Proceedings of the 4th Symposium of the Hellenic Society for Archaeometry: National Hellenic Research Foundation, Athens 28-31 May 2003 9781407301884, 9781407332338

The 4th Symposium of the Hellenic Society for Archaeometry was held at the National Hellenic Research Foundation in Athe

279 41 126MB

English Pages [699] Year 2008

Report DMCA / Copyright


Polecaj historie

Proceedings of the 4th Symposium of the Hellenic Society for Archaeometry: National Hellenic Research Foundation, Athens 28-31 May 2003
 9781407301884, 9781407332338

Table of contents :
Cover Page
Copyright Page

Citation preview



Proceedings of the 4th Symposium of the Hellenic Society for Archaeometry National Hellenic Research Foundation, Athens 28-31 May 2003

Edited by

Yorgos Facorellis Nikos Zacharias Kiki Polikreti

BAR International Series 1746 2008

Proceedings of the 4th Symposium of the Hellenic Society for Archaeometry National Hellenic Research Foundation, Athens 28-31 May 2003 Edited by

Yorgos Facorellis Nikos Zacharias Kiki Polikreti

BAR International Series 1746 2008

Published in 2016 by BAR Publishing, Oxford BAR International Series 1746 Proceedings of the 4th Symposium of the Hellenic Society for Archaeometry © The editors and contributors severally and the Publisher 2008 The authors' moral rights under the 1988 UK Copyright, Designs and Patents Act are hereby expressly asserted. All rights reserved. No part of this work may be copied, reproduced, stored, sold, distributed, scanned, saved in any form of digital format or transmitted in any form digitally, without the written permission of the Publisher.

,6%1SDSHUEDFN ,6%1HIRUPDW '2,KWWSVGRLRUJ $FDWDORJXHUHFRUGIRUWKLVERRNLVDYDLODEOHIURPWKH%ULWLVK/LEUDU\ BAR Publishing is the trading name of British Archaeological Reports (Oxford) Ltd. British Archaeological Reports was first incorporated in 1974 to publish the BAR Series, International and British. In 1992 Hadrian Books Ltd became part of the BAR group. This volume was originally published by Archaeopress in conjunction with British Archaeological Reports (Oxford) Ltd / Hadrian Books Ltd, the Series principal publisher, in 2008. This present volume is published by BAR Publishing, 2016.

BAR PUBLISHING BAR titles are available from: BAR Publishing 122 Banbury Rd, Oxford, OX2 7BP, UK E MAIL [email protected] P HONE +44 (0)1865 310431 F AX +44 (0)1865 316916

EDITORIAL The 4th Symposium of the Hellenic Society for Archaeometry was held at the National Hellenic Research Foundation in Athens, from May 28 to 31, 2003. The scientific program contained oral and poster contributions. The meeting was attended by 298 participants from 25 different countries. The 86 selected contributions presented in this volume, cover a wide spectrum of topics from novel methods in Archaeometry to applications of traditional techniques to various kinds of archaeological material. The official languages of the Symposium were Greek and English and an abstract in both languages is attached to all the papers presented in here. The submitted manuscripts were peer-reviewed for their originality, significance and technical validity, by two independent referees. The thematic sessions of the Symposium correspond to the chapters of this edition and included: Dating – Authenticity, Geophysical Prospection, Geoarchaeology, Palaeoenvironment – Palaeodiet, Palaeoanthropology, Material Characterisation Techniques, Ceramics, Glass, Stones, Mortars, Metals, Painting Media, Organics and Conservation The abstracts of the Symposium presentations and also pictures of the events can still be seen at the web page of the HSA The web page will soon host information for the next 5th HSA Symposium scheduled for October 2008, in Athens. The editors would like to take this opportunity to thank the following colleges for undertaking the slavish task of paper reviewing usually more than one contribution per referee: G. Adamiec, E. Aloupi, D. Anglos, B. Argyropoulos, Y. Bassiakos, C. Beck, H. Bocherens, Ch. Bouras, J. Buxeda I Garrigos, G. Chrysikos, M. Colombini, E. Dimou, S. Fox, C. Gaffney, Α. Hauptmann, A. Hein, E. Ioakimoglou, R. Jones, L. Joyner, I. Kakouli, N. Kallithrakas-Kontos, E. Kiriatzi, M. Korres, K. Kotsakis, I. Liritzis, Y. Maniatis, F. McCoy, J. P. Northover, G. Papadimitriou, St. Papamarinopoulos, V. Perdikatsis, Ε. Pernicka, M. Pollard, Τ. Rehren, A. Sarpaki, A. Sarris, V. Shumak, G. Theodorou, C. Trapalis, G. Tsokas, G. Vakoulis, E. Vardala-Theodorou, M. Varti-Mataranga, G. Wagner and I. Whitbread. An acknowledgement for financial support goes to the following organisations: Ministry of Culture, Ministry of the Aegean and Island Policy, National Hellenic Research Foundation, Technical Chamber of Greece, The American School of Classical Studies in Athens, Hippocampus Hotel – Naoussa – Paros. Grateful thanks go to the members of the Organising Committee, Y. Bassiakos, E. Aloupi, V. Kiriatzi, T. Vakoulis and M. Avgouli as well as the speakers and the participants for their contribution to the scientific success of the conference. The editors Yorgos Facorellis Nikos Zacharias Kiki Polikreti





Dating – Authenticity Individual Age Estimation Based on the D-aspartic and D-glutamic Acid of Teeth J. Csapó, Zs. Csapó-Kiss, É. Varga-Visi, G. Pohn and M. Collins New Prospects in Obsidian Hydration Dating: An Integrated Approach I.Liritzis, C.M. Stevenson, S.W. Novak, I.Abdelrehim, V.Perdikatsis and M. Bonini Luminescence Dating Beyond Pottery: a Review N. Zacharias Hunting out Forgers of Archaeological Marble Objects: Reviewing the Archaeometrist’s Role as a Detective K. Polikreti Accelerator Mass Spectrometry Dating of Epicurus “De Natura” papyrus from Herculaneum C. Lubritto, F. Terrasi, A. D’Onofrio, C. Sabbarese, F. Marzaioli, I. Passariello, A. Palmieri, G.Casa, D. Rogalla, M. Rubino, G. Imbriani, M. Romano, L. Gialanella, V. Roca, C. Rolfs, M. GIancaspro, and A. Travaglione. A Single Aliquot Polymineral TL Protocol Applied for Fired Materials C.T. Michael, N. Zacharias, I. Lefakis and D. Dimotikali Comparative Studies on the Determination of Environmental Radiation Doses N. Zacharias, M. Štuhec, V. Kilikoglou, Y. Bassiakos, C.T. Michael1, E. Vardala-Theodorou, G. Theodorou4, D. Dimotikali Altered Archaeological Pottery and the Effect on TL Dating N. Zacharias, C.T. Michael, A. Schwedt and H. Mommsen

1 9 23

31 37

41 45


Geophysical Prospection The Canal of Xerxes in Northern Greece: Fact or Fiction? Recent Geophysical and Geoarchaeological Investigations 59 B.S.J. Isserlin, M. Arvanitis, R.E. Jones, V. Karastathis, S.P. Papamarinopoulos, P. Stephanopoulos, G. Syrides and J. Uren The Reconstruction of Prehistoric Shorelines in Dokos Island, Aegean Sea, Using Remote Sensing Techniques 65 G. Papatheodorou, M. Geraga and G. Ferentinos Identifying Burned Mud Brick Buildings using Fluxgate Magnetometry 73 M. Boyd and N. Brodie Application of Near-surface Geophysical Tools and GIS for Mapping the Ancient City of Lefkas 77 A. Sarris, S. Topouzi, F. Triantafyllidis, S. Soetens and G. Pliakou Geophysical Research at Syntagma Square, Athens – Relationship to Geology and Ancient Texts 85 M.G. Papaioannou. S. P. Papamarinopoulos and P. Stefanopoulos Integrated Geophysical Studies at Ancient Itanos (Greece) 89 A. Vafidis, A. Sarris and TH. Kalpaxis The Geophysical Research at the Church of St. Theodora at Vasta Megalopolis, Greece 99 St. P. Papamarinopoulos, P. Stefanopoulos, M. G. Papaioannou, N. Charkiolakis and Ch. Vachliotis Ιnvestigation of a Monumental Macedonian Tumulus by Three Dimensional Seismic Tomography 107 L. Polymenakos, S. Papamarinopoulos, A. Liossis and Ch. Koukouli – Chryssanthaki Geophysical Investigations in Pelasgia & Andron 113 A. Sarris, S. Topouzi, S. Soetens, A. Baba, M-F. Papakonstantinou and K. Psarogianni


Use of Remote Sensing and GIS Methods in the Reconstruction of Coastal Palaeogeography of Alexandria, Egypt A. Chalari, D. Christodoulou, G. Papatheodorou, M. Geraga, A. Stefatos and G. Ferentinos Prospecting of the City Wall at the Asea Prehistoric Site (Greece) Using Geoelectric and GPR Techniques M. Dogan, St.P. Papamarinopoulos and P. Stefanopoulos



Geoarchaeology The Most Famous Eruption of the Thera Volcano: a Review of More Than Sixty Years of Studies V. M. Francaviglia Between Peak and Palace. Reinterpretation of the Minoan Cultural Landscape in Space and Time S. Soetens, A. Sarris and K. Vansteenhuyse Geological Contribution to the Stratigraphic Study of the Palaeolithic Sediments at the Lakonis Beach, Gythion, Greece E. Chiotis Troy – Geomorphologic Changes in the Valley of the River Scamander P. Malfas

141 153

163 171

Palaeoenvironment – Palaeodiet Cooking Ingredients from Bronze Age Archondiko: the Archaeobotanical Evidence 187 S.M. Valamoti, A. Papanthimou and A. Pilali Seasonality and Radiology: a Pilot Application on Red Deer (Cervus Elaphus) Dentaries from the Upper Palaeolithic Cave of Kastritsa, NW Greece 195 E. Kotjabopoulou and C.N. Kaftantzis From Sea to Land: the Exploitation of Molluscs in the Geometric Artisan Site of Skala Oropou, Attica 205 T. Theodoropoulou Elemental Analysis of Bones and Diet Reconstruction of the Inhabitants of Thracia (Bulgaria) in the Hellenistic Period 215 B. Zlateva, I. Kuleff, R. Djingova, M. Kuzmanov and D. Gergova

Palaeoanthropology In Search of Mothers and Daughters in 3000 Year Old Burials in Northern Greece L. Kovatsi, S. Triantafyllou, J. Eliopoulos, M. Besios, S. Andreou, A. S. Tsaftaris, A. Dimitriadou and S. Kouidou Ancient DNA Extraction and Amplification of Human Bone Samples from the Area of Delphi: a Pilot Case Study Κ. Κoniavitou, Chr. Κroupis, Α. Κapsalis, L. Karali, Pl. Petridis and F. Μavridis Pygmies, Sciapodes, Macrocephali, Argippeans: mythical Or Existent Tribes? Euterpe Bazopoulou-Kyrkanidou


233 241

Ceramics And Glass The Significance of Fabric Diversity in Neolithic Knossos Ceramics S. Dimitriadis Early Helladic II Pottery From Thebes: an Integrated Typological, Technological and Provenance Study J. Hilditch, E. Kiriatzi, K. Psaraki and V. Aravantinos Regional Production Centers of Iron Age Pottery Wares in Philistia: a Provenance Study Employing Inductively-Coupled Plasma (ICP-AES and ICP-MS) D. Ben-Shlomo Non-destructive Quantitative XRF Analysis on Greek Archaic Pottery of the Alaimo Sanctuary in Lentini (Sicily) iv





L. Grasso, L. Pappalardo and F.P. Romano On the Origin of Stamped Amphorae from Thracian Cities in Bulgaria I. Kuleff, E. Pernicka and T. Stoyanov Minoan and Mycenaean Metallurgical Ceramics: Analysis on Crucibles and Mould from Kommos, South Crete C. Oberweiler, Y. Maniatis and J. Shaw Thin Sectioning Neolithic Identities: the Red Monochrome Ware (A1) from Middle Neolithic Sesklo, Thessaly A. Pentedeka and K. Kotsakis Imported and Local Pottery in Eastern Mediterranean Sites During the Early Byzantine Period: the Case of Eleutherna, Crete A. G. Yangaki, E. Aloupi, V. Kilikoglou, A. Tsolakidou and P. Themelis New Evidence for Amphora Production in East Crete in the Hellenistic Period. an Integrated Approach of Typology and Thin Section Petrography N. Vogeikoff-Brogan, E. Nodarou and M. C. Boileau Understanding the Technology of Large Storage Jars: Raw Materials and Techniques in Hellenistic and Modern Messenia M. Giannopoulou and E. Kiriatzi The Study of Bricks of Roman Monuments I. Papayianni and M. Stefanidou Medieval Glass Bracelets from South East Bulgaria R. Georgieva, Y. Dimitriev, B. Samuneva, E. Kashchieva and L. Doncheva-Petkova A Natron Source For Glass Making In Greece? Preliminary Results E. Dotsika, Y. Maniatis and D. Ignatiadou Composition and Microstructure of Decorative Enamels Included in Roman Fibulae Found in Bulgaria E. Kashchieva, R. Kirov, Y. Dimitriev and S. Tsaneva






335 349 353 359


Stones And Mortars The Material Characterization of Masonry Stones and Mortar From the Archaeological Site in Jerusalem Irene Wasserman Exchange Networks in the Neolithic of Greece: Gabbro and Talc Objects from Drakaina Cave, Kephalonia Island, Western Greece G. Stratouli and V. Melfos Raw Material Procurement in the Middle Palaeolithic of Thessaly: Sourcing Studies P. Κarkanas, E. Panapogopoulou, J. Anoussis and N. Kyparissi-Apostolika Assessment of Aegean Obsidian Sources by a Portable ED-XRF Analyser: Grouping, Provenance and Accuracy I. Liritzis PGAA Analysis of Szeletian Felsitic Porphyry – Non-destructive Analysis of an Important Hungarian Palaeolithic Raw Material A. Markó, Z. Kasztovszky and T. Biró, K. Quantitative Fabric Analysis (QFA) on Marble from West Anatolia: Application of Raster- (Fractal) and Vector-based (Geometric) Approaches Judit Zöldföldi and Balázs Székely Bronze Age Painted Plaster from the Greek Mainland: a Comparative Study of Its Technology by Means of XRD Analysis and Optical Microscopy Techniques A. Brysbaert and V. Perdikatsis Classification of Historical Mortars by Principal Component Analysis, Based on Their Physico-chemical and Mechanical Properties v


381 389






K. Polikreti, A. Moropoulou, A. Bakolas and P. Michailidis Technology of Mortars Used for the Substratum of Mosaics I. Papayianni and V. Pachta


Metals Analysis of Archaeological Objects with LMNTI, a new Transportable LIBS Instrument 441 K. Melessanaki, A. Mastrogiannidou, S. Chlouveraki, S.C. Ferrence, P.P. Betancourt˝and D.Anglos Quantitative Analysis of Athenian Coinage by Pixe 445 Ch. Flament, O. Lateano and G. Demortier Metal Artefacts from Early Bronze Age Poliochni on Lemnos: Archaeometric Analysis in Archaeological Perspective 451 Ch. Flament, O. Lateano and G. Demortier Towards an Integrated Approach for the Study of Ancient Greek Technology: Iron and Archaeology in Aegean Thrace 459 M. Kostoglou The Tréstina Bronzes: Archaeometry and Archaeology 469 M. Ferretti, M. Ferretti, A. Palmieri, E. Formigli, M. Miccio, R. Pecchioli, A. Romualdi, F. Lo Schiavo, E. Macnamara, B. Shefton, L. Lepore and A. Naso Non Destructive Investigation of the Ag Distribution in the Late Roman Folles from about 300 to 310 ad. 481 S. Garraffo, A. Nicolosi, G. Pappalardo, L. Pappalardo and F.P. Romano Metallurgical Study of Three Bronze Rams from Warships, Dating Back to Classical, Hellenistic and Roman Times 487 G. J. Varoufakis A First Valuation of the Excavation of Ancient Washeries at Bertseko Laureotika 493 C.G. Tsaimou The Laurion Shafts of Antiquity: First Investigations and the Problems of Ventilation. Development of a Physical Model. 499 D. Morin and R. Herbach Complex Beauty: the Manufacture of Hellenistic Wreaths 507 E. Asderaki and Th. Rehren Investigating the Presence and Effects of Arsenic in Two Bloomery Iron Clots (2nd – 3rd C. AD) from Kereliai, Lithuania 515 J. Navasaitis, A. Sveikauskaitė and I. Keesmann A Tentative Comparison of Steel-making Crucibles from Central Asia and the Indian Subcontinent 519 O. Papachristou and Th. Rehren The Study of Ancient Copper Slags from Serifos 529 G.D. Papadimitriou and A.Z. Fragiskos Archaeometallurgical Researches in the Othris Mountains, Greece 535 M. Tizzoni, C. Cucini Tizzoni, G. Rebay and M.P. Riccardi New Evidence oν the Manufacturing Technology Used in the Production of Roman Coins (260 – 350 AD) 545 C. Vlachou-Mogire, J.G. McDonnell and R.C. Janaway

Painting Media A Case Study of Purple Adulteration in Hellenistic Textiles E. Kampasakali, E. A. Varella, E. Pavlidou and E. Lambrothanassi The Use of Red and Yellow Ochres as Painting Materials in Ancient Macedonia V. Perdikatsis and H. Brecoulaki The Decoration of the Rhodian White-ground Hadra Hydriae: an Initial Approach vi

555 559 569

A. Giannikouri, K. Andrikopoulos, S. Sotiropoulou, Sister Daniilia and Y. Chrysoulakis Identification of Proteinaceous Binding Media from Post-Byzantine Mural Paintings with Chromatographic Techniques E. Ioakimoglou, E. Kavalieratou, E. Kouloumpi, S. Zevgiti and K. Stoupathis Biodegradation of Art Objects: Impact on Animal Glue Identification A. K. Tsakalof, I. S. Aslani, F. N.Kolisis, K. A. Bairachtari and I. D. Chryssoulakis Spectroscopic Ellipsometry: a Non-Destructive Technique for Measuring the Optical Properties of Varnish Films on Paintings K. Polikreti, A. Othonos, C. Christofidesm and C. De Deyne Τypical Application of Petrographical Study for the Support of Conservation Projects in Μansions at Chios and Inousses E. Polychroniadou, E. Chiotis, E. Dimou and D. Tarenidis The Artistic Traits of Gyzis: a First Diagnostic Approach to His Paintings S. Sotiropoulou, Sister Daniilia, K. Andrikopoulos, G. Karagiannis and Y. Chryssoulakis

579 587


599 603

Organics Analytical Study of the Internal Structure and the State of Fossilization of Skeletal Remains from the Site of Pikermi (Attica, Greece) E. T. Stathopoulou, G. E. Theodorou and V. Psycharis Organic Residues from the Late Neolithic Makriyalos Cooking Pots D. Urem-Kotsou, K. Kotsakis, C.W. Beck and E.C. Stout Development of Fast Derivatisation Procedures for the Chemical Investigation of Organic Residues in Early Byzantine Transport Containers from Ephesos/Turkey R. Linke, S. Stanek, S. Lochner-Metaxas and E. Rosenberg

611 619


Conservation A Comparative Study of Degradation Patterns of Petrified Material from Greek Fossilized Forests V. Lampropoulos, G. Panagiaris, A. Karampotsos and E. Velitzelos Study of Three Consolidants used in Sub - Fossilized Bones that are Going to be Exposed In Situ (in Outdoor Conditions) G. Panagiaris, V. Lampropoulos, I. Kotsifakos, P. Pavlakis and M. Dermitzakis Analysis of the Orange-brown Patina on Marble Surfaces from the Olympieion in Comparison With a Similar Patina from the Propylaea, Acropolis K. Polikreti, Y. Maniatis, Y. Doganis, A. Galanos and P. Valavanis Investigation of Microcracks in Marble from Mt. Penteliby Dielectric Spectroscopy D. Triantis, C. Anastasiadis, I. Stavrakas and F. Vallianatos The Bust of Lenin by Memos Makris: Forms of Corrosion and Conservation Work D. Charalambous, A. Karampotsos and V. Lampropoulos



659 667 671

Ancient Technology Temporary Wooden Bosses: New Remarks on Lifting Devices in the Ancient World A. Nakassis The Structures of V-IV Century B.C. of the Nisyros and Pyrgousa Islands (Dodecanese): Building Techniques, Nature and Sourceland F. Burragato, A. Curuni-Spiridione, M. Di Filippo and B. Toro




INDIVIDUAL AGE ESTIMATION BASED ON THE D-ASPARTIC AND D-GLUTAMIC ACID OF TEETH J. Csapó, Zs. Csapó-Kiss, É. Varga-Visi, G. Pohn University of Kaposvár, Faculty of Animal Science, H-7401 Kaposvár. P.O.Box 16. Guba S. u. 40. Hungary, [email protected]

M. Collins Fossil Fuels and Environmental Geochemistry, NRG, Drummond Building, University of Newcastle-upon-Tyne, Newcastle-upon-Tyne, NE1 7RU, U.K.

J. Csapó Janus Pannonius University Faculty of Natural Sciences, Natural Geography Department, H-7624 Pécs, Ifjúság u. 6. Hungary Abstract:: Last year we presented a new age estimation method of the individual - based on racemisation of aspartic- and glutamic acid in teeth. In the present study we show the correlation between our racemisation method and methods used in anthropology. Calibration curves based on our earlier investigations were used for the age estimation of 65 adults (39 males and 26 females) of unknown individual age from the Avar Period series of Kereki-Homokbánya (Hungary), and for 88 adults (50 males and 38 females) of unknown individual age from 9th-10th century cemetery of Visegrád (Hungary). We tried to apply the calibration diagrams produced by our amino acid racemisation method on tooth samples originating from historical times. The age of individuals from both cemeteries were estimated by the anthropological methods in advance. We presented these ages by linear regression and this way it also presented the correlation of ages estimated by anthropological methods and by the two D-amino acid contents. This correlation of ages estimated by the traditional and the new D-aspartic acid and D-glutamic acid content methods was extremely close. When comparing the results of anthropological age estimation to those of the method based on amino acid racemisation the value of r exceeded 0,9 with both amino acids. For the sake of increasing the accuracy of the method we determined the concentration of the D-aspartic acid and D-glutamic acid and the D/L ratio on both two amino-acids of the incisors, eye teeth, premolar teeth and molar teeth taken from the skull of a female skeleton estimated to be 40-45 years old by anthropological methods. From our research it seems that the incisors contain more glutamic acid than the others, and the aspartic acid content of the first incisor and the first molar tooth is also higher than the others. Examining the D-amino acids and the D/L ratio we ascertained that the D-aspartic acid and the D-glutamic acid content and the D/L ratio of the two incisors are the highest among all the teeth, followed by the eye teeth, the premolar teeth and the molar teeth. We received the least amount of D-amino acid content and D/L ratio at the third molar tooth. From our experiments it seems that the later growing teeth contain less D-amino acids than the ones coming out earlier, which shall be taken into consideration at the age determination based on D-amino acid concentration of teeth. In the comparative experiments it is practical to use teeth of the same age, because the differences among the teeth of the individual can be bigger than the difference between individuals, which makes age determination uncertain. Περιληψη:: Τον τελευταίο χρόνο παρουσιάσαμε μια νέα μέθοδο εκτίμησης ηλικίας ατόμων, η οποία βασίζεται στη ρακεμοποίηση του ασπαρτικού και γλουταμινικού οξέος δοντιών. Στην παρούσα εργασία παρουσιάζουμε τη συσχέτιση μεταξύ της μεθόδου ρακεμοποίησης και των μεθόδων της ανθρωπολογίας. Οι καμπύλες βαθμονόμησης, οι οποίες βασίζονται σε παλαιότερη έρευνά μας, χρησιμοποιήθηκαν για την εκτίμηση της ηλικίας 65 ενηλίκων (39 ανδρών και 26 γυναικών) άγνωστης ηλικίας, από τις σειρές της περιόδου Avar της Homokbánya ánya nya (Ουγγαρία), και 88 ενηλίκων (50 ανδρών και 38 γυναικών) άγνωστης ηλικίας από το νεκροταφείο περιοχής Kereki-Homokbánya -Homokbánya Visegrád ádd (Ουγγαρία) (9ος-10ος αιώνας). Προσπαθήσαμε να εφαρμόσουμε τις καμπύλες βαθμονόμησης της μεθόδου ρακεμοποίησης σε δείγματα δοντιών ιστορικών χρόνων. Η ηλικία των ατόμων και από τα δύο νεκροταφεία εκτιμήθηκε αρχικά με ανθρωπολογικέ μεθόδους. Παρουσιάζουμε τις ηλικίες αυτές σε διάγραμμα γραμμικής παλινδρόμησης, συσχετίζοντας τις ηλικίες που εκτιμήθηκαν με ανθρωπολογικές μεθόδους με την περιεκτικότητα στα δύο D-αμινοξέα. -αμινοξέα. Οι συσχετίσεις αυτές των ηλικιών που εκτιμήθηκαν με την παραδοσιακή μέθοδο και τη νέα μέθοδο περιεκτικότητας σε D-ασπαρτικό -ασπαρτικό και D-γλουταμινικό -γλουταμινικό οξύ, βρίσκονται εξαιρετικά κοντά. Συγκρίνοντας τα αποτελέσματα των ανθρωπολογικών ηλικιών με αυτά της μεθόδου ρακεμοποίησης αμινοξέων βρίσκουμε ότι η τιμή του συντελεστή συσχέτισης r υπερβαίνει το 0,9 και για τα δύο αμινοξέα. Για να αυξήσουμε την ακρίβεια της μεθόδου προσδιορίσαμε την περιεκτικότητα σε D-ασπαρτικό -ασπαρτικό οξύ και D-γλουταμινικό -γλουταμινικό οξύ, καθώς και το λόγο D/L /L L των δύο αμινοξέων, σε δείγματα από τους τομείς, κυνόδοντες, προγόμφιους και γομφίους, κρανίου γυναίκας, ηλικίας 40-45 χρόνων, σύμφωνα με τις ανθρωπολογικές μεθόδους. Από την έρευνά μας συμπεραίνουμε ότι οι τομείς περιέχουν περισσότερο γλουταμινικό οξύ από τα άλλα δόντια και ότι η περιεκτικότητα σε ασπαρτικό οξύ του πρώτου τομέα και του πρώτου γομφίου είναι υψηλότερη από τις υπόλοιπες. Ελέγχοντας τα D-αμινοξέα -αμινοξέα και τους λόγους D/L /L L επιβεβαιώσαμε ότι οι περιεκτικότητες σε D-ασπαρτικό -ασπαρτικό και D-γλουταμινικό -γλουταμινικό οξύ καθώς και ο λόγος D/L /L L των δύο τομέων είναι τα υψηλότερα όλων. Ακολουθούν οι κυνόδοντες, οι προγόμφιοι και οι γομφίοι. Η μικρότερη ποσότητα D-αμινοξέος -αμινοξέος και λόγου D/L /L L βρέθηκε στον τρίτο γομφίο. Από τα πειράματά μας είναι φανερό ότι τα δόντια που αναπτύχθηκαν τελευταία έχουν μικρότερη περιεκτικότητα σε D-αμινοξέα -αμινοξέα από εκείνα που αναπτύχθηκαν νωρίτερα, γεγονός που πρέπει να ληφθεί υπόψη κατά την εκτίμηση ηλικιών με βάση τη συγκέντρωση D-αμινοξέων -αμινοξέων δοντιών. Σε συγκριτικά πειράματα, είναι πρακτικό να χρησιμοποιεί κανείς δόντια ίδιας ηλικίας, επειδή οι διαφορές μεταξύ δοντιών του ίδιου ατόμου μπορεί να είναι μεγαλύτερες από τη διαφορά μεταξύ διαφορετικών ατόμων, έτσι ώστε η τελική εκτίμηση ηλικίας να αποδειχτεί αναξιόπιστη.


J. Csapó, Zs. Csapó-Kiss, É. Varga-Visi, G. Pohn, M. Collins Introduction

atom, the four different groups can occupy two different spatial arrangements that are a nonsuperimposable mirror image of each other. These two forms represent a class of stereoisomers called enantiomers which are optically active, that is they rotate plane polarized light. Special nomenclature has been developed to specify the absolute configuration of the four substituents of asymmetric carbon atom. The absolute configuration of amino acids is specified by the D, L system based on the absolute configuration of the glycerinaldehyde. For all chiral amino acids stereoisomers having a configuration related to that of L-glycerinaldehyde are designated L, and stereoisomers related to D-glycerinaldehyde are designated D.

The possible most accurate estimation of age at the time of death is a significant part of the study of historical populations. In everyday practice age means the time span between the birth and death of any individual human being. It is traditionally measured in years, months, weeks, etc. In other words it is calendar or chronological age. The usual age estimation methods of historical anthropology observe and analyse the changes occurring on the skeleton and on the teeth with the progress of age. Growth, maturation and aging are all processes that leave characteristic traces on human bones and teeth, therefore the so-called skeletal and dental ages can be estimated. Skeletal and dental ages come from the biological age of any given individual. If and when the person’s biological age is close to his or her calendar age our estimate for his or her skeletal (or dental) age can fall very close to his or her calendar age (Ubelaker 1989). This way we are able to produce an indirect estimation of calendar age via biological age determination.

Gillard et al. (1990) analysed the D-Asp content of molars and premolars. They failed to establish significant differences between the D-Asp contents of the same teeth when taking samples from several parts of their crowns and roots. Ohtani and Yamamoto (1992) found significant differences between estimated and actual ages when comparing the D-Asp contents of dental enamel and dentin substance. They detected that racemization (kAsp: 5.75x10-4 year-1) is much faster in dentin than in enamel (kAsp: 4.47x10-4 year-1). They concluded that age could be more accurately estimated on the basis of dentin (±3 years) than on the basis of enamel (±2-11 years). They also calculated post-mortal reaction speed coefficients for an average temperature of 15 oC (dentin: kAsp: 9.70x10-8 year-1 and enamel: kAsp: 1.330x10-7 year-1).

When estimating age the anthropologist analyses ontogenesis, the biological aging of the human organism. These processes are determined by a multitude of external (environmental) and internal (genetic) factors. Chronological age runs at a steady pace, but the passing of biological age can be and really is very varied between individuals and even within one human frame too, among its constituent parts. This is the reason for which we cannot achieve a truly accurate calibration of our age measuring methods compared to the passing of chronological time: this sort of calibration could be nothing else but an individual one. As biological and calendar ages often don’t concur, chronological age in paleoanthropology may be estimated only within certain intervals (5 years at least). There can be no other realistic aim for us but to score within these limits with the greatest plausibility possible. Some constituents of the human body, such as teeth, are easy to examine even centuries after the death of the individual, as their changes are not significantly affected by the environment. Their analysis could significantly improve the accuracy of age estimation.

Ritz et al. (1993) analysed the dentin substance of third molar roots and they arrived at the conclusion that racemization of the root’s dentin was somewhat different to that of the crown’s dentin. They produced a special calibration diagram for those cases where age has to be determined but the crown substance was damaged or impaired. They established that the degree of Asp racemization was multiplied in acid solvent proteins compared to non-acid solvent ones. Ohtani (1994) examined Asp racemization on central and lateral incisors, on first and second molars and he also treated the averages of these. He found a close correlation of actual ages and D/L Asp ratios. Ohtani concluded that racemization within deciduous teeth was a good indicator for the estimation of individual age, but racemization of permanent was far less useful for the same purpose. Ritz et al. (1995) have reported the application of this method to biopsy samples of dentine. Thanks to the strictly regulated nature of the sampling process ages estimated on the basis of Asp racemization presented a very close conformity to actual ages. Ritz et al. (1995) established a margin of error not larger than ±1 year for 46.4% of the cases analysed and the error of age estimation never went beyond the ±5 years limit.

Helfman and Bada (1975) were the first to declare that the aspartic acid racemization process within teeth could be utilised to estimate the age of living animals and of humans. Then they established the reaction speed coefficient of aspartic acid for human teeth at 8.29x10-4 year-1, but they measured it at 7.87x10-4 year-1 a year later (Helfman & Bada 1976). Bada and Brown (1980) produced a calibrating diagram by plotting ln (1+D/ L)/(1-D/L) against time. They found a satisfactory level of conformity between the actual age data and the data estimated on the basis of amino acid racemization. For all the standard amino acids except for glycine, the α carbon is bonded to four different groups; the α carbon atom is thus a chiral center. Because of the tetrahedral arrangement of the bonding orbital around the α carbon

Improvements in chromatographic methods have increased the number of chiral amino acids that can be separated in 2

Individual Age Estimation Based on the D-aspartic and D-glutamic Acid of Teeth a single analysis. Nevertheless the ready separation, rapid racemization and high concentration of aspartic acid in teeth, means that it is still the most widely studied amino acid. We have previously estimated the racemization halflives of different amino acids (Csapó et al 1994, 1998). It was concluded that D-enantiomers of amino acids with faster racemization than that of Asp promise to be equally good indicators of individual age as Asp, and amino acids falling of slower racemization rate could also provide useful information on age.

lysozyme, cytochrom C, fossil bone sample, and individual free amino acids as follows: L-aspartic acid, L-glutamic acid, L-threonine, L-alanine, L-valine, L-phenylalanine, L-histidine and L-tryptophan. The classical Moore and Stein (Moore & Stein 1963) method using 6M hydrochloric acid medium for 24 h at 110 oC was compared with hydrolysis in the same medium but at higher temperature (160-180 o C) with shorter reaction times (15-60 min) using closed Pyrex tubes as reaction vessels. The use of microwave oven was also attempted. The method of Einarson et al. (1987a) was used to monitor the racemization (HPLC column: Kromasil C8 5 μm, 250 x 5.6 I.D.; eluent: gradient system A = 40% methanol in 9.5mM phosphate buffer at pH = 7.05, B = acetonitrile; pre-column derivatization reagent: o-phthalaldehyde/2,3,4,6-tetra-O-acetyl-1-thio-βD-glucopyranoside (OPA/TATG)).

Materials and methods Amino acid determination method The determination of the D/L amino acid ratio in hydrolyzed proteins requires chromatographic methods that are suitable for the simultaneous separation and highly sensitive detection of several amino acids and their enantiomers. In order to be able to measure the changes of the amino acid composition and/or the enantiomeric ratio of the individual amino acids in very small archaeological samples the use of the highly selective and sensitive high-performance liquid chromatographic method (HPLC) is inevitable. Pre-column derivatization of α-amino acids using ophthaldialdehyde (OPA) and various thiols as the reagents leads to fluorescent formed derivatives, while another reagent, 9-fluorenyl-methyl chloroformate (FMOC-Cl) transforms α-amino and imino acids to fluorometrically active derivatives. Both types of derivatives have good chromatographic properties enabling several amino acids to be separated and measured within one chromatographic run at very low concentration levels.

Results and discussion Hydrolysis studies With the use of higher hydrolysis temperature e. g. 160 oC, 170 oC or 180 oC, much shorter reaction time is adequate to achieve complete hydrolysis related to classic hydrolysis conditions when 24 hour heating is required for the same purpose at 110 oC. Even splitting of the most stable bonds adjacent to Val, Ile and Leu is completely accomplished at 160 oC for 60 min, at 170 oC for 45 min and at 180 oC for 30 min. The rate of the racemization in the course of the hydrochloride acid catalyzed protein hydrolysis decreases in the following order: aspartic acid, glutamic acid, threonine, phenylalanine, alanine, valine, histidine, as shown in Table 1 on the examples of lysozime, cytochrom and bone sample. Tryptophan was almost completely decomposed under the examined conditions (Csapó et al 1986).

If the aim is the enantiomeric separation of amino acids, chiral derivatization agent should be used which transforms the amino acid enantiomers to pairs of diastereoisomeric derivatives separable on achiral HPLC columns. The chiral 1-(9-fluorenyl)-ethyl chloroformate (FLEC) reagent (Einarsson et al 1987a), and also the OPA/thiol reagent are suitable for this purpose if a chiral thiol is used in the latter case (Einarsson et al 1987b). Both types of reagents were successfully used in our study.

The rate of the racemization of peptide bound amino acids in 6M hydrochloric acid between 110 and 170 oC is 4-7 times higher than that of the free amino acids. Microwave-promoted hydrolysis is advantageous if racemization occurring due to this sort of treatment has no importance related to the subject of the study. If minimizing racemization is a prerequisite of the success of the research, this method is definitely disadvantageous. As seen in Table 1, the degree of racemization in case of high temperature - short reaction time methods is lower by 20-55% than at 110 oC for 24 h. Two optimal conditions were selected: 160 oC for 60 min or at 170 oC for 45 min, and these were selected for the age estimation studies.

