone 1988; Accordi et al. 1988; Centamore et aL 1992), representative of the materials used in the most important medieval religious buildings of L'Aquila are reported below.

White lithotypes (WL) Genzano: WL1 (Lombardi's quarry). The limestone (Fig. 2a) is pearl grey, compact and classifiable as biointramicrite (packstone). It has an isotropic structure and heterometric clasts (Fig. 2b and c). Fossil fragments, micritic clasts and partially rounded calcitic grains are present. The stone has a primary micropore porosity of 2.57% (see Table 3). The rock has been deposited in a typical sedimentary zone on a shelf edge, perhaps before the Campanian stage (late Cretaceous) (Ghizetti & Vezzani 1992). L'Aquila: WL2 (Collemaggio parking lot). This is a pearl grey macroporous limestone (Fig. 3a). The rock is a calcareous breccia (rudstone) showing an isotropic structure with sparitic, micritic, intramicritic and biomicritic angular clasts of heterometric dimensions ranging from a millimetre to several centimetres (Fig. 3b and c). The voids (macropores) are partially cemented by calcite. The lithotype shows heterometric

325

STONES USED IN MEDIEVAL ARCHITECTURE 15km

/

F_ ~.~ aL, ~ , ~

" " ~

"'--~-S.p . . .

omo PIccOlo I

" ~ ",e, 22o0

29112 &

~ Como G r~m

;89', ~o

_ ;; ¢.,)1~.

i !

$ O~,xlwlo

!

.,,,\ ~,o

~¢km,,'m~

-%

TU~.

'..,\ \

\-

C, C

t~ 2.-'04

Ca;f~Ho

Cagn,, -

Alto ~$ Pm

\ Fig. 1. Location of the quarries of white limestones and red marly limestones in the surroundings of L'Aquila: 1, L'Aquila; 2, Roio: Ciccozzi quarry; 3, Genzano: Colle Roale; 4, Coppito: Colle del Grillo; 5, S. Giuliano; 6, Monticchio: Cavalletto mountain; 7, San Gregorio; 8, Fossa; 9, Poggio Picenze, 'La Petrara'; 10, Santa Croce; 11, Santa Croce (west); 12, Santa Croce (east); 13, Casamaina: at Faucicchio; 14, Casamaina: Libri S. Giovanni; 15, Casamaina: Coppa Matteo; 16, Tornimparte: Gentilucci quarry; 17, Vigliano: Sarra quarry; 18, Cese: Cesavecchia quarry; 19, Amiterno; 20, Cavallari: Fonte Rua; 21, Cavallari: Fonte Rua; 22, Arischia: at Cafaio; 23, Arischia: at Cafaio (east); 24, Capannelle: 'red quarry'; 25, Capannelle: Ponte della Lama.

pores. The limestone is classified as an oligomict orthoconglomerate belonging to a Quaternary subaerial environment.

Poggio Picenze: WL3 (La Petrara quarry). This is a beige limestone with excellent workability (Fig. 4a). The lithotype (Fig. 4b and c) is a biointramicrite (wackestone) with an isotropic structure. Fossil fragments, rounded micritic clasts and a small amount of calcitic grains are present. Bryozoa and Algae (Lithothamnium), and fragments of Bivalvia and Echinodermata, can be also observed (Giampaolo et al. 2000). The high primary porosity of the rock, 18.93% (Table 3), is composed of micro- and mesopores. The stone has been related to the Miocene

'Calcareniti a Briozoi e Litotamni' formation (Serravallian-Langhian) (Ghizetti & Vezzani 1992). Cavallari: WL4. The limestone is white and very compact with isotropic structure (Fig. 5a) classified as an intrabiomicrite (wackestone) (Fig. 5b and c). It shows subangular and partially rounded calcitic grains. The deposition zone was on the Eocene pro parte (pp)-Cretaceous shelf edge (Ghizetti & Vezzani 1992).

Vigliano: WL5 (lmpredatora and Tana del Lupo quarries). The limestone is white, very compact with an isotropic structure (Fig. 6a) and it is characterized by its workability. The material is an intrabiomicrite (wackestone) (Fig. 6b and c)

~_.

~

~'~_~ ~ . A . ,..0

WL1, Genzano WL2, L'Aquila WL3, Poggio Picenze WL4, Cavailari WL5, Vigliano WL6, Roio RL 1, Genzano RL2, Arischia RL3, Cavallari RL4, Capannelle

Lithotype, quarry

2.661 2.453 2.210 2.291 2.558 2.561 2.635 2.665 2.631 2.620

( g c m -3)

( g c m -3)

2.731 2.698 2.723 2.728 2.713 2.755 2.719 2.736 2.719 2.723

Apparent density

Real density

0.97 0.91 0.81 0.84 0.94 0.93 0.97 0.97 0.97 0.96

-

Compactness

2.57 9.08 18.93 6.25 5.69 7.05 3.20 2.1 ! 3.25 4.27

(%)

Porosity

5384 4049 4033 5040 5261 2931 5497 5394 5402 4521

(ms -I)

Ultrasonic velocity

t a b l e 3. Physical and mechanical characteristics of the white and red lithot.vpes

z

~a ~.ea

~'~" ~'~

0.11 2.19 5.58 2.08 1.98 2.92 0.26 0.22 0.20 0.96

(%)

Imbibition coefficient

2,77 2.30 6.10 3.71 !.11 6.79 5.89 4.24 1.73 3.77

× 10 -5 >< 10 -3 × 10 -3 x 10 -3 x 10 -3 × 10 -4 x 10 -5 x 10 -4 x 10 -3 x 10 -4

Capillarity absorption coefficient ',ml cm -2 s -°'5)

105.5 17.4 38.4 54.6 66.7 48,4 101.6 93.7 89.0 146.3

_ 45 ___ 8 ___ 6 ___ 13 _ 7 _+ 15 + 9 + IO _ 7 +_ 8

(MPa)

Iml cm -2) 0.013 0.194 0.476 0.043 0.024 0.008 0 0.003 0 0.006

2ompressive strength kbsorption (30 min)

5.5+1 5.9+_ 1 15.4+ 1 6.1+2 28.1 +_1 11.4___ 1 12.7___ 1 24.1 + 1

25.1 _+5

(MPa)

Flexurai strength

39 32 29 33 36 29 32 32 32 32

000 000 000 000 000 000 000 000 000 600

(MPa)

Elastic modulus

._..,

>

0 >

STONES USED IN MEDIEVAL ARCHITECTURE (a)

(a) ~

327

,~.; ¢

.

(b)

(b)

(c)

(c)

lcm

I iI'lal

Fig. 3. Lithotype WL2: (a) macroscopic aspect of the calcareous breccia; (b, e) thin-section optical micrographs of the specimen sampled at Collemaggio parking lot in L'Aquila.

Fig. 4. Lithotype WL3: (a) macroscopic aspect of the limestone; (b, c) thin-section optical micrographs of the specimen sampled at La Petrara quarry in Poggio Picenze.

with unidentifiable fossil fragments and subangular sparitic and micritic clasts. The material has a primary porosity of 5.69% (see Table 3), mainly composed of mesopores. Considering the morphology and typology of clasts and the

stratigraphic sequence, it is likely that the sedimentary zone of deposition was located on the Eocene pp-Cretaceous shelf edge. Roio: WL6 (Ciccozzi's quarry). This compact marly limestone is pearl grey (Fig. 7a). It is

328

(a)

R. QUARESIMA ET AL.

(a)

.

o

1cm

1 cm

.w

(b)

(b)

(c)

(c)

Fig. 5. Lithotype WL4: (a) macroscopic aspect of the limestone; (b, c) thin-section optical micrographs of the specimen sampled in the quarry of Cavallari,

Fig. 6. Lithotype WL5: (a) macroscopic aspect of the limestone; (b, c) thin-section optical micrographs of the specimen sampled at Impredatora quarry in Vigliano.

classified as an intrabiomicrite (wackestone) (Fig. 7b and c) with isotropic structure. The sparitic clasts and calcite crystals have angular or partially rounded shapes. Iron hydroxides, oxides and some glauconite are also present. The fossil fauna is composed of Foraminifera (Amphistegina, Orbulina, Globigerinids,

Nodosariae) and fragments of bivalves. The stone is rather porous (7.05%; mesopores) and the fractures are cemented by calcite. The presence of planktonic (Orbulina, Globigerinids) with benthonic (Amphistegina) Foraminifera suggests a shelf facies (Chiocchini et al. 1994).

STONES USED IN MEDIEVAL ARCHITECTURE

(a)

329

(a)

~

2:)!

,/

I ,~-. (b)

(c)

(b)

w

Fig. 7. Lithotype WL6: (a) macroscopic aspect of the marly limestone; (b, c) thin-section optical micrographs of the specimen sampled in Ciccozzi's quarry in Roio.