The prerequisite of a successful chromatographic analysis is the complete hydrolysis of the protein content of teeth samples. All hydrolyses were carried out using 6M hydrochloric acid. If the aim is the determination of the enantiomeric ratio, it is very important to keep the extent of racemization at the lowest possible level. The effect of the reaction times and temperatures on the extent of the hydrolysis and racemization has been carefully investigated (Csapó et al 1997). Hydrolysis of proteins with reduced racemization

Age estimation of teeth based on the D-aspartic and D-glutamic acid contents

The following materials were used for testing the racemization during hydrolysis: bovine ribonuclease,

Our research project was carried out in collaboration with the Institute of Odontology and Earth Sciences 3

J. Csapó, Zs. Csapó-Kiss, É. Varga-Visi, G. Pohn, M. Collins Amino acid Asp Glu Thr Ala Val Phe His Amino acid Asp Glu Thr Ala Val Phe His

110 oC for 24 h A 6.62 4.58 3.62 2.99 2.11 3.31 1.83

A 3.29 2.81 2.11 1.72 1.71 2.89 1.52

160 oC for 60 min. B 7.01 4.61 3.74 3.21 2.24 3.42 1.89 170 oC for 45 min. B 3.57 2.89 2.23 1.77 1.82 3.11 1.60

C 7.89 5.93 4.38 4.02 2.53 3.64 2.38

A 3.27 2.79 2.29 1.69 1.69 3.19 1.64

C 4.42 3.74 3.04 2.11 2.27 3.60 1.99

A 3.84 3.51 2.87 2.81 2.54 2.97 1.79

B 3.42 2.84 2.31 1.65 1.84 3.37 1.67

C 4.15 3.61 3.14 2.13 2.33 3.57 2.01

180 oC for 30 min. B 3.99 3.63 3.14 2.89 2.57 2.83 1.93

C 4.67 3.92 3.42 3.04 2.82 3.20 2.11

Table 1. D-amino acid content of lysozyme (A), cytochrome (B) and fossil bone (C) hydrolysed by 6 M hydrochloric acid at different temperatures for different times1

Centre of Gothenburg University. We analysed two recent tooth sample series to establish the so-called calibrating diagrams. In 1998-99 we gathered 22 teeth from the dental surgery of Pannon Agricultural University’s Faculty of Animal Sciences in Kaposvár and we measured the Dand L-aspartic acid contents of them. When planning the sample we attempted to include individuals with in the largest possible age envelope (17-62 years), and we also tried to select enough individuals from each age group to have a comprehensively representative sample.

Age (year) 17 20 21 22 22 24 24 25 27 28 31 32 35 40 42 43 43 44 46 53 53 62

Table 2 contains the data produced by analysing the Kaposvár dental sample of 1998. We calculated ln(1+D/ L)(1-D/L) correlations both for aspartic and glutamic acids besides D/L aspartic and D/L glutamic acid ratios. D/L ratios as well as the ln(1+D/L)(1-D/L) function were presented as a function of age. We calculated correlations of known ages and the D/L ratios of the two amino acids by linear regression. We found a very close positive relation between D/L ratio and age in case of aspartic acid contents. The value of r was 0.91 for the D/L ratio as well as for the calculated function. When analysing glutamic acid we concluded that the values of r fell between 0.980.99 for the D/L ratio as well for the calculated function. Our examination of this dental sample of 22 teeth also led us to the conclusion that D-aspartic acid is a useful indicator for the estimation of individual age if treated to the analytical methods (protein hydrolysis, derivative production, separation and identification of D- and Lenantiomers) we applied. We also drew the conclusion that D-glutamic acid content is also suitable for accurate age estimation beside D-aspartic acid, though D-glutamic acid is present in teeth in a smaller concentration because of its different racemization half-period. That is the reason why

D/L ratio Asp 0.034 0.035 0.036 0.037 0.038 0.039 0.038 0.041 0.042 0.043 0.044 0.044 0.047 0.050 0.052 0.053 0.053 0.053 0.055 0.059 0.060 0.065

ln(1+D/L)/(1-D/L) Glu 0.017 0.017 0.019 0.019 0.020 0.021 0.019 0.021 0.021 0.021 0.022 0.022 0.024 0.026 0.026 0.027 0.028 0.027 0.028 0.031 0.030 0.033

Asp 0.068 0.070 0.072 0.074 0.076 0.078 0.076 0.082 0.084 0.086 0.088 0.088 0.094 0.100 0.104 0.106 0.106 0.106 0.110 0.118 0.120 0.130

Glu 0.034 0.034 0.038 0.038 0.040 0.042 0.038 0.042 0.042 0.042 0.044 0.044 0.048 0.052 0.052 0.054 0.056 0.054 0.056 0.062 0.060 0.066

Table 2. D-amino acid content of teeth of different age

it is more difficult to measure and its scoring is a more demanding job for researchers. Our first conclusions were based on a numerically small sample but we supported them by analysing 102 dental samples in 1999. At the same time we opened up our field of research from comparatively young age groups towards older ones. Our 1999 examinations produced r=0.93

The values refer to the percentage of racemization expressed as the ratio [D/(D+L)]x100. Each value is the mean of triplicate determinations.



Individual Age Estimation Based on the D-aspartic and D-glutamic Acid of Teeth 40-86 years on the basis of the D-aspartic and D-glutamic content of his or her tooth. The results of our 1999 research work confirmed those arrived at in 1998, so we may state the existence of an extremely close link between any individual’s age and the D-aspartic acid content of his or her tooth, and the D/L aspartic acid ratio on the basis of the analysis of a numerically large sample of teeth. Another one of our assumptions was also proven correct: it was not just D-aspartic acid content but also D-glutamic acid content that could be used to estimate the age of an individual if a sufficiently sensitive method of analysis was provided to measure the small concentration of Dglutamic acid present. In the third phase of our work we tried to apply the calibration diagrams produced by our amino acid racemization method on tooth samples originating in historical times. The age of individuals from Avar cemetery were estimated by anthropological methods in advance. The average of their age estimated by “traditional” anthropological methods, D/L aspartic and D/L glutamic ratios as well the age of these bone samples calculated on the basis of D/L aspartic acid and D/L glutamic acid was presented in Figure 3 (unbroken line = Asp, broken line = Glu). In Figure 3 we also presented these ages by linear regression and this way it also presented the correlation of ages estimated by anthropological and by the two D-amino acid contents. It was evident from this figure that this correlation of ages estimated by the traditional and the new D-aspartic acid content methods was extremely close. When comparing the results of anthropological age estimation to those of the method based on amino acid racemization the value of r exceeded 0.9 with both amino acids.

Figure 1. Linear regression between the age of life of the individuals and the D/L aspartic acid ratio of their teeth.

Figure 2. Linear regression between the age of life of the individuals and the D/L glutamic acid ratio of their teeth.

positive correlations between D/L ratios and individual ages both in case of D-aspartic and D-glutamic acids. The relation of an individual’s age and the D-aspartic acid content of his or her tooth was presented in Figure 1, the correlation of age and D-glutamic acid was presented in Figure 2. These two correlations are eminently suitable to estimate the age of any individual in the age envelope of

Figure 3. Linear regression between the age of life estimated by anthropological methods and D amino-acid content.


J. Csapó, Zs. Csapó-Kiss, É. Varga-Visi, G. Pohn, M. Collins We investigated whether are there any differences in the D-aspartic acid and D-glutamic acid content of teeth from the same skull. For the sake of increasing the accuracy of the method we determined the concentration of the Daspartic acid and D-glutamic acid and the D/L ratio on both two amino-acids of the incisors, eye teeth, premolar teeth and molar teeth taken from the skull of a female skeleton estimated to be 40-45 years old by anthropological method. From our researches it seems that the incisors contain more glutamic acid than the others, and the aspartic acid content of the first incisor and the first molar tooth is also higher than the others (Figure 4). Examining the D-amino acids and the D/L ratio we ascertained that the D-aspartic acid and the D-glutamic acid content and the D/L ratio of the two incisors are the highest among all the teeth, followed by the eye teeth, the premolar teeth and the molar teeth (Figure 5). We received the least amount of D-amino acid content and D/L ratio at the third molar tooth. From our experiments it seems that the later growing teeth contain less D-amino acids than the ones erupting earlier, which shall be taken into consideration in the age determination based on D-amino acid concentration of teeth. In the comparative experiments it is practical to use teeth from the same age, because the differences among the teeth of the individual can be bigger than the difference between individuals, which makes age determination uncertain.


L-amino acids (g/100g)

1,2 1,0 0,8 0,6 0,4 0,2 0,0

L-aspartic acid









Figure 4. L-aspartic acid and L-glutamic acid content of different teeth samples from the same skull (I1, I2 = incisor; C = canine; P1, P2 = premolar; M1, M2, M3 = molar).

D/L ratio; D-amino acid (g/100g)

0,06 0,05 0,04 0,03 0,02


The significance of the anthropological utilisation of the amino acid racemization based age estimation method can be summarised as followings: A single tooth or even a fragment with some enamel is sufficient for amino acid based analysis. It could be an especially useful characteristic when treating poorly preserved remains. The amino acid racemization based age estimation method is built on an exact foundation of natural sciences. Its application was clear-cut, and therefore free of intra- and interpersonal errors. Amino acid racemization provides help where it is the most urgently needed: in the estimation of age within the most difficult adult group. The results produced by the amino acid racemization based method presented a fine correlation to those of other “classic” anthropological methods. When used in combination they could confirm each other’s results. The errors of the analysis can occur in sampling of the tooth, the basic or heavy metal environment of the samples which influences the racemization, racemization during hydrolysis of the protein and errors of the analytical method. The errors of the method can be diminished if other calibration curves are formulated based on the method described here.

D-aspartic acid D/L Asp ratio D-glutamic acid





P1 4





Figure 5.Composition of the different teeth samples from the same skull (I1, I2=incisor, C=canine, P1, P2=premolar, M1, M2 ,M3=molar).

life and genetical facilities. In fact, amino acid racemization measured the passing of chronological time and not that of biological time, and therefore it gave a completely new meaning to the word ‘age’ in historical anthropology. The advantage of the method elaborated by us compared to the previous ones can be summarized as follows: - We do not only determine the age of the individual by the aspartic acid, but by the D/L ratio of the aspartic acid and the glutamic acid, and the average of the two estimations are considered as the real age. - If different teeth are available from the different individuals, then we can correct the age of the individual at the time of death by analysing the teeth from the same skull. The accuracy of the analysis can be improved if different teeth are available.

For theoretical consideration, this method was developed within a different framework compared to all routine age determination methods in anthropology. In contrast to all the other methods, it did not take into consideration the genetically programmed evolution of the organism, nor its responses to the environment or its physiological adaptation. It measured the structural alterations of amino acids, which are processes independent of circumstances of

Conclusion After reviewing previous attempts to use the extent of amino acid racemization for the determination of the age of individuals, the authors describe their own approach. After an optimised protein hydrolysis with low racemization, the 6

Individual Age Estimation Based on the D-aspartic and D-glutamic Acid of Teeth D- and L-aspartic acid content in teeth samples of known age (anthropological methods) was determined by HPLC after precolumn derivatization of the two amino acids. Plotting the D/L ratio as a function of time for the two amino acids, calibration curves were obtained which can be used for the age determination of individuals in the range of 16-80 years.

Csapó, J., Csapó-Kiss, ZS., Csapó, J.Jr., 1998, Use of the amino acids and amino acid racemization for age determination in Archaeometry, Trends in Analytical Chemistry, 17, 140−148. Einarsson, S., Folestad, S., Josefsson, B., 1987a, Separation of amino acid enantiomers using precolumn derivatization with o-phthalaldehyde and 2,3,4,6,-tetra-O-acetyl-1-thio-β-Dglucopyranoside, Journal of Liquid Chromatography, 10, 1589−1596. Einarsson, S., Josefsson, B., Möller, P., Sanchez, D., 1987b, Separation of amino acid enantiomers and chiral amines using precolumn derivatization with (+)-1-(9-fluorenyl)ethyl chloroformate and reversed-phase liquid chromatography, Analytical Chemistry, 59, 1191-1198. Gillard, R.D., Pollard, A.M., Sutton, P.A., Whittaker, D.K., 1990, An improved method for age at death determination from the measurement of D-aspartic acid in dental collagen, Archaeometry, 32, 61−70. Helfman, P.M., Bada, J.L., 1975, Aspartic acid racemization in tooth enamel from living humans, Proceedings of the Natural Academy of Sciences, USA, 72, 2891−2894. Helfman, P.M., Bada, J.L., 1976, Aspartic acid racemization in dentine as a measure of ageing, Nature, 262, 279−281. Moore, S., Stein, W.H., 1963, Chromatographic determination of amino acids by the use of automatic recording equipment, In Methods in Enzymology (S.P. Colowick and N.O. Kaplan eds.), 6, 819-831. Ohtani, S. and Yamamoto, K., 1992, Estimation of age from a tooth by means of racemization of an amino acid, especially aspartic acid - Comparison of enamel and dentin, Journal of Forensic Sciences, 37, 1061−1067. Ohtani, S., 1994, Age estimation by aspartic acid racemization in dentin of deciduous teeth, Forensic Science International, 68, 77−82. Ritz, S., Schütz, H.W., Peper, C., 1993, Postmortem estimation of age at death based on aspartic acid racemization in dentin. Its applicability for root dentin, International Journal of Legal Medicine, 105, 289−293. Ritz, S., Stock, R., Schütz, H.W., Kaatsch, H.J., 1995, Age estimation in biopsy specimens of dentin, International Journal of Legal Medicine, 108, 135−139. Ubelaker, D.H., 1989, Human Skeletal Remains, Excavation, Analysis, Interpretation. Taraxacum, Washington, 63-95.

In addition to aspartic acid, D-glutamic acid was also found to be suitable for the estimation of age. Calibration curves based on these investigations were used for the age estimation of more than 200 skeletons of unknown age from the different historical periods. The correlation coefficient between the results and those obtained using standard paleoanthropological methods was over 0.9. From the experiments it seems that the later forming teeth racemized more slowly than those which erupt earlier, and this must be taken into consideration in age determination. Comparative experiments suggest that differences in extent of racemization of the same individual can be larger than the difference between individuals. References Bada, J.L. and Brown, S.E., 1980, Amino acid racemization in living mammals: biochronological applications, Trends in Biochemical Sciences, 5, R3−R5. Csapó, J., Tóth-Pósfai, I., Csapó-Kiss, Zs., 1986, Optimization of hydrolysis at determination of amino acid content in food and feed products, Acta Alimentaria, 15, 3-21. Csapó, J., Némethy, S., Folestad, S., Tivesten, A., Martin, T.G., Csapó-Kiss, Zs., 1994, Age determination based on amino acid racemization. A new possibility, Amino Acids, 7, 317−325. Csapó, J., Csapó-Kiss, Zs., Wágner, L., Tálos, T., Martin, T.G., Némethy, S., Folestad, S., Tivesten, A., 1997, Hydrolysis of proteins performed at high temperatures and for short times with reduced racemization, in order to determine the enantiomers of D- and L-amino acids, Analytica Chimica Acta, 339, 99−107.


NEW PROSPECTS IN OBSIDIAN HYDRATION DATING: AN INTEGRATED APPROACH I.Liritzis,1 C.M. Stevenson,2 S.W. Novak,3 I.Abdelrehim,3 V. Perdikatsis,4 and M. Bonini5 1

Laboratory of Archaeometry, Department of Mediterranean Studies, University of the Aegean, 1 Demokratias Avenue, Rhodes 85100 GreeFH[email protected] 2

Virginia Department of Historic Resources, 2801 Kensington Avenue, Richmond, VA 23221 U.S.A. 3 4

Evans East, 104 Windsor Center, Suite 101, East Windsor, NJ 08520 U.S.A.

Technical University of Crete, Dept. of Mineral Resources Engineering, Chania, Crete, Greece 5

Dept. of Chemistry & CSGI Consortium, University of Florence, Italy

Abstract: The new approach to dating ancient obsidian artifacts (SIMS-SS) based on the modeling of water diffusion profiles is strengthened using eleven archaeological test cases of known age; three new and eight revised results. Hydrogen profiles from hydrated obsidian surfaces have been collected by secondary ion mass spectrometry (SIMS). The H2O concentration versus depth profiles are modeled and diffusion ages have been produced. SIMS based dates for the three obsidian specimens of well-known age, ranging from 600-7000 years before present, have been compared with radiocarbon ages. The convergence between the two dating methods is excellent and validates the new dating approach. Preliminary microscopic investigation employing PLM, AFS and SEM-EDAX of obsidian surfaces from Melos sources indicate presence of mainly feldspars of various sizes and signs of weathering, as well as leaching / enrichment of cations, but also reveal flat surfaces for SIMS profiles. Περιληψη: Η νέα προσέγγιση (SIMS-SS) στην χρονολόγηση αρχαιολογικών αντικειμένων από οψιανό και η οποία βασίζεται στην μοντελοποίηση των κατανομών διάχυσης του νερού στο υλικό, έχει ενισχυθεί με την χρήση αρχαιολογικών δειγμάτων ελέγχου γνωστής ηλικίας. Οι ηλικίες που προέκυψαν με χρήση της τεχνικής SIMS-SS σε τρία δείγματα ηλικίας 600–7000 ετών, συγκρίθηκαν με ηλικίες ραδιοάνθρακα. Η συμφωνία στα αποτελέσματα των δυο τεχνικών είναι εξαιρετική και πιστοποιεί την εγκυρότητα της νέας μεθόδου. Η χρήση τεχνικών μικροσκοπίας (PLM, AFS και SEM/EDAX) σε επιφάνειες δειγμάτων οψιανού από την Μήλο, υποδηλώνουν παρουσία κύρια αστρίων, κυμαινόμενων διαστάσεων και βαθμού διάβρωσης, καθώς επίσης και διαφυγή/εμπλουτισμό σε κατιόντα, αλλά επίσης προέκυψαν και επίπεδες επιφάνειες για μελέτη κατανομών SIMS.


Laboratory of Archaeometry of the University of the Aegean, Rhodes, Greece (Liritzis and Diakostamatiou 2002; Liritzis et al. 2003; Liritzis and Diakostamatiou, 2003).

The obsidian hydration dating (OHD) method is based upon modeling the rate of water diffusion into a natural glass surface and establishing a diffusion coefficient for this process. It is accepted that the rate of water diffusion, the diffusion coefficient, is exponentially dependent on temperature and exhibits an Arrhenius type behavior. A variety of strategies have been developed over the years to calibrate the movement of ambient water into a glass. Many of these approaches have developed procedures for controlling the chemical composition of the glass and modeling the environmental history of the artifact context (e.g., temperature, humidity) (Friedman 1976; Ambrose 1984; Mazer et al. 1991; Friedman et al. 1994; Stevenson et al. 1998; Stevenson et al. 2002). However, the development of calibrations to compensate for variation in external variables has proven to be difficult. This has been the major impediment to making OHD a fully chronometric dating method comparable to radiocarbon dating.

Both groups rely on the modeling of the H2O concentration profile as a function of hydration depth, but following different ways. As a result the ODDSIMS and the SIMSSS coined versions were produced. The ODDSIMS rely entirely on time constraint points via 14C calibration of each site considered, the SIMS-SS seems independent of any calibration. In this paper, we present further results of our SIMS-SS dating approach which strengthens the new procedure for obsidian age estimation that is based upon the depth and shape of the hydrogen diffusion profile as determined by secondary ion mass spectrometry (SIMS) (Liritzis and Diakostamatiou, 2002; Liritzis et al., 2003; Liritzis, 2006).

However, recently a reviving of OHD has been made (Liritzis 2002) especially by the two leading groups – Oak Ridge National Laboratory and Tennessee University (Anovitz et al. 1999; Riciputi et al., 2002), and the

We have termed this approach SIMS-SS since the primary input variable, which controls the water in the mass of obsidian, is the saturation achieved in the surface layer (SS). This method is similar to ODD-SIMS (Anovitz et 9

I.Liritzis, C.M. Stevenson, S.W. Novak, I.Abdelrehim, V.Perdikatsis and M. Bonini Sims-ss modeling

al. 1999; Riciputi et al., 2002) but differs in the manner in which profile is modeled.

Theoretical background There are three advantages to our new fully intrinsic procedure: (1) the final shape of the hydrogen profile incorporates two of the principal external, and highly variable, environmental parameters, those of temperature and humidity; (2) the use of SIMS instrumentation to measure surface hydration layers results in high precision thickness values with an error of 0.02-0.05 μm depending upon the degree of surface roughness; and (3) absolute age estimates may be calculated.

The diffusion of environmental water into obsidians depends on random molecular motion but with the tendency for water to move in the direction of decreasing water concentration. The rate of this movement also depends on temperature, relative humidity and the location and arrangement of ions within the obsidian structure. The result is a sigmoid shape water (hydrogen ions) concentration profile that incorporates all variations of the external parameters into the final profile shape (Fig.1). Using this end product of diffusion, we have developed a phenomenological model, based on certain initial and boundary conditions and physicochemical mechanisms, that expresses the H2O concentration versus depth profile as a diffusion/time equation. This allows chronometric ages to be calculated.

The archaeological context of each example is discussed with particular attention placed on establishing the integrity of the context and the relationship between the radiocarbon event and the deposition of the obsidian artifact. It is important that both forms of material within the deposit (e.g., carbon, obsidian) reflect, as closely as possible, the same depositional event and depositional history.

We model this diffusion process as one-dimensional phenomena, where water molecules enter a semi-infinite medium in a direction perpendicular to the surface. The model is based on the idea that in the saturated surface (SS) layer encountered near the exterior of the sample (i.e., in the first half of the sigmoid curve), the concentration (C) of molecular water is assumed as constant along a very short distance. Thereafter, C decreases gradually following the trend of the sigmoid curve

Water diffusion in obsidian Based upon infrared spectra of the developing obsidian hydration layer (Stevenson et al. 2000), it appears that molecular water is the diffusing species that originates on the obsidian surface. As the molecular water moves through the glass the newly formed OH groups remain relatively fixed. It is assumed that OH group formation lags well behind the movement of mobile water molecules but eventually reaches equilibrium.

The mathematical theory of diffusion in isotropic substances is based on the observation that the rate of transfer of the diffusing substance, through a unit area, is proportional to

It is also well documented that natural glasses contain structural water concentrations that may range up to several tenths of a percent. This will exponentially increase the diffusion coefficient as concentration rises (Stevenson et al. 1998; Zhang et al. 1991; Zhang & Behrens 2000). This is believed to occur because the occurrence of greater amounts of SiOH in the network (referred to the intrinsic water molecules which may be present as OH with the silicious network) may weaken the neighboring Si-O-Si bonds, causing them to easily break, and therefore lower the overall activation energy (Bruckner 1970; McMillan et al. 1994). Under ambient temperatures (0-30oC) the diffusion of water in obsidian is not a steady state diffusion process with a constant D, and cannot be mathematically estimated with Fick’s first law. Instead, as water enters the glass network the structure is depolymerized and allows additional water to enter the glass at a faster rate. This changing diffusion coefficient results in the formation of a characteristic Sshaped concentration-depth profile (Fig.1). As a result, alternate modeling procedures are required. Although the theoretical background is published elsewhere (Liritzis et al. 2003; Liritzis, 2006), we provide a step-bystep outline of the new model to ease readership.

Figure 1: Water diffusion in obsidian illustrated by a hydrogen diffusion profile, showing the saturation layer and location of xs and Cs


New Prospects in Obsidian Hydration Dating: An Integrated Approach the concentration gradient measured normal to the section. It is expressed by the equation:

structural water concentration of the glass; x equals the diffusion depth, where x>0; for a diffusion time, t=0.

Fx = -D ∂C/∂x

Boundary condition: C=Cs the obsidian surface (saturated) water concentration, for x=0 and t>0.


where: Fx is the rate of transfer of water moles per unit area (kmol m-2 s-1) along x-direction, C the concentration gradient of diffusing substance (mass per unit volume) acting as the driving force, x the distance coordinate measured normal to the section, and the proportionality constant D is called the diffusion coefficient (m2 s-1) or less commonly as molecular mass diffusivity or just mass diffusivity. The negative sign arises because diffusion occurs in the direction opposite to that of increasing concentration.

However, in concentration-dependent diffusion, D is not constant during the hydration process but depends on changes in C. A family of curves that relates concentration-to-distance for an exponential diffusion coefficient during sorption is given in Figure 2, which, through linear interpolation, are based on those given by Crank (1975, p.172). Additional curves have been added (Fig. 3) and developed to accurately estimate k-value (of ek). They represent calculated results for variable diffusion coefficients for non-steady-state conditions. It follows that the form of exponential dependence of the C from D is given by exponential equation:

For the use of diffusion for dating it is essential to establish the time that diffusion has occurred. For this, complex dynamic or time-dependent analysis is required rather than simple steady state. Hence, the concept of property balance is introduced, whereas, input (∂C/∂t) + generation = output (C) + accumulation (∂(D∂C/∂x)) (Brodkey & Liritzis, 2004). In considering this case of one-dimensional diffusion through an obsidian surface, where there is a concentration gradient along the x-axis, the diffusion coefficient (D), as a function of change in C with time, will be variable, and it is derived from the partial differential equation: ∂C/∂t = ∂ (D∂C/∂x)/ ∂x

D=Ds exp (kC/Cs)= Dsexp[β(C-Cs)]


where: k= a constant, Ds the diffusion coefficient for C=Cs and β= a constant. In fact, Boltzmann showed that for certain boundary conditions, provided D is a function of C only, C may be expressed in terms of a single variable and that Equation (2) may be reduced to an ordinary differential equation by the introduction of a new variable.

(2) Based upon Crank’s (1975, p.105) numerical solutions of diffusion, and Boltzmann’s transformation, the following auxiliary variables are introduced:

which is a combination of eq. (1) for non-steady state condition with the balance equation concept for no flow or generation, where: t=time and x = distance. It denotes the rate of change of C with time (t), which equals a diffusion coefficient by the rate of change of C along X-axis. In fact D relates C(t) with C(x), and has the significance that it is the contribution to the rate of transfer in the x-direction due to the component of C gradient in this direction.



c=(C-Cs)/ (Ci-Cs)


where Ci the initial concentration (pristine water) in obsidian prior to diffusion of water, and,

For constant D equation (2) becomes: γ=β(Ci-Cs) ∂C/∂t = D(∂2C/∂x2)


(3) Then eq.(4) becomes,

that is, there is a gradient of C only along the x-axis, and the rate of change of C with time (t) equals the rate of transferred water molecules per unit distance (x) times the rate of change of C divided by x. Equations (1) and (3) are usually referred to as Fick’s 1st and 2nd laws of diffusion by direct analogy with the equations of heat conduction.

D=Dieγ(Di at C=Ci)


Substitution of equations (5), (7) and (8) into (2) we obtain d(eγdγ/dys)/dys+(2ys)(dγ/dys) = 0

It has been shown (Crank 1975) that Equation (3) produces several numerical solutions depending on the system’s initial and boundary conditions. For the present obsidianwater system, an exponential diffusion coefficient is modeled under the constraints (Figure 1), where:


In view of the boundary condition, eq.(5), and eq.(7), the integration of eq.(9) is to be performed with γ=0 and ys=0. Thus, we can use the values of γ calculated from eq.(9). In view of initial condition for C=Ci, the eq.(6) and eq.(7), we have,

Initial conditions: C=Co=Ci, where Ci is the initial or 11

I.Liritzis, C.M. Stevenson, S.W. Novak, I.Abdelrehim, V.Perdikatsis and M. Bonini

Figure 2: Non-dimensional plot, where K values presented as ek are derived from a family curves (based upon Crank, 1975).

Figure3: Estimation of K-values from family curves of ek versus x/xs for certain C/Cs, produced by interpolation from Fig. 2.



and eq. (6), the flux across the surface x=0 of eq.(1) becomes:

c = γ/r

(11) -Ds(∂C/∂x)x=0 = -1/2(Ds/t)1/2(Ci-Cs)(dc/dys)ys=0 (15)

Here, it is, r = log( DC = Ci / D C = Cs)

From g=(dγ/dy), eq.(11) and eq.(14), it follows that the dc/dys for ys=0 is given by:


( Indeed, D(C=Ci)=D(C=Cs)exp(β(Ci-Cs)) → D(C=Ci) / D(C=Cs) = exp(β(Ci - Cs)), and log[D(C=Ci) / D(C=Cs)] = β(Ci- Cs) = r)

dc / dys = g / r = 1.128 / (1 - 0.177r)

From eq.(12) and eq.(4) we define

Equation 15 relates the flux across obsidian’s surface at x=0 with the time the flux has occurred, the Cs, the Ci and Ds for Cs.

r = k C / Cs

where, r is calculated from eq.(12).


It is only when, a) the initial and boundary conditions are expressible in terms of ys alone, b) the x and t are not involved separately, c) diffusion occurs in semi-infinite media, and d) initial water concentrations are uniform (and may be zero) as it occurs in obsidian, that the transformation of eq.(5) and eq.(9) can be used.

Therefore, we use the exponential form D ( C = Ci ) = D ( C = Cs ) exp( kCi / Cs ) upon which the concentrationdistance curves, of Fig.2 and 5, are based (see below, Curve fitting and Age calculation). Plotting the values of r (i.e. the limiting values of γ at y=∞), as a function of auxiliary parameter g used for the integration of eq.(9) (for u=eγ), numerical values of c are compiled. Solutions of eq.(9) from y=- ∞ to y = + ∞, for the initial conditions, γ=0, u=1, y=0, were obtained for different values of g given by g=(dγ/dy).(for details, see Crank 1975, p.113 & p.137).

It is necessary to clarify that the variation of C during sorption depends on various factors (environmental, such as, temperature and humidity, atomic structure and chemistry in obsidian, and the chemistry between burial environment and obsidian surface). This variation reflects a respective variation in D. But, the variation of C with distance (hence time) presumably incorporates all these factors. We consider that it is the obtained sigmoid shape of, C versus x, which incorporates all these unknown factors as an end product of all processes, which lead the diffusion. It is also assumed that since we are dealing with a compact solid material, the diffusion profile has not changed significantly with time, the material being buried in the natural environment. In this case the environmental

However, instead of using such plot one may calculate the g values with the aid of the following empirical interpolation formula: g = ( 1.128r ) / ( 1 - 0.177r )



Following Boltzmann’s transformation, and using eq.(5) 12

New Prospects in Obsidian Hydration Dating: An Integrated Approach and concentration that reflect different values of K. In fact, the non-dimensional plot of x/xo versus C/Co given by Crank (1975) (his Fig. 9.9), includes some curves, where the parameter that defines each curve is the term ek. For a full coverage of k-values, we have produced intermediate curves by linear interpolation (Figure 2). Exact estimation of K-values is produced from family curves of ek versus x/xs for certain C/Cs, by interpolation from Fig.2 Our SIMS profiles were re-plotted as non-dimensional (standardized to 1) diagrams of C/Cs versus x/xs and were matched to the appropriate profile of our produced curves guided by the end point of C along the x-axis (Fig. 5a,b,c). This way the k-value can be calculated from any particular sigmoid shape. A curve is then fitted to the SIMS data (Figure 6a,b,c) using TABLECURVE 2D software and takes the form of a polynomial with exponential terms: C=exp (a+bx+cx2+dx3)


The choice of the fitted curve is important for the age determination of the SS layer, due to the following three criteria: 1) our goal is to achieve the best fitting curve especially for the first portion, or upper half, of the S-like curve. This is a required in order to closely model the shape of the theoretical exponential diffusion curves of the C versus x non-dimensional plots of Fig. 5; 2) the upper half represents the (calculated) gradient of the fitted curve, which corresponds to the surface diffusion coefficient Ds, (i.e., the coefficient for the constant concentration of the saturated layer); and 3) the fitting should be congruent with the tail of the sigmoid curve. Needless to say that this part follows a different diffusion mechanism than the latter part to the end and that in comparing the experimental (Fig.5) and theoretical (Fig.2) curves, the initial points near the surface must be discarded (Brodkey et al. 2003). To do this, the first points on the SIMS profile, for x>0, are removed gradually and up to the point that concentration ceases to drop, searching for the fitted curve which satisfies the above criteria. When the 3rd degree exponential polynomial curve is found, the SS is defined from the sliding successive linear regressions of the data sets (Table 1). The first linear regression has as the first point that one remaining after the removal of the initial data points and up to the points that approach a constant value. The successive regressions start from the second, third, and so on point, and with data lengths gradually increasing till the inflection point of the profile. Thereafter, the gradient apparently becomes negative. The smallest gradients (ideally should be a zero gradient within error range of the data points) are met with linear regressions that define the plateau (SS layer) level. This procedure is facilitated through the determination of peaks from derivatives of the profile and the diffused initial part of the profile combined

Figure 4: Concentration of water versus hydration depth for a) DL-1983-92, b) DL-1983-98, c) DL-1983-106.

variations are not so radical and so long, in order to cause a significant alteration to the diffusion profile (Fig.4). Curve fitting and age calculation In order to model the form of the diffusion profile a value for the constant K must first be determined. In Liritzis and Diakostamatiou (2002) this was fixed by the shape of the experimental curve that closely corresponds to Crank’s (1975) theoretical curves. For the non-steady state condition considered here, a collection of sigmoid shaped curves have been produced for non-dimensional distance 13

I.Liritzis, C.M. Stevenson, S.W. Novak, I.Abdelrehim, V.Perdikatsis and M. Bonini with successive regressions, fully computerised (Liritzis and Ganetsos, 2006). In fact, these appear along a band of less than 10% spread. The successive regressions give almost a stable end point xs for the respective sliding data sets, which corresponds to a constant concentration value, Cs. The SS layer is defined either when the gradient is near zero or changes from positive to negative values, which presumes diffusion within the obsidian (i.e. direction towards decreasing C values). The average + standard deviation of xs and Cs values within the band define the end of the SS layer, and these are used in the age calculation (Table 2). Occasionally, a second SS layer is obtained, however, with larger gradient values, attributed at present to possible reuse of the tool. The exclusion of some initial points to achieve the best fitting with the first half of S-curve is justified from the near surface disturbances due to either sputtering and/or variable environmental effects (Stevenson et al. 2002; Liritzis & Diakostamatiou 2002). The choice of a 3rd order polynomial, rather than higher order exponential polynomial, was chosen to satisfy the three criteria for fitting of the initial and end points, as well as the first half of the S-curve. Overall, the higher order polynomial fitting seems better but at the expense of the criteria, particularly with the unavoidable presence of more oscillations. This was reinforced in practice during the applied trials and blind tests whereas the 3rd order polynomial gave the best results. As a first approach in the earlier publication (Liritzis & Diakostamatiou 2002, p.12), the gradient dC/dx, or the tangent at a certain point of the fitted curve, or from another point of view, the rate of change, was attributed to the diffusion coefficient: dC/dx= (b+2cx+3dx2) exp (a+bx+cx2+dx3)


and D was found and expressed from the calculated data of eq.(18).

Figure 5: Non-dimensional plot for samples a) DL-1983-92, b) DL-1983-98, c) DL-1983-106.

For C=Cs and x=xs then we set D=Ds as shown below.

is the inverse of gradientx10E-11 years, assuming flux is constant and taken as unity, that is:

From eq.(16) the age is calculated: T= [(Ci-Cs)2 (1.128/(1-0.177kCi/Cs)2] / [4Ds(b exp(a)) 2 ]

(∂C/∂x)x=0 = bexp(a). In the calculation of D the 10E-11 is been done in order to convert the units of D from the calculated μm2/1000 years to cm2/year, which is the units used in SIMS-SS.


Where, Ci and Cs (both in gram moles of water per cm2), as well as xs (in cm) are measured parameters, while Ds, eff= (aDs + b) / (1022*Ds) (empirically derived from more than 20 well dated samples), a= 8.051*106, b=0.0999 (r2=0.999), Ds (=flux/gradient) and D = Ds*exp(kC/Cs), and k are calculated, and (a) and (b) are constant coefficients. In fact according to Fick’s law Ds

We now evaluate the ability of our model to estimate the ages of prehistoric artifacts using SIMS profiles recorded from ancient samples. To do this, the water diffusion coefficient at any particular moment is expressed by the first derivative of the hydrogen profile. This is the apparent hydration rate. The average Xs and Cs obtained from 14

New Prospects in Obsidian Hydration Dating: An Integrated Approach primary ion beam with an impact angle of 60o with respect to surface normal was used and negative secondary ions were detected. Charge build up during profiling was compensated for by use of an electron beam. The ion signal was maximized by using the electron beam resulting in the least amount of charging. The measurements were performed using a 300 x 300 micron ion beam raster, which results in very little visual disruption to the sample surface. Generally, the SIMS depth scale accuracy is within 5-10%. This translates into an estimated error of +0.05 μm. This value is not equivalent to the +0.01-0.03 μm standard deviation usually associated with SIMS because of the irregular surface topography present on naturally cleaved samples. Archaeological samples and contexts Hopewell Site, Ohio, USA The obsidian artifacts from the eastern United States come from the Hopewell Site near Chillicothe, OH. They originated from within a large 136-kilogram obsidian cache placed within an earthen mound (Mound 11). Excavation of this Mound 11 (Shetrone 1926) resulted in the identification of a cremated person associated with pearl beads, mica and copper artifacts. The cache was located away from the cremation pit and appeared not to have been effected by the burning of the individual. Obsidian flakes were removed from the cache for SIMS analysis. The mound was recently dated by Cowan and Greber (2002) using small carbon particles that had adhered to obsidian artefacts retained from the cache. Two age estimates were obtained that returned age estimates of 1740+/-40 BP (2-sigma 217-411 AD) (Beta 170560) and 1800+/-40 BP (2-sigma 94-340 AD) (Beta 170561). This places the securely within the Middle Woodland Period (200 BC - 500 AD, 2150-1450 BP). Samples assigned reference numbers DL-1983-92, DL-1983-98, DL-1983106, DL-1983-99, and DL-1983-103, fall within the 2sigma age range of C-14 dates (Table 3). Chalco, Mound 65, Mexico The prehistoric site of Chalco was located on the shore of Lake Chalco within the Basin of Mexico, Mexico. The city was occupied from the Early Post Classic (200-750 AD) through the Late Postclassic (1200-1521 AD) and emerged as one of the larger urban centers in the region. Within this urban setting test excavations were conducted at Mound 65, a stratigraphically complex deposit with up to 20 cultural levels that included patios and living floors. Ceramic associations and radiocarbon dates revealed that the mound was initially formed during the Epiclassic (750-950 AD) with later occupation within the Aztec I-III periods (1150-1519 AD. An occupational hiatus occurred from 900-1100 AD (Riciputi et al. 2002). We have utilized two published SIMS profile from Anovitz et al. (1999) and

Figure 6: SIMS hydrogen profile (dots) with a fitted curve (line) for samples a) DL-1983-92, b) DL-1983-98, c) DL-1983-106.

the determination of the SS layer gives the overall error attached to the SIMS-SS ages (Table 2). Secondary ion mass spectrometry SIMS analyses were conducted at the commercial laboratory of Evans East, East Windsor, NJ, the profiles were collected using PHI Model 6300 and 6600 quadrupolebased secondary ion mass spectrometers. A 5.0 KeV Cs+ 15

I.Liritzis, C.M. Stevenson, S.W. Novak, I.Abdelrehim, V.Perdikatsis and M. Bonini

Table 1a. Determination of the surface saturation layer with the successive linear regressions for the sample DL-1983-92. 1-55 are columns for the location of successive slops. Bold numbers define the points where positive slops terminate.

Riciputi et al (2002). Sample CH0081 is from Level L9f2, Unit B, and CHO086 from level L9, Unit B. (Table 2). The stratigraphically calibrated C14 date for these layers (Riciputi et al. 2002, p.1059) are, 1270-1440 BP (1σ) and 1290-1440 BP (Table 3).

trenching operations (Facorellis and Maniatis 2002, p.309-315). The age span of this settlement is 4500-5100 BC (7050-6450 BP) calibrated C14 years (Table 3).

Ftelia and Philakopi, Greece

The weathered, and possibly eroded, obsidian tool surface should be thoroughly examined prior to any SIMS profile (Pawlikowski et al. 2003) since the condition may impact the outcome of the surface analysis. A number of obsidian flakes and tools were thin sectioned to create a crosssection profile from the surface to a depth of one millimeter. Preliminary observations were carried out with the use of Polarized Light Microscopy (PLM) and Atomic Force Spectroscopy. During chopping of an obsidian nodule, flakes were produced as a waste scattered on the ground. The geological surface of these flakes was examined as an extreme case of weathering. This was done in order to

Microscopic investigations

Melos island hosts the most important obsidian source in the Aegean, comprised of at least two sub-sources, Adamas and Demenegaki (Stevenson et al. 2002; Renfrew & Wagstaff 1982). The three samples from the Philakopi settlement on Melos Island (DL-2000-145, 146 and 147) are of Early Bronze Age III period (2000-2500 BC, 44503950 BP) and derive from an excavated trench within the site boundaries. One sample come from the Ftelia Neolithic settlement on the Cycladic island of Mykonos (DL-2000148) was derived from living floors encountered during 16

New Prospects in Obsidian Hydration Dating: An Integrated Approach

Table 1b. Determination of the surface saturation layer with the successive linear regressions for the sample DL-1983-98. 1-19 are columns for the location of successive slops. Bold numbers define the points where positive slops terminate.

Table 1c. Determination of the surface saturation layer with the successive linear regressions for the sample DL-1983-106. 1-22 are columns for the location of successive slops. Bold numbers define the points where positive slops terminate.

understand the weathering process over a long period of time, so that we can compare and understand better the results in smaller time and weathering scale, such as in tools.

light microscope. Our observations indicate that it is possible to distinguish between the two sources by a simple microscopic investigation. The glassy matrix of the obsidian from both sources contains an abundance of crystallites of various sizes. The crystallites are mainly feldspars (the type of feldspars is discussed later with the use of SEM), and minor amounts of quartz and biotite are also present. The feldspar crystallites are of the 1st and 2nd generation, that is, there are two sizes of crystallites. The main difference between the two sources is the size of the

Polarized light microscopy Thin sections of obsidian flakes and tools present surface of the two Melos sources (Nychia and Demenegaki) were prepared and then examined through a polarized 17

I.Liritzis, C.M. Stevenson, S.W. Novak, I.Abdelrehim, V.Perdikatsis and M. Bonini

SAMPLE REF. No DL-1983-92 DL-1983-98 DL-1983-106 DL-2000-145 DL-2000-146 DL-2000-147 DL-2000-148 CHO086 CHO081 DL-1983-99 DL-1983-103

Xav+sd Cm 0.000204443 +/- 1.61E-06 0.000106645 +/- 8.67E-07

Cav +sd gmol/cc 0.000223181 +/- 3.88E-06 0.00028845 +/- 6.86E-06

0.000275814 +/- 9.24E-06 0.00010192 +/- 6.02E-06 0.000128701 +/ -8.76E-06 0.000147785 +/- 1.10E-06 0.000184732 +/- 1.06E-05 0.000133799 +/- 7.10E-06 0.000163166 +/-6.85E-06 8.00E-05 +/- 8.25E-07 0.000122055 +/- 9.98E-06

0.00022833 +/- 1.03E-05 0.000654082 +/- 9.77E-06 0.00052221 +/- 7.20E-06 0.00048927 +/- 5.61E-06 0.00093023 +/- 1.26E-05 0.13450331 +/- 0.001796 0.11684581 +/- 0.000956 0.00031011 +/- 4.30E-06 0.00031367 +/- 7.96E-06





Ds (cm2/year) 5.17E-13































AGE SIMS-SS Years BP* +/- sd** 1759 +75/-65 1714 +160/-147 1855 +167/-143 4025 +203/-195 4516 +195/-188 4226 +167/-161 7099 +176/-172 689 +21/-16 630 +118/-70 1927 +85/-83 1658 +126/-167

Table 2. Basic Data of SIMS-SS Age Equation by Sample for Age Calculation

second generation crystallites. In the obsidian from Nychia, these crystallites are very small and sometimes they cannot be detected under the PLM, while in Demenegaki they are easily observed.

obsidians and to more closely examine the surfaces of the obsidian flakes and tools (examined at the Univ. of Crete, Chania). Through SEM-EDAX the 2nd generation crystallites were identified as K-rich feldspars, while the 1st generation crystallites are probably both plagioclases and K-rich feldspars. On a preliminary examination, in flakes with the widest weathered zone, enrichment of Ti, Fe, Mg and Al was detected on the surface, while Si, Na and K have been leached out. On tools, the enrichment is not as obvious as the leaching of Si, Na and K. The apparent difference between tools and flakes is due to aging differences.

In order to understand the weathering process the weathered surface was examined at a high magnification to resolve surface features that were 20-100 μm in size. It was noted that it likely contained oxidized products of iron and it had cracked in many places, both perpendicular and horizontal to the surface. The horizontal cracks caused the loss of the first μm of the surface in several areas (Photo 1). The surface of the tools, as expected, seems more homogenous and not so weathered, but the size of the weathered surface is only up to 15 μm in some tools and details are hard to distinguish through PLM, but possible with SEM or Atomic Force Spectroscopy (ASF) (see below). The hydration layer is not even in size across a sample and this may be due either to the weathering processes or damage suffered during the preparation of the thin sections. However, a notable feature is that crystallites are often found inside the hydration layer (Photo 2).

With AFS, details were observed on the surface that could not be detected using PLM.or SEM These features consisted of cracks and voids, which were sometimes within the first 10μm of the tool surface (Photos 3 a, b, c). It is our opinion that this would make any SIMS measurement problematic if the opening was inadvertently targeted. However, there are some areas free of crystals, voids and cracks, and these regions are ideal for SIMS. Atomic force microscopy (AFM) imaging was performed with an Explorer TMX

Scanning electron microscopy and atomic force spectroscopy

2000 microscope (Topometrix) using a 100x100 μm xy, 10 μm z air scanner and a silicon “V”-shaped cantilever with an integrated pyramidal tip for contact imaging. Images were taken at room temperature in contact mode and, in order to remove the background slope, they were processed by flattening.

SEM was applied to identify the crystallites found in the The BP (Before Present) for SIMS-SS is year 2000, while for C14 is 1950 ** Standard Deviation *


New Prospects in Obsidian Hydration Dating: An Integrated Approach




Hopewell Ohio State USA Hopewell Ohio State, USA

SIMS-SS Age (BP)* 1759

Calibrated C14 or Archaeological Age (BP)


Hopewell Ohio State, USA



Melos Greece Melos Greece

4025 Liritzis & Diakostamatiou (2002) 4516 Liritzis & Diakostamatiou (2002)

1450-2150** 1539-1856*** Cowan et al. (2002). 1450-2150** 1539-1856*** Cowan et al. (2002). 1450-2150** 1539-1856*** Cowan et al. (2002). 4000-4500 Liritzis & Diakostamatiou (2002) 4000-4500 Liritzis & Diakostamatiou (2002)


Melos Greece

4226 Liritzis & Diakostamatiou (2002)

4000-4500 Liritzis & Diakostamatiou (2002)


Myconos Greece

7099 Liritzis & Diakostamatiou (2002)

6500-7100 Liritzis & Diakostamatiou (2002)

7-26H2 (CHO086)

Mexico, Chalco, (Pachuca source)

689 Liritzis & Diakostamatiou (2002)

660-530 Riciputi et al. (2002) and Anovitz et al. (1999)

7-25H4 (CHO081)

Mexico, Chalco (Paredon source)

630 (Liritzis & Diakostamatiou (2002)

640-590 Riciputi et al. (2002) and Anovitz et al. (1999)


Hopewell, Ohio State, USA

1927 Liritzis & Diakostamatiou (2002)


Hopewell, Ohio State, USA

1658 Liritzis & Diakostamatiou (2002)

1450-2150** 1539-1856*** Cowan et al. (2002). 1450-2150** 1539-1856*** Cowan et al. (2002).




Table 3. Obsidian Hydration Ages and Radiocarbon Dated Contexts

Discussion and conclusions

Results obtained on the tool MYC-1 (DL-2000-155) (hereby referred to simply as MYC for Mykonos) have shown that the surface profile is pretty irregular, with large almost-flat areas separated by deep holes (Photo 4 a).

The ages for the eleven obsidian samples were calculated using Equation 20 and compared with radiocarbon dates from the same or adjacent occupation layers (Table 3, Figure 7). The obsidian dates are in very good agreement with the radiocarbon determinations in all cases, for comparisons made on the 2σ probability range for radiocarbon. Earlier age estimations were slightly different from present revision but within the error bars (Table 3).

The holes have not a regular shape and also the depth is not homogeneous, ranging from a few tens of nanometers up to a maximum of about 1 μm (Photo 4 b). It’s important to take in mind that the scanner used to perform the measurements has a maximum z excursion of 10 μm, well beyond the higher detected depth of the holes.

The microscopic examination indicates that most surfaces are weathered and cracked, and that crystallites and voids are often present within the first μm of the original surface.

The almost-flat areas are constituted by globular structures with a typical diameter of the smaller sub-units ranging from 50 to 100 nm (Photo 4 c). In view of the SIMS analysis, this sample appears to be reliable with the condition that the spot investigated by the SIMS is in the flat region. Since the size of the holes is always larger than 1 μm (see, Photo 4 a), this could be ensured by using an optical microscope to choose the right position on the sample.

* BP (before present) is 2000 for SIMS-SS and 1950 for 14C. * *Age estimates for the Hopewell site represent the time span of this cultural period (see also, Liritzis et al., 2004). *** 2-Sigma Calibrated. These results date the construction of the mound from which our samples were taken.

Photo 1: a) The weathered surface of a scattered obsidian flake from Nychia source in Melos under PLM (length of picture: 1.35 mm), b) The same area as in (a) (cross nicols). Crystallites of feldspars, mica (biotite) and quartz are found in the glassy matrix, while the white zone is the hydrated surface layer.


I.Liritzis, C.M. Stevenson, S.W. Novak, I.Abdelrehim, V.Perdikatsis and M. Bonini

Photo 3: a) SEM picture of a blade from Ftelia, a Late Neolithic Settlement in Myconos. Photo 2: The weathered/hydrated surface of an obsidian blade from Ftelia Late Neolithic Settlement in Myconos. As in Photo1 crystallites of feldspars, mica (biotite) and quartz are found in the glassy matrix, while the white zone of hydration is not as obvious as in the scattered samples.

Photo 3: b) An elemental content variation line scan that follows the arrow in Photo 3a, made on the surface of the sample. The surface is on the right hand side of the diagrams. Leaching of silica, potassium, sodium and some iron on the surface is observed. Enrichment of calcium and aluminium is present.

Figure 7: Comparison plot between 14C and SIMS-SS dates of the eleven samples.

However, there are locations with little alteration where SIMS can be carried out. It is recommended that the leaching / enrichment of cations needs further investigation at a number of sites to determine the conditions that initiate this process. Obsidian hydration dating based upon SIMS hydrogen profiling of the surface saturated layer is emerging to be a reliable chronological dating method for archaeological applications. The errors involved are less than 5% with some cases between 5-10%, which is a very high level of precision, equivalent to, or less than, that of conventional calibrated radiocarbon dating. An additional advantage is that SIMS-SS does not suffer from the problem of alternate age ranges like those generated by multiple intercepts

Photo 3: c) The flat section of a blade from Ftelia suitable for SIMS that is located below the white hydrated zone. The inner obsidian matrix is on the lower left side.


New Prospects in Obsidian Hydration Dating: An Integrated Approach

Photo 4: a) Topography image (left) and its three-dimensional representation (right) of a 16.84 X 16.84 μm spot of the MYC blade.

Photo 4: c) Topography image (left) and its three-dimensional representation (right) of a 5.02 X 5.02 μm spot of the MYC sample.

on the radiocarbon calibration curve. More often than not, the probabilities assigned to the alternate outcomes of C14ages, coupled with site context information, can alleviate this problem, but it is often a concern. Furthermore, SIMS-SS age determinations have the potential for to produce age determinations within the last centuries and thus useful for historic investigations. However, the SIMS-SS is a new approach and will require additional refinements in the modeling process. We expect that a larger number of analyses and comparisons with dated contexts will demonstrate the utility of this approach. Acknowledgements Result:

point1: point2: Diff: Length: Pt Angle:

X(μm) 3.17 10.34 7.17

Line1 Y(μm) 14 14 0 7.225μm

We are grateful to the Ministry of Culture, Greece, for granting permission, to the Municipality of Rhodes for financial assistance, and to Prof. M.Tomozawa, Dr H. Behrens and Prof. R. Brodkey for useful comments and correspondence, Dr M. Diakostamatiou for making some calculations.

Z(nm) 853.9 4.1 -849.7


6.75o Ambrose, W. 1984, ‘Soil temperature monitoring at Lake Mungo: Implications for Racemization dating’, Australian Archaeology, 19, pp.64-74.

Photo 4: b) Line profile analysis of a 20.00 X 20.00 μm spot of the MYC sample.


I.Liritzis, C.M. Stevenson, S.W. Novak, I.Abdelrehim, V.Perdikatsis and M. Bonini Liritzis, I and Ganetsos, T, (2006) Obsidian hydration dating from SIMS H+ profiling based on saturated surface (SS) layer using new software. Appl. Surface Science, 252, 19, 7144-7147. Mazer, J., Stevenson, C., Ebert, W. & Bates, 1991, ‘The exponential hydration of obsidian as a function of relative humidity and temperature’, American Antiquity, vol.56, pp.504-513. McMillan, P.F., Poe, B.T., Gillet, P. and B. Reynard, 1992, ‘A Study of SiO2 Glass and Supercooled Liquid to 1950 K via High Temperature Raman spectroscopy’. American Mineralogist, vol. 8, pp1145-1156 Pawlikowski, M, Liritzis.I, Tsamasfyrou.I and Perdikatsis.V. 2003, ‘Surface microscopic investigations of obsidians. International Specialized Workshop, Recent Advances in Obsidian Dating and Characterization, 2-5 July 2003, Melos Island, Aegean, Greece University of the Aegean and IAOS, Rhodes, (Abstract Book), pp.26. Renfrew, C. & Wagstaff, M. 1982, An Island Polity. The Archaeology of Exploitation in Melos. Cambridge University Press, Cambridge. Riciputi, L.R, Elam, J.M, Anovitz, L.M & Cole, D.R. 2002, ‘Obsidian diffusion dating by secondary ion mass spectrometry: a test using results from mount 65, Chalco, Mexico’, Journal of Archaeological Science, vol. 29, pp.1055-1075. Shetrone, H.C., 1926, ‘Explorations of the Hopewell Group of Prehistoric Earthworks’. Ohio State Archaeological and Historical Quarterly, vol. 35, pp1-122. Stevenson, C.M., Mazer, J. & Scheetz, B. 1998, ‘Laboratory obsidian hydration rates: theory, method and applications’, In S.Shackley, (ed.) Archaeological Obsidian Studies: Method and Theory. Advances in Archaeological and Museum Science, New York, Plenum Press, vol.3, pp.181-204. Stevenson, C.M., Liritzis, I., Diakostamatiou, M. & Novak, S.W. 2002, ‘Investigation towards the hydration dating of Aegean obsidian’, Mediterranean Archaeology and Archaeometry, vol.2, no.1, 93-109. Stevenson, C.M, Liritzis, I, Diakostamatiou, M, Novak, S.W, and Abdelrehim.I. 2003, ‘The dating of hydrated obsidian surfaces by SIMS-SS: an evaluation with data from Aegean and Easter Islands, Mexico and USA’, International Workshop, ‘Recent Advances in Obsidian Dating and Characterization’, Melos island, Greece, 2-6 July, 2003, University of the Aegean and IAOS, Rhodes, (Abstract Book), pp.14-15. Zhang, Y. & Behrens, H. 2000, ‘H2O diffusion in rhyolitic melts and glasses’, Chemical Geology, vol. 169, pp.243-262. Zhang, Y., Stolper, E.M. & Wasserburg, G. 1991, ‘Diffusion of water in rhyolitic glasses’, Geochimica et Cosmochimica Acta, vol.55, pp.441-456.

Anovitz, L.M., Elam, J.M., Riciputi, L.R & Cole, D.R. 1999, ‘The failure of obsidian hydration dating: sources, implications and new directions’, Journal of Archaeological Science, vol.26, pp.735-752. Brodkey, R.S., Liritzis, I., and Diakostamatiou, M. 2003, ‘Transport phenomena and archaeology: application to obsidian hydration dating’, International Workshop, ‘Recent Advances in Obsidian Dating and Characterization’, Melos Island, Greece, 2-6 July, 2003, University of the Aegean and IAOS, Rhodes, (Abstract Book), pp.13-14. Brodkey.R.S and Liritzis.I. 2004, ‘The dating of obsidian: a possible application for transport phenomena (a tutorial)’. Mediterranean Archaeology & Archaeometry, vol.4, no.2, 67-82 Bruckner, R., 1970, ‘Properties and Structure of Vitreous Silica’. Journal of Non-Crystalline Solids, vol. 5, pp123-216. Cowan, F.L. and Greber, N.B. (2002) ‘Hopewell Mound 11: Yet another look at an old collection. The Newsletter of Hopewell’. Archaeology in the Ohio River Valley, 5, pp. 7-11. Crank.J. 1975, The Mathematics of Diffusion, 2nd Edition, Oxford Science Publication, Oxford University Press, Oxford. Facorellis.G and Maniatis.Y. 2002, ‘Radiocarbon dating of the Neolithic settlement of Ftelia on Myconos, calculation of the marine reservoir effect in the cyclades’. In Sampson.A.(editor), The Neolithic Settlement at Ftelia, Mykonos., University of the Aegean, Department of Mediterranean Studies, Rhodes, pp.309-315. Friedman, I. & Long, W. 1976, ‘Hydration rate of obsidian’, Science, 191, pp.347-52. Friedman. I, Trembour, F.W., Smith, F. & Smith, G. 1994, ‘Is obsidian hydration dating affected by relative humidity?’, Quaternary Research, 41, pp.185-190. Liritzis,I. 2002, ‘Reviving obsidian studies’, Mediterranean Archaeology & Archaeometry, vol.2, no.2. Liritzis, I, Diakostamatiou.M, Stevenson C.M, Novak S.W and Abdelrehim I. 2003, ‘The dating of hydrated obsidian surfaces by SIMS-SS’. Journal of Radioanalytical & Nuclear Chemistry, Vol. 261, No.1, 51-60. Liritzis, I and Diakostamatiou, M. 2003, ‘Potential and limitations of the novel SIMS-SS obsidian dating method’, International Workshop, ‘Recent Advances in Obsidian Dating and Characterization’, Melos island, Greece, 2-6 July, 2003, University of the Aegean and IAOS, Rhodes, (Abstract Book), pp.7 Liritzis, I. & Diakostamatiou, M. 2002, ‘Towards a new method of obsidian hydration dating with secondary ion mass spectrometry via a surface saturation layer approach’, Mediterranean Archaeology and Archaeometry, vol. 2 , no.1, pp.3-20. Liritzis, I, (2006) SIMS-SS, a new obsidian hydration dating method: analysis and theoretical principles. Archaeometry , 48 (3), 533-547.


LUMINESCENCE DATING BEYOND POTTERY: A REVIEW N. Zacharias Institute of Materials Science,N.C.S.R. Demokritos, 15 310 Ag. Paraskevi, Attiki, Greece [email protected] Abstract: Applications of Luminescence techniques have undergone remarkable development and continuous refinement during the last 30 years since the first applications of Thermoluminescence dating on heated material like ceramic, bricks and burned soil. Despite that luminescence dating techniques generally do not reach high precision chronological information, they serve as of the most attractive in archaeology and geology in terms of age range and scope of material. This review surveys an up to date status on Luminescence studies and points the future research directions. Additionally it stands as a guide to the selection of the more appropriate laboratory procedures for particular dating applications. Περιληψη: Η χρήση και εφαρμογή των τεχνικών Φωταύγειας για την χρονολόγηση υλικών έχει διάγει μια εντυπωσιακή ανάπτυξη τις τελευταίες δεκαετίες. Η αρχική χρήση της Θερμοφωταύγειας για την χρονολόγηση θερμασμένων υλικών και κύρια της κεραμικής, έχει δώσει την θέση της στην παράλληλη χρήση της Οπτικής Φωταύγειας για την χρονολόγηση ιζημάτων όλων των τύπων και στην επέκταση των χρονολογήσεων σε πληθώρα υλικών. Στην παρούσα εργασία παρουσιάζονται τα υλικά, οι τεχνικές και οι χρήσεις της Φωταύγειας, ενώ παράλληλα γίνεται αναφορά στις σύγχρονες ερευνητικές τάσεις και σε μελλοντικές εφαρμογές.

Ιntroduction ntroduction

The minerals are acting as natural radiation dosimeters: when a mineral is formed or reset, all electrons are in the ground state (valence band). Naturally occurring radioactive isotopes from U and Th series, 40K and cosmic irradiation emit a variety of rays which ionize atoms. Negatively charged electrons are knocked off atoms in the valence band and transferred to a higher energy state (conduction band, Fig. 1); positively charged holes remain near the valence band. After a short time of diffusion most electrons recombine with holes, thus returning the mineral back to its original state. However, all natural minerals contain defect sites (e.g. lattice defects, interstitial atoms, etc.) at which electrons and holes can be trapped. For the measurement of a luminescence signal, the trapped electrons have to be either thermally (heating) or optically (by light exposure) activated. This activates the electrons in those traps that are light sensitive and while returning to the conduction band a certain number –that is material

A number of techniques are included under the term of luminescence dating, which mainly refer to the phenomena of Thermoluminescence (TL; stimulated by heat) and Optically Stimulated Luminescence (OSL; stimulated by light). Luminescence dating requires solution of the equation Age = De /Dr, where De (measured in Gy) refers to the absorbed radiation dose, usually referred as equivalent dose and Dr (measured in Gy/ka) to the dose rate, or annual dose, resulting dates in thousand years (ka) within a time span of some hundreds up to 300ka and potentially 1Ma that typically range between 5-10% of total error. By using a luminescence technique, with respect to the material to be dated and the infrastructure availability, both research and routine services are possible for pottery, bricks, porcelain, sediments, coastal sand deposits, naturally occurred calcite formations, slags, mortars, burned flint, and obsidian artefacts. Within the last years, the application of Luminescence for extracting additional information is also in use in characterization and provenance studies, e.g. for pottery, archaeological glass, chert/flint, mosaic and steatite material. Theoretical and dating principals Both TL and OSL Luminescence dating methods have played a major role in the establishment of chronologies in archaeology particularly in the pre-radiocarbon time range. TL and OSL methods, are also referred, together with the Electron Spin Resonance method, as trapped charge techniques and are based on the same physical principles; namely the time-dependent accumulation of electrons and holes in the crystal lattice of certain minerals.

Figure 1. Schematic diagram of the luminescence energy levels and main processes involved (after Aitken, 1985), where E is the energy gap; T electron trap and L luminescence centre. i) ionization of electrons ii) storage of trapped electron population, iii) thermal of optical stimulation releases trapped population and subsequent emission of light during recombination of electrons and luminescence centers/holes.


N. Zacharias

Figure 2. TL glow curves (upper plot) and OSL shine-down curves (lower plot) recorded from fired and optically bleached quartz samples respectively.

depended- will recombine with the holes. If such holes are luminescence centers, light emission (luminescence) is observed and recorded either as glow curves in TL or shine down curves in OSL; in Figure 2 typical luminescence signals are presented. The above described energy model is rather simplified since more and usually competitive processes are present during luminescence readings; for a detailed presentation McKeever (1985) and McKeever and Chen (1997), should be addressed.

very similar. Light in a narrow frequency range (blue or green light for measuring quartz, infra-red for measuring feldspars) is shone on to the sample. The colour filters in front of the photo-multiplier are used to eliminate quantitatively the light emitted from the light source; in OSL measurements the emission is plotted against the time elapsed after the light source was switched on, resulting the shine down curve. In most cases, the light emission for dating lies in the ultra violet range, but other colors have also been investigated (Krbetschek et al., 1997). For more details on the instrumentation used in luminescence dating see also Aitken (1998) and Bøtter-Jensen et al. (2003).

Figure 3 shows graphically a laboratory set used for TL and OSL measurements. A mineral component of the sample is deposited on a disk which is placed on a heater for TL measurement. The photons evicted are converted into electric pulses with a photomultiplier. The light emission is then plotted versus the heating temperature, resulted the glow curve. The configuration for OSL measurement is

Estimation of the equivalent dose, De Since the luminescence signal should be reliable, the following properties have to be provided: (i) when the 24

Luminescence Dating Beyond Pottery: a Review

Figure 3. Drawing of the Risø TL/OSL luminescence reader (permission by Dr. L. Bøetter-Jensen). The set up is equipped with a beta 90Sr/90Y source for artificial irradiation of the samples and is automated controlled enabling measurements of both TL and OSL for up to 48 sample aliquots.

sample is reset it may contains an initial portion; that can be either experimentally determined or assumed to be zero, (ii) the signal intensity grows linearly versus the dose received, (iii) the number of traps is constant or changes in a predictable manner, (iv) recrystallization, crystal growth of phase transitions must not have occurred, (v) anomalous fading should not be present or accurate estimation should be possible and (vi) the signal is not influenced by sample preparation (triboluminescence, exposure to laboratory light, etc.). There are several techniques for the determination of the De value. The additive dose method is most widely used: after measuring the natural dose, several aliquots made from the same sample is then irradiated with known elevating doses, thus generating more trapped electrons and increasing the thermally activated signal. Finally the signal growth is plotted against the received laboratory dose and these data points are used to extrapolate to the initial signal yielding the De value. A modified additive dose method is of the foil technique (Michael et al., 1997) which produces a normalized growth curve each point corresponds to the ratio of the TL intensity of the first TL measurement to the intensity after given the same reference dose fro every aliquot.

Figure 4. Histograms and associated probabilities resulted from De values measured using single aliquot SAR samples (upper plot) and radial plot (Galbraith et al., 1999) resulted from single grain SAR measurements (lower plot).

is measured iun better detail and, furthermore, the final estimation is little dependent on the mathematical model used for the fitting of the data points.

The regeneration technique is an alternative for De estimation. First the natural signal is measured and the subsequent aliquots are reset. These aliquots are then irradiated are the projection of the natural signal on to the regenerated growth curve yields De. This method has the advantage of involving smaller errors since each point

The single aliquot regeneration (SAR; Murray and Wintle, 2000) method stands as of the most applied ones in luminescence dating since the SAR protocol carries all measurements out on the same aliquot. First, the natural intensity is measured; then the sample is irradiated with 25

N. Zacharias elevating doses. Sensitivity changes are monitored and corrected for by giving the sample a reference dose and measuring its response to it. An extensive review of all SAR modifications and protocols in use can be found in Wintle and Murray (2006).

The techniques employed for the estimation of U and Th concentration are usually alpha- beta-counting and low level gamma-spectrometry (Prescott and Hutton 1995). For converting the elemental analysis into dose rates the factors provided by Liritzis and Kokoris (1992) are used.

Aliquots are made either by the fine grain polymineral fraction or in case that the provided sample is in abundance, coarse grain aliquots are preferable. Since single grain measurements is now available using laboratory attachments to the main TL/OSL set, the main percentage of luminescence dating is currently performed, by using SAR modified protocols on single aliquot or single grain measurements. As a strong benefit, statistical evaluation of large De data sets (Fig. 4) collected for every sample can be applied thus achieving dense dose distributions and finally lower total errors for the provided ages.

New developments allow the detection and possibly estimation of disequilibrium in soil samples (Michael and Zacharias, 2000) which presence if not detected and corrected for can cause serious errors in the final ages; this is easily understood when quartz is measured that is free of any radioactivity and most of the measured dose comes from the surroundings.

Estimation of the dose rate, Dr

Luminescence dating was first practiced on archaeological pottery. Even though TL on pottery is provided worldwide in a routine base, possible implications regarding high fired pottery, where K leaching is well documented (Schwedt et al., 2006) aware for additional examination in order to overcome potentially age overestimation (Zacharias et al., 2005).

Dating applications pottery/bricks

The dose rate can be expressed as follows: Dr = Dα+Dβ+Dγ+Dcos where α- β- and γ- refer to type of rays that are summed up to the total dose rate and cos stands for the cosmic irradiation contribution.

burnt flint

The concentrations of the radioactive elements (U, Th and 40K) in the sample are usually very different from those of its surroundings. Thus, internal and external dose rates have to be assessed independently. Furthermore, it is necessary to estimate the cosmic dose which decreases with depth below ground, and is also depended on altitude as well as geographic latitude (Prescott and Hutton, 1994). The environmental dose is best measured by employing laboratory phosphors preferably the Al2O3:C type (Akselrod, 1998) to be buried for up to a period of 3-4 months since this controls possible seasonal variations in the cosmic dose and humidity.

In nowadays, TL dating of burnt flint/chert artifacts is of the main chronological applications, of especially prehistoric archaeological contexts, covering the time span of the controlled use of fire by humans. TL dating of burnt flint artefacts from Lower and Middle Palaeolithic of the Near East, is widely used as a reference in debates on the evolution of Palaeolithic industries and on the origin of modern humans and their relationship to the Neanderthals (Mercier and Valladas, 2003). A detailed TL dating study (Valladas et al., 2007) on burned flints from the lowest Middle Palaeolithic stratigraphic unit of Theopetra cave at Thessaly (Greece), verified that these layers are much older than postulated on the basis of earlier radiocarbon dates and also that Theopetra contains the oldest, so far dated, lithic artefact deposits of the Greek Middle Palaeolithic. slags, metallurgical remains Ancient slag heaps present a valuable source for the technological and archaeological study of early metal production processes. Crucial to the understanding of these sites is their dating, often complicated by the rarity of diagnostic pottery fragments necessitating the use of appropriate scientific techniques (Fig. 5). TL has a great potential for absolute dating of archaeometallurgical remains, both slags and kiln fragments (Godfrey-Smith and

Figure 5. View from a site at Seriphos Island, Cyclades (Greece) with remains indicating archaeometallurgical activity.


Luminescence Dating Beyond Pottery: a Review Casey, 2003), due to the fine resetting of the ‘luminescence clock’ caused by the temperature of the smelting process. Direct dating of slag by means of luminescence has proven problematic in the past either with TL (Elitzsch et al., 1983) or with OSL (Gautier, 2001) protocols, mainly due to the complex composition of the material and inaccuracies in microdosimetry calculations. In Zacharias et al. (2006a; 2006b), TL dating of archaeometallurgical kiln wall fragments from two Cycladic islands (Kythnos and Seriphos) is reported. For that dating was applied on selective layers recovered form the fragments TL dates resulted indicating that the metallurgical activities for both islands took place broadly in the first half of the third millennium BC, which corresponds to the Aegean Early Bronze Age periods.

Figure 6. Photos of sediment dating applications in various environments. From the upper left to the lower right: Sand-dune at Attiki, Greece / beach rock formations / geoglyps at Palpa, Perou / Calcite marine fossileferous soils at Corinth, Greece.

sediments The establishment of sediment dating was trailed behind the development of OSL dating. OSL can reliably provide chronologies for almost all kind of sediment material (Fig. 6), ranging from the easily transported/bleached aeolian to fluvial and colluvium formations (Liritzis, 2000; Wallinga, 2002; Lian and Roberts, 2006).

sediments from the Middle Stone Age occupation levels (Henshilwood et al., 2002), and for two sterile sands, one underlying the archaeological sediment and one overlying the Later Stone Age occupation levels and conclude that the periods of occupation were determined by changes in sea level, with abundant sources of seafood available in times of high sea level and with the cave being closed by the accumulation of large dunes during periods of low sea level, such as during oxygen isotope stages 4 and 6.

In all of the sediment material used for dating purposes the minerals employed is either quartz or feldspars. Although feldspars emit much more luminescence than does quartz, and also the luminescence from feldspars usually saturates at higher doses compared to that from quartz, there are several reasons that luminescence specialists using preferably quartz. The reasons is that feldspars display some malign properties; e.g. K-feldspars when using the SAR protocol shows sensitivity changes and as most significant obstacle is the so-called anomalous fading (Wintle, 1973; Spooner, 1994), which refers to the loss of electrons from traps that are thermally stable at ambient temperatures.

- Eitel et al. (2001) worked in NW Namibia where locally produced and allochthonous dust blown silty deposits are widespread. They concluded that between about 30 and 8 ka loessic alluvial deposits several metres in thickness accumulated in valleys and basins and that a change from a dry to a moister semi-arid climate, additionally supported by sedimentological studies, occurred at about 9–8 ka. beach-rocks The term refers to cemented, usually by calcite, sand and shingles, appeared in coastal areas. They usually provide valuable information on the sea-level fluctuations and generally changes of marine, coastal and continental environments; in Mediterranean areas indicate for Upper Pleistocene – Holocene periods (Frechen, 2004; Mauz and Hassler, 2000).