Red lithotypes (RL) Genzano: RL1 (Lombardi's quarry). This rust brown marly limestone (Fig. 8a) is very compact. The lithotype is classified as a biointramicrite (wackestone) (Fig. 8b and c) with isotropic structure. Fossil fragments, micritic clasts,

(c)

Fig. 8. Lithotype RLI: (a) macroscopicaspect of the marly limestone;(b, c) thin-sectionoptical micrographsof the specimen sampled in Lombardi's quarry at Genzano.

subangular calcitic grains and iron hydroxides are present. The fossil fauna is composed of algal fragments, and planktonic (Globigerinids, Globigerinatheka, Hantkenina) and benthonic (Nummulites) Foraminifera. Fractures are partially filled with iron oxides. The sedimentary

R. QUARESIMA ET AL.

330

zone was on the mid- to late Eocene shelf edge or on a nearby slope. Arischia: RL2 (St. Nicola's quarry). This compact marly limestone is mainly brick red with some white calcareous layers (Fig. 9a). The stone is classified as a biointramicrite (mudstone (Fig. 9b and c) and locally wackestone (Fig. 9b' and c')) with isotropic structure.

Intrabiosparitic (with Orbitoides) and intrabiomicritic (with Globotruncanids) rock fragments and iron oxides are present. Fossils such as Morozovella velascoensis, Morozovella aequa and Globigerinids can be also observed. Fractures are filled with calcite and iron oxides. The late Paleocene sedimentary zone was located on a shelf slope.

(a) .

(b)

(b') •

,,

(,,

.

•

.~

•

(c)

~,

.

i

m,.

(c')

Fig. 9. Lithotype RL2: (a) macroscopic aspect of the marly limestone; thin-section optical micrographs of the specimen sampled in the quarry of St. Nicola in Arischia: (b, c) micritic component; (b', c') clastic component.

STONES USED IN MEDIEVAL ARCHITECTURE

331

Cavallari: RL3. This marly limestone has a terracotta colour (Fig. 10a) and is very compact. The stone is classified as a biointramicrite (mudstone) (Fig. 10b and c). Clasts of white facies (packstone) can be observed. Algae, various fragments and Foraminifera such as

Amphistegina, Lepidocyclina and Operculina of

(a)

(a)

the mid- to late Oligocene are present. Capannelle: RL4. This compact marly limestone has a burgundy colour (Fig. 1 l a). The lithotype is classified as an intrabiomicrite (mudstone) (Fig. 1 lb and c) and has an isotropic

I cm n,

(b)

(b)

(c)

(c)

Fig. 10. Lithotype RL3: (a) macroscopic aspect of the marly limestone; (b, c) thin-section optical micrographs of the specimen sampled in the quarry of Cavallari.

Fig. 11. Lithotype RL4: (a) macroscopic aspect of the marly limestone; (b, c) thin-section optical micrographs of the specimen sampled in the quarry of Capannelle.

332

R. QUARESIMA ET AI_,

Table 4. Main environmental and anthropogenic causes and factors of decay Causes of decay Freeze -thaw Salt crystallization Pollution Humidity Biodeterioration Irregular conservation treatments

Factors Winter climate conditions De-icing salts (NaC1-CaCI2) Traffic and domestic heating Rainwater seepage through roofs, faqades, gutters; rising damp and sewage Lichens (Aspicilia calcarea, Aspicilia radiosa, Caloplaca cirrochroa, Caloplaca citrina, Caloplaca aurantia e flavescens, Lecanora albescens, Lecanora muralis), algae (chlorophytae: Chlorella, Chlorochoccum, Scenedesmus) and weeds Sandblasting of WL3 and WL5 tithotypes; reintegration (WL4) or pointing (RL1 and RL3) of columns and basements with cementitious mortars

structure. Fossil fragments, Globigerinids, calcite rhombohedra, rounded calcitic grains, opaque minerals and iron hydroxides are present. The sedimentary zone was on the Eocene shelf edge or on a nearby slope. The stones were quarried from shallow excavations located at the base of the hillside. The most important ancient quarries were located in the surroundings of the city within a range of 15 km. Several quarries of 'breccia' were also located within the urban perimeter. Moreover, as well as the geological aspects, the location of the quarries and the exposure of their fronts must be related to environmental and economical requirements, such as conditions of light (duration of daylight during autumn and winter, presence of high mountains nearby), the prevalent wind directions during snow storms, the great availability of materials and, last but not least, the need to cut down transport costs. The material cropping out along the quarry front has a massive or stratified nature. In the first case, quarry activity was favoured by the presence of fractures caused by tectonic events that produced medium and large ( 1 - 3 m ) boulders. In the second case, the presence of calcareous white layers alternating with red marly limestone layers (the latter of around 4 0 - 6 0 cm

thickness in the 'Scaglia Rossa' formation) makes easier the splitting of the blocks along this bedding planes. Moreover, in addition to the facts that the quarries were close to the city of L'Aquila, the road system was good and quarrying (blocks or bedding planes) was easy, the extensive use of those stones can be also explained in the light of the definition, 'Petra gentile del Poggio' ('gentle stone from Poggio'), given by Silvestro d'Ariscola, a master stonemason who lived and worked in Abruzzo during the 15th century, known as a pupil of Donatello. The adjective 'gentle' contains in itself the main characteristic that explains the use of this stone in major commercial, public and artistic buildings: its workability. This is due to the overall characteristics of the rock and its geological formation (see Poggio Picenze, WL3). Large-scale and effective applications such as those of the stone of Poggio Picenze were not so easy with the compact limestones quarried in Vigliano, Fossa and Assergi, or with the friable sandstones quarried in Montereale and Campotosto, towns located farther to the north. Physical and mechanical properties The main physical and mechanical properties of the rock are summarized in Table 3. With the

Table 5. Correlation between weathering and ~pologies of stones Decay form Exfoliation Spalling Cracking Alveolar disease Differential degradation Erosion Patina White: carbonate crust Black: black crust Yellow: calcium oxalate films

White limestones WLI, WL2, WL3, WLI, WL2, WL3, WL3 WL1, WL2, WL3, WL1, WL2, WL3, WL1, WL2, WL3, WL1, WL2, WL3,

WL4, WL5 WL4, WL5 WL4, WL5 WL4, WL5 WL4, WL5 WL4, WL5

Red marly limestones RL1, RL4 RL1, RL3, RL4 RL1, RL2, RL3, RL4 RL 1 RLI, RL2, RL3, RL4 RLI, RL2, RL3, RL4 RLI, RL2, RL3, RL4 RL1, RL2, RL3, RL4

STONES USED IN MEDIEVAL ARCHITECTURE

(a)

(b)

(c)

(d)

(e)

(f)

333

Fig. 12. Presence of clay and iron oxides (a) in the matrix (100x, plane-polarized light) and (b) in the veins (50x, plane-polarized light) of the RL 1 lithotype. (c) Fracture along the iron oxide veins as a result of freeze-thaw phenomena (funnel-shaped pores) (sample collected from the faqade of the Basilica of S. Maria di Collemaggio, in the year 2000, during restoration work (50 x, plane-polarized light). Different use of the bedding planes (d) as a jambstone in the main portal of the church of S. Domenico or (e) as a column in the main portal of the church of S. Marco. (f) Typical decay of the red marly limestones (exfoliation) as a result of the presence of iron oxides and/or clay distributed in veins.

334

R. QUARESIMA E T A L

exception of WLI, the white limestones have compressive strengths lower than those obtained for the reddish stones (50-60 MPa compared with 100MPa); in contrast, the compactness and workability are very high. The different mechanical resistances of the reddish and white stones account for the state of decay observed in the portals and faqades of the religious buildings of the city. The decay originates from the detachment of the decorative and construction elements from the masonry and the way the architectonic elements (ashlars, columns, etc.) were originally placed with respect to the loads. Weathering and decay f o r m s

Tables 4 and 5 report the main environmental and anthropogenic decay factors in L'Aquila (Quaresima et al. 1995b; Quaresima et al. 1995a, 1997, 1999, 2002a,b; Quaresima & Volpe 1998) correlated to the typologies of the stones identified (Quaresima et al. 2003). As compared with the white stones, the worse decay of the red ones is due to the following factors: (1) the presence of clay and iron oxides homogeneously distributed in the matrix or concentrated in layers; (2) the placing of the architectonic elements in the buildings. It was observed that the red stones, which generally show bedding planes, were placed with the stratigraphic plane parallel to the loads, to achieve architectonic elements of remarkable dimensions or with a homogeneous colour. Some correlations between these factors and some forms of decay are reported in Figure 12.