An anthology of large scale sediment dating applications that significantly contribute to our knowledge of past environmentally evolutions should include the work done by: - Bateman and Murton (2006) who presented a study of the chronology of sand sheets and sand dunes situated in Tuktoyaktuk Coastlands of the Arctic Coastal Plain in Canada that span the last 40 ka. Using quartz and the standard SAR technique, they managed to make important links between the nature and timing of sediment supply, various styles of sedimentation, landform stability, and what is already known about fluctuations in past climate.

volcanic material TL/OSL can successfully be applied for dating sediments affected by lava flows and various products of volcanic activity. An updated review, practises and approaches encountered with the luminescence dating of such or related material can be found at Fattahi and Stokes (2003) and Watanuki et al. (2005).

- Jacobs et al. (2006) applied OSL on single aliquots and single quartz grains from deposits within Blombos Cave, South Africa. Ages have been obtained for six 27

N. Zacharias mortars

Other uses of luminescence

Lime mortars mixed with sand are used extensively in architectural and decorative works during the last 4.000 years. The first application of quartz OSL dating extracted from mortars referred to a Byzantine church monument dated to the 10th century (Zacharias et al., 2002; Goedicke, 2003). The same principals can applied to any cemented materials with the requirement to contain quartz.

Luminescence is also in use through a variety of applications in personal, medical and retrospective (accident) dosimetry since the development of the technique (Bøtter-Jensen and Murray, 2002; Akselrod et al., 2006). One critical issue in space exploration programs is the achievement of best control of radiation sources received by the astronauts; luminescence is playing a major role by contributing to the development of organologies, experimental parameters and ultra-sensitive radiation phosphors (Yukihara, et al., 2006).

Innovative applications stone courses

Finally, luminescence signals have the potential to contribute to characterisation and technological studies of archaeological material, like glass, obsidian and steatite; this is achieved by on a combination of the thermal prehistory, radiation dose received and the geological origin of the material (Akridge and Benoit, 2001; Galli et al., 2004; Zacharias et al., in press)

The potential use of luminescence techniques to date the initial or reconstruction phases of stone blocks, walls, buildings and ancient temples intrigued the field since the ‘90s (Liritzis et al., 1997; Liritzis and Galloway, 1999) finally provided reliable conventional (Rink and Bartoll, 2005; Vafiadou et al., 2007) or non-destructive approaches (Greilich et al., 2005).

Conclusions rock painting The wide material and case studies topics, spanning from almost all inorganic material found or formed on the Earth’s crust, and the age range that can potentially reach 1Ma, illustrates the flexibility and robustness of luminescence dating methods and their ability to solve long-standing, as well as new, problems in Archaeology and Quaternary research.

Here, luminescence contributes to the efforts of dating rock art since this stands as an important behavioural hallmark of modern humans. The method uses quartz grains removed from fossil mud-wasp nests underlying the paintings (Roberts et al., 1997; Yoshida et al., 2003). tsunamis

Luminescence dating ‘… should lead to a revolution in our understanding of archaeological processes similar to the introduction of radiocarbon dating in the ‘50s and ‘60s’ (Pollard and Brothwell, 2002).

From time to time tsunami deposits in their wake; these take the form of sheets of sand or gravel carried forward by the tsunami and dumped up shore. On the assumption that the materials were adequately exposed to daylight before transport for long enough to reset the dating ‘clock’ and were then covered by tidal or other deposits, they can be successfully dated (Ollerhead et al., 2001; Reinhardt et al., 2006).

References Aitken, M.J. 1985 Thermoluminescence Dating, Academic Press, London Aitken, M.J. 1998. An introduction to Optical Dating, Oxford University Press, Oxford. Akridge, D.G., 2001 P.H. Benoit, Luminescence properties of chert and some archaeological applications, Journal of Archaeological Science 28 (2), 143-151. Akselrod, M.S., A.C. Lucas, J.C. Polf, S.W.S. McKeever, 1998. Optically stimulated luminescence of Al2O3, Radiation Measurements 29 (3-4), 391-399. Akselrod, M.S., 2006 L. Bøtter-Jensen, S.W.S. McKeever, Optically stimulated luminescence and its use in medical dosimetry, Radiation Measurements 41 (SUPPL. 1), S78S99. Bateman, M.D,. J.B. Murton, 2006. The chronostratigraphy of Late Pleistocene glacial and periglacial aeolian activity in the Tuktoyaktuk coastlands, NWT, Canada, Quaternary Science Reviews 25 (19-20), 2552-2568. Bøtter-Jensen, L. A.S. Murray, 2002. Optically stimulated luminescence in retrospective dosimetry 2002 Radiation Protection Dosimetry 101 (1-4), 309-314. Bøtter-Jensen, L., S.W.S. McKeever, A. Wintle, 2003. Optically Stimulated Luminescence Dosimetry, Elsevier, Amsterdam. Cheong, C.S., D.G. Hong, K.S. Lee, J.W. Kim, J.H. Choi, A.S.

fault movements Fault slip-rates and the recurrence interval between earthquakes can be determined from reliable dating of deformed and associated deposits, either wind-blown or colluvial and fluvial transported sediments. Within the last decade completed dating case studies related with known and still active fault zones are those of the Late Pleistocene alluvial and colluvial deposits from Nahal Shehoret, at Arava Valley, Israel (Porat et al., 1997), Late Pleistocene fluvial deposits from the Wangsan fault, SE Korea (Cheong, et al., 2003), the Sabzevar thrust fault in NE Iran provided Holocene based sequences (Fattahi et al., 2006), and more.


Luminescence Dating Beyond Pottery: a Review bleaching of thermoluminescence of ancient marble, Journal of Radioanalytical and Nuclear Chemistry 241 (2), 361-368. Liritzis, I. 2000. Advances in thermo- and opto-luminescence dating of environmental materials (sedimentary deposits) Part I: techniques, Global Nest (Part II: Applications) 29-49, 3-27. Mauz, B., U. Hassler, 2000. Luminescence chronology of late pleistocene raised beaches in southern Italy: New data of relative sea-level changes, Marine Geology 170 (1-2), 187203. McKeever, S.W.S. 1985. Thermoluminescence in Solids, Cambridge University Press, Cambridge. McKeever, S.W.S., Chen, R., 1997. Luminescence models, Radiation Measurements 27 (5-6), 625-661. Mercier, N., H. Valladas, 2003. Reassessment of TL age estimates of burnt flints from the Paleolithic site of Tabun Cave, Israel, Journal of Human Evolution 45 (5), 401-409. Michael, C.T., N. Zacharias, D. Dimotikali, Y. Maniatis, 1997. A new technique for measuring the natural dose in TL dating and its application in the dating of a mortar containing ceramic grains, Ancient TL 15 (2-3), 36-42. Michael, C.T., N. Zacharias, 2000. A new technique for thicksource alpha counting determination of U and Th, Nuclear Instruments and Methods 439 (1), 167-177. Murray, A.S., A.G. Wintle, 2000. Luminescence dating of quartz using an improved single-aliquot regenerative-dose protocol, Radiation Measurements 32 (1), 57-73. Ollerhead, J., D.J. Huntley, A.R. Nelson, H.M. Kelsey, 2001. Optical dating of tsunami-laid sand from an Oregon coastal lake 2001 Quaternary Science Reviews 20 (18), 1915-1926. Pollard, Α.Μ., D.R. Brothwell, 2002. Archaeological Sciences, Wiley editions., U.K. Prescott, J.R., J.T. Hutton, 1994. Cosmic ray contributions to dose rates for luminescence and ESR dating: large depths and long-term time variations, Radiation Measurements 23 (2-3) 497-500. Prescott, J.R., J.T. Hutton, 1995. Environmental dose rates and radioactive disequilibrium from some Australian luminescence dating sites, Quaternary Science Reviews 14 (4), 439-448. Porat, R. N., Amit, E. Zilberman, Y. Enzel, 1997. Luminescence dating of fault-related alluvial fan sediments in the Southern Arava Valley, Israel, Quaternary Science Reviews 16 (3-5), 397-402. Reinhardt, E.G., B.N. Goodman, J.I. Boyce, G. Lopez, P. Van Hengstum, W.J. Rink, Y.. Mart, A. Raban, 2006. The tsunami of 13 December A.D. 115 and the destruction of Herod the Great’s harbor at Caesarea Maritima, Israel, Geology 34 (12), 1061-1064. Rink, W.J., J. Bartoll, 2005. Dating the geometric Nasca lines in the Peruvian desert, Antiquity 79, 390-401. Roberts, R., G. Walsh, A. Murray, J. Olley, R. Jones, M. Morwood, C. Tuniz, E. Lawson, M. Macphail, D. Bovudery, I. Naumann, 1997. Luminescence dating of rock art and past environments using mud-wasp nests in northern Australia, Nature 387 (6634), 696-699. Schwedt, Α., H. Mommsen, N. Zacharias, J. Buxeda i Garrigós, 2006. Analcime crystallization and compositional profiles – comparing approaches to detect post-depositional alterations in archaeological pottery, Archaeometry 48 (3), 237-251. Spooner, N.A. 1994. The anomalous fading of infrared-stimulated luminescence from feldspars, Radiation Measurements 23, 625–632. Yukihara, E.G., G.O. Sawakuchi, S. Guduru, S.W.S. McKeever, R. Gaza, E.R. Benton, N. Yasuda, Y. Uchihori, H. Kitamura, 2006. Application of the optically stimulated luminescence (OSL) technique in space dosimetry, Radiation Measurements 41 (9-10), 1126-1135. Vafiadou, A., A.S. Murray, I. Liritzis, 2007. Optically stimulated luminescence (OSL) dating investigations of

Murray, U. Chwa, C.B. Im, C.J. Chang, H.W. Chang, 2003. Determination of slip rate by optical dating of fluvial deposits from the Wangsan fault, SE Korea, Quaternary Science Reviews 22 (10-13), 1207-1211. Eitel, B., W. Dieter Blümel, K. Hüser, B. Mauz, 2001. Dust and loessic alluvial deposits in Northwestern Namibia (Damaraland, Kaokoveld): Sedimentology and palaeoclimatic evidence based on luminescence data, Quaternary International 76-77, 57-65. Elitzsch,C., E. Pernicka, G.A. Wagner, 1983. Thermoluminescence dating of archaeometallurgical slags, PACT 9, 271-286. Fattahi, M., S. Stokes, 2003. Dating volcanic and related sediments by luminescence methods: A review, Earth-Science Reviews 62 (3-4), 229-264. Fattahi, M., R. Walker, J. Hollingsworth, A. Bahroudi, H. Nazari, M. Talebian, S. Armitage, S., Stokes, 2006. Holocene sliprate on the Sabzevar thrust fault, NE Iran, determined using optically stimulated luminescence (OSL), Earth and Planetary Science Letters 245 (3-4), 673-684. Frechen, M., A. Neber, A. Tsatskin, W. Boenigk, A. Ronen, 2004.Chronology of Pleistocene sedimentary cycles in the Carmel Coastal Plain of Israel, Quaternary International 121 (1), 41-52. Galbraith, R.F., R.G. Roberts, G.M. Laslett, H. Yoshida, J.M. Olley, 1999. Optical dating of single and multiple grains of quartz from Jinmium Rock Shelter, Northern Australia: part I, experimental design and statistical models, Archaeometry 41, 339–364. Galli, A., M. Martini, C. Montanari, E. Sibilia, 2004. Thermally and optically stimulated luminescence of early medieval blue-green glass mosaics, Radiation Measurements 38 (4-6), 799-803. Gautier, A. 2001. Luminescence dating of archaeometallurgical slag: use of the SAR technique for determination of the burial dose, Quaternary Science Reviews 20, 973-980. Godfrey-Smith, D.I., J.L. Casey, 2003. Direct thermoluminescence chronology for Early Iron Age smelting technology on the Gambaga Escarpment, Ghana, Journal of Archaeological Science 30, 1037-1050. Goedicke, C. 2003. Dating historical calcite mortar by blue OSL: Results from known age samples, Radiation Measurements 37 (4-5), 409-415. Greilich, S., U.A. Glasmacher, G.A. 2005. Wagner, Optical dating of granitic stone surfaces, Archaeometry 47 (3), 645-665. Henshilwood, C.S., F. d’Errico, R. Yates, Z. Jacobs, C. Tribolo, G.AT. Duller, N. Mercier, J.C. Sealy, H. Valladas, I. Watts, A.G. Wintle, A.G., 2002. Emergence of modern human behavior: Middle stone age engravings from South Africa, Science 295 (5558), 1278-1280. Jacobs, Z., G.A.T. Duller, A.G. Wintle, C.S. Henshilwood, 2006. Extending the chronology of deposits at Blombos Cave, South Africa, back to 140 ka using optical dating of single and multiple grains of quartz , Journal of Human Evolution 51 (3), 255-273. Krbetschek, M.R., J. Götze, A. Dietrich, T. Trautmann, 1997. Spectral information from minerals relevant for luminescence dating, Radiation Measurements 27 (5-6), 695-748. Lian, O.B., R.G. Roberts, 2006. Dating the Quaternary: progress in luminescence dating of sediments, Quaternary Science Reviews 25 (19-20), 2449-2468. Liritzis, I., M. Kokoris, 1992. Revised dose-rate data for thermoluminescence and ESR dating, Nuclear Geophysics 6 (3), 423-443. Liritizis, I., P. Guibert, F. Foti, M. Schvører, 1997. The Temple of Apollo (Delphi) Strengthens Novel Thermoluminescence Dating Method, Geoarchaeology - An International Journal 12 (5), 479-496. Liritzis, I., R.B. Galloway, 1999. Dating implications from solar


N. Zacharias Zacharias, N., B. Mauz, C. T. Michael, 2002. Luminescence quartz dating of lime mortars, A first research approach, Radiation Protection Dosimetry, 101 (1-4), 379-382. Zacharias, N., J. Buxeda i Garrigos, H. Mommsen, A. Schwedt, V. Kilikoglou, 2005. Implications of burial alterations on luminescence dating of archaeological ceramics, Journal of Archaeological Science 32 (1), 49-57. Zacharias, N., C.T. Michael, O. Philaniotou-Hadjiannastasiou, A. Hein, Y. Bassiakos, 2006a. Fine-grain TL dating of archaeometallurgical furmace walls, Journal of Cultural Heritage 7, 23-29. Zacharias, N., C.T. Michael, M. Georgakopoulou, V. Kilikoglou, Y. Bassiakos, 2006b. Quartz TL dating on selected layers from archaeometallurgical kiln fragments: A proposed procedure to overcome age dispersion, Geochronometria 25, 29-36. Zacharias, N., A. Oikonomou, K. Beltsios, A. Karydas, Y. Bassiakos, C.T. Michael, Ch. Zarkadas, (in press) Solid-State Luminescence Techniques Applied for the Examination of Archaeological Glass Beads, Journal of Optical Materials.

rock and underlying soil from three case studies, Journal of Archaeological Science 34 (10), 1659-1669. Valladas, H., N. Mercier, L. Froget, J.-L. Joron, J.-L. Reyss, P. Karkanas, E. Panagopoulou, N. Kyparissi-Apostolika, 2007. TL age-estimates for the Middle Palaeolithic layers at Theopetra cave (Greece), Quaternary Geochronology 2, 303–308. Wallinga, J. 2002. Optically stimulated luminescence dating of fluvial deposits: A review, Boreas 31 (4), 303-322. Watanuki, T., A.S. Murray, S. Tsukamoto, 2005. Quartz and polymineral luminescence dating of Japanese loess over the last 0.6 Ma: Comparison with an independent chronology, Earth and Planetary Science Letters 240 (3-4), 774-789. Wintle, A.G. 1973. Anomalous fading of thermoluminescence in mineral samples, Nature 245, 143–144. Wintle, A.G., A.S. Murray, 2006. A review of quartz optically stimulated luminescence characteristics and their relevance in single-aliquot regeneration dating protocols, Radiation Measurements 41 (4), 369-391. Yoshida, H., R.G. Roberts, J.M. Olley, 2003. Progress towards single-grain optical dating of fossil mud-wasp nests and associated rock art in northern Australia, Quaternary Science Reviews 22 (10-13), 1273-1278.


HUNTING OUT FORGERS OF ARCHAEOLOGICAL MARBLE OBJECTS: REVIEWING THE ARCHAEOMETRIST’S ROLE AS A DETECTIVE K. Polikreti Photonics and Optoelectronics Laboratory, Department of Physics, University of Cyprus, P.O. Box 20537, Nicosia 1678, Cyprus, [email protected] Abstract: When C-14 and thermoluminescence dating were applied on archaeological objects, organic material and ceramics correspondingly, archaeology was at last disburdened of the fear of forgers. Marble forgers though are still active, since no marble authenticating technique is yet available. The problem is that, in principle, we are interested in the age of the marble surface, i.e. the date of the surface creation. The age of marble itself, even if we could have it precisely measured, would give no useful information. Till the early fifties, the usual way to detect forgeries was mainly by stylistic criteria. But the forgers had become more sophisticated and museums and galleries needed a reliable methodology for authenticity testing. Ultraviolet induced - visible photography is the technique used today by most museums all over the world, in order to verify marble object authenticity. This technique though, is based on empirical – statistical observations and its results have not yet been scientifically explained nor proved reliable. Due to the lack of a widely approved test, archaeometrists investigate the marble provenance, as an indirect way to detect a fake. They also study every single detail of the marble surface. A tiny toolmark could reveal modern origin, the presence of a peculiar mineral on the surface could imply “cooking” of the object with acids etc, the profile of elemental distribution from the outer surface to the marble bulk could give evidence on an artificial paste applied for creating a “dull” or “aged” appearance. All types of surface information are gathered, seeking the decisive factor against or in favor of authenticity. In certain cases, this decisive factor cannot be found, and the ambiguity remains unresolved through the years. The famous ‘Getty kouros’, made on Thassian marble is suspected of being a fake since 1980 and has not yet been decisively authenticated nor decisively unmasked. Fakes of Cycladic idols were so numerous in the 1970s that the level of looting of the Cycladic sites was reduced those years and several cases are questioned again and again. A new authentication approach, based on thermoluminescence, is still in its infancy and cannot yet be applied on museum items. This methodology can be applied on buried objects, to date the “date of burial” or distinguish marble objects exposed to sunlight since antiquity from those exposed recently. The on going work is extended on artifacts buried and then exposed to sunlight, a category including most of the known objects of ambiguous authenticity. Περίληψη: Είναι γνωστό, ότι δεν υπάρχει μέχρι σήμερα αξιόπιστη φυσικοχημική τεχνική για τον έλεγχο της αυθεντικότητας μαρμάρινων αλλά και γενικότερα πέτρινων αντικειμένων. Μέχρι τις αρχές της δεκαετίας του 1950, ο συνήθης τρόπος ανίχνευσης των πλαστών αρχαίων αντικειμένων από μάρμαρο ήταν τα στυλιστικά καλλιτεχνικά κριτήρια. Λόγω της έλλειψης μιας αξιόπιστης τεχνικής πιστοποίησης αυθεντικότητας, οι ερευνητές αρχικά διερευνούν την προέλευση του μαρμάρου. Επίσης μελετούν την παραμικρή λεπτομέρεια της επιφάνειας για τυχόν ανίχνευση ιχνών από σύγχρονα εργαλεία ή την παρουσία κάποιου ορυκτού που υποδηλώνει την παρουσία τεχνητής πάστας. Επίσης ελέγχουν την υφή της επιφάνειας για πιθανή δράση με οξύ, την πρόσφυση της πατίνας στην επιφάνεια, τη στοιχειακή κατανομή από την επιφάνεια της πατίνας προς το υγιές μάρμαρο κ.ά. Συλλέγονται δηλαδή όλες οι πληροφορίες για τη σύσταση και τη μορφολογία της πατίνας, σε μια αναζήτηση ενός κρίσιμου παράγοντα υπέρ ή εναντίον της αυθεντικότητας του αντικειμένου. Η φωτογραφία φθορισμού στο υπεριώδες φως είναι η τεχνική που χρησιμοποιείται συνήθως από μουσεία και συλλέκτες ανά τον κόσμο, για έλεγχο αυθεντικότητας. Η τεχνική αυτή όμως βασίζεται σε εμπειρικές παρατηρήσεις και η αξιοπιστία της είναι αμφισβητήσιμη. Τα τελευταία χρόνια αναπτύσσεται η τεχνική της θερμοφωταύγειας, η οποία βασίζεται στις κρυσταλλικές ατέλειες του μαρμάρου και την αλληλεπίδρασή τους με τις φυσικές ακτινοβολίες. Τα αποτελέσματα είναι ακόμα σε πειραματικό στάδιο αλλά η έρευνα συνεχίζεται.


Although, the exact definition of the term “forgery” is beyond the scope of this paper, it is well accepted that a forged work of art possesses the intent to deceive, usually for financial gain, by proffering an art object as representing something other than it is. Forged artworks may be: • Deliberate imitations offered as originals • Genuine objects altered by reworking to increase their value • Early copies (not initially intended to deceive) passed off later as originals • Pastiches comprised of original parts of different works passed off as homogenous originals

Products of a workshop that have been erroneously attributed to the master of the workshop.

The discussion here aims to make a very brief review of the techniques that can be used in authenticity testing. A few examples of famous marble objects of disputable authenticity will also be referred to during the discussion. Although archaeometry can solve many archaeological problems there is no archaeometric technique to ensure reliable answers to marble (or stone in general) authenticity problems. In cases of disputable authenticity the archaeometrist act as a detective. He carries out all applicable examinations and analyses, hunting out traces 31

K. Polikreti in favor or against authenticity. The result may be a clear answer or not, depending on the particular case and nature of the marble surface. In the last 20 years many controversies over important artifacts have remained unresolved for years due to lack of conclusive evidence.

Treatment with acid (hydrochloric or sulphuric) was the forger’s favorite method to imitate weathered marble surfaces during the 1950s and is still used today. It gives the surface a “dull and jelly” appearance. The surface is usually covered by loosely adhering material, the nature of which depends on the “burial treatment”, the subsequent stage (soil, cinders or a paste of various compositions). An example of an acid treated forgery is a Kouros Torso, purchased by the Getty Museum in 1990 (Summary of Scientific Research on the Getty Kouros 1992). However, in certain cases, the acid treatment can be a result of cleaning, which was a common practice of conservators in the past (Von Bothmer 1964).

Techniques used in authenticity investigations The techniques used in authentication studies can be classified into the following categories: 1. 2. 3.

Macro- and microscopic examination: Root-marks, tool-marks and depositions. Non-destructive techniques: UV-induced fluorescence photography. Destructive techniques: Marble provenance investigation, analytical techniques for studying the patina composition and morphology and thermoluminescence.

Marble provenance investigation In most cases, the first step in authenticating a work of sculpture is to identify the quarry from which the stone was extracted. If the marble source proves to be irrelevant with the object’s historical context, then we have a decisive indication against its authenticity. Isotopic analysis for example, revealed that a head of Achilles owned by the Getty Museum and supposed to come from a Tegea Temple (Peloponnesus), was carved in Parian marble (Margolis 1989; Spier 1990). If the source is historically reasonable, authenticity is still uncertain since most of the marble types used in antiquity are still available. Provenance determination can also serve as a test for correct assembly of broken fragments (Herz and Wenner 1978). Stable isotope analysis for example, showed that a Livia portrait in the Ny Carlsberg Glyptotek had an original Parian head but skullcap of Ephessian and nose of Carrara marble (Fig. 1) (Herz 1990).

All these commonly used techniques will be discussed in the following along with a few examples of their use. Visual examination and optical microscopy The term “root-marks” represents dendrite–like formations, originating from alive or rotting plant roots. The root outline is etched on the surface of contact due to the acidic liquid contained in the alive roots or the humic acids produced by rotten roots. As the process continues, re-crystallisation of calcite builds a deposit upon the etched root-shaped designs (Young and Ashmole 1968). Rootmarks are usually a positive sign in case of a disputable object although it is said that wrapping an object in wholewheat spaghetti creates marks like those left by plant roots (Gill and Chippindale 1993).

Some of the numerous physical and geochemical techniques have been proved successful in marble provenance investigation studies over the last two decades are: petrographic examination under the microscope,

As to the original carvers’ tool-marks, there are three important points that should be kept in mind: 1. 2. 3.

Changes of the tool-mark shape or type may indicate recent reworking. The use of certain tools is unknown before certain periods. Sharp edges are not expected after long term burial.

The marks of a running-drill for example, a technique unknown till the mid-fifth century B.C., were used as a conclusive proof that an “Archaic Greek Kore” purchased by the Metropolitan Museum in 1926 was a fake (Ashmole 1961). A Julius Caesar marble head acquired by the British museum in 1918 was for a long period the most famous and widely reproduced portrait of Julius Caesar until 1961, when it was concluded to be a forgery. Ashmole (1961) pointed out that the surface had been artificially aged perhaps by pounding with a nail-studded piece of wood and stained.

Figure 1. The stable isotope analysis of “Livia” head, Copenhagen, Ny Carlsberg Glyptotek: The head (H) made on Efessos marble, the skullcap (S) made on Parian marble and the nose (N) made on Carrara marble. Carbon and oxygen isotopic fields from Gorgoni et al. 2002.


Hunting out Forgers of Archaeological Marble Objects The use of CL as an accessory to stable isotopes can be very useful in the provenance of white marble (Barbin et al. 1992). Comparison of stable isotopes of marble surface and bulk The isotopic composition of Mediterranean white marble varies from -1‰ to +6‰ for δ13C and from -11‰ to +1‰ for δ18O. However, long exposure to atmospheric or ground water, results in enrichment of the light isotopes in the marble surface layer, due to dissolution-reprecipitation processes (Ulens et al. 1994). According to Ulens et al. (1994) quarry samples show large differentiations from bulk δ-values, while ancient surfaces show smaller variations. The technique is not considered indisputable since it depends on the specific weathering conditions and conservation history of the artifact (acid cleaning for example).

Figure 2. Isotopic shift between marble bulk and surface observed on quarry and archaeological samples. Reproduced from Ulens et al. 1995.

Chemical, mineralogical composition and morphology of the patina

Cathodoluminescence (CL), Electron Paramagnetic Resonance (EPR), Instrumental Neutron Activation Analysis (INAA) and Stable Isotope ratio analysis. Each technique has advantages and disadvantages and no one of them can give reliable and accurate results in all cases. A combination of techniques seems to be the best approach (Moens et al. 1992).

The general term “patina” describes the altered surface layers of marble. The thickness of these layers is a function of age, and varies from 10 μm to 10 mm depending on marble type, grain size, porosity, environmental acidity and humidity. The question of authenticity involves the possible faking of the patina and specific analyses need to be carried out to check this hypothesis. The techniques typically employed are: thin section petrography, X-ray diffraction and Scanning Electron Microscopy (SEM) with microprobe analysis (EPMA) (Margolis 1989).

The examination of marble thin sections under the petrographic microscope (Schmid et al. 1999) is probably the oldest technique. The need for a large amount of material for analysis makes the technique unviable for most artifacts. Today MGS is usually mapped together with other analytical data such as EPR (Polikreti and Maniatis 2002) or isotopic data (Gorgoni et al. 2002).

Elemental distribution or simple profiles from the surface to the marble bulk give very important information because a natural patina shows a continuous, gradual decrease in elemental concentration. Laser ablation-Induced Coupled Plasma-Mass Spectrometry (LA-ICP-MS), introduced by Ulens et al. (1995) has been used for sampling successive layers towards the bulk and the vaporised material is transported in an ICP-MS unit. According to Ulens et al. (1995) La, Ba, Mn, Pb, Al, Fe and Mg concentrations increase in the patina, while Sr concentration decreases. Al, Si and Fe, show the highest concentration, most probably due to the presence of clay minerals.

At present, the most popular technique for identifying quarry sources is isotopic analysis of oxygen and carbon (Herz 1990). Intensive sampling of quarries has lead to a large databank of standardized oxygen and carbon values (Fig. 2), which is typically depicted in the form of isotopic quarry fields (Gorgoni et al. 2002). Sometimes fields overlap, so other techniques (Pollini et al. 1998) must be employed to make a reliable distinction between sources. EPR spectroscopy has also been used with great success in marble provenance investigations (Polikreti and Maniatis 2002). The greatest advantage of this technique is that the EPR spectrum contains a lot of information which can be quantified in at least 10 parameters. The technique has been applied to a large number of archaeological and arthistorical cases, with a promisingly high rate of success (Pollini et al. 1998, Maniatis and Polikreti 2000).

Due to the complexity of the patina formation mechanisms, building up a “patina sample bank” is useless. Only a few general rules are accepted by all researchers in the field: • • •

Instrumental Neutron Activation Analysis (INAA, Matthews 1997) has also been tried but many trace elements vary by factors of over a hundred within the same quarry. 33

A typical natural patina consists mainly of iron oxides, quartz, clay minerals, calcite and dolomite. A micrite layer of recrystallised calcite is always present on authentic ancient surfaces. The penetration of soil material well into the matrix of marble is characteristic of authentic surfaces.

K. Polikreti surfaces increases, the amount of iron (Fe3+) in the CaCO3 lattice decreases by percentages from 7% to practically 100% (Fig. 3, Polikreti and Maniatis 2004). The main cause of this depletion of iron is the dissolution and recrystallisation of CaCO3. The problem is that in the 20 patina samples examined (Polikreti and Maniatis 2004), the alteration depths as determined by SEM are always less than 2mm, while Fe3+ depletion extends up to 4-7 mm. If re-crystallisation stops at 4-7 mm, what other phenomenon affects the outer marble layers? How Fe3+ is removed from the healthy marble crystals located in depths between 2 and 4-7 mm? According to the bibliography (Stipp et al. 1998), iron may follow the behavior of other ions present in the calcite lattice: Fe3+ is entrapped in Ca2+ sites during metamorphosis. But it is not stable because Fe3+ has a smaller radius and higher charge than Ca2+. When we curve a marble surface, this sensitive equilibrium is destroyed and Fe3+ ions start diffusing through the lattice, in a search for more stable positions that is the surface, grain boundaries and micro fractures. For other ions reported in literature, the diffusion rates are in the scale of nanometers in weeks or months. This gives us a very good and very promising “clock mechanism” for authenticity and dating but a lot of work is ahead to be done.

Figure 3. Profile of Mn2+ and Fe3+ with distance from the weathered marble surface for an ancient piece. The difference between alteration depth and Fe3+ depletion depth is shown. Figure taken form Polikreti and Maniatis 2004.

As an example of a very extensively studied patina, we note the patina of the Getty Kouros, which consists of calcium oxalates (~90% whewellite), which are common on marble surfaces exposed to the atmosphere and are usually related to biological action. However, the Kouros patina is unusually continuous in comparison with wellknown, biological calcium oxalate patinas, which have a very complex morphology and composition. This continuity, “suggests that the Kouros may have been exposed to later treatment or was weathered in a fairly uniform environment” (Summary of Scientific Research on the Getty Kouros 1992).

Ultra violet – induced fluorescence photography Rorimer (1931) observed that under ultraviolet light, newly worked marble, displays an even red-violet colour, while ancient sculptures show a spotty bluish-white colour. Although, the reliability of the technique has not been ensured, it is commonly used by museums and private collectors due to its non-destructive character (Von Bothmer 1964; Young and Ashmole 1968; Riederer 1977; Margolis 1989).

We have to point out here, that all the previous techniques give total iron concentrations. EPR spectroscopy is the only technique, which has shown that while the total amount of iron (oxides and silicates included) on marble Recently cut marble

As a characteristic misleading example we refer to the surface of the Getty Kouros, which displays a purple colour under ultraviolet light. This violet colour indicates

Exposed to the sun for several years

Buried for more than a thousand years


Geological thermoluminescence – homogenously distributed

The surface layers show thermoluminescence bleaching

Exposed to the sun again



The surface TL increases again due to the radioactivity of the ground (40K, 238U, 232Th)

The surface shows TL bleaching again

Figure 4. The basic idea of authenticating marble sculpture by thermoluminescence.


Hunting out Forgers of Archaeological Marble Objects B.C. The calculated TL age was 2570 ± 410 years BP, which is encouragingly close to the archaeological age. The “type C” class includes most of the ambiguous pieces (Fig. 4) but has not been studied yet. Conclusions •

The difficulties in authenticating ancient marble are a direct result of the complexity of the phenomena governing the patina formation. No reliable result can be obtained without examining cross-sections which give invaluable information on the morphology of the interface between weathered layers and healthy marble (optical or SE microscopy). Such an examination needs a rather large sample, which is in some cases prohibitive, especially for small or precious works of art.

TL seems to provide a natural (light) clock: when a marble surface is exposed to the sun the clock is set to zero, while in burial environment the clock counts years of age but the work has to go on in order for the whole phenomenon to be fully understood.

Marble authentication investigations should be entrusted to well-experienced experts, because the task is not just the application of physicochemical techniques on a weathered material.

Figure 5. Thermoluminescence profiles with distance from the marble surface for various exposure times. Figure taken from Polikreti et al. 2002.

modern surfaces. This result is not at all useful because the light is produced by the thick calcium oxalate layer and not by the weathered marble (Summary of Scientific Research on the Getty Kouros 1992). Thermoluminescence Thermoluminescence (TL) is a new approach in marble authenticity testing which is still in the experimental stage (Theocaris et al. 1997, Polikreti et al. 2002, 2003). The basic idea is that long-term burial increases TL signals while exposure to sunlight bleaches them. Accordingly, we can then classify marble objects into three types (Fig. 4): those continuously exposed to sunlight since antiquity (minimum TL signals – type A), those exposed and then continuously buried (intense surface signals proportional to the burial time – type B) and those exposed, buried and then exposed again (type C). Only the first two types have been studied extensively and the results are promising.

References Ashmole, B., 1961, Forgeries of ancient sculpture in marble: Creation and detection, The First J.L Myres Memorial Lecture, Oxford, p.3. Barbin V., Ramseyer K., Decrouez D., Burns S.V., Chamay J. and Maier J.L., 1992, Cathodoluminescence of white marble: An overview, Archaeometry, 34, 175-183. Gill, D. W. J. and C. Chippindale, 1993, Material and intellectual consequences of esteem for Cycladic figures, American Journal of Archaeology, 106 (4), 601-659. Gorgoni, C.L., Lazzarini, L., Palante, P. and Turi, B., 2002, An updated and detailed mineropetrographic and C-O stable isotopic reference database for the main Mediterranean marbles used in antiquity, In Interdisciplinary studies in Ancient Stone. 5th ASMOSIA Conference (J.L. Herrmann, N. Herz and R. Newman, eds), Boston, 1998, London: Archetype Publications, 114-131. Herz, N. and Wenner, D. B., 1978, Assembly of Greek marble inscriptions by isotopic methods, Science, 199 (4333), 10701072. Herz, N., 1990, Stable Isotope Analysis of Greek and Roman Marble: Provenance, Association and Authenticity, In Marble, Art Historical and Scientific Perspectives on Ancient Sculpture (M. True and J. Podany, eds), Malibu: The J. Paul Getty Museum Publications, 101-112. Maniatis, Y. and Polikreti,, K.,, 2000, The Characterisation and Discrimination of Parian Marble in the Aegean Region, In Proceedings of 1st International Conference on the Archaeology onn “Paria Lithos” (D. ( Skillardi and D. Katsonopoulou, eds.),, Parikia, Paros, 2-5 October 1997, The Paros and Cycladic Institute of Archaeology, Athens, 575-584.