Conclusions A general strategic approach, based on a diagnostic phase and an application phase, seems to be applicable for the conservation of the notable religious buildings of the l lth to 15th centuries of the city of L'Aquila. The main quarries of white limestones and red marly limestones around the city were identified; in particular, those of Poggio Picenze (Petrara), Genzano (Lombardi) and Cavallari were recognized as being of historical relevance. Rather than just a single quarry, Genzano is an interesting group of quarries featuring several lithotypes. During inspections of these quarries unique stones were observed, very similar to those observed on the portals of the Basilica of S. Mafia of Collemaggio in L'Aquila (highly compacted white limestone with bedding planes). Investigations and studies of the quarry district of Genzano are in progress. Moreover, it is possible that other materials were quarried

several centuries later (in the 19th and 20th centuries) from the quarries of Vigliano, Arischia and Capannelle. The quarries supplied calcareous materials of two typologies: (1) irregular boulders of medium and large dimensions ( 1 - 3 m ) ; (2) regular blocks of 4 0 - 6 0 cm thickness ('Scaglia Rossa' formation). In both cases, it was easy to quarry or to split the stones because of the fractures produced by tectonic events or of the presence of bedding planes. The petrographical, physical, mechanical and technological characteristics of the lithotypes that allow the recognition and identification of the stones present in the monuments have also been evaluated. Considering the state of decay of the stone, 10 typologies of lithotypes were detected in the religious monuments and attributed to several quarries. The forms of alteration and decay recorded were grouped into 10 classes, and the typologies of stones have been related to each of them. White stones have fewer conservation problems than the reddish ones. In fact, the latter show some serious decay (exfoliation and differential degradation) as a result of the presence of clays and iron oxides homogeneously distributed in the matrix or located in layers, or as a result of the way in which the stone elements were placed in the buildings (ashlars, columns, etc.). For constructive or aesthetic reasons the red lithotypes were often placed with the bedding planes parallel to the loads, to achieve ashlars or columns of greater dimensions or homogeneous colours. In the light of the different compressive strengths recorded for the white and the red stones, the way in which they were used accounts for the abundance of cracks and fractures caused by overloading (related to seismic events or detachment of the ashlars from the masonry) observed on the portals and faqades of the religious buildings of the city.

References ANTONINI,O. 1999a. Architettura Religiosa Aquilana, 1. Gallo Cedrone, L'Aquila. ANTONINI,O. 1999b. Architettura Religiosa Aquilana, 2. Gallo Cedrone, L'Aquila. ACCOROI,B. & CARBONE,F. 1988. Sequenze carbonatiche meso-cenozoiche. In: Note illustrative relative alia Carta delle Litofacies del LazioAbruzzo ed aree limitrofe. Quademi della Ricerca Scientifica CNR, 114, 12-88. ACCORDI,B., CARBONE,F., CIVITELLI,G., et al. 1988. Lithofacies map of the Latium-Abruzzi and neighbouring areas--Carta delle Litofacies del Lazio-Abruzzo ed aree limitrofe. Quaderni della Ricerca Scientifica CNR, 114, 170-195.

STONES USED IN MEDIEVAL ARCHITECTURE CENTAMORE, E., ADAMOLI,L., BERTI, D., ETAL. 1992. Carta Geologica dei bacini della Laga e del Cellino e dei rilievi carbonatici circostanti. Studi Geologici Camerti, Volume Speciale, 1991(2). CHIOCCHINI, M., EARINACCI, m., MANCINELLI, N., MOLINARI, V. & POTETTI, M. 1994. Biostratigrafia a foraminiferi, dasicladali e calpionelle delle successioni carbonatiche mesozoiche dell'Appennino Centrale. Studi Geologici Camerti, Volume Speciale 'Biostratigrafia dell'Italia centrale', 123-168. GIAMPAOLO, C., ADANTI, B., DI PACE, A. & BARTOLINI, G. 2000. 'Pietra di Poggio Picenze' e 'Calcare di Vigliano-Scoppito' dal sito Internet di 'Italithos' (www.italithos.uniroma3.it). GHIZETTI, F. tf)z VEZZANI, L. 1992. Carta Geologica della Regione Abruzzo. Carte Regionali. Regione Abruzzo. SEL.CA, Firenze. GREENSMITH, J. T. 1989. Petrology of the Sedimentary Rocks. Unwin Hyman, London. JERVtS, G. 1889. I tesori sotterranei dell'Italia. Loescher, Torino. PONZI, G. & MASI, F. 1873. Catalogo ragionato dei prodotti minerali italiani ad uso edilizio e decorativo. Coltellini e Bassi, Roma. QUARESIMA, R. 8z VOLPE, R. 1998. Conservation and weathering of the stone used in the Basilica of Collemaggio in L'Aquila (Italy). IV International Congress, Rehabilitacion del Patrimonio Arquitectonico y Edification (CUBA '98), 175-176. QUARESIMA, R., PASANISI, A. (~ SCARSELLA, C. 1995a. Patine di ossalati e croste here: indicazioni su possibili interventi conservativi. XI National

335

Congress 'La Pulitura delle Superfici dell'Architettura', Bressanone, 179-186. QUARESIMA, R., DI GIUSEPPE, E. & VOLPE, R. 1995b. Impiego di biocidi per la rimozione della microflora algale. XI National Congress 'La Pulitura delle Superfici dell'Architettura', Bressanone, 267-275. QUARESIMA, R., SCOCCIA, G. & VOLPE, R. 1997. Studio dei trattamenti conservativi per la pietra della basilica di S. Maria di Collemaggio a L'Aquila. In: National Proceedings, 'Materiali e tecniche per il restauro', Cassino, 73-85. QUARESIMA, R., SCOCCIA, G. (~ VOLPE, R. 1999. Comportamento di alcuni litotipi trattati e non a cicli di cristallizzazione salina. H National Congress 'Materiali e tecniche per il restauro', Cassino, 1-10. QUARESIMA, R., SCOCCIA, G. & VOLPE, R. 2002a. Behaviour of Different Treated and Untreated Stones under Salt Crystallization Test, Balkema, Rotterdam, 455-460. QUARESIMA, R., SCOCCIA, G. (~ VOLPE, R. 2002b. La gelivit~ dei materiali lapidei naturali: problematiche metodologiche e conservative. VI National Congress AIMAT, Modena, 108-117. QUARESIMA, R., SCOCCIA, G., TAGLIERI, G. & VOLPE, R. 2003. La caratterizzazione dei materiali e delle forme di degrado dell'architettura religiosa aquilana. Ili Convegno 'Materiali e tecniche per il restauro', Cassino, 214-224. RODOLICO, F. 1953. Le pietre delle cittgt d'Italia. Le Monier, Firenze.

Textural analysis of ancient plasters and mortars: reliability of image analysis approaches F. C A R O 1, A. DI G I U L I O l & R. M A R M O a

lDipartimento di Scienze della Terra, Universith degli Studi di Pavia, Strada Ferrata, 1, 27100 Pavia, Italy (e-mail: federico.caro @manhattan. unipv.it) aDipartimento di Informatica e Sistemistica, Universitdt di Pavia, Strada Ferrata, 1, 27100 Pavia, Italy Abstract: Different image analysis (IA) methods have been developed to compute textural

parameters of ancient plasters and mortars using standard petrographic thin sections. These IA routines were applied to samples of materials with different technological characteristics from three historical buildings in the city of Pavia (Northern Italy), covering a period from the 12th to the 19th century. The IA techniques tested in this study belong both to classical digital image processing and to neural network modelling. In the first case, analyses were performed by commercial IA software whereas in the second case a Multi-Layer Perceptron neural network (MLP) was tested. Digital image analysis was performed on images taken by means of a petrographic microscope; additionally, analysis of back-scattered electron (BSE) images was performed. Textural data obtained through the IA applied to thin sections were compared with the data from traditional point counting and mechanical sieve analysis of disaggregated samples of the same materials. The results show that the IA of thin sections provides robust results in a fast and easy way. However, the reliability of the analyses can be prejudiced by textural and compositional heterogeneity of the samples.

Studies of structures and textures of both loose and lithified clastic sediments involve examination of grain size, morphology, surface texture and fabrics (Tucker 2001). The same approach can be easily extended to the study of structures and textures of man-made similar materials, such as plasters and mortars. Together with compositional features, structural and textural properties play a fundamental role in determining mortar and plaster performance. Therefore, textural studies are necessary for the design of new building materials as well as for the characterization of ancient plasters and mortars. The determination of textural parameters of ancient plasters and mortars is normally achieved either through sieving of previously disaggregated material or by microscopical observations of thin sections. The choice between these two different approaches depends upon the particular case study and it is mainly related to the quantity of material available. The introduction of automatic or semiautomatic image analysis (IA) techniques to process digital images and extract quantitative information is improving the capability of collecting measurements from thin sections of

geomaterials (Chermant et al. 2001; Pirard 2002). In particular, the modal composition of the aggregate fraction, its texture and dispersion, the homogeneity of the materials, the characteristics of the pore network, and the shape and orientation of cracks are some of the most common parameters that are investigated using IA procedures (Chermant et al. 2001). Nevertheless, problems related to sample representativeness of images, the accuracy of the method and reliability still exist (Francus 1998; Pirard 2002; Perring et al. 2004). In the case of ancient plasters and mortars, additional difficulties in automatic segmentation arise when constituents of complex and various nature are present (Montana 1995). This paper investigates the possibility of applying IA procedures to the textural characterization of such materials. Special attention was given to: (1) the reliability of quantitative IA results compared with alternative analytical method; (2) the use of different sources of thinsection imaging; (3) the use of different IA approaches. In particular, procedures based on convolution and non-convolution algorithms and on grey-level thresholding were used to enhance and identify the aggregate fraction on

From: MAGGETTI,M. & MESSIGA,B. (eds) 2006. Geomaterials in Cultural Heritage. Geological Society, London, Special Publications, 257, 337-345. 0305-8719/06/$15.00 © The Geological Society of London 2006.