In more detail, a fresh marble object shows an intense TL signal throughout its volume (Fig. 4). After a certain time of exposure to sunlight the TL of the outermost surface is minimized (bleached). The bleaching continues deeper into the surface layers after prolonged exposure and a characteristic profile is created depending on the exposure time (Polikreti et al. 2002, Figure 5). It is obvious that the ancient surfaces can be very easily discriminated from the modern ones. If the object is buried and remains in situ for hundreds or thousands of years, the TL of the surface is restored again due to the radioactive nuclides in the ground. The TL intensity is then proportional to the “age of burial” as in the case of pottery (type B, Fig. 4). The method has been used to date an excavated architectural marble fragment from Vergina, Macedonia, Greece (Polikreti et al. 2003). The fragment was dated by archaeologists to the 2nd century


K. Polikreti Margolis, S. V., 1989, Authenticating Ancient Marble Sculpture, Scientific American, 260 (6), 104-110. Matthews, K.J., 1997, The establishment of a data base of neutron activation analyses of white marble, Archaeometry, 39 (2), 321-332. Moens, L., De Paepe, P. and Waelkens, M., 1992, Multidisciplinary Research and Cooperation: Keys to a Successful Provenance, Determination of White Marble, In Ancient Stones: Quarrying, Trade and Provenance: Interdisciplinary Studies on Stones and Stone Technology in Europe and Near East from the Prehistoric to the Christian Period (M. Waelkens, N. Herz, and L. Moens, eds.), Leuven: Leuven University Press, 247252. Polikreti, K. and Maniatis, Y., 2002, A new methodology for marble provenance based on EPR spectroscopy. Archaeometry, 44 (1), 1-21. Polikreti, K. and Maniatis, Y., 2004, Distribution changes of Mn2+ and Fe3+ on weathered marble surfaces measured by EPR spectroscopy, Atmospheric Environment, 38, 3617-3624. Polikreti, K., Michael, C. T. and Maniatis, Y., 2002, Authenticating marble sculpture with thermoluminescence, Ancient TL, 20 (1), 11-18. Polikreti, K., Michael, C. T. and Maniatis, Y., 2003, Thermoluminescence characteristics of marble and dating of freshly excavated marble objects, Radiation Measurements, 37, 87-94. Pollini, J., Herz. N., Polikreti, K. and Maniatis, Y., 1998, Parian lychnites and the Prima Porta statue: new scientific tests and the symbolic value of the marble, Journal of Roman Archaeology, 11, 275-284. Riederer, J., 1977, Forgeries of Cycladic marble objects, In Art and Culture in the Cyclades: Handbook of an Ancient Civilisation (J. Thimme, ed.), Karlsruhe, 92-93. Rorimer, J.J., 1931, Ultraviolet rays and their use in the

examination of works of art, Boston: Metropolitan Museum of Fine Arts. Schmid, J., Ambühl, M., Decrouez, D., Müller, S. and Pamseyer, K., 1999, A quantitative fabric analysis approach to the discrimination of white marbles, Archaeometry, 41 (2), 239252. Spier, J., 1990, Blinded with science: the abuse of science in the detection of false antiquities, The Burlington Magazine, CXXXII (1050), 623-631. Stipp, S.L.S., Konnerup-Madsen, J., Franzereb, K., Kulik, A. and Mathieu, H.J., 1998, Spontaneous movement of ions through calcite at standard temperature and pressure, Nature, 396, 356–359. Summary of Scientific Research on the Getty Kouros: Results of the Getty Kouros project (directed by Preusser F. and Podany J.), 1992, Unpublished paper handed to the participants of “The Getty Kouros colloquium”, Athens 25-27 May 1992. Theocaris, P.S., Liritzis, I. and Galloway, R.B., 1997, Dating of two Hellenic pyramids by a novel application of thermoluminescence, Journal of Archaeological Science, 24, 399-405. Ulens, K., Moens, L. and Dams, R., 1995, Analytical methods in authenticating ancient marble sculptures, In 3rd International Symposium of ASMOSIA (Y. Maniatis, N. Herz and Y. Bassiakos, eds.), Athens, 17-19 May 1993. Archetype, London, 199-205. Ulens, K., Moens, L., Dams, R. and De Paepe, P., 1994, Study of the patina of ancient marble sculptures by stable isotope analysis, Science of the Total Environment, 158, 63-69. Von Bothmer, D., 1964, The head of an Archaic Greek Kouros, Archaölogischer Anzeiger, 615-627. Young, W. J. and Ashmole, B., 1968, The Boston relief and the Ludovici Throne, Boston Museum of Fine Arts Bulletin, 66, 124-166.


ACCELERATOR MASS SPECTROMETRY DATING OF EPICURUS “DE NATURA” PAPYRUS FROM HERCULANEUM C. Lubritto*, F. Terrasi, A. D’onofrio, C. Sabbarese, F. Marzaioli, I. Passariello, A. Palmieri, G.Casa, D. Rogalla+, M. Rubino Dipartimento di Scienze Ambientali, Seconda Università di Napoli – Via Vivaldi, 43 – 81100 Caserta - Italy

G. Imbriani, M. Romano, L. Gialanella, V. Roca Dipartimento di Scienze Fisiche and INFN Sezione di Napoli – 80125 Napoli -Italy

C. Rolfs Institut für Physik mit Ionenstrahlen, Ruhr-Universität Bochum – Bochum - Germany

M. Giancaspro, A. Travaglione Biblioteca Nazionale di Napoli – 80100 Napoli - Italy Abstract: The age of a fragment of the Epicurus “De Natura” papyrus has been measured, for the first time, by means of the 14C AMS technique, a non- or micro-destructive analytical technique. A radiocarbon age of 2275±80 yr BP was obtained. This result is coherent with palaeographic studies which had given a date between the end of the 3d and the beginning of the 2d century B.C. Περιληψη: Προσδιορίσθηκε η ηλικία ενός δείγματος παπύρου, του Επίκουρου “De Natura” για πρώτη φορά, με χρήση της μη- έως ελάχιστα-καταστροφικής τεχνικής του Φασματογράφου Επιταχυντή Μάζας (AMS) της μεθόδου του ραδιοάνθρακα 14 C. Η ηλικία που προσδιορίσθηκε είναι 2275±80 χρόνια BP και είναι σε συμφωνία με παλαιογραφικές μελέτες οι οποίες έδιναν μια ηλικία μεταξύ του τέλους του 3ου και στις αρχές του 2ου αιώνα π.Χ.


technique (Fifield 2000; Terrasi et al., 1990; Campajola et al. 1987), which consists of the extraction and counting of the 14C content of a microsample, which in turn gives information on the time when the “papyrus plant” was cut in order to produce the “sheet”. The sample, prepared at the Environmental Science Department of the Second University of Naples, in Caserta (I), was then successfully measured at the Dynamitron Tandem Laboratory (DTL) in Bochum (D). A brief description of the sample preparation procedure will be given, together with a brief illustration of the experimental set-up. The measurement results and data analysis will be described. The comparison between AMS results with palaeographic studies will then be discussed.

The catastrophic Vesuvius eruption of 79 AD “sealed” the city of Herculaneum. After 17 centuries the excavations promoted by the Bourbon kings of Naples started to unveil the social and cultural life of those times. Priceless information on the cultural production has been supplied by the papyrus collection found in the so called “Villa dei Papiri”, which has not yet been completely excavated, due to technical difficulties. The Herculaneum papyruses, now housed in the Naples National Library, represent the oldest collection of manuscripts in the world. Since the middle of the 18th century to nowadays several de-unrolling methods have been employed that allowed to obtain more than 1880 papyrus fragments. In such a way it has been possible to recover compositions of Filodemus from Gadara, of Metrodorus and Calote from Lampsacus and not known fragments of the Epicurus “De Natura”.

Sampling procedure and sample preparation A few crumbs of material (around 8 mg in weight) were collected from the bottom of the glass box preserving the precious item and treated in the mass spectrometry laboratory of the Environmental Science Department of the Second University of Naples, in Caserta (I) (Fig.1).

So far a dating of these items based on chemicalphysical analytical methods was never attempted, due to the destructive characteristics of “classical” analytical methods. The development of non- or micro-destructive analytical techniques, like, e.g., ultrasensitive Accelerator Mass Spectrometry (AMS) has changed the situation drastically.

The main steps in the sample preparation are (Lubritto et al. 2004): •

In this way, for the first time, the age of the “De Natura” papyrus has been measured by means of the 14C AMS *

corresponding author; e-mail: [email protected]


Marie Curie fellow, contract N. HPMD-CT-2001-00088

• 37

Pretreatment: it consists of the removal of extraneous material and chemical procedures to extract carbon of organic origin from the sample. The method and the intensity of the pretreatment depend on the type, quality, quantity, likely age and surrounding environment of the sample. Combustion: pretreated organic samples are converted

C. Lubritto et al.

Figure 1 - An Herculaneum papyrus conserved in the Naples National Library (left). Treatment of a 4 mg sample of the “De Natura” papyrus (right).

to CO2 by combustion at 700°C, burning few milligrams of sample together with silver wire and CuO in the quartz combustion. Graphitization: in order to obtain better accelerator ion source performances, graphite is formed via hydrogen reduction over an iron catalyst (Vogel et al. 1984) at 900°C, starting from CO2 produced in the combustion stage.

procedure consisted in the standard AAA (Acid-AlkaliAcid) treatment. Then the samples were transformed to CO2 in the combustion line and finally “graphitized” for the use in the sputter source of the accelerator. From an aliquot of CO2 the isotopic ratio 13C/12C was measured and used for the fractionation correction of the 14C/12C isotopic ratio (Fifield 2000). Results of ams measurements and discussion

In the present case practically no pretreatment was needed, due to the homogeneity of the samples. The chemical

Accelerator mass spectrometry (AMS) is, at present, a

Figure 2 – Block diagram of the Bochum AMS system used in the present work (from ref. Lubritto, 2004.)


Accelerator Mass Spectrometry Dating of Epicurus “De Natura” papyrus from Herculaneum Conclusion

very powerful method to measure isotopic ratios of rare radioisotopes; so that it can encompass problems such as long half-lives and/or very low concentrations and small sample availability.

In this work it has been shown that the AMS analytical technique can be used to gain information on the age of extremely “delicate” items like the Epicurus “De Natura” papyrus.

The AMS method uses a particle accelerator (Fig. 2) to achieve energies high enough (of the order of several MeV) to remove or detect contaminating isobars and isotopes (Lubritto, 2004). This method allows rapid sample throughput and smaller sample sizes compared to the conventional method of decay counting (about 1 mg of sample for 14C, which is 10000 times smaller than what is needed for decay counting with the same accuracy).

In such a way it is possible to complement information from palaeographic studies and burst the interdisciplinary approach to the field of archaeological investigations. The interval corresponding to the largest probability (320-210 yr BC) is in good agreement with palaeographic studies due to Prof. G. Cavallo (Giancaspro, personal communication), which had given a date between the end of the 3d and the beginning of the 2d century B.C.

Two samples, treated at Caserta as described above (4 mg each), were measured at Bochum DTL. Repeated measurements of the two samples gave consistent results. The 14C/12C isotopic ratio was determined relative to an ANU (Australian National University) sucrose secondary standard, yielding an average value of 75.32 ± 0.75 PMC (percent modern Carbon), including background and fractionation corrections. The latter were determined using the measured stable isotope ratio δ13C=-16.3±0.1‰. The radiocarbon age turned out to be 2275± 80 yr BP. This value was converted to calibrated age using the Stuiver and Reimer CALIB code (Stuiver and Reimer, 1993). The resulting calibrated age consists, at level of one sigma, in two intervals: 1. 2.

450-350 yr BC 320-210 yr BC

References Campajola, L., et al., 1987. The Italian AMS program: Measurements with the Naples TTT-3 tandem accelerator, Nucl. Instr. And Meth. In Phys. Res B, 29, pp. 129-132. Fifield, L.K., 2000. Advances in accelerator mass spectrometry, Nucl. Instr. And Meth. In Phys. Res. B, 172, 134-143. Lubritto, C., et al., 2004, Accelerator mass spectrometry at the 4 MV Dynamitron Tandem in Bochum, Nucl. Instr. and Meth. in Phys. Res. B, 222, pp. 255-260. Stuiver, M. and Reimer, P.J., 1993, Extended 14C Data Base and Revised CALIB 3.0 14C Age Calibration Program, Radiocarbon, 35, 215-230. Terrasi, F., et al., 1990, AMS at the TTT-3 tandem accelerator in Naples, Nucl. Instr. and Meth. in Phys. Res. B, 52, 259-262. Vogel, J.S., et al., 1984, Performance of catalytically condensed carbon for use in accelerator mass spectrometry Nucl. Instr. And Meth. B, 5, 289-293.

(35% probability) (65% probability)




Laboratory of Archaeometry, Institute of Materials Science, N.C.S.R. Demokritos, 15 310 Ag. Paraskevi, Attiki, Greece [email protected]

Department of Materials Science and Engineering, School of Chemical Engineering, N.T.U.A., Zografou Campus, Athens, Greece

Abstract: A single aliquot polymineral protocol for the estimation of the natural dose in Thermoluminescence (TL) dating is presented that makes use of polymineral fine grains extracted from ceramic materials. This protocol is demonstrated using aliquots made of two Neolithic sherds and is compared against an additive dose polymineral technique. The results of both techniques are in close agreement and highlight the potential advantages of the new procedure when especially sample availability is restricted, e.g. dating of small pottery fragments, or in cases of authenticity testing. Περίληψη: Περιγράφεται ένα πειραματικό χρονολογικό πρωτόκολλο κεραμικών υλικών, για τον προσδιορισμό της φυσικής δόσης κατά την χρονολόγηση με την μέθοδο της Θερμοφωταύγειας (ΘΦ, TL). Το πρωτόκολλο αυτό εμπίπτει στις τεχνικές της αναγεννώμενης δόσης του ενός δείγματος και κάνει χρήση του πολυορυκτολογικού κλάσματος των λεπτών κόκκων του υλικού για την παρασκευή των δειγμάτων. Παρουσιάζονται τα αποτελέσματα εφαρμογής του σε δυο νεολιθικά κεραμικά και συγκρίνονται με τα αποτελέσματα ενός πρωτοκόλλου προσθετικής τεχνικής στα ίδια κεραμικά. Η συμφωνία στα αποτελέσματα των δυο τεχνικών καταδεικνύει την εφαρμοσιμότητα του νέου πρωτοκόλλου και αναδεικνύει την χρησιμότητά της ειδικά σε περιπτώσεις όπου η διατιθέμενη ποσότητα του υλικού είναι περιορισμένη (π.χ. έλεγχοι αυθεντικότητας).


Tl measurements

Single aliquot procedures, for the estimation of natural doses in luminescence dating techniques are of increased research interest during recent years (Wintle, 1997), since unlike to the multiple aliquot ones, a set of aliquots usually results to an equal number of dose estimations.

Two sherds of Neolithic pottery excavated at Mandalo, Macedonia, Greece, were selected, labelled with the code numbers 18003b and 7251bb. An external layer of ca. 0.3 cm was removed from the sherds with the use of a hand drill and then they were gently crushed in a vice. The coarse product was left to settle in an ethanol column following the procedure described at Michael et al., 1997, in order to receive the 2-8 μm m fine fractions. For sample preparation and the measurements the procedure of the foil technique was followed since it reduces significantly spurious TL contributions, as the measurements can be made in vacuum, due to a good heating contact between sample and heater plate (Michael et al., 1999).

A first step of a TL single aliquot technique, that is, measuring of the natural luminescence signal intensity (1st glow) and comparing it with the 2nd glow signal of the same aliquot, after exposure to a beta or gamma reference dose, usually provides only an approximate value of the natural dose (paleodose) P. That is because of sensitivity changes, caused by the pre-dose effect, due to the natural dose and the subsequent heating. The new approach for the estimation of the natural dose provides the function, which describes the relation between the signal intensity of successive irradiation and measurement-cycles and the measurement cycle number of the same aliquot and thus finally correcting for sensitivity changes.

TL glow-curves were recorded with the use of a Littlemore TL 711 set, equipped with an EMI 9635QA photomultiplier and a Corning 7-59 blue filter. Measurements run at a heating rate of 10oC s-1 and the irradiations administered with a calibrated 90Sr beta-source providing 0.53 Gy. min-1.

The new single aliquot protocol is tested against the fine grain additive dose polymineral foil technique (Michael et. al., 1997) by applying both procedures to two sherds of archaeological pottery. The provided natural dose estimations are in close agreement and within the calculated errors. From the presented study it became evident that the new protocol sounds as very useful for TL dating and authenticity testing of fired archaeological materials.

The TL suitability of the samples was checked using the plateau test by performing initial test measurements and also for possible anomalous fading effects following three months storage of the samples.


C.T. Michael, N. Zacharias, I. Lefakis and D. Dimotikali

Figure 1.TL glow-curves of repeated irradiation and measurement cycles (D=15 Gy) for sample No 18003b. It can be seen that the level of the TL intensity is increasing with increasing measurement cycle number reaching to a level of saturation.

Single aliquot Using one aliquot the natural TL signal is recorded and by comparing it with the TL signal intensity of the same aliquot after giving a laboratory dose, estimation (first approximation) of the natural dose D is resulted. Then a second aliquot is used to perform the repeated irradiation and measurement cycles, by providing the same laboratory dose equal to D after the measurement of its natural TL (Figure 1). Usually, a regeneration cycle of 6 - 7 measurements was found as adequate for the natural dose estimation. In Figure 1, it can be seen that the level of the TL intensity is increasing in every cycle according to the expected change of sensitivity caused by the irradiation and the subsequent heating applied (Aitken, 1985; Tso and Li, 1994). This increasing goes on asymptotically reaching to a level of saturation for all the samples measured as yet (Figures 2.a. and 3.a.).

Figure 2. Sample No18003b. a. Glow-curve intensities at 350 oC for the single aliquot protocol (D=15 Gy).

(2) The relation that describes the TL intensity T in terms of the measurement cycle number X is given by the equation 1, where the parameters of A, B, C are provided using a computing program.

where NTL is the level of the natural TL. Equation 2 results from the obvious proportionality in this case between TL and irradiation dose if the sample does not exhibit supralinearity, as our samples. The case of samples exhibiting supralinearity will be explicitly discussed in a forthcoming paper.

(1) It is evident that for X=0 we have T=A=DTL where DTL is the level of TL for the laboratory dose D in case the sample could retain the initial sensitivity, i.e. without pre-dose effect. Thus the natural dose P is provided by the following equation 2,

Foil technique The additive dose foil technique uses a number of aliquots for the estimation of the natural TL intensity and another 42

A Single Aliquot Polymineral Tl Protocol Applied For Fired Materials set of aliquots for monitoring the natural plus an additional laboratory dose, usually reaching to a level of ca. 3 times the natural TL intensity. The normalised growth curve in the FT, for the estimation of the P value is then produced as follows: each point of the curve (rL) is given by the ratio of the natural TL intensity, or the natural TL plus the laboratory dose (DL), to the TL intensity of the 2nd glow of the same aliquot after irradiation with a reference dose (ρ) for inter-aliquot normalisation. By inserting Equation 3 in a software program and the paired values rL, DL and also the value of ρ the values of P and k are calculated were k is a specific sensitivity enhancement factor that accounts for the pre-dose effect. Figure 2. Sample No18003b. b. The normalised growth curve in the foil technique for palaeodose estimation P.


Results of the p value estimation For the sample No 18003b the single aliquot protocol provided: 16.29±0.30 Gy, and 17.17±1.18 Gy for the foil technique. For the sample No 7251bb the calculated values were: 48.1±4.9 Gy, and 50.6±1.3 Gy for the two techniques respectively. Conlcusions Although the accuracy to be achieved with the proposed new protocol needs more experimental testing, nevertheless, the applicability of the protocol sounds promising in cases of sample deficiency such as dating of small sherds or valuable objects which demands quasi-destructive analysis.

Figure 3. Sample No7251b. a. Glow-curve intensities at 350 oC for the single aliquot protocol (D=24.6 Gy).

Furthermore, the protocol can be used in a routine base to provide estimations of received natural doses also for coarse-grain aliquots (e.g. quartz samples) by following the same approach for fitting of the experimental to numerical values. References Aitken, M. J., 1985. Thermoluminescence dating. Academic Press. Michael, C. T., Zacharias, N., Dimotikali, D., Maniatis, Y., 1997. A new technique (foil technique) for measuring the natural dose in TL dating and its application in the dating of a mortar containing ceramic fragments. Ancient TL Vol.15, No.2-3, 36-42. Michael, C. T., Zacharias, N., Polikreti, K., Pagonis, V., 1999. Minimising the spurious TL of recently fired ceramics using the foil technique. Radiation Protection Dosimetry, Vol.84, Nos.1-4, 499-502. Tso, M-N., W., Li, S-H., 1994. Equivalent dose estimation for pottery by single disc regeneration method. Radiation Measurements, 23, Nos. 2/3, 451-454. Wintle, A.G., 1997. Luminescence dating: Laboratory procedures and protocols. Radiation Measurements, 27, 769-818.

Figure 3. Sample No7251b. b. The normalised growth curve in the foil technique for palaeodose estimation P.


COMPARATIVE STUDIES ON THE DETERMINATION OF ENVIRONMENTAL RADIATION DOSES N.Zacharias,1 M. Štuhec,2 V. Kilikoglou,1 Y. Bassiakos,1 C.T. Michael,1 E. Vardala-Theodorou,3 G. Theodorou,4 D. Dimotikali5 1

Laboratory of Archaeometry, Institute of Materials Science, N.C.S.R. Demokritos, 15 310 Ag. Paraskevi, Attiki, Greece][email protected] 2

Jožef Stefan Institute, Jamova 39, 1000 Ljubljana, Slovenia

G.N.H.M.- GAIA Centre, Othonos 100,14562 Kifissia, Greece Department of History, Geology and Paleontology, National and Kapodistrian University of Athens, 3


15784 Zografou Campus, Greece 5

Department of Materials Science and Engineering, School of Chemical Engineering, N.T.U.A., 15780 Zografou Campus, Athens, Greece

Abstract: The aim of the study is to present environmental dosimetry data performed at the Peninsula of Perachora near Loutraki/ Korinth, Greece. The measurements employed Thermoluminescence dosimetry, in situ spectrometry and also dosimetry estimations resulted from chemical analysis data on associated sediment samples. Six sampling sites in the studied area were selected based on the variety of dominant geological formations (from highly calcareous to alumino-silicate marine sediments) and the diversity of fossil shells and other remains of invertebrates that they include. The resulting values are presented and the potential implications on dosimetry and radiometric dating studies are discussed. Περιληψη: Η παρούσα εργασία εστιάζει στον ακριβή προσδιορισμό της περιβαλλοντικής δόσης ραδιενέργειας, η οποία πραγματοποιήθηκε στην περιοχή της Περαχώρας Λουτρακίου, Κορινθίας. Ο προσδιορισμός της περιβαλλοντικής δόσης (γάμμα- και κοσμικής-ακτινοβολίας) συνεισφέρει σημαντικά στις απόλυτες χρονολογικές εργασίες με τεχνικές Φωταύγειας και Ηλεκτρονικού Συντονισμού Στροφορμής. Παρουσιάζονται τα αποτελέσματα της συνδυασμένης χρήσης των τεχνικών του φορητού σπινθηρομέτρου, της ανάλυσης με Νετρονική Ενεργοποίηση και της χρήσης δοσιμέτρων Φωταύγειας. Παράλληλα συζητείται η συνεισφορά στην ακρίβεια των τεχνικών και στο συνολικά αποδιδόμενο σφάλμα των εξαγόμενων ηλικιών.


displacement of some of the fossiliferous sections, especially for those at heights of 100m. Therefore, the absolute dating techniques of Thermoluminescence (TL) and Electron Spin Resonance (ESR) were applied on mollusk species that are characteristic indicators for Tyrrhenian deposits.

During the last years, palaeoenvironmental and dating studies on fossil mollusc shells have been carried on in the area of Perachora, Loutraki at Korinth, Greece. The area (Fig. 1) is well known for its tectonic activity and the very rich fossilferous marine Quaternary sediments. These sediments compose of marine terraces deposited on the alpine substrate. At present, the terraces occur from sea level to the height of more than 100m (Stiros, et al., 1998; Vardala-Theodorou, 1999), with the occurrence of these resulting from combined eustatic and tectonic activity. Eustatic activity alone cannot explain deposition at heights of more than 30 m during the warm transgression phases of the Quaternary (Tyrrhenian, Eutyrrhenian, or Neotyrrhenian) which are well documented by the occurrence of the molluscs Natica lactea, Spondylus gaederopus and corals Cladocora caespitosa. Deposition of Eutyrrhenian sea terraces has been documented also by Morewood and Roberts (1997) by providing a date of 126 ka ago for sediments of the north coast of Perachora.

Within the application of both techniques and for the estimation of the total dose rate values, there is a need for accurate determination of the environmental dose since this makes a contribution to the final age results of ca.1525% (Galloway, 2001; Zacharias et al., 2005)). The accurate and precise estimation of the environmental radiation is usually constrained from a number of parameters, like possible inhomogeneous distribution of radioactive nuclides on both a macroscopic and microscopic level (Liritzis, 2001; Kalchgruber et al., 2003), variations of water context and disequilibria in the U and Th decay chains (Aitken, 1985; Olley et al., 1997). Six sites within the studied area were selected for the present work to result the estimation of the environmental doses. The selection of the sites was decided by the presence of those fossils on which TL/ESR dating was in

The use of biostratigraphical data alone is not adequate to explain sufficiently the geological and palaeogeographical evolution of the area and the strong tectonic vertical 45

N. Zacharias at al.

Figure 1. A map of Greece (left) where Athens (A) and Loutraki (L) are pointed; on the right the studied area of Perachora (pointed by arrow) with the code names of the sampling points.

progress. Additionally, the studied sites are representative of the macro-morphological and elevation variations of the Peninsula (Fig. 2).

(Klemiz et al., 1999). Overall uncertainty of measured doses was 5%. Portable scintillator

Dosimetry techniques Field measurements were practised with a portable Tldoped NaI scintillator unit (SCINTREX SPP2-NF, 50cm2 in diameter) to monitor in situ the natural radioactivity of the sampling points. The scinltillator and TLD phosphors were placed at the exact point were the fossil shells to be dated, had been previously removed. The measurement time for the scintillator in order to reach stability ranged from 5- 8 minutes in average.

TL dosimetry (TLD400) TL dosimetry was employed since it potentially provides accurate estimation of the environmental dose independent from seasonal variations and inhomogeneous radiation field effects. The phosphors used were of the TLD4300 type (CaF2: Mn). This type of ceramic phosphor produced at J. Stefan Institute, Ljubljana, is highly sensitive and has negligible fading. Doses measured with the home built measuring system IJS-MR200 proved to have good repeatability (Breznik et al., 2001; Jezeršek et al., 2003). A pre-heat at 100 oC for 20 minutes was used at a heating rate of 10.5 C s-1. The maximum temperature reached was 300 oC where glow peak is located concerning calibration conditions. Since TLD400s were transported from J. Stefan to Demokritos and then again to J. Stefan to be measured, 4 pills were only posted between the two Laboratories without burial and another 4 pills were stayed at Demokritos laboratory for a certain period of time to receive environmental dose. In this way, both the travel dose and doses due to a daily to the mail procedure were accurately estimated.

Neutron activation analysis (NAA) The application of chemical analysis and specifically NAA to provide dose rate values is not commonly used since the method does not provide any indication about possible dissequilibrium effects in the U and Th radioactive series plus the time-consuming and relatively expensive application of the techniques. Instead, low-energy gamma spectrometry, or thick-source a-counting is usually employed for environmental and radiometric dating studies (Aitken, 1985; Bøtter-Jensen et al., 2003). NAA was initiated in the present study under the assumption of radioactive equilibrium state for the sediments formations based on the age estimation (Tyrrhenian - Neotyrrhenian formation; Morewood and Roberts, 1997) and also to provide the chemical fingerprint of the material.

The dosimeters consisted of two tablets in 1.5 mm copper alpha/beta shielding and black plastic water resistant envelope. The tablets have individual calibration factor, which was checked before and after irradiation on experimental sites. Response of the dosimeters to environmental conditions was checked independently by international intercomparison for environmental dosimetry

Sample preparation for NAA. In order to reduce natural variations in the chemistry of the soil samples and a more representative estimation of the gamma-radiation field to result (an area of half sphere and within a 50 cm radius, centred at the sampling points), 46

Comparative Studies on the Determination of Environmental Radiation Doses

Figure 2. The sampling points of the study. The calcite fossilferous nature for most of the sampling points is dominant. F105 is mainly a coral rock.

two samples from every sampling point were prepared as follows:

estimate the cosmic dose rate of the sites, the geographical coordinates and altitude of the sites were recorded with a portable GPS unit. Certainly, the current average sea-level of the sites is not representative of the geological period covered by the formations due to the tectonism of the area. Also, although surface-picking up to a depth of ca. 20cm were applied for the samples, it is assumed that the sampling points should not always have been remained to the surface. Therefore, the cosmic rate values employed for the calculations of the total environmental dose were estimated by assuming half values for the present altitude and then by introducing these to the approach given by Prescott and Hutton (1994).

An average of ca. 100 g of each sampled material was homogenized using grinded in a mortar resulting powder samples of grain size less than 50μm. Additionally, rock fragments from each site were powdered forming separate lumps of material. Finally, two aliquots were prepared from each sample/site, to be measured with NAA: the one as having 100% material from the first grinding procedure and another by mixing 80% and 20% respectively from the first and second preparation process Table 1). NAA at the NCSR Demokritos requires about 130 mg of the powdered and dried sample. The samples are irradiated for 45 min in the Demokritos’ reactor, at a thermal neutron flux of ca. 3 x 1013 n cm-2 s-1, including two SOIL-7 standard samples. Each sample is measured twice, 7 and ca. 20 d after the irradiation. A Ge γ-detector covering the energy range 80-1600 keV is used for the measurements. NAA results provided the concentrations of the entire set of 27 elements (Hein et al., 2002) of the measured soil samples (Table 1).

Results – discussion The chemical analysis of the samples verified the highly calcite nature of the formations which is responsible for the low environmental doses reported in the study. In Table 3 the combined results given by the three approaches are given; it can be noticed the high agreement between the TLD and NaI data, although the later is given within large errors (Fig. 3). Although NAA acquired radiation doses fall within experimental errors with the dosimetry values a large variation should be highlighted for the NAA values.

By employing the conversion factors of Adamiec and Aitken (1998), the U, Th concentrations of the samples provide the gamma-radiation dose rates (Table 2). To 47


Ba 96.9 94.6 128 105 111 129 87.7 121 119 92.3 125 134

Sm 1.46 1.9 1.1 1.17 1.12 1.26 0.771 1.04 0.625 0.694 2.43 3.72

Nd 6.26 11.8 4.06 6.78 5.13 9.34 6.88 5.58 4.09 4.31 15.9 18.7

Lu 0.116 0.127 0.129 0.092 0.071 0.097 0.091 0.09 0.095 0.091 0.145 0.186

Zr 56.5 68.2 81.8 99.5 48.3 55.8 49.5 57.9 57.4 55.3 82.3 71.2

U 1.77 1.93 6.07 3.62 2.57 2.66 4.13 3.77 4.82 4.51 2.62 2.31 Cs 1.32 1.74 0.632 0.804 0.562 0.74 0.838 1.02 1.17 1.48 1.9 2.77

Yb 0.865 0.902 0.467 0.671 0.517 0.561 0.486 0.545 0.289 0.323 1 1.37 Ni 142 205 83.1 128 87.7 104 80 113 24.1 24 37.1 53.7

As 5 5.89 3.08 3.3 2.14 2.5 6.77 7.19 8.22 15.4 6.65 7.63 Tb 0.183 0.252 0.127 0.159 0.123 0.173 0.113 0.108 0.081 0.068 0.316 0.453

Sb 0.352 0.404 0.204 0.193 0.117 0.163 0.281 0.369 0.278 0.455 0.363 0.483 Sc 3.6 4.63 2.27 3.19 2.29 2.63 2.1 2.67 1.32 1.53 3.32 4.57

Ca (%) 27.7 23.2 32.1 29.4 32.9 31.4 32.5 29.5 37.9 34.1 33.4 27.9 Rb 17.1 24.6 9.54 13.4 8.73 8.63 8.31 13.1 9.41 9.65 27.3 38.1

Fe (%) 1.11 1.47 0.598 0.871 0.576 0.663 0.596 0.825 0.432 0.596 0.939 1.35

Na (%) 0.23 0.274 0.245 0.244 0.269 0.271 0.184 0.217 0.524 0.61 0.209 0.251 Zn 23.9 35.6 14.6 19.4 16.2 18.4 18.9 28 11.8 13.9 29.5 40.2

K (%) 0.302 0.528 0.132 0.236 0.184 0.157 0.229 0.271 0.154 0.145 0.414 0.629 Ta 0.205 0.298 0.107 0.155 0.115 0.132 0.102 0.158 0.055 0.085 0.263 0.321

La 7.31 10.2 4.73 5.58 5.23 5.82 3.87 5.68 3.06 3.64 12.1 19.1 Co 10.3 13.3 5.58 7.69 5.44 6.28 7.3 9.66 4.17 5.41 7.64 9.95

Ce 15.8 22 11.4 12.6 12.3 13.6 8.73 13.3 7.17 8.28 21.8 31.6

Cr 369 694 380 462 554 599 316 953 23.9 28.5 105 91.1 Eu 0.332 0.416 0.231 0.272 0.221 0.283 0.177 0.228 0.12 0.138 0.489 0.704

Th 1.91 2.93 1.15 1.58 1.55 1.4 0.975 1.6 0.837 1.01 2.62 3.81

Table 1. NAA results of the analysed samples. Unless indicated otherwise, concentrations are given in ppm (=μg/g). The sample indication –A stands for 100% soil samples while –X for 80% soil plus 20% rock powder.

Hf 1.65 1.78 1.11 1.62 1.01 0.992 0.556 1.02 0.223 0.315 1.27 1.65

F35_A F35_X F105_A F105_X F110_A F110_X F160_A F160_X F501A_A F501A_X F501B_A F501B_X

N. Zacharias at al.

Comparative Studies on the Determination of Environmental Radiation Doses


Gamma dose




altitude (m)

average environ. dose

F35_A 35_A _A F35_X 35_X _X F105_A _A F105_X 105_X _X F110_A 110_A _A F110_X 110_X _X F160_A _A A F160_X _X X F501A_A _A A F501A_X _X X F501B_A 1B_A _A A F501B_X _X X

320 0


230 0







12 2







360 0

10 0













500 00




230 0








230 0






Table 2. Gamma- cosmic- and average environmental-doses for the measured samples estimated using the NAA data. Given values are in μGy/a. Gy/a.

The use of chemical analysis is valuable when applying to provide the chemical profiling and thus to assist at the knowledge on the nature and development of the area. Noteworthy, chemical analysis does not indicate for possible secular disequilibrium effects. Since the achievement of accurately estimated radiation doses is essential in both dosimetry and absolute dating studies, the use of calibrated TLD phosphors serves as the choice with the most sensitive, inexpensive and precise results.





F35 F105 F110 F160 F501A F501B
















697±25 7±25 ±25



Table 3. Final environmental dose data and their errors; values in μGy/a. Gy/a.

Acknowledgments The authors would like to thank P. Bakandrea for assistance during sample preparation for NAA measurements at NCSR Demokritos and E. Stathopoulou during a field trip. The study forms part of a bilateral Slovenia - Greece research project for the period 2002-2004, funded by EU. References Aitken, M., 1985, Thermoluminecence dating, Academic Press, London. Adamiec, G., and Aitken, M.J., 1998, Dose-rate conversion factors: update, Ancient TL, 16, 37-50. Bøtter-Jensen, L., McKeever, S.W.S., Wintle, A., Optically Stimulated Luminescence Dosimetry, Elsevier, Amsterdam, 2003. Breznik B, Janzekovic H, Stuhec M, Zdesar U., 2001, Development of a quality assurance programme for external personal dosimetry in Slovenia, Radiation Protection Dosimetry, 96 (1-3), 49-52, 2001. Galloway, R.B., 2001, Luminescence dating: Limitations to accuracy attainable, Journal of Radioanalytical and Nuclear Chemistry, 247,3, 679-683. Hein, A., Tsolakidou, A., Iliopoulos, I., Mommsen, H., Buxeda i Garrigós, J., Montana, G., Kilikoglou, V., 2002. Standardisation of elemental analytical techniques applied to provenance studies of archaeological ceramics: an inter laboratory calibration study. Analyst 127, 542-553. Jezeršek D., Miheli, M., Rupnik Z., Štuhec M., Zorko B., (2003), The new IJS MR 200 TLD measuring system in Proceedings of the 5th Symposium of the Croatian Radiation Protection

Figure 3. Diagram where the environmental doses estimated using TLD400 phoshpors, NaI scintillator and dose rates resulted from NAA analysis are plotted. Triangle marks indicate TLD400 vs. NaI and square marks TLD400 vs. NAA; values in μGy/a. Association, (Eds. I. Krajcar Broni, S. Miljani, B. Obeli (HDZZ-CRPA, Zagreb) p. 121. Kalchgruber, R., Fuchs, M., Murray, A.S., and Wagner, G.A. (2003). Evaluating dose-rate distributions in natural sediments using α-Al2O3: C grains, Radiation Measurements, 37, 293-297. Klemic G., Shobe J., Sengupta S., Shebell P., Miller K., Carolan P.T., Holeman G., Kahnhauser H., Lamperti P., Soares C., Azziz N., Moscovitch M., (1999). State of the Art of Environmental Dosimetry: 11th International Intercomparison and Proposed


N. Zacharias at al. Performance Tests, Radiation Protection Dosimetry, 85, 201206. Liritzis, I., 2000. Advances in thermo- and opto-luminescence dating of environmental materials (sedimentary deposits) Part I: techniques. Global Nest, 3-27 (Part II: Applications. Global Nest 29-49). Moorewood, N.C., Roberts, G.P., (1997). Geometry, kinematics and rates of deformation in a normal fault segment boundary, Central Greece. Geophysical Research letters 24(23), London, p.3081-3084. Olley, J.M., Roberts, R.G., and Murray, A.S., (1997). Disequilibria in the uranium decay series in sedimentary deposits at Allen’s cave, nullarbor plain, Australia: Implications for dose rate determinations, Radiation Measurements, 27,2 433-443. Prescott, J.R., Hutton, J.T., (1994). Cosmic ray contributions to

dose rates for luminescence and ESR dating: Large depths and long-term time variations, Radiation Measurements, 23, 497-500. Stiros, S.C. and Pirazzoli, P.A., (1998). Lake Quaternary coastal changes in the Gulf of Korinth, Greece. INQUA Shorelines Commission Guidebook, Patras, Greece, p.1-50. Vardala-Theodorou, G.E., (1998). Study of the recent and fossil (Quaternary) bethink Mollusca of Vouliagmeni Lake at Perachora. Ph.D Thesis, Nat. Capodistrian Univ. of Athens (in in Greek), p.1-480. Zacharias, N., Buxeda i Garrigos, J., Mommsen, H., Schwedt, A., Kilikoglou, V., Implications of burial alterations on luminescence dating of archaeological ceramics, Journal of Archaeological Science, Vol.32,1, 49-57, 2005.