338

F. CAR0

both plane-polarized light images and backscattered electron (BSE) images of the same samples. In addition, a Multi-Layer Perceptron neural network (MLP) was used to analyse BSE images. Study m a t e r i a l s The experimentation was undertaken on 22 mortars and l l plasters sampled from three historical buildings in Pavia, Northern Italy. The buildings studied are: (1) the Lardirago Castle (Fig. l a) erected 7 km NE of the city of Pavia in the l lth century; (2) the Broletto Palace, the 12th century city hall, which is in the centre of the city of Pavia facing the main city square (Fig. lb); (3) the San Tommaso Convent Complex (Fig. lc), a late 14th century. building in the centre of Pavia. The collected samples cover a period from the 12th to the 19th century. The sampling strategy was designed to collect a wide spectrum of building materials with various technical purposes, and different textural and structural characteristics. Thus, mortars have been sampled from coarse filling masonries, wide structural walls, narrow superimposed walls, and fine brick arches and columns. Also, coarse extemal plasters devoid of finishing layers and fine inner pigmented surfaces have been sampled.

Petrography of samples All materials studied by IA techniques are composed of an aggregate fraction and carbonate inorganic binder. Their structure is comparable with that of clastic sedimentary rocks with carbonate cement. According to the Pettijohn classification, all sand used as aggregates can be classified as lithic sands. In particular, average QFRf parameters of samples from Lardirago are: Q--- 16.3_+5.4%, F = 4 . 7 + 2 . 2 % and

ETAL R f = 79 + 5.9%; samples from the Broletto palace have Q -- 32.6 +_ 9.6%, F = 10.9 +__5.8% and R f = 56.5 _+ 11.9%; and samples from the San Tommaso Convent have Q = 34.1 _ 11.4%, F = 7.9 +_ 3.1% and Rf = 58.0 + 12.8%. None of the samples contained pozzolanic materials such as crushed brick fragments or natural pozzolanic additives. The binder fraction of all samples consists of a range of carbonate particles of micro- to cryptocrystalline dimensions grouped into clusters. Generally, the binder is non-homogeneous, with areas of differing compactness and crystallinity as well as dispersed lumps of various dimensions and aspects.

Experimental methods If possible, i.e. if enough material was available, all the techniques commonly used in textural analysis were applied to each sample. This approach permits us to compare diverse textural data and hence helps to establish the reliability of IA-based methods. Textural analyses were performed by (1) mechanical dry sieving of previously disaggregated material; (2) point counting on petrographical thin sections by means of a petrographic microscope; (3) analysis of digital images captured by a camera mounted on the petrographic microscope; (4) the analysis of back-scattered electron images. After separation of the aggregate fraction from the bulk sample by means of gentle mechanical disaggregation followed by dissolution of lime binder with HC1 solution, a series of ISO 565 sieves were used with a spacing of 1 phi (~) unit. The diameters of meshes ranged from 4000 la,m to 32 ixm. Each fraction of sieved sediment was weighed to an accuracy of 0.01 g and cumulative grain-size distributions were recorded.

Im

Fig. 1. (a) Lardirago Castle; (b) facade of Broletto Palace seen from the city-square; (c) bell tower and part of the San Tommaso Convent Complex seen from the cloister.

ANCIENT PLASTERS AND MORTARS

339

Table 1. Analytical equipmentused Analytical method Point counting Sieve analysis Image analysis Neural network-based image analysis

Instrument Leitz Laborlux polarizing microscope coupled with an Olympus Camedia Z4040 digital camera ISO 565 sieves Media Cffbernetics Image-Pro Plus® v.4.5 MatLab'~ (MathWorks Inc.)

Point counting was performed on petrographical thin sections using a micrometric eyepiece. The maximum and minimum diameters passing through the centre of mass of each grain encountered at grid intersections along random parallel directions were measured, and the arithmetic mean size was used to determine the cumulative grain-size distribution. Up to six images were analysed in each thin section using a commercial IA softwarebased macro and an MLP. The details of each analytical method are reported in Table 1. Grain-size curves obtained by the abovementioned procedures were characterized by an Excel-based statistical method described below.

Image analysis techniques Image acquisition A first set of images was taken by a petrographic microscope in plane-polarized light, with an objective of 1.6x magnification. Images were saved in .tiff format with a resolution of 2048 pixels x 1536 pixels. No image compression was used. The resulting area covered by each image is 7.6 m m x 5.7 mm and the pixel resolution is 3.7 ixm. Up to six images were collected in each thin section. A simple staining technique with a solution of alizarin red-S in 2% HC1 was used to distinguish the carbonate composition of the binder and the silicate composition of the aggregate (Humphries 1992). As a result, the main contrast is between colourless or lightly coloured elements (mainly aggregate grains and voids) and dark elements (mainly the binder fraction) (Fig. 2). Colourless grains under plane-polarized light are the most abundant family of aggregates. The most frequent colours displayed by grains range from yellow to brown passing through green hues. Rock fragments can appear both colourless and with differently coloured areas. The binder displays a homogeneous deep red-purple colour varying with the thickness of the thin section. The variable compactness and crystallinity of the binder and possible differences in sample

Fig. 2. Optical subset image of mortar sample B5. preparation can cause some variations in the appearance of the binder. A Jeol JSM-6400 scanning electron microscope coupled to a BSE divided annular solidstate detector was used to collect grey-level BSE images at 12x, 14x and 16× magnifications. The images supported by the BSE imaging system have a resolution of 1024 pixels x 1024 pixels, resulting in pixel resolution varying from 6.3 to 5.5 p,m. In such images the distribution of phases of different composition, i.e. with a different mean atomic number, was recorded by different grey tones (Dilks & Graham 1984). In this way, images characterized by a good contrast between aggregate and binder fractions can be obtained, making the identification of grains easy (Fig. 3).

Traditional image processing Image processing provides an enhanced image where single grains of the aggregate fraction are easily discernible from the neighbouring ones and from the binder fraction. At the end of the procedure, the constituents are readily thresholded by analysing the intensity histogram of each image. The optical and BSE images were processed using the commercial IA software Image-Pro

340

F. CAR0 ETAL.

of the aggregate or binder fraction. The output of the image processing is thus a binary image where the aggregate grains are white and the binder is black. The MLP used is a three-layer feed-forward network. The neurons are grouped into input, output and hidden (i.e. those units that are neither input nor output) layers. Each neuron of a given layer is connected to all neurons of the next one (Egmont-Petersen et al. 2002). The input layer has nine neurons corresponding to the analysed pixels, the hidden layer has four neurons, and the output layer has two neurons corresponding to grain and binder classes. Supervised training must be used in the MLP whereby the network learns from a training set consisting of features input and the desired class outputs. The training set is composed of 280 pixels (140 for each class). Fig. 3. BSE subset image of mortar sample ST2.

Image analysis Plus v.4.5 (Media Cybernetics), with the following main steps: (1) enhancement of the original image and extraction of two colour channels; (2) creation of an image by averaging two colour channels and extraction of the edges of grains; (3) creation of an image with sharpened grains; (4) thresholding of the intensity histogram. After the automatic thresholding of the aggregate fraction, some guided semi-automatic and manual interactions were allowed to check rapidly the reliability of the analysis and to correct possible oversights, mainly related to the separation of touching particles and the merging of incomplete or missing grains (Russ 1999). The IA procedure has been validated in a previous study (Carb & Di Giulio 2004) by using laboratory mortars with known composition and grading curves. The macro expressly tailored for the image analysis of BSE images is directly derived from the above-mentioned procedure, the main difference being that colour processing of the images is not required as the BSE images are greylevel images.

MLP image processing Neural networks are models to express knowledge using a connectionist paradigm inspired by the working mechanisms of the human brain (Bishop 1998). A neural network model was also tested to substitute the automatic thresholding. The MLP analysed regions of 3 pixels x 3 pixels of the images, classifying them as part

Once objects are defined, the image analysis allows us to quantify the information from the images. Measures were collected through the sum of grains detected in all images captured in each thin section to obtain a single larger population of grains for each sample. In this study four parameters were used to characterize the aggregate percentages and the grain-size distribution: (1) the area composed of all pixels within the object perimeter; (2)(4) the maximum, mean and minimum Feret diameters, which were calculated by physically rotating the object outline a few degrees at a time and measuring the widths of the largest and smallest rectangles enclosing the object. The apparent volume of each grain was calculated as the volume of the equivalent ellipsoid that has the three Feret diameters as axes, and the volume-based grain-size distribution was determined by assigning to each size class the relative volume amount. The transformation from apparent to real volume-based grain-size distributions was accomplished using the linear correction factor of Car6 & Di Giulio (2004). Textural parameters were computed by the method of momentum measures (Krumbein & Pettijohn 1963). An Excel spreadsheet derived from the Excel sheet GRANPLOTS.xls devised by Balsillie et al. (2002) was used, allowing an automatic computation of mean, median, main and secondary modes, sorting, skewness and kurtosis. Finally, the binder v. aggregate ratio (B/A ratio) was derived from the ratio of the area of the counted objects to the total area, following the principle of stereology (Russ & DeHoff

ANCIENT PLASTERS A N D M O R T A R S 1999), w h i c h states that the v o l u m e f r a c t i o n o f a p h a s e w i t h i n a structure is a m e a s u r e o f the area fraction on the i m a g e , i.e. Vv = AA ( D e l e s s e ' s principle).

Results O f the 33 s a m p l e s , 22 w e r e s i e v e d ; for the r e m a i n i n g 11 s a m p l e s the a v a i l a b l e q u a n t i t y o f m a t e r i a l w a s insufficient f o r analysis. T h e c o m p l e t e set o f o p t i c a l i m a g e s w a s a n a l y s e d b y p o i n t c o u n t i n g a n d the traditional IA p r o c e d u r e , w i t h the e x c e p t i o n o f s a m p l e s L 2 4 a n d L 3 1 , b e c a u s e o f an u n s u c c e s s f u l e t c h i n g treatment. Six s a m p l e s w e r e a n a l y s e d b o t h b y traditional a n d M L P IA o f B S E i m a g e s . In T a b l e 2 the m a i n textural p a r a m e t e r s o f the s a m p l e s are r e p o r t e d f o r e a c h analytical a p p r o a c h . T a b l e 3 r e p o r t s the m a i n textural p a r a m e t e r s o f the

Table 2. Sample

LI lb L13 L24 L12 L23 L25 L31 B1 B2 B3 B4 B5 B6 B7 B8 B9 B 10 Bll ST1 ST2 ST3 ST4 L 10 Llla L14 L18a L18b B3b ST0 ST5 ST6 ST7 ST8

341

six s a m p l e s that w e r e a n a l y s e d b y m e a n s o f the two different thresholding methods. T h e s i e v e data w e r e t a k e n as r e f e r e n c e d a t a a g a i n s t w h i c h the reliability o f the p o i n t c o u n t i n g and I A p r o c e d u r e s w a s e s t i m a t e d . T h e a n a l y s e s were necessarily performed on different portions o f the s a m e original f r a g m e n t , w h i c h c o u l d e x h i b i t slightly d i f f e r e n t grain p o p u l a t i o n s . I A o f optical and B S E i m a g e s w a s p e r f o r m e d o n d i f f e r e n t s u b i m a g e s o f the s a m e thin sections, a n d the s a m e B S E i m a g e s w e r e a n a l y s e d b o t h b y traditional a n d M L P - b a s e d I A p r o c e d u r e s .

Discussion IA of optical images A s r e p o r t e d in T a b l e 2 a n d s h o w n g r a p h i c a l l y in F i g u r e 4, the textural p a r a m e t e r s c o m p u t e d b y I A are m o r e or less c o n s i s t e n t w i t h s i e v e

Main textural parameters computed using the three tested analytical procedures Type

Mortar Mortar Mortar Mortar Mortar Mortar Mortar Mortar Mortar Mortar Mortar Mortar Mortar Mortar Mortar Mortar Mortar Mortar Mortar Mortar Mortar Mortar Plaster Plaster Plaster Plaster Plaster Plaster Plaster Plaster Plaster Plaster Plaster

Sieve analysis

Point counting

Image analysis

Mean (mm)

Sorting (tr)

B/A ratio

Mean (ram)

Sorting (09

B/A ratio

Mean (ram)

Sorting (09

B/A ratio

0.76 0.26 0.74 0.45 0.70 0.79 0.75 1.02 0.42 0.41 0.43 0.34 0.99 0.88 0.41 0.54 0.57 0.51 0.27 0.29 0.28 0.65 -

1.24 0.84 1.53 1.20 1.27 1.52 1.82 1.82 1.08 1.23 1.39 0.71 2.08 1.93 1.00 1.34 1.39 1.33 0.90 0.91 0.91 1.34 -

-

0.39 0.54 0.25 0.54 0.25 0.68 0.42 0.44 0.46 0.39 0.29 0.34 0.34 0.47 0.34 0.32 0.45 0.37 0.36 0.34 0.38 0.31 0.26 0.26 0.30 0.35 0.44 0.46 0.50 0.28 0.32 0.60 0.48

0.82 1.14 0.98 1.14 0.98 1.24 1.55 0.84 1.36 0.85 0.99 0.93 1.21 1.04 0.51 0.61 0.95 0.76 1.06 1.16 1.92 1.49 1.04 1.09 0.94 1.03 0.86 1.06 1.08 1.01 1.08 1.17 1.03

0.91 1.50 1.41 1.50 1.41 0.48 0.39 0.69 0.35 0.46 0.70 0.87 1.14 0.56 0.68 0.66 1.02 1.13 0.60 0.50 0.50 0.80 1.18 1.01 0.62 3.63 2.54 0.70 0.50 0.50 0.60 0.50 0.50

0.46 0.23 0.88 1.42 0.50 0.98 2.60 0.58 0.38 0.31 0.42 0.76 0.30 1.10 0.41 0.41 0.62 0.65 0.68 0.60 0.28 0.31 0.29 0.42 0.43 0.90 0.78 0.29 0.72 0.97 0.80

0.88 0.76 1.08 0.88 0.87 1.72 1.22 1.15 0.95 0.75 1.29 1.29 0.74 1.96 0.84 0.89 1.31 1.36 0.92 1.31 0.77 0.84 0.75 0.91 0.79 1.17 1.12 0.76 1.43 1.18 1.33

0.90 0.80 1.23 0.50 0.80 0.62 0.80 0.72 0.61 0.66 0.88 0.44 0.39 0.64 0.40 0.63 0.37 0.42 0.34 0.44 0.64 0.56 0.34 3.06 1.28 0.74 0.44 0.60 0.50 0.40 0.50

342

F. CARO ET AL.

Table 3. Main textural parameters computed using the three tested IA procedures Sample

L13 L31 ST2 ST3 B1 B 11

Type

Mortar Mortar Mortar Mortar Mortar Mortar

Traditional optical IA

Mean (mm)

Sorting (or)

B/A ratio

Mean (ram)

Sorting (09

B/A ratio

Mean (ram)

Sorting (~r)

B/A ratio

0.23 0.65 0.68 0.98 0.41

0.76 1.36 0.92 1.72 0.89

0.80 0.42 0.34 0.62 0.63

0.25 0.53 0.72 1.23 0.33 0.41

0.70 1.34 1.64 1.39 0.66 0.92

0.60 0.53 0.67 0.60 0.42 0.66

0.24 0.55 0.63 1.25 0.35 0.42

0.70 1.25 1.27 1.41 0.73 0.89

0.53 0.30 0.40 0.59 0.36 0.57

references except for some cases (relative percentage error from 15% to >50%), suggesting that some limitations exist. As these noticeable deviations cannot be completely ascribed to misidentification of aggregate grains, a different explanation must be found. Figure 5 reports the sorting of samples analysed through sieve, point counting and image analyses. The aggregates range from moderately to very poorly sorted. Taking the sieve data as reference, with decreased the sorting the scatter of the results of thin-section analyses increases for point counting and image analysis. Sorting computed from thin-section analysis is generally lower than the sieve reference, indicating that the grain population is not completely represented. In these cases, the number of analysed images needs to be increased. Point counting, which affects a lower number of counted grains compared with image analysis, shows greater differences in sorting computation. 2.50

BSE MLP IA

Traditional BSE IA

; .

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

Looking at the grain-size distribution and the textural data for the aggregates, the accuracy of results diminishes as grain-size distributions become strictly bimodal and, in the cases studied, grains fall into the range of granules and pebbles. The reliability of textural analyses is thus related to the grain size and sorting of the aggregate, i.e. to the nature of the material. Volume-based grain-size distributions may not be meaningful when the aggregate is too poorly sorted and the number of counted grains is not adequate. Point counting is less sensitive to this bias as it is based on number frequencies. From these considerations, it can be concluded that the reliability of the proposed IA procedures is linked to the representativeness of the samples. In particular, results are reliable when (1) the aggregate ranges from very fine to coarse sands, and (2) the aggregate varies from very well to moderately sorted. If we look at the textural characteristics of the samples studied, it

.

i • Sieve nlmage Analysis 2,00

v

[ ! P°int C°un~n9__

1.5o

g 1.oo

0.50

0,00 ...I

-..I

r-...I

...I

m

._.1

T-_..1

..,I

O0

¢/~

[XI

03

,.,I

on

Sam pie

Fig. 4. Mean grain size of samples determined by the three analytical procedures tested. The sorting of samples increases to the right (see Fig. 5).

ANCIENT PLASTERS AND MORTARS

343

2.5

vps

Sieve

2.0

i •

Point Counting

~

~ o

Image Apa!},sis

J I III IIJI o

1.5

•

A

I

I

I

~

-¢

I

~

I

I

,¢

~

¢

~

I

¢

•

I

,,

ms rows

0.5

................................................................................................................................................................................................ 0.0

,

,

,

|

,

,

,

,

,

,

,

,

,

,

"vwS ,

,

Sample

Fig. 5. Sorting of ancient plasters and mortars determined by three analytical procedures, vws, very well sorted; ws, well sorted; mws, moderately well sorted; ms, moderately sorted; ps, poorly sorted; vps, very poorly sorted. becomes obvious that poorly sorted or very poorly sorted aggregates are commonly used for the preparation of bedding mortar mixtures. In these cases, even the analysis of the complete thin-section surface may not guarantee a reliable analysis. Of course, some oversights in grain identification always occur, because of the complexity of imaging thin sections of mortars, resulting in the erroneous merging touching grains and the artificial disaggregation of polymineralic grains (Fig. 6a). These errors can be tolerated when they do not represent a statistically significant component of the quantitative analysis.