ALTERED ARCHAEOLOGICAL POTTERY AND THE EFFECT ON TL DATING N. Zacharias, C.T. Michael Laboratory of Archaeometry, Institute of Materials Science, NCSR Demokritos, 15 310 Aghia Paraskevi, Attiki, Greece

A. Schwedt, H. Mommsen Gruppe Archäometrie, Institut für Strahlen- und Kernphysik, Universität Bonn, Nussallee 14-16, D 53115 Bonn, Germany Abstract: Since the 1970s, analyses of archaeological pottery reported alterations in several alkali metal concentrations. According to the results of recent studies, leaching of potassium from the glassy phase produced in high fired/overfired calcareous pottery has been well established. Within this study, fine-grain thermoluminescence (TL) dating has been applied on a set of calcareous pottery excavated at a Roman site in Mallorca, Spain which was well dated by archaeological criteria. The TL ages resulted as significantly overestimated, thus strengthening the importance of the alteration effects on luminescence dating. Furthermore, the study demonstrates a step-like model to provide an estimation of the unaffected burial period. Περιληψη: Απο την δεκαετία του 1970, έχει τεκμηριωθεί σε αναλυτικές εργασίες αρχαιολογικής κεραμικής ότι η πολύ υψηλά θερμασμένη ασβεστιούχα αρχαιολογική κεραμική εμφανίζει εξαλλοιώσεις και κύρια διαφυγή αλκαλίων από το κεραμικό σώμα. Στην παρούσα εργασία παρουσιάζονται οι επιπλοκές των φαινομένων εξαλλοίωσης στην χρονολόγηση κεραμικής των ρωμαικών χρόνων από την Mallorca.. Παράλληλα, παρουσιάζεται ένα μοντέλο για την χρονολογική εκτίμηση της περιόδου κατά την οποία εμφανίσθηκε/ενεργοποιήθηκε η εξαλλοίωση.


TL measurements performed on sherds not affected by this alteration provided an estimation of the value of the environmental radiation dose, which was then used for the TL dating of the affected sherds. The calculations of the internal dose rates using the K concentrations measured today and without considering K-leaching effects resulted in overestimated TL dates for the affected samples.

In luminescence dating (Thermoluminescence or Optically Stimulated Luminescence), the many parameters incorporated in the estimation of the dose rate, like inhomogeneous distribution of radioactive nuclides (Kalchgruber et al., 2003), variations of water content and disequilibria in the U and Th chains (Aitken, 1985; Olley et al., 1997), allow for resulted dates with an error at approximately 10% (1σ).

Using the archaeological age of the sherds and assuming that the alteration was a rather fast process, it was then possible to provide an estimation of a time span after which the environmental alteration effects occurred.

Early mineralogical studies (Picon, 1976; Olin et al., 1978) on high/over-fired calcareous archaeological pottery samples indicated the significant variations in alkali metal concentrations due to environmental alterations during burial. For especially the K concentration, recent studies (Buxeda et al., 2002) provided evidence for a leaching process that occurred from the glassy phase. Zacharias et al. (2005), examined this effect on Greek Bronze Age pottery and the resulting implications in luminescence dating studies have been reported and discussed therein.

Chemical and mineralogical characterization of the mcf pottery group In a first study (Tsantini et al., 2004), the chemical reference group of Sa Mesquida kiln site was established. The study revealed unexpectedly large spreads in the K, Na and Rb content and the existence of different equivalent firing temperatures (EFT), according to the mineral phases identified by X-Ray Diffraction (XRD). These results point to the existence of an alteration of the glassy phase for those high/over-fired sherds, together with the crystallization of the Na-zeolite analcime (Na[AlSi2O6]·6 H2O), a mineral phase which must have formed sometime after firing because of its high water content.

In the present study, archaeological pottery recovered from the excavations of the Roman kiln site of Sa Mesquida (Mallorca, Spain), called further on as MCF pottery and dated back to the 1st century AD, was used to perform TL measurements. Neutron Activation Analysis (NAA) of the complete pottery group provided the concentrations of bulk and profile measurements (Schwedt, 2004) and showed that a variation in the alkali element concentrations is connected to the post-burial alteration of the glassy phase and to the crystallization of the zeolite analcime (cf. below).

In order to investigate these alteration processes further, seven samples (MCF22, MCF29, MCF33, MCF34, MCF35, MCF38 and MCF59), representing unaffected and affected cases, were selected to study a possible formation of concentration profiles. The samples were 51

N. Zacharias, C.T. Michael, A. Schwedt and H. Mommsen

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

MCF samples 19 slices M σ (%) 4.97 (21.) 357. (15.) 14.2 (10.) 66.5 (1.9) 9.65 (4.7) 70.3 (9.0) 10.6 (4.8) 1.11 (4.3) 3.08 (2.7) 21.6 (7.1) 4.98 (3.8) 1.56 (46.) 35.4 (1.6) 0.38 (4.8) 0.93 (43.) 27.8 (12.) 73.8 (53.) 101. (43.) 0.77 (5.9) 13.4 (1.9) 5.07 (12.) 1.11 (4.2) 0.76 (5.6) 12.4 (2.2) 0.47 (26.) 2.56 (7.9) 2.89 (7.7) 2.74 (3.8) 94.4 (8.2) 99.1 (25.)

Figure 1. Dependency of Na and K concentrations on the extent of analcime crystallization, semiquantified by the intensity of the 5.61 Å peak.

adequately sized were used for TL measurements. In three of them (MCF34, MCF38 and MCF59), XRD showed the presence of analcime, while in the other two (MCF29 and MCF35) this phase was not detected. According to Tsantini et al. (2004), the thermodiffractometric experiments (heating rate 100 °C/h, maximum temperature kept for 1 h) yielded the different EFT for the samples: 850-950°C (MCF35), 950-1000°C (MCF29), 1000-1050°C (MCF38 and MCF59) and 1050-1100°C (MCF34). Thus, obviously, the alteration of the glassy phase, with the leaching of K, and the analcime crystallization, with the enrichment of Na, is related to the high-/over-fired estimated EFT in such calcareous clays. Figure 1 clearly shows that the more analcime has crystallized, the lower the K and the higher the Na concentration are found. Moreover, the measured concentration profiles show that this chemical alteration takes place quite homogeneously across the whole sherd respectively (Fig 2). Therefore, the today measured K concentration does not represent the original value needed for the TL dating.

Table 1. Average concentrations M and spreads (standard deviations) σ of the analysed sample slices. Unless indicated otherwise, concentrations are given in μg/g, spreads in % of M.

cut into about 1mm thick slices, and several samples per sherd were analysed by NAA (Schwedt, 2004). Table 1 shows the mean values of these analyses. As can be seen, most of the measured elements are very homogeneously distributed in all analysed slices pointing to all sherds have being produced from an initially very homogeneous high calcareous paste. It must be taken into consideration that As, Ba and Ca are usually scattered within pottery, and Ni, Ti and Zr can only be measured with high statistical uncertainties in Bonn NAA set up (Mommsen, 2001).

Thermoluminescence and dose rate measurements For the TL measurements an external layer of 0.3-0.4 cm was removed from the five sherds with the use of an electric drill rotating at low speed to avoid tribo-TL effects. The remaining essays were gently crushed in a vice and the coarse product was left to settle in an ethyl alcohol column. The 2-8 μm fine grain fraction was then obtained for the preparation of several aliquots (20-25) from each sample.

Finally, in order to study the implications on TL dating, out of the previous seven samples used in the profile study, five 52

Altered Archaeological Pottery and the Effect on TL Dating

samples MCF29 MCF35

Na (wt %) 0.52±0.003 0.45±0.002

K (wt %) 2.30±0.08 2.40±0.02

Rb [μg g-1] 147±8.3 148±2.5

U [μg g-1] 2.50±0.08 2.28±0.08

Th [μg g-1] 13.00±0.28 12.00±0.28


1.33±0.004 1.31±0.004 1.103±0.004

0.76±0.03 1.10±0.04 1.10±0.06

46.2.5 83±2.5 77±5/7

2.90±0.36 2.50±0.10 2.60±0.12

13.00±0.29 12.00±0.22 12.00±0.16

Table 2. Concentration values and experimental errors for the TL studied MCF samples. The alkali metals of Na, K, and Rb were measured using NAA analysis and for the rare earth elements of U and Th, alpha-counting was used.

Figure 2. Na and K concentration profiles in the analysed samples. All values expressed as relative to the average concentrations of the core slices for the unaffected samples.

2000) since this allows for the detection of possible disequilibrium, which found as not present in the examined samples. The results for the bulk samples are given in Table 2 and also the K, Rb and Na concentrations from NAA analysis. By employing the conversion factors of Adamiec and Aitken (1998), the alpha- and beta-dose rate values of the samples were estimated.

TL glows were recorded with the use of a Littlemore TL 711 set, equipped with an EMI 9635QA photomultiplier and a Corning 7-59 blue pass filter. Measurements ran at a heating rate of 10oC s-1 and the irradiations administered with a 60Co gamma-source providing 1.73 Gy min-1 and an 241 Am alpha-source for the a-value estimation. The TL suitability of the samples was checked for plateau by performing initial test measurements. Using laboratory irradiated aliquots stored for three months the samples were checked against possible fading effects.

Since the samples were provided without an estimation of the environmental dose value it was indirectly calculated; archaeologically it was well established that the age was 1.95 (± 0.05) ka (M.A. Cau Ontiveros and M. Orfila Pons, personal communication), thus the measurement of the equivalent dose allowed, after the subtraction of the alphaand beta- contribution, the estimation of the environmental dose (Figure 3, Step 1 to 3). In that way, the TL measurement of the unaffected samples MCF29 and MCF35 enabled the estimation of an average environmental dose of 0.73 ± 0.40 Gy ka-1 (Table 3a).

For the estimation of the equivalent dose (De) an additive dose multiple-aliquot procedure (foil technique; Michael et al., 1997) was used. Using the foil technique, after the 1st TL read out, each aliquot is irradiated with the same test dose and the 2nd TL signal is recorded for the normalisation of the measurements. The U and Th values were estimated using a PIPS thick source alpha-counting technique (Michael and Zacharias

The relatively close estimated environmental dose values 53




1.95±0.05 1.95±0.05 1.95±0.05


7.91±0.90 8.55±1.05 8.43±0.95

De [Gy]

7.91±0.90 8.55±1.05 8.43±0.95



De [Gy]


0.16 0.17 0.17


3.41±0.47 3.65±0.48 3.69±0.46


[Gy ka-1] 4.73±0.43



[Gy ka-1] 1.44±0.28 1.51±0.30 1.54±0.26

D alpha

0.16 0.17 0.17





[Gy ka-1] 0.70±0.04 0.63±0.02 0.64±0.02

D beta(U,Th) [Gy ka-1] 0.73±0.35 0.73±0.35 0.73±0.35


D env-ave

0.11±0.02 0.13±0.02 0.13±0.02




[Gy ka-1] 4.06 4.38 4.32



1.44±0.28 1.51±0.30 1.54±0.26


[Gy ka-1] 1.71±0.34


D alpha

[Gy ka-1] 1.19 1.51 1.41


D beta(K)

1.24±0.08 1.41±0.08 1.42±0.08


[Gy ka-1] 2.28±0.15


D beta

unaffected burial period [ka] 1.12 1.62 1.39

1.67 2.14 1.99


TL_AGEave [ka, ±1σ]


[Gy ka-1]


D env-ave

Kave (wt %)

2.32±0.44 2.34±0.45 2.28±0.40

TL_AGE [ka, ±1σ]


[Gy ka-1] 0.74±0.53


D env


Table 3. TL dating results for MCF samples. W factor stands for the saturation content, given by the ratio. (saturation wet weight – dry weight)/(dry weight). For the .F value (fractional uptake . the effective alpha-dose rate, the beta-dose rate, D the total dose-rate value of water) a 0.50±0.25 was used. 3.a. The a-value corresponds to the effectiveness of the alpha particles, D D . alpha beta and D env the environmental dose-rate. By applying the TL dating procedure (cf. Figure 3) for the samples 29, 35, an average environmental dose-rate resulted which was then used for the TL age estimation of the samples MCF34, MCF38, MCF59. 3.b. Estimation of the K beta dose-rate ( D beta(K)) which provides an average (during burial period) K concentration (Kave) and by following equation (2) the ‘unaffected’ burial period estimation.




[Gy ka ] 0.73±0.35 0.73±0.35 0.73±0.35

D env-ave








N. Zacharias, C.T. Michael, A. Schwedt and H. Mommsen

Altered Archaeological Pottery and the Effect on TL Dating


Step 1

Estimation of the total dose rates ( D ) for the ‘unaffected’ samples using the relation


D =Equivalent Dose (De) / Archaeological Age


Step 2


. .

D value provides an environmental dose ( D env) for every used sample, since


. .

D env = D - D alpha - D beta


Step 3

Estimation of an average environmental dose ( D env-ave), resulting from the


D env values of the ‘unaffected’ samples

Step 4 TL age estimation for the ‘affected’ samples using the



D env-ave value

Step 5

Estimation of the K-dose rates ( D beta(K)) of the ‘affected’ samples like in step (1) and by using the step (3), thus providing

Step 6 the average K concentrations (Kave), of the ‘affected’ samples, during burial

Step 7 and finally an estimation of the ‘unaffected burial period’ given by the equation (2).




D env-ave value of

N. Zacharias, C.T. Michael, A. Schwedt and H. Mommsen for the two unaffected samples are reasonable since the samples were recovered from the same stratigraphical horizon and thus the average estimated value could then be used for the TL dating procedure of the affected samples MCF34, MCF38 and MCF59 (Figure 3, Step 4). For the presented altered samples, TL dating resulted deviating absolute ages with a mean TL age of 2.31±0.26 ka. This value, which is ca. 18% higher in comparison with the archaeologically provided age, indicates that potassium leaching occurred during burial.

the average K concentration measured from the examined unaffected MCF samples, which is 2.35±0.05%. This value is used under the assumption of an initial very similar chemical composition for the whole group. The final concentration Kfin corresponds to the actual determined content for each one of the affected samples. Thus, by using equation (2), we can estimate the duration of the ‘unaffected’ burial part for each one of the affected sherds (Table 3b) pointing to an alteration occurred at ca. 300 – 900 years B.P.

Estimation of the alteration period Conclusions Once verified that a severe alteration effect has taken place, it was considered that the K-leaching hypothetically could have happened as a relatively fast process at some point during burial. This seems to be suggested by the nonexistence of K content profiles (Schwedt, 2004), but also by considering the occurrence of a significant K-leaching (ca. three quarters out of the original mean of 4.63% of K2O) in a 17th century AD majolica assemblage (Buxeda et al., 2001).

The studied altered pottery, experience a leaching of ca. 50% of their original K content, thus resulting in a systematic age overestimation in luminescence measurements. This reinforces the need of additional mineralogical analysis when luminescence dating is attempted on sherds that show signs of over-firing. Although the overestimated TL ages for the altered samples overlap with the archaeological age when considering a 2σ error bars, the presented study alerts for the significance of the leaching effects when conventional luminescence service is performed.

Therefore, we could consider the burial period approximately as the sum of two parts (t1, t2), the internal irradiation status of which is differing only at their Kdoses. Thus, the luminescence equation Age = De / D can be formulated as in (1)






D = t1 D ini + t2 D fin,


Considering that the total burial period can be effectively divided in two parts, before and after a ‘fast’ alteration process of the glassy phase, the estimation of an average K concentration for the affected samples can be used to approximate the time during which the alteration took place. The possibility to correlate this time of significant leaching with a main change induced in the burial conditions of the site is of significant research interest and should be further investigated.


where D ini refers to the dose rate having the . contribution from an initial K concentration and D fin with the contribution from the measured today K concentration. Following the steps 5 and 6 of Figure 3, we estimate the average K concentration (Kave) for every affected sample, which accounts for the measured dose considering the archaeological age. This is provided by calculating the total dose rate and subtracting the contributions of the environmental dose and alpha and beta doses coming from U and Th, which are assumed as constant for the whole burial period. This assumption was additionally supported from the equilibrium state found for the decay series of both radioisotopes. Therefore, changes in the dose rates given in equation (1) can be given as changes in K concentrations. Since the K contribution of the two parts must be equal to that of the average K, equation (1) can be rewritten as follows

Acknowledments The authors are indebted to M.A. Cau Ontiveros and M. Orfila Pons, directors of the excavation of Sa Mesquida, for enabling the study and providing the samples. The authors would like to thank P. Bakandrea for the preparation of the TL samples and assisting with the experimental work. References Adamiec, G., and Aitken, M.J., 1998, Dose-rate conversion factors: update, Ancient TL, 16, 37-50. Aitken, M., 1985, Thermoluminecence dating, Academic Press, London. Buxeda i Garrigós, J., Madrid Fernández, M., and Gurt i Esparraguera, J. M., 2001, Provinença i tecnologia de les ceràmiques de „Pisa“ i d’ „Obra de Manises“ del dipòsit de la Plaça Gran de Mataró, in J. A. Cerdà i Mellado (Ed.): La ceràmica Catalana del segle XVII trobada a la Plaça Gran (Mataró), Associació Catalana de ceràmica decorada i terrissa, Barcelona, 156 – 170. Buxeda i Garrigós, J., Mommsen, H., and Tsolakidou, A., 2002, Alteration of Na-, K-, and Rb-concentrations in Mycenaean

Age Kave= t1Kini + t2Kfin Finally, the time t1 could be given in the form of equation (2)

t1 =

( Kave − Kfin ) Age ( Kini − Kfin )


The initial concentration Kini corresponds in equation (2) to 56

Altered Archaeological Pottery and the Effect on TL Dating pottery and a proposed explanation using X-Ray Diffraction, Archaeometry, 44, 187 – 198. Kalchgruber, R., Fuchs, M., Murray, A.S., and Wagner, G.A., 2003, Evaluating dose-rate distributions in natural sediments using α-Al2O3: C grains, Radiation Measurements, 37, 293297. Michael, C.T., Zacharias, N., Dimotikali, D., and Maniatis, Y., 1997, A new technique (foil technique) for measuring the natural dose in TL dating and its application in the dating of a mortar containing ceramic grains, Ancient TL, 15, 36-42. Michael, C.T. and Zacharias, N., 2000, A new technique for thicksource alpha counting determination of U and Th,, Nuclear Instruments and Methods, 439,1, 167-177. Mommsen, H., 2001, Provenance determination of pottery by trace element analysis: Problems, solutions and applications, Journal of Radioanalytical and Nuclear Chemistry, 247, 657662. Olley, J.M., Roberts, R.G., and Murray, A.S., 1997, Disequilibria in the uranium decay series in sedimentary deposits at Allen’s cave, nullarbor plain, Australia: Implications for dose rate determinations, Radiation Measurements, 27,2 433-443. Olin, J.S., Hartbottle, G., and Sayre, E.V., 1978, Elemental

Compositions of Spanish and Spanish-Colonial Majolica Ceramics in the Identification of Provenience, in Archaeological Chemistry II, (ed. G.F. Carter), 200-229, Advances in Chemistry Series, 171, American Chemical Society, Washington D.C. Picon, M., 1976, Remarques preliminaries sur deux types d’alteration de la composition chimique des ceramiques au cours du temps, Figlina 1, 159 – 166. Schwedt, A., 2004, Untersuchung von (Spuren-) Elementkonzentrationsprofilen in archäologischer Keramik mittels Neutronenaktivierungsanalyse, Dissertation Universität Bonn, math_nat_fak/2004/schwedt alexander/0336.pdf Tsantini, E., Buxeda i Garrigós, J., Cau Ontiveros, M. A., and Orfilia Pons, M., 2004, Caracterización arqueométrica de la cerámica común producida en la villa romana de Sa Mesquida (Mallorca), Pyrenae, 35, 157 – 186. Zacharias, N., Buxeda I Garrigós, J., Mommsen, H., Schwedt, A., and Kilikoglou, V., 2005, Implications of burial alterations on luminescence dating of archaeological ceramics, Journal of Archaeological Science, 32,1, 49-57.


THE CANAL OF XERXES IN NORTHERN GREECE: FACT OR FICTION? RECENT GEOPHYSICAL AND GEOARCHAEOLOGICAL INVESTIGATIONS B.S.J. Isserlin,1 M. Arvanitis,2 R.E. Jones,3 V. Karastathis,2 S.P. Papamarinopoulos,4 P. Stephanopoulos,4 G. Syrides5 and J. Uren6 University of Leeds, UK [email protected]


Geodynamics Institute, National Observatory of Athens


Department of Archaeology, University of Glasgow, UK

3 4

Laboratory of Geophysics, Department of Geology, University of Patras 5 6

Department of Geology, University of Thessaloniki

Department of Civil Engineering, University of Leeds, UK

Abstract: Herodotus describes how a canal was constructed in northern Greece by King Xerxes in around 480 BC to allow the Persian fleet into the Aegean in advance of its invasion of Greece. If so, this canal must have been a remarkable engineering operation for its time. This paper reports on the results of a non-invasive investigation at the supposed site of the canal on the 2 km wide isthmus of the Mount Athos peninsula. Geophysical, especially seismic survey, topographic survey and analysis of borehole sediments have played complementary roles in demonstrating that Herodotus’ account was very probably correct. Περιληψη: Ο Ηρόδοτος περιγράφει τον τρόπο με τον οποίο έγινε η διάνοιξη του διώρυγας στην Βόρειο Ελλάδα από τον βασιλιά Ξέρξη, για να καταστεί δυνατή η διάβαση στο Αιγαίο του Περσικού στόλου, κατά την εισβολή που επιχειρήθηκε 480 π.Χ. Εάν αυτό αληθεύει, πρόκειται για ένα αξιοθαύμαστο τεχνολογικό επίτευγμα για την εποχή του. Η εργασία αναφέρεται στα αποτελέσματα της γεωφυσικής διασκόπισης στην διερευνούμενη περιοχή της διώρυγας και σε πλάτος 2 χλμ. στο στενό που διαγράφεται επί της χερσονήσου του Αγ. Όρους.. Τα αποτελέσματα των γεωφυσικών, σεισμοτεκτονικών και τοπογραφικών αποτελεσμάτων της έρευνας, καθώς και η ανάλυση των ιζημάτων των πυρήνων που λήφθηκαν, δείχνουν να επιβεβαιώνουν τις αναφορές που παρατίθενται από τον Ηρόδοτο. Keywords: Geophysical survey; seismics; Xerxes; canal; coring; sediment analysis; ground-penetrating radar; radiocarbon dating


complementary roles in the investigation in demonstrating the veracity of Herodotus’ account. The results of the project have been published with different audiences in mind: Isserlin et al (2003) have reported the project’s final stage and discussed the findings as a whole in an archaeological context, while the more technical aspects have been explored by Karastathis, Papamarinopoulos and Jones (2001) and Jones et al (2000). The present paper aims to give an overview of the project now that its current phase of field work is complete, to demonstrate the role of integrated input from geophysical and topographic surveys and analysis of sediments from cores, and to consider the progress of possible future work at the canal. In our mind, two main factors justify the need for such a paper; first, the results deserve to be presented to as wide an audience as possible since the object of this investigation - the canal - is unique (not only in Greece), and furthermore it is unlikely for logistical as well as financial reasons to be excavated by conventional archaeological means at least in the near future (but see Discussion; and Isserlin et al 2003, Appendix 1); second, the project can be viewed as a case study within the context of the increasing impact that geophysical and related survey is now making in Greek and more widely in Mediterranean archaeology (Sarris and Jones 2000).

The ancient Greek historian Herodotus (Histories VII 22-4) describes how a canal was constructed in northern Greece on the orders the Persian King Xerxes in 483-1 BC to allow the Persian fleet safe passage into the Aegean in advance of its invasion of Greece; he records that it was some 2 km long and was wide enough for two triremes to pass side by side. If this account is true, its construction must have been a remarkable engineering operation for its time. Although the canal has attracted scholarly attention over the years, no archaeological fieldwork was carried out to verify whether indeed it was a canal linked at both ends to the sea, or whether, as some have postulated, there was instead a slipway (diolkos) across the isthmus. Today the only visible remains are a depression in the central part of the isthmus. To resolve this question, a British-Greek team has carried out over the last decade a non-invasive investigation at the supposed site of the canal which is situated on the Athos peninsula of the Chalkidiki (Fig. 1 inset). Geophysical (seismic, radar, electric and magnetic) and topographic surveys and analysis (and radiocarbon dating) of sediments from boreholes have all played important, 59

B.S.J. Isserlin et al.

Figure 1(a). Plan of the canal represented by the centreline of the present-day lowest ground level (as estimated in 1991), showing the locations of the geophysical surveys. The site is marked on the inset map.

Figure 1(b). The locations of the boreholes (BH) and later seismic profiles along the course of the canal as estimated in 1991 and following revision in 2000.

The location of the canal

constructed. But having read Herodotus’ account, it comes as a disappointment to find so little on the ground today that can be linked to this structure. Apart from the broad depression along the central sector, there is an apparent absence of, for example, any infrastructure that might be expected to accompany the operation of a canal, from buildings, breakwaters to harbour installations. In this situation, the counter-suggestion that there was instead of a canal, a slipway (diolkos) along which ships could be dragged takes on more credibility. Indeed, Demetrius of Skepsis writing some three centuries later than Herodotus, claimed that the canal did not traverse the whole isthmus because of the hard, rocky terrain at the southern end at Tripiti. In recent times, opinions have been divided; some North European travellers to Greece in the 18th to early 20th century, such as Choiseul-Gouffier (1809) (see Fig. 2b), followed Herodotus, while Cousinéry (1831: 153ff) believed there was a diolkos. Local views have been more divided between the two models.

The presumed course of the canal lies at the narrowest point of the Mount Athos isthmus, amid agricultural land. Today, from the sea at its northern end it passes close to the village of Nea Roda across an area of open flat, formerly marshy land; entering the central sector where the land level rises to a maximum elevation of 15.7 m above sea level, there is a long natural depression which gives way to a stream bed that passes between two hills close to the sea at Tripiti. This picture of the canal’s topographic environment is complicated by the fact that the area is one of high tectonic activity. Sands, silts, red beds and alluvium are the main surface geological features at the isthmus, and below them, generally but not exclusively at a depth of at least 10 m, lie red bed sediments (Syrides 1990). The choice of this location to construct the putative canal was a good one since not only was it at the narrowest and lowest part of the isthmus, but the sediments would have been relatively easy to dig out. To the northwest limestone and granite outcrop, and there is gneiss to the south-east.

The survey The project developed a flexible and often experimental strategy, summarized in rough chronological order in Table 1 that was capable of detecting and describing on the one hand a large, potentially deeply buried structure and, on the other, buildings associated with that structure. Fig. 1a, b shows the locations of most of the geophysical

Archaeological – historical considerations Herodotus describes not only the events that led Xerxes to order the construction of the canal but also how it was 60

The Canal of Xerxes in Northern Greece: Fact or Fiction?



1. Walk-over survey

1. Roman pottery near the southern end of canal; Byzantine structures on the southern hill at Tripiti. Many ancient sites in the vicinity: ancient Acanthus, Sani and Ouranopoulis. 2. c. 6000 3-D readings, 1:2000 contoured map. Identification of a likely course of the canal based on present-day lowest ground level. 3. Several Greek Air Force overhead and oblique photographs No clear demarcation of the canal’s sides up to a depth of 30 m

2. Topographic

3. Aerial 1. Resistivity soundings: 60m profiles; Schlumberger array at 5 m intervals 2. Electrical imaging 1. Magnetic (gradiometer) mapping mainly at the northern end to detect buildings 2. Magnetic (gradiometer) mapping over a building structure observed in an air photograph, situated to the east of the canal in its central sector 3. Magnetic profile (with proton magnetometer) of the canal Ground-penetrating radar (GPR): 80 and 120 MHz antennae, ten 50-120 m profiles 1. Seismic reflection and refraction profiles 2. Seismic tomography Sedimentological analysis of nine borehole cores, ranging in depth from 10 to 24 m Radiocarbon dating of four sediments Satellite (LANDSAT-TM false colour) image

Publication in addition to Isserlin et al 1994; Jones et al 2000 1. Isserlin (1991)

1. Anomalies were more likely associated with modern soil disturbance than with buildings 2. No anomalies found; the building may have been ploughed out 3. See text

3. Isserlin et al (2003)

See text See text

1. Karastathis et al (2001) 2. Isserlin et al (2003)

See text See text See text

Isserlin et al (1996) Isserlin et al (2003)

Table 1: The techniques used in the Xerxes Canal Project

surveys and boreholes; the former were generally aligned as profiles at right angles to the supposed course of the canal. Operating conditions of the survey techniques and data processing methods are given in the papers referred to in Table 1. The most informative results emerged from survey in the central sector of the isthmus.

Instead, several episodes of (natural) infilling of the depression were evident, as was the apparent asymmetry of this depression which was also detected in some of the seismic data. Following the initial promising seismic refraction and reflection profiles in the central sector, the technique was employed at points along the whole course of the canal. The results overall were highly satisfactory generally, and also specifically owing to the lower velocity values obtained in the infill as compared to those of the host material: Fig. 2a shows clearly the profile of the canal at one position in the central sector, and the picture resulting from 2-D inverse modelling techniques on data from a series of five profiles at the northern end appears in Fig. 3. Furthermore, the direction of the canal apparent in Fig. 3 differs from that originally determined on the basis of present-day lowest ground level (Fig. 1a) such that it is now likely the canal connected with the sea at least 50 m further east (Fig. 1b: 2000 route). The combination of seismic survey and analysis of borehole cores proved to be powerful for two reasons. First, the former estimated the

Beginning with the electrical survey, the pseudosections from resistivity soundings displayed at the top a 1-2 m thick high resistance layer of colluvium infill, giving way to progressively lower resistance values in the thick sand/ silt layer to a depth of c. 30 m. There was no indication of a canal-like structure. In the corresponding magnetic profile across the canal, the infill gave a low intensity anomaly. The magnetic anomalies detected at the northern end were interpreted as the result of modern disturbance rather than as building structures associated with the canal. The depth of penetration of the GPR survey did not exceed 10 m owing to the relatively high water table; this, coupled with the lack of dielectric permittivity contrast between the infill and the surrounding more compacted host sediments, resulted in a failure to detect a complete canal-like outline. 61

B.S.J. Isserlin et al.

Figure 2. (a) Representation of the canal section (a) at Profile D based on the high resolution seismic refraction data following velocity analysis: velocity in the host material (shaded) 2170 m/sec, in the infill 1390 m/sec and in the top layer 380 m/sec. (b) according to de Choiseul-Gouffier (1809) (Isserlin et al 1996)


depth of the base of the canal below present-day ground surface in the central sector to be c. 15 m, a value which coincides remarkably with the depth of the interface of the clay/silt layers and the more compact red beds (Groups A and B sediments respectively in Fig. 4) as observed in the sequence of sediments in borehole 3 (BH3 in Fig. 1b); this interface surely represents the base of the canal (Fig. 4). Second, the combined methods resolved the issue of the diolkos model at the southern end: not only did there seem to be no serious obstacle from shallow bedrock there that would have prevented the canal being extended all the way to the sea, but there were no indications of a diolkos anywhere. Two notable features of the core sediments taken in the central sector were the absence of marine indicators that hints that the canal’s direct links with the sea were limited in duration, and the presence of indicators that the canal fell into disuse probably rapidly by side collapse creating in places stillwater marshy environments. Finally, although nothing is evident in the air photographs of the canal beyond the central depression, it is noteworthy that among the drainage areas appearing in yellow in a LANDSAT-TM false-colour satellite image of the region is one traversing the isthmus in a direction remarkably similar to that proposed by the combined survey evidence (marked 2000 route in Fig. 1b).

Archaeological interpretation of the project’s results can be summarised succinctly: across the 2 km isthmus a canal was very likely dug, its top width being c. 30 m and its base now buried some 15 m below present ground level; the canal would have been shallow, but more generally apparently around 3 m deep. This description conforms fairly well to that given by Herodotus, and it harmonises with the model according to one of the early travellers, de Choiseul-Gouffier (1809), shown in Fig. 2b. At both ends, the canal connected with the sea; there is no good reason to suppose there was a diolkos at the southern end. The canal seems not to have been of uniform construction, and nor does it appear to have had a long life; its main function fulfilled, it was probably not maintained and so fell into disuse rapidly. In this light it is perhaps not surprising that no evidence was found for buildings that could be associated with the canal’s operation; no doubt such buildings existed, but they were temporary and have long since disappeared. Overall, although the project’s ambitious aims have been largely fulfilled, the quality of the results was not uniform across the isthmus; that the clearest results were consistently obtained in the central sector has been frequently alluded to, in contrast first to the 62

The Canal of Xerxes in Northern Greece: Fact or Fiction?

1.8 km/s Interface -1.6 -1.4 -1.2 -1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1 1.2 1.4

Figure 3. A visualization of the seismic refraction data obtained in the flat area at the northern part of the isthmus (SP 1-5/00 in Fig. 1b). It is based on a scheme connecting the subsurface points with the same acoustic velocity (1.8 km/s). The colour scale indicates the elevation (in m) of these points, the blue representing points below present-day sea level. On the right, an arrow marks the proposed course of the canal.

situation around GT A1-A2 in Fig. 1a where the modern stream bed joins the canal depression towards Tripiti, and second to what is mentioned below in the plain at the northern end. Turning to the methodologies adopted, seismic survey was very well suited to the problem in hand, while the contribution from the other geophysical techniques was useful but less crucial. Borehole coring was not only necessary in such a project but the results that this approach provided from the sediment analyses were essential. That said, the borehole coring confronted two problems; first, the nature of the sediments was such that core loss was significant in some places. Second, whereas the decision where to locate the boreholes in the central and southern sectors was relatively straightforward, the situation in the wide flat plain at the northern end was different; here, the partial picture obtained from the exploratory few cores taken had little archaeological significance. The same remarks apply to the exploratory use of radiocarbon dating to try and date the construction of the canal. As expected, the dates from sediments taken just above the base of the canal at two locations in the central sector were considerably older than the likely time of construction because of the presence of ‘older’ organic components in the sediments. The point is that to obtain archaeologically meaningful dates would require sediment samples for radiocarbon dating that would be too numerous to be justified in terms of cost and effort.