IA of BSE images Traditional IA of BSE images can be applied to overcome the above-mentioned problems (Fig. 6b). In these experiments, the use of BSE images allows the analysis of samples showing inhomogeneities and/or features hampering the analysis of microscopic images (i.e. samples L31 and L24). Textural data from BSE images (Table 3) have been collected with minor interaction of the operator. The results are consistent with the sieve reference although some variance remains depending on the representativeness of the collected images.

I lO00om I

b

Fig. 6. (a) Example of aggregate grains as threshoided by IA of optical images; (b) example of aggregate grains as thresholded by IA of BSE images. Errors are indicated by circles.

344

F. CARO ET AL.

However, variations in brightness intensity of the BSE images can complicate the detection of aggregate and binder fractions by traditional histogram-based thresholding. Additionally, some difficulties in the image analysis arise because of the similar backscattering coefficient of some components.

MLP-based IA of BSE images The accuracy of the MLP thresholding was first assessed on samples ST1, ST3 and L13; these samples were selected as they show slightly low, medium and high contrast between aggregate and binder fractions. To estimate the classification performance, the MLP thresholding was compared with the manual classification of the aggregate and binder fractions of the same samples from 3 pixel × 3 pixel regions randomly sampled from the images. The test set was composed of 631 pixel neighbourhoods; MLP classified correctly 335 of 335 neighbourhoods of aggregate grains (100.0%) and 283 on 296 neighbourhoods of binder (96.0%). The results were obtained using a classification strategy based on the choice of the class suggested by the highest output. Textural results are presented in Table 3 and indicate that this approach can be successfully implemented to identify the aggregate fraction of plasters and mortars when it is difficult to establish a fixed threshold that can be used in all images (Egmont-Petersen et al. 2002). By using the MLP technology it was possible to establish a more flexible threshold that fitted different BSE images. The chance of reducing the number of applied algorithms, i.e. transformations of the original image, increased the accuracy of thresholded features, as aggregate grains that disappeared or were reduced in size during the traditional IA procedure are now preserved. Although this fact does not influence the computation of grain-size parameters, it does noticeably influence the determination of the B / A ratio (Table 3).

Conclusions Image analysis showed some capabilities that can be successfully applied at different levels of approximation for the characterization of plasters and mortars. However, the reliability of the analyses proved to be dependent on: (1) the textural characteristics of the analysed material; (2) the compositional nature of the aggregate fraction.

The textural characteristics of the studied samples revealed that poorly sorted or very poorly sorted aggregates were commonly used for the preparation of bedding mortar mixtures whereas moderately well sorted or moderately sorted aggregates were used for plasters and fine mortars. In the first case, the analysis of the complete thin-section surface may not guarantee a reliable analysis. In contrast, plasters and fine mortars showed textural characteristics that guarantee reliable thin-section analyses. In this case, IA proves to be an indispensable tool for the characterization of such materials where, for technical and historical-artistical reasons, only small amount of samples may be taken, which would be insufficient for sieve analysis. Additionally, the experiments show that much attention must be given to the development of more adaptable procedures to overcome failures in grain detection, which are typical of complex and variable systems. The use of BSE images combined with MLP technologies represents a promising solution to the problem. The authors are grateful to P. Jacobs and V. Cnudde of the Department of Sedimentary Geology and Engineering Geology of Gent University for their valuable help in collecting the BSE images. The research has been supported by grants provided by FAR funds (Pavia University, to. M. Cobianchi).

References BALSILLIE, J. H., DONOGHUE, J. F., BUTLER, K. M. &

KOCH, J. L. 2002. Plotting equation for gaussian percentiles and a spreadsheet program for generating probability plots. Journal of Sedimentary Research, 72, 929-943. BISHOP,C. M. 1998. Neural Networks for Pattern Recognition. Clarendon Press, Oxford. fARO, F. t~ Di GIULIO,A. 2004. Reliability of textural analysis of ancient plasters and mortars through automated image analysis. Materials Characterization, 53, 243-257. CHERMANT, J.-L., CHERMANT, L., COSTER, M., DEQUIEDT, A.-S. & REDON, C. 2001. Some field

of applications of automatic image analysis in civil engineering. Cement and Concrete Composites, 23, 157-169. DILKS, A. & GRAHAM, S. C. 1984. Quantitative mineralogical characterization of sandstones by back-scattered electron image analysis. Journal of Sedimentary Petrology, 55, 347-355. EGMONT-PETERSEN, M., DE RIDDER, D. & HANDELS, H. 2002. Image processing with neural networks-a review. Pattern Recognition, 35, 2279-2301. FRANCUS, P. 1998. An image-analysis technique to measure grain-size variation in thin sections of soft clastic sediments. Sedimentary Geology, 121, 289-298.

ANCIENT PLASTERS AND MORTARS HUMPHRIES, D. W. 1992. The Preparation of Thin Sections of Rocks, Minerals, and Ceramics. Microscopy Handbooks 24. Oxford University Press, Oxford. KRUMBEIN, W. C. & PETTIJOHN, F. J. 1963. Manual of Sedimentary Petrography. Appleton-CenturyCrofts, New York. MONTANA, G. 1995. Mineralogical-petrographic characterization of plasters by BSE images and their digital processing. Science and Technology for Cultural Heritage, 4(2), 23-31. PERRING, C. S., BARNES, S. J., VERRALL,M. & HILL, R. E. T. 2004. Using automated digital image

345

analysis to provide quantitative petrographic data on olivine-phyric basalts. Computers and Geosciences, 30, 183-195. PIRARD, E. 2002. Quantitative micro-imaging of building materials: from reality to dream. In: Workshop CSTC, KU Leuven, La microscopie appliqu~e aux mat~riaux de construction, 1-3. Russ, J. C. 1999. The Image Processing Handbook, 3rd edn. CRC Press, Boca Raton, FL. Russ, J. C. & DEHOFF, R. T. 1999. Practical Stereology, 2nd edn. Plenum, New York. TUCKER, M. E. 2001. Sedimentary Petrology, 3rd ed. Blackwell Science, Oxford.

Index Page numbers in italic denote figures. Page numbers in bold denote tables. Abri Pendimoun geological setting 35 Neolithic pottery 33-47 analytical methods 36-38 ceramic finds 36, 39-42 chemical content 44, 45 mineralogy 43 petrography 39 raw materials 35, 37, 42-43 recipes 43, 45, 46 temper 34, 39-41, 45, 46-47 acerc6 170 adze, Italian, prehistoric 260, 261,262, 264-267 Alps Central Italian, pietra ollare 229-238 Italian, high pressure meta-ophiolites 258, 264 alum, cobaltiferous 92, 95, 97-98 Amarna, cobalt blue pottery 91, 92, 93, 94, 95, 96-99 amphibole, Ferrara bricks 135 amphorae Egyptian 93 Micaceous, Tas-Silg 82, 83, 87, 89 anatase 22 antimonate, glass opacifier 178-9, 180-184, 185 Apennines, high pressure meta-ophiolites 258, 259, 264 aplite 39, 41, 45, 46 Apulia geological setting 50-51, 52 Neolithic pottery 49-61 analytical methods 50 firing 59 mineralogy 60 Murge plateau 51, 52, 53-57, 58-59 raw materials 51-57 preparation 57-59 Tavoliere Plain 50-51, 52, 53, 55-56, 58 temper 58-59 archaeometry l - 3 problems 2 - 3 Archaeoraman 9, 10, 11, 15, 16, 27 Archaic, Late, fibre-tempered pottery 119 Argentina see obsidian, northwestern Patagonia Argille Subappennine 50, 51, 153 arsenic, in pigment 159-160 ash, in glassmaking tree 187, 217 plant 96, 204-208, 212-213 axe Guatemalan, jade, RM 19, 21 Meso-American, RM 10, 18 blades, Italian, prehistoric 260, 261,262, 264-267

Belus River see Levant, glassmaking, sand Bern, La Trne ceramics 63, 65, 66-67, 69-70, 75-76 bindheimite 184 Biscuit Ware 82, 83, 86 Borg in-Nadur Ware 82, 83, 86, 87 bracelets, stone ring, Neolithic 260, 261,262, 267 Bradanic Cycle see Argille Subappennine Bradanic Trough 52, 56, 57 bricks, medieval, Ferrara 127-140 Bricky Red Ware 82, 83, 86, 89 Bronze Age glass 201,202 pottery, Tas-Silg 82, 84, 87 stone tools 257-268 calcite, sparry 39-41, 45, 46 calcium antimonate, in glass 178-179, 182, 185 Campania felix, Roman glass production 175 cave paintings, prehistoric, RM 17-18, 24, 26 ceramics Iron Age technology, Galilee 101-116 Late La Trne, Switzerland 63-80 vitrified wall, RM 19, see also La T~ne; majolica; pottery Ceramiraman 10, 15, 16 Chassey Culture 33, 34, 37, 46 chert 257 Bedoulian 33, 34 tools, western USA 307-320 geochemical composition 308-309, 310-311, 313, 316-318, 319, 320 geology 308 petrography 312, 313,314, 315-316, 320 chisels, stone, Neolithic 260, 261,262, 264 Cholila, Patagonia, obsidian 241,242, 243-244, 245, 247,248, 250, 252-253 clay calcareous, storage pots 101, 103-109, 113-114 carbonate 42, 43, 44, 45 marly 42, 43, 44, 45 non-calcareous, cooking pots 101, 113, 114, see also Argille Subappennine cobalt blue Egyptian pottery 91-99 Laterza majolica 155, 158-160, 161 RM 20 Roman glass 183 coins, copper, RM 10, 19-20 conservation, medieval stone architecture 323-334 contamination, chemical, La Trne pottery 65 copper coins, RM 10, 19-20 Egyptian glass 91, 93 Roman glass 182, 183, 185

348 Copper Age, stone tools 257-268 corrosion, metals, RM 10, 19-20 cotto see terracotta Crescent site see Stallings Island Culture Crisp Ware 82, 83, 87 crystal orientation 21-22 cullet 21 l, 212 Dakhla Oasis, cobalt pigment 92, 93, 95, 97 Derriere Sairoche glassworks 187-198 Hupper sands 188-198 chemical composition 190, 191 grain size 190, 192, 193, 194-195, 197, 198 mineralogy 189-190 devitrification 208 Drab Coarse Ware 82, 83 eclogite 261,262, 263, 264-265, 266, 266, 267 axe 18 Egypt glassmaking 91,208, 209 New Kingdom, cobalt blue pottery 9 1 - 9 9 scarab, RM 10, 21-22 el-Raqqa, glassmaking 206, 212-213 Elko, Nevada, chert tools 308-309, 310-311,312, 313, 315. 316-320 enstatite 2 l Enviroraman 15, 16 epidote Ferrara bricks 136 Etruscan age, stelae, Valdelsa Valley 274-281 faience 151, 152 cobalt blue 91-92 Ferrara 128 medieval and Renaissance bricks 127-140 chemical composition 127-128, 129-130, 131 firing phases 132-136 mineralogical composition 131 - 136, 137-139 raw materials 128, 131, 132, 136 temper 128 firing phases, Ferrara historical bricks 132-136 firing techniques Grkeytip pottery t43. 148-149 Iron Age, Galilee 101 - 102, 109- l I l, 113 Neolithic 59, 61 flashing, stained glass 220, 221-222, 223 flint 257 Frescoraman 10, 15, 16 flit 96 fritting 170, 172 Galilee Iron Age pottery technology 101 - 116 ceramic consolidation 113-114 ceramic matrix composition 106-109, 110-112 firing 101-102, 1 0 9 - I l l , ll3 manufacturing technology 114-115 origin 114 petrography 103-106 raw materials 101-103 tempering 111 - 113 garnet 10, 24, 25 Gemmoraman 10, 15, 16

INDEX gemstones Florentine tables, RM lO, 25 Medieval cloisonn6 gold, RM 10, 24-25 Moghul, RM 10, 23, 24 Navaratna, RM 10, 23 Roman intaglios, RM 10, 18 Gen~ve, La T~ne ceramics 63, 67-68. 70-73, 76-77 glass Egyptian cobalt blue 91-92 Late Antiquity and Islamic 201-213 fuel supply 21 l primary and secondary workshops 202 soda-lime-silica glass types 202-205, 203 high-magnesia 204-205, 212-213 H1MT 207, 208, 21 l low-magnesia 204. 207. 208-212 sources of lime 206-208 sources of silica 205-206 sources of soda 203-205 pre-industrial, Derriere Sairoche glassworks 187-198 recycling 185, 209. 210 Rhenish 207 Roman 206, 211-212 game counters, Pompeii 175-185 analysis 176, 178 blue and blue-green 179-180, 182. 183 chemical composition 176. 177, 178-184 colourless 182 production cycle 184-185 red 181 white 178-179, 182 yellow and yellow-green 180-181, 183-184 opaque 176, 177, 178-182 transparent 177. 182, 201.211 stained Medieval, Europe 217 Pavia Certosa 218-226 analysis 218-220 chemical composition 223-225 flashed 220. 221-222, 223 glass types 225-226 RM I0, 20-21 glauconite pellets 37, 39, 40, 42, 44. 45 glaucophane 260, 262, 263. 265 glaze cobalt blue 91 Laterza majolica 156-158, 161 lead, Islamic and mtidejar Spain 163-172 ceramic body 165 chemical composition 166, 167 colour 166. 168-169 glaze interface 165. 166-168 microstructure 165. 166 tin-glaze 163. 169-170. 171. 172 gneiss. G6keytip pottery 141-150 G6keytip, Turkey 142 geology 141. 143 golden mica cooking pottery 141-150 chemical composition 145, 146. 147, 148-149 firing 143. 148 mineralogical composition 144-146, 148 production 142-143 gold, cloisonn6, Medieval 24-25

INDEX greenschist 229, 261,262, 265 greenstone 34, 257, 258, 261 Grotte du Four, La T~ne ceramics 63, 68, 73, 77

Haifa, Bay of, glassmaking sand 206 Holocene, western USA chert tools 307-320 geology 308 Hupper sands, Swiss Jura 188-198 chemical composition 190, 191 grain size 190, 192, 193, 194-195, 197, 198 mineralogy 189-190 Iconoraman 10, 15, 16 image analysis plaster and mortar 323-344 Multi-Layer Perceptron neural network 340, 341,342, 344 inclusions, micro 21 Inspector Raman 24, 26, 27 intaglios, Roman, RM 10, 18 iron, RM 10, 20 Iron Age, pottery technology, Galilee 10t-116 Italy Chassey Culture 33, 34 Etruscan stelae, Valdelsa Valley 273 281 medieval architectural stone, L'Aquila 323-334 medieval bricks, Ferrara 127-140 medieval stained glass, Pavia 217-226 modern majolica, Laterza 151-161 Neolithic pottery, Apulia 49-61 pietra ollare, Central Alps 229-238 plaster and mortar, Pavia, image analysis 337-344 prehistoric polished stone tools 257-268, 258 jadeite 19, 21, 23,266 jades Neolithic tools 261,262, 263, 264, 265, 266, 267 RM 10, 19, 21, 23, 24 Jura, Derriere Sairoche glassworks 187-198 kaolinite 1 0 1 - 102 Karnak, cobalt blue pottery 93-94 Kharga Oasis, cobalt pigment 92, 93, 95, 97 knapping, obsidian 247, 252, 253

La TSne, Late ceramics 63-80 analytical methods 64 Bern 63, 65, 66-67, 69-70, 75-76 chemical contamination 65 Gen~ve 63, 67-68, 70-73, 76-77 Grotte du Four 63, 68, 73, 77 La T~ne 63, 68, 73, 77 Matin 63, 68, 73, 77-78 regional or local production 63, 79-80 St.Triphon-Massongex 63, 68, 73, 78 Yverdon 63, 68, 74, 78 lapis lazuli 25 L'Aquila, medieval architectural stone 323-334 Laterza

modern majolica 151-161,152 raw materials 153, 154 lazurite 25 lead in glaze Islamic and mtidejar, Spain 163-172 Laterza majolica 156-158, 161 RM 10, 20 in Roman glass 180-184 lead antimonate, glass opacifier 180-184, 185 Levant blue painted pottery 93 glassmaking 208-209, 210, 211 sand 205-209 lime, in glassmaking 206-208

magnesia, in glassmaking 204-205 majolica, modern Laterza 151-161,152 ceramic body composition 154, 155-156, 160 chemical composition 154-156, 158-159, 160 glaze 156-158, 161 mineralogical composition 154-155 pigment 155, 158-160, 161 raw materials 153, 154, 156 Malkata, Thebes, cobalt blue pottery 91, 93-99 Malta, pottery 81-89 Matin, La Tbne ceramics 63, 68, 73, 77-78 Massongex, La T~ne ceramics 63, 68, 73, 78 Memphis, cobalt blue pottery 93 meta-ophiolites, high pressure Northern Italy 257, 258, 259, 261-262, 262, 263, 264, 2 6 6 axe blades 264-267 Metalloraman 10, 15, 16 metals, corroded copper coins, RM 10, 19-20 iron ingot, RM 10, 20 lead plates, RM 10, 20 mica Ferrara bricks 134 Grkeyiip pottery 142-150 Laterza majolica 155 Micaceous Ware amphorae 82, 83, 87, 89 micromapping, Raman 9, 21-22 microscopy optical 49 Raman 9-27 advantages 13-14 classification 15-16, 15 disadvantages 14-15 immobile, micromapping 21-22 immobile vertical microscope 16-21 mobile 9, 14 horizontal microscope 22-23 optical fibre under air 23-25 under glass 25-26 under water 26 ultra-mobile 24, 26 Middle Ages, Spain, lead glaze 163-172