Figure 4. Lithological column of the borehole BH3 (see Fig. 1b). Ground level is at c. 14.1 m above present sea level. The Group A sediments are typically grey-brown to dark brown8 with blackish silty-clayey sands, silts-clays and coarse sands scattered small pebbles, while Group B sediments are reddish brown coarse silty sands with varying proportions of clay and scattered small pebbles; they are more compact sediments than A. It is proposed the base of the canal lies at the interface of the Groups A and B sediments, c. 3.80 m below present sea level which is marked with a horizontal arrow.

reconnaissance of the two ends of the canal for indications of harbour installations and breakwaters, as well as for information supplementing that already obtained by Papangelos and Kambouroglou (1999) on sea level changes. Second, the opportunity for exploiting the potential of core sediments for environmental reconstruction has not been lost; Drivaliari and Syrides (2003) have already presented their preliminary environmental findings which they expect to publish at a later date; magnetic susceptibility measurements on the cores may also be valuable (D. Kontopoulou, pers. comm.). Third, having defined the course of the canal with some confidence by geophysical survey there is sufficient associated topographic data

Looking to the future, although the project’s field work as reported here is now completed, there are a number of avenues for further work. One that has been recognised for some time is the desirability of an underwater 63

B.S.J. Isserlin et al. to make a 3-D reconstruction of the canal. This could have direct educational value, as well as enhancing the presentation of the canal to visitors to the region, especially in the event that the proposed ‘archaeological park’ at Nea Roda becomes a reality. But, fourth and most provocative of all would the excavation of a test section of the canal, as outlined by Isserlin et al (2003: Appendix 1), probably in the central sector. This would be a very challenging but in our opinion worthwhile enterprise. Finally, we note with interest the current Persian War Shipwreck Survey off Mt Athos which is a collaborative effort of the Greek Ephorate of Underwater Antiquities, the Canadian Archaeological Institute at Athens and the Hellenic Centre for Marine Research (Whitley 2004: 54).

Isserlin, B.S.J., Jones, R.E., Papamarinopoulos, S.P. and Uren, J. 1994. The Canal of Xerxes: preliminary investigations in 1991 and 1992. Annual of the British School at Athens 89: 277-84. Isserlin, B.S.J., Jones, R.E., Papamarinopoulos, S., Syrides, G.E., Maniatis, Y., Facorellis, G. and Uren, J. 1996. The Canal of Xerxes: investigations in 1993-1994. Annual of the British School at Athens 91: 329-40. Isserlin, B.S.J., Jones, R.E., Karastathis, V., Papamarinopoulos, S.P., Syrides, G.E. and Uren, J. 2003. The Canal of Xerxes: Summary of Investigations 1991-2001. Annual of the British School at Athens 98: 369-87. Jones, R.E., Isserlin, B.S.J., Karastathis, V., Papamarinopoulos, S.P., Syrides, G.E., Uren, J. with Balatsas, I. and Kapopoulos, Ch. 2000. Exploration of the Canal of Xerxes, Northern Greece: the role of geophysical and other techniques. Archaeological Prospection 7: 147-170. Karastathis, V., Papamarinopoulos, S.P. and Jones, R.E. 2001. 2-D Velocity structure of the buried ancient canal of Xerxes: an application of seismic methods in archaeology. Journal of Applied Geophysics 47(1): 29-43. Papangelos, I. and Kambouroglou, E. 1999. Archaeogeomorphological researches for the Xerxes Canal at Athos peninsula. Proc. 5th Panhellenic Geographical Soc, Athens. Sarris, A. and Jones, R.E. 2000. Geophysical and related techniques applied to archaeological survey in the Mediterranean: a review. J. Mediterranean Archaeology 13 (1). 3-75. Syrides, G. 1990. Lithostratigraphic, biostratigraphic and palaeogeographic study of the Neogene-Quaternary sedimentary deposits of Chalkidiki peninsula, Macedonia, Greece. PhD thesis, Scientific Annals, School of Geology 1 (11). Aristotle University of Thessaloniki. (In Greek). Whitley, J. 2004. Archaeological Reports for 2003-2004. Society for Promotion of Hellenic Studies and British School at Athens, London.

Acknowledgments Many organisations and individuals have made this project possible. Here we wish to thank the Ministry of Culture for permission to carry out this project, the British School at Athens, and NATO’s Science for Stability program. References Cousinéry, E.M. 1831. Voyage dans la Macédoine, Paris. De Choiseul-Gouffier, G.A.P. 1809. Voyage pittoresque en Grèce II, Paris. Drivaliari and Syrides, G.. 2003. Bio-indicators in transitional archaeo-environments. a case study at Xerxes canal, Abstract, Fourth Symposium on Archaeometry (Hellenic Society of Archaeometry) (Athens 2003). Isserlin, B.S.J. 1991. The Canal Xerxes: facts and problems. Annual of the British School at Athens 86: 83-91.


THE RECONSTRUCTION OF PREHISTORIC SHORELINES IN DOKOS ISLAND, AEGEAN SEA, USING REMOTE SENSING TECHNIQUES1 G. Papatheodorou, M. Geraga and G. Ferentinos Laboratory of Marine Geology and Physical Oceanography, Department of Geology, University of Patras, [email protected] Abstract: The evolution of the coastline configuration around Dokos Island over the last 18000yrs BP is herewith presented in detail, based on the mapping of palaeoshorelines features, using marine remote sensing techniques. The formation age of the identified palaeoshorelines based on the global sea-level change curves shows that they coincide with the occurrence of global climatic changes. The shoreline configuration shows that Dokos Island was connected to the mainland from the 18000yrs BP to at least 9600yrs BP. During that time Dokos was a low height headland exposed to winds and waves thus making it inhospitable to the mariners of that time. There are indicators that changes to the shoreline configuration at 6000yrs BP and 4500yrs BP may be associated with the establishment of the Neolithic settlement on the island and, may have been a contributing factor in the sinking of the ship, the wreck of which has been found on the seafloor. Περιληψη: Το Εργαστήριο Θαλάσσιας Γεωλογίας και Φυσικής Ωκεανογραφίας (Ε.ΘΑ.ΓΕ.Φ.Ω) Τμήματος Γεωλογίας Παν/μίου Πατρών προσκλήθηκε το 1992 από το Ινστιτούτο Εναλίων Αρχαιολογικών Ερευνών (ΙΕΝΑΕ) να αναλάβει τη θαλάσσια γεωφυσική έρευνα στη βόρεια υφαλοκρηπίδα της νήσου Δοκού με τη χρήση συστημάτων θαλάσσιας τηλεπισκόπησης (τομογράφο υποδομής πυθμένα 3.5kHz και ηχοβολιστή πλευρικής σάρωσης). Στη θέση αυτή αναπαύεται αρχαίο ναυάγιο περίπου 5000 ετών, το οποίο είχε μελετηθεί με ιδιαίτερη λεπτομέρεια από το ΙΕΝΑΕ. Σκοπός της γεωφυσικής έρευνας ήταν ο εντοπισμός των παλαιοακτών και η ανάπλαση της παράκτιας παλαιογεωγραφίας της περιοχής τα τελευταία 18000 χρόνια και ιδιαιτέρως τη χρονική περίοδο που συνέβη το ναυάγιο. Κατά τη διάρκεια των ερευνών εντοπίστηκε πληθώρα καταβυθισμένων αναβαθμών που αποδόθηκε στη δημιουργία προσωρινών παλαιοακτών. Βάσει της καμπύλης της παγκόσμιας ανύψωσης της θάλασσας κατά Fairbanks και της επιφανειακής κατανομής των βαθών εμφάνισης των αναβαθμών κατέστη δυνατή η αποτύπωση των παλαιοακτών της νήσου του Δοκού κατά τη διάρκεια της Παλαιολιθικής, Μεσολιθικής και Νεολιθικής περιόδου. Ο κόλπος Σκιντού υπολογίζεται ότι διαμορφώθηκε στην αρχή της Νεολιθικής περιόδου και ακολούθως περίπου στα 6000yrs BP ένας κολπίσκος σχηματίστηκε στην βόρεια ακτή του Σκιντού στη θέση όπου εντοπίστηκε το αρχαίο ναυάγιο. Η συνεχής άνοδος της θάλασσας φαίνεται ότι κατέκλυσε αυτόν τον κολπίσκο περίπου στα 4000yrs BP που ίσως χρησίμευε ως φυσικό καταφύγιο στους θαλασσοπόρους και πιθανώς αυτό το γεγονός να σχετίζεται με την βύθιση του πλοίου.


During prehistoric times coastal environments had hosted human settlements and thus the identification and the mapping of the coastal paleogeography is of great importance to the archeological studies (Kraft et al. 1982). Over the last decades the use of remote sensing techniques for the reconstruction of the ancient coastal environments has become a common practice because these techniques permit the high accuracy identification and mapping of submerged palaeoshorelines which are either on the seafloor surface or are buried under loose post-glacial sediments (van Andel & Shacleton 1982, van Andel & Lianos 1984, Wiedicke et al. 1999).

The change in climate over the last 18000yrs BP caused significant changes to sea level and thus to coastal palaeogeography (Fairbanks 1989). At 18000yrs BP, during the last glacial maximum the sea level was 120m below the present level, exposing large areas of land. The melting of the ice sheets which followed caused a rise in sea level and the submersion of the coastal low lands. The rise was rapid but pulsating. Short term cold events (stadials) have caused still-stands which have resulted in the formation of a series of palaeoshorelines. Their imprints on the coastal zone depend mainly on the sediment supply. In littoral environments with adequate sediment supply the sea level stands are imprinted as changes in coastal trangressive deposits, whilst in environments with limited sediment supply the still stands are imprinted as geomorphologic features, usually scarps and ridges (van Andel & Lianos 1984, Lobo et al. 2001, Karisiddaiah et al. 2002).

The purpose of this survey is twofold: firstly to reconstruct the geomorphological condition in the embayment where the Dokos wreck was found (Papathanasopoulos et al. 1989-1992) and secondly to reconstruct the Late Paleolithic, Mesolithic and Neolithic shorelines around Dokos island (Fig. 1). Dokos Island lies on the shelf of the Argolida peninsula. At 18000yrs BP Dokos and the nearby islands (Spetses and Hydra) were connected to the mainland forming

This paper is dedicated to N. Tsouhlos, former president of H.I.M.A., who passed away. His name will certainly remain as a pioneer of underwater archaeology in Greece. 1


G. Papatheodorou, M. Geraga and G. Ferentinos a “semicontinent” (van Andel & Shackleton 1982, Perissoratis & Conispoliatis 2003). The transgression of the sea, that followed, flooded the lowlands and thus the present day islands were shaped. This occurred at ~9000yrs BP (van Andel & Shackleton 1982). Methodology The marine remote sensing survey around Dokos island was carried out using a 3.5kHz Geopulse sub-bottom profiler and an EG &G side scan sonar. The sub-bottom profiler system emits medium to high frequency acoustic pulse in the form of acoustic conical beams. The acoustic pulses, which penetrate under the seabed are reflected back by the underlying layers and are recorded on a continuously running chart, thus providing a geological profile (tomography) of the sub-bottom beneath the path over which the system is towed. The side scan sonar system emits acoustic pulses which scan the seafloor. The reflected acoustic pulses from the seafloor are graphically recorded on a running chart thus providing a plan view seafloor acoustic image.

Figure 1. General map of the studied area. The grey area indicates the area where the marine remote-sensing survey was carried out. The arrow indicates the position of the wreck in Skindou embayment.

A Magelan GPS system with an accuracy of about 30m was used for the navigation and the positioning. The ship tracks were plotted in the hydrographic chart of the area (Chart) and a detailed bathymetric map was prepared using the depth data from the sub-bottom profiles and the bathymetric information of the chart (Fig. 2). Data presentation The study of the acoustic profiles based on the definitions of Damuth (1975), has shown that, the sea-bottom in the records is characterized by four different echo types each one corresponding to different seabed material (Fig.3).

Figure 2. Bathymetric map of the studied area based on bathymetry data collecting during the survey. The index map presents the ship tracks.

Echo Type I: is characterized by a very prolonged echo with no sub-bottom reflections. Locally the prolonged echo exhibits low relief diffraction hyperbolae. Echo Type I extends from the present day shoreline to a depth of about 20 to 30m (Fig.3). The acoustic echo corresponds to hard rock seafloor which generally has a smooth surface. The rocky seafloor continues onshore and outcrops on the island. The rocky outcrops on the island were identified as Cretaceous limestones. Echo Type II: is characterized by a continuous sharp bottom echo with one prolonged sub-bottom reflector. Echo Type II extends from the 20m isobath to the 100m isobath (Fig.3). This echo type corresponds to a surficial homogenous non-layered stratum probably consisting of sandy silt. This stratum overlies the aforementioned Cretaceous limestone and its thickness increases from 0.5 to 3m seawards. The homogeneus character indicates that the deposition of the sediments probably took place under

Figure 3. Spatial distribution of the Echo Types (I-IV) identified in the studied area. In the left side of the Figure representative 3.5kHz records of Echo Types (I-IV) are shown.


The Reconstruction of Prehistoric Shorelines in Dokos Island, Aegean Sea, Using Remote Sensing Techniques

Figure 5. (A) 3.5kHz profile showing a well-shaped scarp, buried under loose post-glacial sediments and (B) Simplified drawing showing the formation of the scarp.

prominent feature in all the records is the limestone bedrock. This limestone bedrock can be traced from the shoreline to a water depth more than 110m. At water depths less than 20 to 30m the limestone bedrock outcrops on the seafloor. Beyond the 20 to 30m water depth the limestone is covered by sediments and occurs at sub-bottom depths from a few cm to 15m increasing seawards. The above suggests that the limestone bedrock in the area under investigation lies at its seaward ends about 125m below the present day sea level. The morphology of the limestone surface exhibited in the records indicates a “basal trangressive” surface. This basal trangressive surface has also been identified around the coastline of the Argolida peninsula by Van Andel & Lianos (1984).

Figure 4. Selected seismic profiles from the studied area. The arrows indicate the geomorphologic features which correspond to surficial or buried traces of submerged palaeoshorelines (scarps).

high wave energy regime inside the nearshore zone, whilst the small thickness of the stratum indicates a slow rate of deposition probably caused by the rapid transgression of the sea.

Detailed examination of the profiles on the limestone bedrock surface reveals numerous small scarps which appear as sharp breaks in the overall slope gradient of the limestone surface (Fig. 4). These scarps on the limestone bedrocks appear to have been sculptured by the erosional power of the waves when the sea level was at a stand –still for a short period of time during the sea transgression (Fig. 5). It can therefore be regarded as evidence of palaeoshoreline.

Echo Type III: is characterized by regular hyperbolae with varying vertex elevations and no sub-bottom reflectors. ET III appears within the areal extent of ET II and between the depth waters of 60 to 90m (Fig.3). Echo Type III corresponds to area whereas the hard rock seafloor appears in steep slopes.

The side scan sonar survey has shown that these palaeoshorelines on the seafloor are continuous for long distances along the coastline. Similar scarps which have also been considered as palaeoshorelines markers have been found along the coastline of the Argolid peninsula (Van Andel & Lianos 1984). If we assume that the sea level at the time of their formation corresponded to the bases of the scarps, the position of these past sea levels can be determined to within a 2 to 3m accuracy.

Echo Type IV: is characterized by a continuous sharp bottom echo with up to 15m of conformable sharp continuous reflectors. This echo type borders Echo Type II and extends beyond the 100m isobath (Fig.3). This echo type corresponds to a layered sedimentary sequence which overlies the limestone bedrock. This layered sequence thickens seawards attaining a maximum thickness of about 15m. The layered character of this sequence suggests that the sediments were deposited continuously and slowly in a low energy environment beyond a nearshore zone.

To examine the evolution of the past glacial sea level rise, a percentage occurrence diagram of the identified scarps

The study of the seismic records suggests that the most 67

G. Papatheodorou, M. Geraga and G. Ferentinos

Figure 6. Histogram of submerged shore features (scarps) observed in the seismic records of the studied area, plotted against depth below present sea level.

versus the water depth in which they were detected, was drawn (Fig. 6). The study of the diagram shows that the scarps are not randomly distributed but rather gathered in clusters which correspond to selective water depths. The water depth in which the scarps are clustered are at 112m, 98m, 61-62m, 51m, 45-48m, 35-38m, 31-33m, 20-24m and 10-11m. The above suggests that the transgression of the sea around Dokos island was not continuous during the Late Glacial-Holocene period but was interrupted for short time intervals. If we assume that eustasy was the only parameter controlling the above sea level changes then their age of formation can be determined using the sea-level change curve. Table I presents the age of the above mentioned palaeoshorelines according to the curve constructed by Fairbanks (1989) (Fig.7).

Figure 7. Diagram showing the depths of the Dokos paleoshorelines plotted on the sea level change curve, by Fairbanks (1989).

(1984). Furthermore, the age of the palaeoshorelines based on the same sea level change curve (Fairbanks 1989) agrees well with the age of the palaeoshorelines around Dokos island (Table I). A good agreement also exists between the water depth of the present study shorelines for the 18000yrs BP, 10000yrs BP and 6000yrs BP and the water depth of the same age palaeoshorelines around the Argolida peninsula, as estimated by Lambeck (1996) who took into consideration the contribution of the isostatic effect caused by the ice melting.

Discussion The study of the palaeoshorelines around Dokos island reveals that the water depths they have formed correlates well with the water depth of the palaeoshorelines detected around the Argolida peninsula by van Andel & Lianos N. Dokos (Present Study) Depth of palaeoshorelines Below present sea level (m)

Argolid (Van Andel and Lianos, 1984)3 Fairbanks, 1989 Age (ka BP)

112 98

15.1 12.7

61-62 51 45-48 35-38 31-33 20-24 10-11

10.4 9.6 9.5 9.0 8.6 7.6-8.0 6.0

Depth of palaeoshoreline

Argolid shore clusters

Fairbanks, 1989 Age (ka BP)




97 75-74 62-65 52 45-48 37-39 27-28 18-16 10-11


12.7 11.5 10.5 9.6 9.5 9.0 8.4 7.5 6.0


Table 1 68

The Reconstruction of Prehistoric Shorelines in Dokos Island, Aegean Sea, Using Remote Sensing Techniques

N. Dokos (present study)

112 98 61-62 51 45-48 35-38 31-33 20-24

Age (ka BP) (Fairbanks, 1989) 15.1 12.7 10.4 9.6 9.5 9.0 8.6 7.6-8.0



Depth of palaeoshorelines (m)

Stadials obtained at high latitudes of the Northern Hemisphere and the Mediterranean region Heinrich 1 Tectonic event? Younger Dryas 1st Holocene stadial Tectonic event ? Sapropel S1a Tectonic event? 2nd Holocene stadial (interruption of the S1 sapropel) 3rd Holocene stadial (final end of the S1 sapropel)

Table 2 Finally, the very last portion of the rise which formed the 10-11m palaeoshoreline formed around 6000yrs BP is in agreement with the Neolithic artifacts which are dated at between 7610 to 6200yrs in a water depth of 11m in the southern Argolida shelf (Gifford 1983).

indicate that one of the palaeoshorelines was also caused by a tectonic event. The palaeoshorelines at 20-24m and 10-11m coincide with the 2nd and 3rd Holocene stadials at about 7600-8000yrs BP and 6000yrs BP, respectively. The age and water depth of the latter palaeoshoreline also agrees well with the Neolithic artifacts dated between 7610 and 6200years and found submerged under a water depth of 11m (Gifford 1983).

The formation age of these palaeoshorelines matches well with the age of stadials (cold short-term events) prevailing during that period in the high latitudes by the Northern Hemisphere (Bond et al. 1992, Bond and Lotti 1995, Bond et al. 1997) and in Mediterranean areas (Matthews & Aylon 1997, Rohling et al. 1998, Geraga et al. 2000, Karkanas 2001). A detailed comparison of the age of the formation of the palaeoshorelines and the occurrence of the paleoclimatic instabilities shows the following:

After 6000yrs, according to archaeological data, the sea level rose at about 4000yrs BP and reached 5 to 7m below present-day sea level rise (Van Andel & Lianos, 1984). Based on the above presented distribution of the palaeoshorelines and the archaeological records for the prehistoric period as deducted by detailed studies in the Franchthi cave located in the Argolida (Perles 1987a & b) an estimation of the areal extent of the lowlands can be made for the surrounding area. The sea level during the Upper Paleolithic at about 18000yrs BP was at about -120m and rose to –(61-62)m at about 10400yrs BP which coincides with the end of the Upper Paleolithic. During this time span the mean rising rate was about 4.4m/ka. During the Mesolithic from 10000yrs BP to 8000yrs BP the sea level rose another 30m with a mean rising rate of 15m/ka. The transition from Mesolithic to Neolithic coincide with the formation of the palaeoshoreline at -24m whilst during Late Neolithic the palaeoshoreline was at –(10-11)m.

The palaeoshoreline at 112m can be well correlated with the occurrence of the Heinrich 1 event between 13000 to 16000yrs BP (Table 2) whilst the palaeoshoreline at 61-62m with the occurrence of Younger Dryas event at 10000-11000yrs BP (Table 2). The palaeoshoreline at 98m formed at about 12700yrs BP does not correlate with any known climatic change, therefore it can be suggested that the palaeoshoreline has probably been formed by a local tectonic event. The palaeoshorelines at 51m and 45-48m are located in the most inclined part of the curve which indicates a very fast rate of sea level rise for a short period. These two palaeoshorelines seem to be associated with the 1st Holocene stadial at about 9600yrs BP. The presence of two palaeoshorelines with a height difference of about 4 to 5m corresponding to a time span of less than 2 to 3 hundred years may indicate that one of the two palaeoshorelines was probably caused by a tectonic event.

The study of the palaeogeographic maps (Fig. 8, Fig. 9) show how different the coastal configuration was between the Upper Paleolithic, Mesolithic and Neolithic Ages. The most striking feature in the Upper Paleolithic was the extensive lowlands which in the Neolithic period were severely reduced. Dokos island was connected with the Argolida peninsula as were the two other islands Hydra and Spetses. All these three islands at that time formed isolated hills.

The palaeoshorelines at 35-38m and 31-33m were formed during the sapropel S1a layer between 9500 and 9000yrs BP during a period of warm and humid climate. The formation of these two shorelines in such a short time span during which no abrupt climate change was observed,


G. Papatheodorou, M. Geraga and G. Ferentinos The data indicates that in the Upper Paleolithic a vast area around the island was probably occupied by coastal plains. These lowlands might have been biologically productive and may have had large herds of animals. At that time Dokos island as well as the islands of Spetses and Hydra were connected to the mainland forming low height headlands. Dokos was disconnected from the mainland between 8600 and 9000yrs BP when the sea level was at about –33 to m below present sea level. At that time most of the low-land was covered by the sea. The impact of such land loss may be illustrated in the case of the nearby Franchthi cave. The human occupation of the cave became more intense and continuous after 10000yrs and lasted until Final Neolithic. This might have been caused due to the shortage of food resources as the lowlands were covered by the sea. The study of the shorelines around Dokos Island shows that from 15100yrs BP to 9500yrs BP there was no indentations along the coastline and that the coastline was exposed to winds and waves making the island in-hospitable to mariners of that time. At about the beginning of the Neolithic, when the sea level rose to –(20-24)m below present, an embayment formed in the northern coastline of the island “the Skindou embayment” (Fig. 9). The coastal configuration of the Skindou embayment shows that it was exposed to the westerly winds with a fetch of about 10km, thus making the embayment unsafe for the seafarers for protecting their small crafts.

Figure 8. Map showing the palaeoshorelines around Dokos Island at 15100yrs BP, 12700yrs BP, 10400yrsBP and 9600yrsBP.

At about 6000yrs BP the rise of the sea level to 10-11m below present sea level changed the configuration of the shoreline in the Skindou embayment forming a small cove (Fig.9, Fig.10) which was protected in all weathers. This cove would have made an ideal shelter for the mariners carrying goods and moving to and fro in the area. This probably led to the establishment of the Neolithic settlement found on the island. At about 4000yrs BP the sea level rose to about 7m below present, causing the submergence of the small cove, making the former cove area no longer suitable for sheltering as the coastline was now straighter and exposed to the westerly winds. This change probably contributed to the sinking of the boat whose wreck was found on the seabed.

Figure 9. Map showing the palaeoshorelines around Dokos Island and Skindou embayment between 15100yrs BP and 6000yrs BP. The grey circle indicates the location of the wreck.

References Bar Matthews, M., and Ayalon A., 1997, Late Quaternary palaeoclimate in the Western Mediterranean region from stable isotope. Analysis of speleothems at Soreq Cave, Israel, Quaternary Research, 47, 155-168. Bond, G., Heinrich, H., Broecker, W., Labeyrie, L., McManus, J., Andrews, J., Huon, S., Jantschik, R., Clasen, S., Simet, C., Tedesco, K., Klas, M., Bonani, G., and Ivy, S., 1992, Evidence for massive discharges of icebergs into the North Atlantic ocean during the last glacial period. Nature, 360, 245-249. Bond, G.C. and Lotti, R., 1995. Iceberg discharges into the North Atlantic on millennial time scales during the last glaciation. Science, 267, 1005-1010. Bond, G., Showers, W., Cheseby, M., Lotti, R., Almasi, P., deMencal, P., Priore, P., Cullen, H., Hajdas, I., and Bonani,

Figure 10. Side scan sonar sonograph showing: (a) the paleoshoreline of 11-14m, (b) the paleoshoreline of 22-24m and (c) the Dokos wreck location. G., 1997, A pervasive millennial scale cycle in North Atlantic holocene and glacial climates, Science, 278, 1257-1266. Damuth, J.E., 1975. Echo character of the western equatorial Atlantic floor and its relationships to the dispersal and distribution of terrigenous sediments. Marine Geology, 18, 17-45. De Rijk, S., Hayes, A., and Rohling E.J., 1999, Eastern


The Reconstruction of Prehistoric Shorelines in Dokos Island, Aegean Sea, Using Remote Sensing Techniques Papathanasopoulos, G., Vichos, Y., Lolos, Y., Tsouhlos, N., Antonopoulos, F., Kritzas, H. and other members of H.I.M.A., 1989 – 1992. The project of Dokos. Enalia I-V. Perles, C., 1987a, Les industries lithiques Taillées de Franchthi. Tome I. Présentation générale et Industries Paléolithiques, in Excavations at Franchthi Cave, Greece (ed. T.W. Jacobsen), Indiana University Press, Bloomington and Indianapolis, fasc. 5. Perles, C., 1987b, Les industries lithiques Taillées de Franchthi (Argolide, Gréce), Tome II. Les Industries du Mésolithique et du Néolithique initial, in Excavations at Franchthi Cave, Greece (ed. T.W. Jacobsen), Indiana University Press, Bloomington and Indianapolis, fasc. 5. Perissoratis, C., and Conispoliatis, N., 2003, The impacts of sealevel changes during latest Pleistocene and Holocene times on the morphology of the Ionian and Aegean seas (SE Alpine Europe), Marine Geology, 196, 145-156. Rohling, E.J., Hayes, A., de Rijk, S., Kroon, D., Zachariasse, W.J. and Eisma, D., 1998, Abrupt cold spells in the northwest Mediterranean, Paleoceanography, 13, 316-322. Rossignol-Strick, M., 1995, Sea-land correlation of pollen records in the Eastern Mediterranean for the Glacial-Interglacial transition: Biostratigraphy versus radiometric time-scale, Quaternary Science Reviews, 14, 893-915. Van Andel, T.H. and Lianos, N., 1984, High Resolution seismic reflection profiles for the reconstruction of post-glacial trangressive shorelines: An example from Greece, Quaternary Research, 22, 31-45. Van Andel, T.H., and Shackleton, J.C., 1982, Late Paleolithic and Mesolithic coastlines of Greece and the Aegean, Journal of Field Archaeology, 9, 445-454. Wiedicke, M., Kudrass, H-R., and Hubscher, Ch., 1999, Oolitic beach barriers of the last Glacial sea level lowstand at the outer Bengal shelf, Marine Geology, 157, 7-18.

Mediterranean sapropel S1 interruption: an expression of the onset of climatic deterioration around 7kyrs BP, Marine Geology, 153, 337-343. Fairbanks, R.G., 1989, A 17,000 year glacio-eustatic sea level record: influence of glacial melting rates on the Younger Dryas event and deep-ocean circulation, Nature, 342, 637-642. Geraga, M., Tsaila-Monopoli, St., Ioakim, Ch., Papatheodorou, G., and Ferentinos G., 2000, An evaluation of palaeoenvironmental changes during the last 18000yrs BP in the Myrtoon Basin, SW Aegean Sea, Paleogeography, Paleoclimatography, Paleoecology, 156, 1-17. Gifford, J., 1983, Coring sampling of a Holocene marine sedimentary sequence and underlying Neolithic cultural material of Franchthi Cave, Greece, in Quaternary coastlines and Marine Archaeology (Eds. P. Maslers and N. Flemming), Academic Press. Karkanas, P., 2001. Site Formation Processes in Theopetra Cave: a record of climatic change during the Late Pleistocene and Early Holocene in Thessaly, Greece, Geoarchaeology, 16, 373-399. Karisiddaiah, S.M., Veerayya, M., Vora, K.H., 2002, Seismic and sequence stratigraphy of the central western continental margin of India: Late-Quaternary evolution, Marine Geology, 192, 335-353. Kraft, J.C., Aschenbrenner, S.E., Rapp, G.Jr., 1977, Paleogeographic reconstruction of coastal Aegean Archaeological sites, Science, 195(4282), 941-947. Lambeck, K., 1996. Sea level change and shore-line evolution in Aegean Greece since Upper Paleolithic time, Antiquity, 70, 588-611. Lobo, F.J., Hernandez-Molina, F.J., Somoza, L., and Diaz del Rio, V., 2001, The sedimentary record of the post-glacial trangression on the Gulf of Cadiz continental shelf (Southwest Spain), Marine Geology, 178, 171-195.


IDENTIFYING BURNED MUD BRICK BUILDINGS USING FLUXGATE MAGNETOMETRY M. Boyd Fitch Laboratory, British School at Athens, Greece [email protected]

N. Brodie McDonald Institute for Archaeological Research, University of Cambridge, U.K. Abstract: Geophysical techniques of archaeological investigation are well established but the interpretation of anomalies is hampered by the lack of excavated examples. This paper describes magnetic anomalies on two prehistoric sites that might be due to burnt clay surviving from mudbrick or wattle-and-daub architecture. In theory baked clay surviving from a burned mud-brick structure should carry a thermoremanent magnetisation that will be detectable with magnetic techniques. Clay was a common construction material through all periods of Aegean prehistory and numerous published accounts of buildings destroyed by fire suggest that burnt destruction deposits should be a relatively frequent occurrence in the archaeological record. Thus it is to be expected that burnt structures should be discovered routinely during magnetic surveys but because of differential patterns of building destruction and collapse their recognition might not be immediate or straightforward. This paper presents a preliminary investigation of the magnetic characteristics of burnt clay architecture. Large and unusual magnetic anomalies discovered by fluxgate gradiometer surveys at the prehistoric sites of Koufóvouno in Lakonía and Iklena in Messinía are presented and evidence derived from excavation and surface survey that supports their interpretation as burned mud-brick or wattleand-daub features is described and discussed. Περιληψη: Η αποτελεσματικότητα των γεωφυσικών τεχνικών σε αρχαιολογικές έρευνες είναι πλέον αδιαμφισβήτητη αλλά η ερμηνεία πιθανών ανωμαλιών στα δεδομένα δεν είναι επαρκής λόγω της έλλειψης περιπτώσεων όπου οι περιοχές με ανωμαλίες να έχουν ανασκαφεί. Η εργασία αυτή περιγράφει τις μαγνητικές ανωμαλίες σε δύο προϊστορικές θέσεις, οι οποίες μπορεί να προέρχονται από ψημένο πηλό που διασώζεται από πλινθόκτιστα οικοδομήματα ή δομές από κλαδιά επιχρισμένα με πηλό. Θεωρητικά, ο ψημένος πηλός που διασώζεται από πλινθοδομές αναμένεται να φέρει θερμο-παραμένουσα μαγνήτιση, η οποία θα είναι ανιχνεύσιμη με μαγνητικές τεχνικές. Ο πηλός ήταν σύνηθες οικοδομικό υλικό σε όλες τις περιόδους της Αιγαιακής προϊστορίας και ο μεγάλος αριθμός δημοσιευμένων οικοδομημάτων που έχουν καταστραφεί από πυρκαγιά, υποδηλώνει ότι οι αποθέσεις καταστροφής από πυρκαγιά θα πρέπει να συναντώνται σχετικά συχνά στις αρχαιολογικές καταγραφές. Πρέπει λοιπόν να θεωρούμε αναμενόμενη, την ανίχνευση καμένων οικοδομημάτων με ρυθμό ρουτίνας, κατά τη διάρκεια μαγνητικών διασκοπήσεων. Όμως, η αναγνώριση των καμένων οικοδομημάτων μπορεί να μην είναι άμεση ή προφανής εξαιτίας των διαφορών στον τρόπο με τον οποίο τα οικοδομήματα αυτά έχουν καταστραφεί και καταρρεύσει. Η εργασία αυτή παρουσιάζει μια πρώτη διερεύνηση των μαγνητικών χαρακτηριστικών οικοδομημάτων με πλίνθους από ψημένο πηλό. Παρουσιάζονται εδώ κάποιες μεγάλες και ασυνήθιστες μαγνητικές ανωμαλίες που εμφανίστηκαν κατά τις διασκοπήσεις με διαφορικό μαγνητόμετρο ροής, στις προϊστορικές θέσεις Κουφόβουνο Λακωνίας και Ίκλαινα Μεσσηνίας. Συζητώνται επίσης τα δεδομένα από την ανασκαφή και την επιφανειακή έρευνα, οι οποίες υποστηρίζουν την ερμηνεία των παραπάνω ανωμαλιών ως ενδείξεων πλινθοδομών από ψημένους πλίνθους ή κλαδιά επιχρισμένα με πηλό.


whether as mudbrick, pisé or wattle-and-daub. Excavation accounts suggest the widespread use of these techniques and point to their common preservation by burning during the destruction of a building or site. A geophysical technique that could detect and map burned clay buildings would go a long way toward establishing a standardised approach to prehistoric sites.

The use of geophysical prospection in Greek archaeological contexts is now well established. Major surveys over the past 20 years, such as those at Stymphalos (Williams 1985, Papamarinopoulos et al 1988), Evropos (Tsokas et al 1993) and Philippi (Boyd & Provost 2001, Provost & Boyd 2002), have shown the potential of the technique in urban environments, while much experimental work has taken place in relation to other types of site. It remains the case, however, that standardised approaches to geophysical interpretation, based on the kinds of archaeology and geology present in Greece, are not well developed. In this context, this paper considers the problems associated with the detection and identification of burned clay-built structures.

In theory magnetic techniques should be well suited to the detection of these features. Iron minerals demagnetise when their Curie points are reached (565°C for magnetite, 675°C for haematite) and upon cooling acquire a thermoremanent magnetism in alignment with the earth’s field. Clay bakes at temperatures probably in excess of the Curie points. It is generally reported that temperatures in the range 800–900°C are needed to convert clay into a durable ceramic (Rice 1987: 86) and the survival after burning of inscribed clay tablets from several Bronze Age palaces shows that temperatures at least as high as

For many periods in Greece, especially the prehistoric, one very common building technique involved the use of clay,


M. Boyd and N. Brodie this could be reached in burned destructions, as does the appearance of baked daub or mudbrick on other prehistoric sites (including Koufóvouno, which is reported here). This confirms that temperatures reached by destructive fires could often exceed Curie points. Thus, in theory, claybased architectural features should be relatively easy to detect magnetically when they have been destroyed by fire, as they should present a strong magnetic anomaly. The magnetisation of fired clay is of course the foundation of the magnetometer’s success in detecting small industrial installations such as kilns and furnaces (Clark 1990: 17).

is, what does the magnetic signature of a burned mudbrick building look like? The large magnetic anomaly attributed to a destruction event at Evropos illustrates the type of feature that might be detected (Tsokas et al 1993), while at Çatal Höyök in Central Anatolia some magnetic anomalies were later shown to be associated with burned destruction deposits (Shell 1996). The two sites discussed in this paper may also prove to include good examples of burned mudbrick structures.

However, in practice, things are not so simple. Destruction and post-destruction factors introduce numerous qualifying effects. There are two processes of which we have to be aware. First, there is the nature of the destruction episode itself. Presumably, a burned structure will collapse in whole or in part, and it is the collapse deposit that will acquire the thermoremanent magnetism. Thus the magnetic feature detected by gradiometer survey may not be regular and may not be recognisably architectural.