349

350 monocrystals, Aztec, RM 23 mortar see plaster and mortar moss, Spanish see Tillandsia usnedoides Multi-Layer Perceptron neural network 340, 341,342, 344 Murge plateau 51, 52, 53-57, 58-59

natron 92, 96, 187, 204 Neolithic pottery Abri Pendimoun 33-47 Apulian 49-61 Tas-Silg 82 stone tools 257-268 nephrite 19, 23, 259, 263, 265-267 neural network model 340, 341,342, 344 neutron activation analysis Tas-Silg pottery 81-89 procedure 82-84 Nile Valley clays 96-97, 98 pottery 93

obsidian 34 Mediterranean 257 northwestern Patagonia 241-254 analysis 244, 245-246, 247 chemical composition 247, 248-251,252 geology 242-243 knapping 247, 252, 253 sources 242-243, 253 Old Red Sandstone, Lower 284-285 petrology 288, 289-298, 299 omphacite 261,262, 263, 265, 266 opacifier, glass 178-179, 180-184, 185 Orange ware, fibre-tempered pottery 120, 122, 124 oxidation, G6keyiip golden mica pottery 149-150

palmetto 122, 124 Patagonia, obsidian 241-254, 242 Pavia, ancient plaster and mortar, image analysis 337-344 Pavia Certosa medieval stained glass 218 analysis 218-220, 219 chemical composition 223-225 flashed glass 220, 2 2 1 - 2 2 2 , 223 glass types 225-226 pestles, stone, Neolithic 267 Petroraman 10, 15, 16 Piedra Parada, Patagonia, obsidian 241,242, 243, 244, 245-246, 248-250, 252-253 pietra ollare 229-238 analysis 230-231 artefacts 230, 238 composition 231-233 porosity 233, 234, 234, 238 thermal properties 233-236, 237, 238 pietre verdi see greenstone pigments Aztec, RM 22, 23 Egyptian

INDEX cobalt blue 91-99 composition 95-97 production 92 Laterza majolica 155, 158-160, 160, 161 Oceanian, RM 25 prehistoric cave paintings, RM 17-18, 24, 26 Roman wall-paintings, RM 17 plaster and mortar, ancient, Pavia textural analysis 337-344 image analysis techniques 339-344 Pliny the Elder, account of Roman glassmaking 175, 185, 206, 207, 211 Pompeii, glass game counters 175-185 pottery Egyptian, New Kingdom cobalt blue 91-99 decorative technique 93 fibre-tempered, Stallings Island Culture 119-124 firing 59 golden mica, G6keyiip 141-150 Iron Age, Galilee 101 - 116, see also Galilee, Iron Age pottery technology Islamic and mtidejar, Spain 163-164 local 33-34, 43, 63, 65, 81-82, 84, 89 Neolithic Abri Pendimoun 33-47 Apulian 49-61 Tas-Silg, Malta 81-89 thermal qualities i 24 Pozzuoli, Roman glass production 175 Punic Period, pottery, Tas-Silg 82, 83, 84, 86, 87, 89

Raman, Sir Chandrasekhara Venkata I l, see also microscopy, Raman Raman effect I l, 13, see also scattering, Raman Stokes Raman spying 9, 27 Ramanita method 14, 18, 25 Raqqa see el-Raqqa Rayleigh tail 13 recycling, glass 185, 209, 210 Resinoraman 15, 16 RM see microscopy, Raman Roucadour cave, ultra-mobile RM 24, 26 ruby 23-25

St. Triphon, La Tbne ceramics 63, 68, 73, 78 Salicornia 204 Salsola 204

Sandy Pink Ware 87 scarab, Egyptian, RM I0, 21-22 scattering Raman Stokes I l, 12 Rayleigh 12, 13 Scotland, medieval carved sculpture 283-304 sculpture, medieval Scotland 283-304 geological .setting 284-286, 287 magnetic susceptibility 288, 299, 301, 302, 303, 304

petrology 286-288 Lower Old Red Sandstone outcrop 287, 288, 289-298, 299

INDEX West Highland outcrop 300-301,301 Pictish 283, 284-286, 284, 287, 299-300, 302, 303 West Highland 283, 285, 286, 287, 301-302, 303 Seg2 chemical group 85-86, 87 Segesta, Sicily, pottery 85-86 serpentinite 259, 261,262, 263, 265, 267 shell, in glassmaking 207 Sid6rolithique, Swiss Jura 187-188 SILA chemical group 84-86, 87, 87 SILB chemical group 85, 86, 89 silica sources, in glassmaking 205-206 silicates, sheet 144, 146, 147, 148-150 slip 92, 95-98, 157 smectite 102, 144 soapstone see pietra ollare soda in glassmaking 202, 217 sources 203-205 Soft Brown Ware 82, 83 South Carolina see Stallings Island Culture Spain, Islamic and mtidejar lead glaze 163-172 Spanish moss see Tillandsia usnedoides spectroscopy infrared 11, 13, 14 Raman 11 - 15, see also Archaeoraman; microscopy, Raman spinel, cobalt 92, 96, 102 springs, hydrothermal 273, 279, 280-281 Square Mouthed Pottery-phase I Culture see VBQ I Stallings Island Culture fibre-tempered pottery 119-124, 120 aplastic components 121-122, 123 fibre components 122-123 paste components 123-124 petrography 120-124, 122 raw materials 124 stelae, Etruscan Valdelsa Valley 274-281,275 carbonate fabric 276-277 depositional evolution 277-281 stone medieval architectural L'Aquila 323-334 properties 326, 332, 334 red lithotypes 326, 329-332 restoration 323, 334 weathering and decay 332, 333, 334 white lithotypes 324-328, 329 polished jade, RM 19, 23, see also jades Meso-American, mobile RM 10, 22-23 Meso-American axe, RM 10, 18 prehistoric tools Italy 257-268 high pressure meta-ophiolites 257, 258, 259, 26l-262, 263, 264, 266 strontium, in glassmaking 207-208, 212-213 Switzerland ceramics, Late La T~ne 63-80 Derriere Sairoche glassworks 187-198 Tas-Silg, Malta, pottery local production 84 neutron activation analysis 81-89

351

Tavoliere Plain 50-51, 52, 55-56, 58 Tel-Hadar Iron Age pottery 101-116, see also Galilee, Iron Age pottery technology telescopy, Raman 9, 27 temper Abri Pendimoun Neolithic pottery 34, 39-41, 45, 46-47 Apulia Neolithic pottery 58-59 calcite 103, 104, 105, 111-113 fibre Orange ware 120, 124 Stallings Island Culture 119-124 Galilee, Iron Age pottery 103, 104, 105, 111-113 gneissic, G6keyiJp 142-150 sand 120, 128 Temple Period Ware 82, 83, 84, 86, 87 terra rossa 42-43, 44, 45, 51, 57, 58, 103, 104, 105, 106 terracotta, Ferrara 127, 129-130, 131, 136 Thebes, cobalt blue pottery 92, 93 Thermi Ware 82, 83, 84, 87 Thorn's Creek Ware 120, 122, 124 Tillandsia usnedoides 122-123, 124 tin in glass 182-184 in glaze Islamic and mtidejar, Spain 163, 169-170 majolica 156-158 Tissueraman 15, 16 Tolt River, Washington, chert tools 308-309, 310-311, 314, 315-320 tools Guatemalan axe, RM 19, 21 Holocene chert, western USA 307-320 Meso-American axe, RM 10, 18 prehistoric stone, Italy 257-268, 262, see also adze; axe blades; chisels travertine 273-274, 280, 281 tufa, calcareous 273-274, 280, 281 Turkey, golden mica cooking pottery, G6key~ip 141 - 150 Tyre, glassmaking 205-206, 212-213 USA, South Carolina see Stallings Island Culture western, chert tools 307-320

Valchiavenna, pietra ollare 230-238 Valdelsa Valley, Etruscan stelae 273-281,274, 280 Valmalenco, pietra ollare 230-238 VBQ 1 33, 34, 37, 46 vitrification bricks 128 walls, RM 19 Vitroraman 10, 15, 16 wall-paintings, Roman, RM 17-18 walls, vitrified, RM 19 West Highland outcrop, petrology 300-301,301 windows, stained glass, Pavia 217-226 wollastonite 202, 208 Woodland, Early, fibre-tempered pottery 119, 120, 122 Yverdon, La T~ne ceramics 63, 68, 74, 78

Geomaterials in Cultural Heritage Edited by M. Maggetti and B. Messiga

This volume gives a broad view of the application of geoscience techniques to the study of monuments and objects from excavations and museums, including their origin, technique of manufacture, age and conservation. It reaffirms the . ~-~ .~.~ ~. important contribution of geosciences in the interdisciplinary approach to the study of complex materials such as minerals, rocks, ,.'~ 41[, r ? . I glass, metals, mortar, plaster, slags and pottery. The papers in this book cover three topics: the study of pottery, glass, stone and mortar; the application of Raman spectroscopy to a wide variety of objects; and the future of archaeometry. Interdisciplinary studies including field geology, geophysics, microscopy, textural analysis, physical methods and geochemistry are used to unlock information from the ancient materials, such as the provenance of the raw materials, the firing technology, the ancient recipes, and the alteration pathways. Visit our online bookshop: http:I/www, geolsoc.org.uk/bookshop Geological Society w e b site: http:/lwww.geolsoc.org.uk

Cover illustration:

Wasters of the amphora workshop of Rosignano, italy. (Width of photo 50crn ) Photographer: G Thierrin-Michael