The surveys were conducted using Geoscan FM-36 gradiometers. At Koufóvouno readings were taken at 1m intervals in 20m x 20m grids; at Iklena, most readings were taken at 0.5m intervals using an automatic sample trigger. Data have been corrected for diurnal drift and the interpolated images presented here were produced by MapInfo and Vertical Mapper software. No further filtering or alteration has been carried out on the data. Koufóvouno

Second, there is what would normally be called postdepositional disturbance. The collapsed building is the pristine archaeological deposit with a coherent magnetisation, but human interference during and especially after burning will disturb it to a greater or lesser extent, and randomise the direction of magnetisation. Various scenarios can be imagined: the previous occupants of a burned house may return and dig around looking for possessions; a burned structure might be scavenged for any useable building materials that may have survived intact; material might be moved aside or levelled in advance of rebuilding; or a burned site might be degraded long after its first destruction by ongoing agricultural operations. The outcome of any of these activities or – more likely – a combination may well be a thoroughly mixed context where the aligning effect of the burning is seriously compromised by the destroyed integrity of the deposit in question, although the presence of burnt clay should nevertheless ensure an increased magnetic susceptibility (Weston 2002: 208). Post-depositional processes will also most likely bury the site to a greater or lesser extent, which will attenuate the magnetic signal to a variable degree.

The site of Koufóvouno is about two kilometres south of Sparta in Lakonía. It is primarily a middle Neolithic–Early Bronze Age site and is currently being excavated by a British-French team under the joint directorship of Bill Cavanagh, Chris Mee, and Josette Renard. Neil Brodie carried out a geophysical survey there in 1999 during surface investigations in advance of excavation. On the eastern and northern areas of the site vertical iron rods supporting young olive trees or irrigation sprinklers gave rise to a number of localised positive anomalies which appear on figure 1 as white, circular features. There are four main archaeological anomalies (figure 1, 1-4). The locations of 1 (about -15nT to 25nT), 2 (about -40nT to 60nT) and – to a lesser extent – 4 (about -25nT to 45nT) are marked by high surface concentrations of Neolithic pottery. Pieces of likely baked clay daub were picked up during surface collection in the areas of 1 and 2 (marked on figure 1 as X) and in 2001–2 excavation at 1 (trench C on figure 1) uncovered the remains of a burned Neolithic house. This house has yet to be published by the excavators and so cannot be discussed here in any detail. Its very existence, though, is significant. Trenches have not been opened over anomalies 2 and 4, although a small trench opened over anomaly 3 showed it to be a Roman structure, which was not fully investigated. It is always possible of course that the Roman structure was itself burned, or that it was built over an area of extensive Neolithic or early Bronze Age burning. Future excavation there might tell. Negative evidence is important too. Four trenches (figure 1: A, B, D and G) were opened in areas with no obvious magnetic anomalies, and no burned deposits were found.

This is not to suggest that burned mud brick is unlikely to be detectable by magnetic prospection. It does suggest, however, that clear architectural plans of walls, as often found with resistance prospection of urban sites, are unlikely to be detectable by magnetic means. Sarris & Jones (2000: 22), in their recent survey of geophysics in the Mediterranean, mention burned buildings along with hearths, kilns and ovens, as easily detectable in magnetic survey, yet few of the surveys they catalogue targeted or located burned buildings and it is clear that they have been much less commonly detected unambiguously in the way that kilns, for example, often are. The question therefore

In view of these results, both positive and negative, it seems reasonable to suggest that anomalies 2 and 4 also 74

Identifying Burned Mud Brick Buildings using Fluxgate Magnetometry area of 3ha. Figure 2 shows the complete magnetic survey. Principal anomalies to be discussed here are located in the eastern zone, which is on an almost flat area of ridge-top. The area to the west, showing fewer magnetic anomalies, is somewhat downslope. Both areas have produced plentiful late Bronze Age sherd material in intensive pick-up. The working hypothesis of the survey is that the site at Iklena is one of the ‘second-order’ major late Bronze Age sites listed in the Pílos tablets; this leads to an expectation of some kind of non-domestic centre at the site. The area to the south is magnetically quiet, with readings ranging from -5nT to 5nT, probably indicating the edge of the site-centre. To the west, readings are more variable, generally ranging within -10nT to 10nT. While certain coherent features are detectable on this western side, such as the double linear anomaly to the north, at least 90m in length, for the most part large features are absent here. In the centre linear features can be clearly associated with buried architectural features. This paper is concerned with results on the eastern side (figure 3). Readings here range from about -20nT to about 30nT, and several coherent features can be discerned. Most obvious is the positive linear anomaly (A on figure 3) running north-south for about 50m and associated with another large anomaly at the northern end of this area (B on figure 3): here a right angle defines the edge of an area at least 30m x 15m and continuing beyond the survey area, a positive anomaly with its edges defined by the highest readings. The right angle can leave little doubt that this feature is anthropogenic.

Figure 1. Fluxgate gradiometer survey at Koufóvouno, Lakonía

The eastern side is dominated, however, by an area about 68m x 45m (C on figure 3), somewhat irregularly shaped, which is again characterised by mostly positive readings, but with a surrounding halo of negative readings. The remarkable internal consistency of this large anomaly has led us to the conclusion that it is caused by a coherent sub-surface phenomenon: the most likely candidate would seem to be the burned destruction of a large building or complex resulting in a relatively homogenous large burned clay deposit.

Figure 2. Fluxgate gradiometer survey at Iklena, Messinía.

mark the position of burned structures. When the site of Koufóvouno is fully excavated, it will become possible to compare the depth and constitution of the burned house with the intensity and configuration of its associated magnetic anomaly. Hopefully, trenches might be opened over other anomalies too. Iklena The site of Iklena is located about 10km south of the ‘Palace of Nestor’ or Bronze Age administrative and storage complex on the west coast of Messinía, in the southwest of the Peloponnese. The site had been known since small-scale trial excavations in the 1950s (Marinátos 1954: 308-310) and is currently the subject of an intensive field survey programme, directed by Professor Michael Cosmopoulos. Geophysical survey is an integral part of this programme and the data presented today were gathered in 2000-2003. The centre of the site, covering an area of 1.6ha, has been surveyed using magnetic techniques, while resistance survey has also been carried out on a larger total

Figure 3. Central anomaly in fluxgate gradiometer survey at Iklena, Messinía.


M. Boyd and N. Brodie Discussion and conclusion

aim, however, to highlight these problems. As excavations at Koufóvouno continue, the depth and extent of burned structures there will be revealed and it is to be hoped that future excavations at Iklena will provide detailed information on the hypothesised burned structures there; this will allow a more informed interpretation of the gradiometer data which will be of use as a comparandum for future surveys on comparable Neolithic and Bronze Age sites. Ultimately it may well be possible to define magnetic signatures for burned mudbrick deposits.

We therefore have several candidates for magnetic anomalies associated with burned destruction at two sites. In the case of Koufóvouno, the correlation has been confirmed by excavation over one of the anomalies, as well as generally by the collection of burned construction material in field survey. At Iklena, although excavations have yet to take place, a significant amount of the early and late Mycenaean pottery collected in field survey appears to have been burned (Shelmerdine pers. comm.).

References Anomaly 1 at Koufóvouno, which on excavation was confirmed as the site of a burned destruction, covers an area of about 18m x 13m. Most positive readings in this area average around 10nT, while there is a surrounding halo of negative readings at about -10nT. A similar, but much less clear-cut, pattern is apparent with anomaly 2. Anomalies 3 & 4 exhibit a much less coherent structure, although with the eye of faith they might be said to fit the same pattern. At Iklena, although the main anomaly is much larger, it exhibits similar characteristics: a positive core anomaly surrounded by a negative halo. The edge of the core anomaly at Iklena often exhibits the highest readings, a pattern repeated with the anomaly to the north, and there is just a hint that a similar situation is present at the first Koufóvouno anomaly.

Boyd, M. J. and Provost, S., 2001, Application de la prospection géophysique à la topographie urbaine I. Philippes, les quartiers Sud-Ouest, Bulletin de Correspondance Hellénique, 125, 453-521. Clark, A. J., 1990, Seeing Beneath the Soil Prospecting methods in archaeology, London: Batsford. Marinátos, S. N., 1954, Ἀνασκαφαὶ ἐν Πύλ ύλλῳ, Πρακτικὰ τῆς ἐν Ἀθήναις Ἀρχαιολογικῆς Ἑταιρείας,, 109, 299-316. Papamarinopoulos, S. P., Jones, R. E. and Williams, H., 1988, Electric resistance survey of the southern part of the buried ancient town of Stymphalos, Geoexploration, 25, 255-61. Provost, S. and Boyd, M. J., 2002, Application de la

prospection géophysique à la topographie urbaine II. Philippes, les quartiers Ouest, Bulletin de Correspondance Hellénique, 126, 431-488.

Rice, P., 1987, Pottery Analysis: A Sourcebook, Chicago: University of Chicago Press. Shell, C.A., 1996, The magnetometric survey at Çatalhöyük east, In On the Surface: Çatalhöyük 1993–5 (I. Hodder ed.), 101–13, Cambridge: McDonald Institute. Sarris, A. and Jones, R. E., 2000, Geophysical and Related Techniques Applied to Archaeological Survey in the Mediterranean: A Review, Journal of Mediterranean Archaeology, 13(1), 1), ), 3-75.. Tsokas, G.N., Giannopoulos, A., Tsourlos, P., Vargemzis, G., Tealby, J.M., Sarris, A., Papazachos, C.B. and T. Savopoulou, 1994, A large scale geophysical survey at the archaeological site of Europos (northern Greece), Journal of Applied Geophysics, 32, 85–98. Weston, D.G., 2002, Soil and susceptibility: aspects of thermallyinduced magnetism within the dynamic pedological system, Archaeological Prospection, 9, 207-15. Williams, H., 1985, Investigations at Stymphalos, 1984, Echos du Monde Classique Classical Views XXIX n. s. 4, 215-224.

The Koufóvouno data do not hint at architectural details, suggesting that collapse and post-depositional factors have led to a mass of undifferentiated, magnetically-active material below the surface. At Iklena, the situation is rather different. While overall the anomaly is a large, shapeless area, internal details on the southwest side are clear. Here at least one right angle and other sharp jogs seem to suggest that some architectural features are preserved in the subsoil. This paper does not pretend to offer a definitive account of the magnetic signatures of burned destruction deposits and the problems associated with their detection. It does


APPLICATION OF NEAR-SURFACE GEOPHYSICAL TOOLS AND GIS FOR MAPPING THE ANCIENT CITY OF LEFKAS A. Sarris, S. Topouzi, F. Triantafyllidis, S. Soetens Laboratory of Geophysical - Satellite Remote Sensing & Archaeo-environment, Institute of Mediterranean Studies - Foundation of Research & Technology (F.O.R.T.H.), Melissinou & Nik. Foka 130, PO. Box 119, Rethymnon 74100, Crete, Greece, [email protected]

G. Pliakou 12th Eforia of Prehistoric and Classical Antiquities, Plateia 25th March 6, Ioannina 45221, Greece Abstract: Whether it was called “Nirikos” or “Lefkas”, the area opposite the Acarnanian coast was inhabited from the early times. At this strategic point the Corinthians founded their colony, “Lefkas”, during the 7th century B.C. Excavations initiated in 1901 by the German Archaeological Institute revealed parts of the ancient wall and the theatre. Rescue excavations, contracted by the Greek Archaeological Service during the last two decades, enabled the partial reconstruction of the classical city’s topography. The city walls are well preserved in the northern and western part, while some eastern parts are visible near the coast. Problematic,, however, is the direction of the southern part of the wall. Geophysical investigations were carried out in two phases, employing vertical magnetic gradient and soil resistance techniques. Mapping was focused in the southern limits of the ancient city. Geophysical data was able to identify a number of characteristics of the ancient city plan. The urban system was verified, consisting of parallel and vertical streets forming large building blocks. Drainage pipes were found to be running to the sides of the roads. Further to the south, the density of architectural remnants decreases, suggesting a potential location for the southern wall of the city. Similarly, a crossroad found in the SE corner of the surveyed region could be projected to lead towards the cemetery to the west and towards the port to the south. Aerial images of the area were registered to the topographic map and enhanced using image processing techniques. A similar methodology was followed for the processing of hyper-spectral satellite imagery (ASTER). All geographical data was imported to a GIS, in which the different geophysical layers were overlaid. Interpretation of the geophysical anomalies (in vector format), together with the resulting images, can provide supplementary information and be used for conservation and development planning. Περίληψη: Η νήσος Λευκάδα βρίσκεται απέναντι από τις ακτές της Ακαρνανίας και κατοικήθηκε από την Νεολιθική Εποχή. Από τον 7ο αι. π.Χ., η Λευκάς ή Νήρικος αποτέλεσε μία σημαντική αποικία των Κορινθίων. Οι πρώτες ανασκαφές που διενεργήθηκαν από την Γερμανική Αρχαιολογική Σχολή έφεραν στο φως τμήματα των αρχαίων τειχών και θεάτρου. Τις τελευταίες 2 δεκαετίες, σωστικές ανασκαφές από την Αρχαιολογική Υπηρεσία έχουν προσφέρει πολύτιμες πληροφορίες σχετικά με το πολεοδομικό σχέδιο της πόλης. Τα τείχη της πόλης βρίσκονται σε καλή κατάσταση διατήρησης προς τα βόρεια και δυτικά, ενώ κάποια τμήματα των τειχών είναι ορατά προς τα ανατολικά. Οι γεωφυσικές διασκοπήσεις που διενεργήθηκαν σε 2 φάσεις κάνοντας χρήση μαγνητικών και ηλεκτρικών τεχνικών εστιάστηκαν στο νότιο τμήμα της πόλης. Οι μετρήσεις είχαν ως αποτέλεσμα τον εντοπισμό παράλληλων και κάθετων δρόμων που σχηματίζουν μεγάλα οικοδομικά τετράγωνα. Σε ορισμένα σημεία, αγωγοί αποχέτευσης φαίνεται να βρίσκονται δίπλα στους δρόμους. Στη νότια πλευρά της περιοχής, η κατανομή των υπεδάφειων στόχων μειώνεται δραστικά, ενώ υποδεικνύεται η πιθανή θέση της νότιας πλευράς των τειχών. Τέλος, μία διασταύρωση στο ΝΑ άκρο της περιοχής φαίνεται να οδηγεί προς το αρχαίο νεκροταφείο (προς τα δυτικά) και προς το αρχαίο λιμάνι της πόλης (προς τα νότια). Η δημιουργία μωσαϊκού από ανορθωμένες αεροφωτογραφίες διαφορετικών περιόδων λήψης, σε συνδυασμό με την επεξεργασία δορυφορικών εικόνων ASTER συνείσφερε στην ανακατασκευή της μορφολογίας του τοπίου. Τα παραπάνω στοιχεία, σε συνδυασμό με την υπέρθεση των αποτελεσμάτων των γεωφυσικών διασκοπήσεων σε περιβάλλον των Γεωγραφικών Συστημάτων Πληροφοριών (GIS), GIS), ), βοήθησαν ουσιαστικά στην εξαγωγή συμπερασμάτων σχετικά με τον πολεοδομικό ιστό της πόλης και την οριοθέτηση των υπεδάφειων μνημείων, ενώ αναμένεται να χρησιμοποιηθούν για την προστασία και ανάδειξη αυτών σε σχέση με τα αναπτυξιακά σχέδια της περιοχής.


The city extends between the modern settlements of Kalligoni to the North and Karyotes to the South. The western part of the city follows the natural relief of the gentle slopes of hill Koulmos, while the eastern part extends to the offshore plain. The east and west sections of the city are divided by the modern provincial road leading from Lefkada to Nydri.

The ancient city of Lefkas was founded in the end of the 7th century. B.C. as a Corinthian colony. The city was located on the north-eastern part of the island of Lefkas, at a distance of about 2.5km from the modern capital of the island (Figure 1). In this way, the port of the city was able to control the narrow sea passage between the northeastern coastline of the island and the Acarnanian coast, and thus the navigation towards the Ionian and the Adriatic Sea (Murray 1982).

The area was not re-occupied after the depopulation of the city in the early 1st century A.D. During the Venetian period land use was limited to olive groves. The 77

A. Sarris, S. Topouzi, F. Triantafyllidis, S. Soetens and G. Pliakou

Figure 1: (Left) Map indicating the location of ancient Lefkas at the north-eastern part of the island. (Right) A section of the settings of the ancient city overlooking the Akarnanian coast.

development of the area, which begun in the 1960s, has been rather intensive during the last few years due to the rapid tourist development of the island. The city walls are preserved today at the north and southwest parts of the city at a height of 2-2.5m. Their estimated perimeter is 4.5km.

to georeference the rest of the aerial photos, provided by the Hellenic Ministry for the Environment, Physical Planning and Public Works (Hellenic Mapping and Cadastral Organization). These 1:6000 scale photos, which were obtained at different dates (16/05/1985 and 12/10/1994), and were scanned in high resolution so as to achieve the optimum result in the quality of every image. These photos were also processed through histogram equalization and geometric registration in order to form a time sequence of aerial mosaics. Similarly, the proposed expanded town plan of the area was employed in order to see the degree of correlation with the geophysical anomalies. A number of thematic maps were created, including the road network, private properties, proposed protection zones, the layout of the geophysical grids, a.o.

Although architectural remnants (from buildings, terraces, roads and defensive wall) of the ancient city are visible at the slopes of Koulmos, those situated on the plain have been covered by shallow deposits. Thus, specific construction constraints were imposed in the area aiming towards the protection of the monuments of the ancient city. Following the same directives and aiming towards a better outline of the architectural relics, it was decided to carry out an extensive shallow-depth geophysical prospection survey which could contribute in mapping the layout of the ancient city in relation to the new design of the urban planning of the area.

Finally, the GIS system included thematic layers based on the available archaeological data, such as digital excavation plans.

Design of the geographical information system Past archaeological research The development of a Geographical Information System was considered essential for the better representation of the results of the geophysical prospection survey in accordance with the new urban planning design. The specific Geographical Information System contains data in both vector and raster format, transformed in the same geodetic reference system (Hellenic Geodetic Reference System, EGSA ’87).

In recent years, information about the relics of the ancient city of Lefkas has been provided by foreign travelers (Leake, Holland, Pouqueville, Goodisson, Dodwell) who visited the island during the 19th century. The German Archaeological Institute was responsible for initiating the systematic excavations in the region of the ancient city during the beginning of the 20th century. Under the direction of W. Dörpfeld, excavation trenches were opened and remnants of the walls and the ancient theatre were revealed (Dörpfeld 1927). In the 1960s, archaeological research was re-initiated in the form of rescue excavations conducted by the local Archaeological Service, uncovering parts of the city’s cemeteries, roads and private buildings and contributing to the reconstruction of the urban planning of ancient Lefkas (Fiedler 1999, Douzougli, 2001, Pliakou 2001, Andreou 2002).

The digital terrain model (in Triangulated Irregular Network / TIN format) was constructed by 3D topographic data, which was transformed from ASCII format to dbf files and then converted to shape-files containing elevation information for the 20m iso-lines. The area coverage of the Digital Terrain Model (DTM) included the extension from the modern city of Lefkada to the modern settlement of Lygia. Three geographically registered ortho-rectified maps (date: 1998, scale 1:5000), provided by the Ministry of Agriculture and covering the same area as the TIN model, were joined creating a photo-mosaic (Figure 2). The mosaic was used

According to the updated results of the archaeological excavations, the city seems to be divided in two parts: one on the hills of Koulmos to the west and one in the plain 78

Application of Near-surface Geophysical Tools and Gis for Mapping the Ancient City of Lefkas


Figure 2: Digital Terrain Model (DTM) and overlay of the aerial orthophoto maps.



A. Sarris, S. Topouzi, F. Triantafyllidis, S. Soetens and G. Pliakou

Figure 3: Aerial image of the wider region of interest and overlay of the outline of the geophysical grids, accompanied by their corresponding codes.

towards the coast to the east. The part that extends on the hills, where (possibly) ritual buildings and the remains of the ancient theatre have been found (Pliakou 1997, Pliakou 2001), has no specific urban planning, while the part at the plain was built according to an orthogonal system, very similar to the Hippodamian one. According to the information retrieved by different excavation trenches, the parallel streets, with average width 4.5m (Agallopoulou, 1971, Kalligas, 1972, Andreou, 1979, 2002), run across the city every 30m in an east to west direction, that is, from the hills to the sea. These streets intersect vertically with others of about 5.5-6m in width (Andreou, 1977, 1984, 2002), creating building plots with their long direction along the east-west axis. The study of the excavation results showed that two private houses, divided by a draining pipe, exist in every building plot (Douzougli, 1993, Pliakou, 1999). Other draining pipes run along the ancient streets (Agallopoulou, 1971, Kalligas, 1972, Andreou, 1979).

Amvrakia (also a Corinthian colony), which is in the area of modern Arta (Andreou, 1993). It is worth mentioning that the orthogonal city plan system in ancient Lefkada was formed almost two centuries before the creation of the Hippodamian system. Due to the increased pressure imposed by the recent expansion of construction and building works in the area, the archaeological service decided the protection of the area imposing certain restrictions in the building activities. At the same time, the use of geophysical investigations was considered of crucial importance, in order to provide further information regarding the degree of conservation of the architectural relics and their density in the southern section of the city. Thus geophysical investigations, including magnetic and soil resistance measurements, were employed in a total area of about 80,000 square meters at the SE section of the ancient city. The specific region is surrounded by visible architectural parts, while it also included excavated

The aforementioned conclusions have been compared with other known examples in the wider area, like the city of 80

Application of Near-surface Geophysical Tools and Gis for Mapping the Ancient City of Lefkas


sections and trenches at the northern and western sections. One of the goals of the research plan was to locate the outline of the southern section of the ancient wall, which has remained without any significant traces, and which would contribute in defining the limits of the ancient city.

the geological trends and diminish the external noise. Measurements had an accuracy of 0.1nT/m. In the measurements of the soil resistance, a Geoscan resistivity meter RM15 with a Twin Probe electrode configuration was used with a spacing of 0.5m between the mobile electrodes (Clark, 1990). Emphasis was given to the wide coverage of the site with a sampling interval of 1m, while high-resolution measurements were carried out in a limited number of grids of specific importance.

Geophysical prospection techniques The geophysical survey of the archaeological site was carried out in two phases (April 13-23 and June 16-27, 2002). In the first phase of the campaign, both magnetic and soil resistance techniques were employed in selected grids in order to test the registered signals originating by the subsurface targets. The second phase of the fieldwork activities was devoted to the extended coverage of the site through the use of magnetic techniques.

Returning to the design of the GIS used for the management of the results of the geophysical research, the outlines of the geophysical grids were also added to the system as a different layer (Figure 3). The location of the geophysical grids was defined through the use of a total station, a work which was carried out by the local archaeological service. Based on these measurements, the final geophysical maps were geometrically corrected and registered to the system as B&W or color images in a

A Geoscan fluxgate gradiometer (FM36) was employed for the measurements of the vertical magnetic gradient. The gradiometer readings were able to smooth away 81

A. Sarris, S. Topouzi, F. Triantafyllidis, S. Soetens and G. Pliakou

Figure 5: (Left) Results of the magnetic measurements in the area coverage of grids C, B, A, U, X & Z. The dynamic range lies from –45 to 35 nT/m. (Right) Comparison between resistivity and magnetic data for grid X.

.jpg format (Figure 4). The processing and manipulation of the geophysical measurements included de-spiking of extreme values, grid and line equalization techniques, compression of the dynamic range of the initial values, application of directional filters in terms of the estimation of the first derivatives, application of high pass and low pass filters, shading relief techniques and construction of 3 dimensional images. The magnetic measurements in grids C, B, Α & U (Fig.3), east of the provincial road from Lefkada to Nydri, indicated the existence of a network of parallel streets, at a SWWNEE direction, lying at a distance of 35m apart (Figure 5). They have a width of about 3.75-4.5m and appear as high anomalies lying within the range of 5-15nT/m above the background magnetic level. The dipole character of the anomalies suggests the existence of a drainage pipe running at the northern part of the roads. A road, perpendicular to the rest, appears at the northern part of grid B (see Fig. 4). Similar evidence comes from grids X and Z, suggesting that the town plan was consisting of a dense network of parallel and perpendicular roads, which divided the ancient city in building blocks.

Figure 6: Details of the results of the magnetic measurements at the SE section of the surveyed region, where a perpendicular road intersection was located. One of the roads is leading to the ancient cemetery (to the west) and the other (to the south) towards the ancient port of Lygia.

Grids X and Z are of particular interest, since there are clear indications of architectural relics within the building blocks in both vertical magnetic gradient and soil resistance data (Figure 5). The outline of the magnetic anomalies suggests the good preservation condition of the particular structural remains. The roads appear as high magnetic anomalies within the ~3-15nT/m range and as high resistance anomalies within the ~8-15 Ohms range. The parallel road system seems to fade out towards the east. A vertical linear anomaly, which appears at the eastern side

of the particular grids, seems to be of a different nature, probably consisting of two sections, and can be considered as a candidate target for the location of the wall surrounding the town. The rest of the grids (VOR1, VOR2, TH1, TH2, TH3) towards the north-east indicate a loose clustering of architectural remnants, which extend to the coast-line. In contrast, areas which were surveyed in the western side of the road connecting Lefkas to Nydri did not indicate any significant anomalies. However, we need to consider that 82

Application of Near-surface Geophysical Tools and Gis for Mapping the Ancient City of Lefkas

Figure 7: Diagrammatic interpretation of the geophysical anomalies and their overlay onto the aerial image of the wider region.

none of the above grids was located at the projection of the ancient road network which was identified in the east section of the surveyed area.

Eastern coast of Lefkada with the Akarnanian coast, where the fortress of Agios Georgios-Plagia is located (Murray 1982). The mole was constructed to protect the southern entrance of the canal, was cut by the Corinthians, and it is probably dated to the same period, namely about 600 B.C. Similarly, the SW-NE road seems to lead to the ancient cemetery, which has been located to the west of the provincial road.

The area to the south of the SWW-NEE dirt-road, which divides the area of interest in north and south sections, exhibits a lower interest, as it is indicated by the decreasing density of geophysical features. The area of grids R1-R5 exhibits an elevation difference of about 3-4m with respect to the dirt road and the north section of the surveyed region. In the projection of the dirt road to the west, large stone blocks may be correlated to the ancient city wall. Similar evidence exists along the dirt road, reinforcing the hypothesis that at least a large section of the city wall is located along its length.

Final remarks Through the use of geophysical survey and GIS techniques it was possible to collect information which can be used to clarify the city plan of the ancient city of Lefkas. A detailed plan of subsurface anomalies was identified enhancing the information context of the up to date excavations and offering an integrated image of the southern section of the ancient city.

Finally, additional evidence of the extension of the city plan was provided by grids Ν1 and Ν2 to the SE (Figure 6). In grid N1, a cross-section is aligned in agreement to the street orientation at the north section of the surveyed region. The NW-SE road seems to be at the projection of the road registered at the NE edge of grid B, while there are also indications from grid N2 that the road continues further to the south. Due to its proximity to the coast, it is very probable that the road may lead to Lygia, where the ancient port facilities of the city were located. More specifically, in Lygia there is a mole used to connect the

A road network, structural relics, drainage pipes, kilns, and wall remains are included among the candidate targets that were registered by the soil resistance and magnetic techniques. The results of the geophysical survey are in perfect agreement to the surface monuments and the excavation results, completing to a large degree the fragmentary image of the ancient city.


A. Sarris, S. Topouzi, F. Triantafyllidis, S. Soetens and G. Pliakou

Figure 8: Overlay of the photomosaic of the region to the DTM and a 3D model of the proposed outline of the ancient wall.

More specifically, the city plan seems to extend to the north of the area of interest (namely, north of the SWWNEE dirt-road and east of the provincial road) (Figure 7). In contrast, the south section of the area exhibits a lower density of architectural remains and an increasing number of candidate kiln structures (anomalies at grids I and K), suggesting workshop activities. The above observations, together with the surface monuments along the dirt-road and the elevation difference between the north and south sections in the other sides of the dirt road, suggest that part of the southern city wall is probably running along the dirt-road.

5HIHUHQFHV Agallopoulou, P., 1971, Archaeologikon Deltion, 26, Β΄: 357. Andreou, I., 1977, Archaeologikon Deltion, 32, Β1: 155 (D. Lazaris plot). Andreou, I., 1979, Archaeologikon Deltion, 34, Β1: 269. Andreou, I., 1984, Archaeologikon Deltion, 39, Β΄: 187-188 (Ar. Lazaris plot). Andreou, I., 1993, Ambracie, une ville se reconstitue peu à peu par les recherches, Colloque International de ClermontFerrand (25-27 Octobre 1990) “L’ Illyrie méridionale et le Épire dans l’ antiquité II”: 91-101. Paris. Andreou, I., 2002, Poleodomika Stoixeia tis Arxaias Lefkados, Archaeologikon Deltion, 35 (1998), Α΄ Meletes: 148-185. Clark, A., 1990, Seeing Beneath the Soil, England, B. T. Batsford Ltd. Dörpfled, W., 1927, Alt –Ithaka. Ein Beitrag zur Homer-Frage, 156-57. Douzougli, A., 1993, Archaeologikon Deltion, 48, Β1: 285-187 and 293-300. Douzougli, Α., 2001. Nekrotafeia arhaias Lefkadas, symposium “The capital of Lefkada”, 45-84. Lefkada, August 1999. Fiedler, M., 1999, Leukas, Wohn und Altagskultur in einer nordwestgriechischen Stadt, in Geschichte des Wohnens in 5000 v. Chr. – 500 n. Chr. Vorgeschichte Fruhgeschichte – Antike (ed. W. Hoepfner), 412-416, Stuttgart.. Kalligas, P., 1972, Archaeologikon Deltion, 27, Β2: 486. Murray, W.M., 1982, The coastal sites of western Akarnania, 226-239 and 243-250. Phd., University of Pensylvania. Pliakou, G., 1997, Nea stoiheia gia to arhaio theatro tis Lefkadas. Mia topographiki prossegissi, Hpeirotika Chronika, 32, 3742. Pliakou, G., 1999, Archaeologikon Deltion, 54, (1999), Chronika, to be published. Pliakou, G., 2001, I archaia poli Lefkas : to asti kai i evriteri periochi tou, symposium “The capital of Lefkada”, 30-34. Lefkada August 1999.

The design of the Geographic Information System was extremely helpful in registering the different information layers and correlating the results of the geophysical survey with the rest cartographic material (Figure 7), including the topographic maps and the aerial mosaics. In this way, it was possible to have an accurate picture of the distribution of the geophysical anomalies and their correlation with the surface relics, while the time sequence of the aerial imagery contributed to the analysis of the evolution of the landscape. A three-dimensional representation of the research area combining the digital elevation model, draped with an aerial photo mosaic and the architectural remnants, found in the previous excavations and geophysical survey has been also constructed (Figure 8). The above approach will contribute in the management of the archaeological relics of ancient Lefkas.


GEOPHYSICAL RESEARCH AT SYNTAGMA SQUARE, ATHENS – RELATIONSHIP TO GEOLOGY AND ANCIENT TEXTS M.G. Papaioannou Sev. Kallisperi 32, 152 34 Chalandri, Athens, Greece, [email protected]

S. P. Papamarinopoulos and P. Stefanopoulos University of Patras, Department of Geology, Laboratory of Geophysics, 261 00 Rio-Patras Abstract: The Geophysical research that took place between 1992-93 at the center of the city of Athens and more precisely in the area of Syntagma, brought to light new archaeological and geological evidence. The use of geophysical methods in an urban environment has been “revolutionary”, fraught with many problems but also producing significant results. This research began during the construction of the Athens Metro and initially aimed at facilitating the construction of Syntagma Station. However, very important geological and archaeological evidence came to light as the investigations progressed. The utilization of georadar, at several frequencies, helped determine the geological and archaeological deposits at various depths, becoming a useful tool in the development of the works. The discovery of the subterranean river, Eridanos, and the subsequent mapping of its course, caused us to consult the ancient texts to investigate the relationship of known and unknown buildings to the course of the river. During excavations on Amalias Avenue, it became apparent that much of the archaeological evidence indicated past human activities directly related to the existence of the river in the vicinity. Another important geological discovery was that of a syncline in the Syntagma Square area. The geology combined with the existence of the river and the syncline feature, can be related to “cosmogonist situations” and with adverse phenomena like floods and earthquakes as described in Plato’s texts and in particular in “Timaeos and Critias” (Bury 1989). Several other writers mention the river Eridanos and relate it to many known and unknown features of the ancient city of Athens. Drillings and excavations in the area verify the geophysical discoveries, proving once again the effectiveness of geophysical methodology in underground surveys, even in an urban environment. Περιληψη: Η γεωφυσική έρευνα που πραγματοποιήθηκε την περίοδο 1992-93 στο κέντρο της πόλης των Αθηνών και συγκεκριμένα στην περιοχή του Συντάγματος, έφερε στο φως νέα αρχαιολογικά και γεωλογικά στοιχεία. Η χρήση των γεωφυσικών μεθόδων σε αστικό περιβάλλον υπήρξε «επαναστατική», με πολλά προβλήματα αλλά και σημαντικά αποτελέσματα. Η έρευνα αυτή έλαβε χώρα στα πλαίσια των εργασιών για την κατασκευή του Μετρό των Αθηνών και είχε αρχικά συγκεκριμένους στόχους ώστε να προγραμματιστούν τα βήματα των εργασιών κατασκευής του σταθμού Συντάγματος και να διευκολυνθεί η πορεία τους. Η συνέχεια όμως υπήρξε εξαιρετικά ενδιαφέρουσα τόσο από αρχαιολογικής όσο και από γεωλογικής άποψης. Η χρήση γεωραντάρ διαφόρων συχνοτήτων οδήγησε στην αποτύπωση της γεωλογικής και αρχαιολογικής πραγματικότητας σε διάφορα βάθη, αποτελώντας σημαντικό βοήθημα στην εξέλιξη του έργου. Η ανακάλυψη του ποταμού Ηριδανού και η χαρτογράφηση της πορείας του ήταν ένα από τα πρώτα στοιχεία που έστρεψαν το ενδιαφέρον στα αρχαία κείμενα και στη συσχέτιση μέσω αυτών, γνωστών και άγνωστων κτιρίων με την πορεία του ποταμού. Πολλές από τις ανθρώπινες δραστηριότητες που ήρθαν στο φως μετά την ανασκαφή που πραγματοποιήθηκε στη Λεωφόρο Αμαλίας, συνδέονται άμεσα με την ύπαρξη του ποταμού στη θέση αυτή. Ένα δεύτερο και εκ πρώτης όψεως καθαρά γεωλογικό στοιχείο, υπήρξε η παρουσία ενός συγκλίνου στην περιοχή της πλατείας Συντάγματος. Η γεωλογία της περιοχής σε συνδυασμό με την ύπαρξη του ποταμού και της συγκλινικής αυτής μορφής, μπορούν να συσχετιστούν με «κοσμογονικές καταστάσεις» και ακραία φαινόμενα όπως κατακλυσμοί και σεισμοί που περιγράφονται στα κείμενα του Πλάτωνα και ιδιαίτερα στο «Τίμαιος και Κριτίας». Πλήθος άλλων συγγραφέων αναφέρουν τον ποταμό Ηριδανό και τον σχετίζουν με πολλά γνωστά και άγνωστα σημεία της πόλης της αρχαίας Αθήνας. Γεωτρήσεις και ανασκαφές στην περιοχή επιβεβαιώνουν τα γεωφυσικά στοιχεία, αποδεικνύοντας για άλλη μια φορά την αποτελεσματικότητα της γεωφυσικής σε υπεδαφικές έρευνες ακόμα και αν πρόκειται για περιβάλλον πόλης.


of information on the location of a lot of important monuments, some of which are accurately positioned and some are related to other landmarks one of them being the ancient river Eridanos.

It was well expected that when the works for the construction of the Metro of Athens commenced at the beginning of the ’90s, one of the most serious and most unpredictable problems would be the archaeological remains buried under the city centre.

In July 1992, Olympic Metro, the constructors of the Athens Metro, assigned to the Laboratory of Geophysics of the University of Patras the task of exploring the area of Syntagma Square. The main purpose was to identify and locate ancient remains, as well as geological evidence and the location of public utility networks.

The archaeological excavations available were very limited especially at Syntagma Square in the heart of the city, where most information both archaeological and geological had come from during the various excavation phases of construction.

The method of Ground Penetrating Radar transmitting electromagnetic waves at frequencies of 80, 300 and 500 Mhz with a bistatic transducer, was utilised for this

On the other hand, ancient texts give a large amount 85

M.G. Papaioannou. S. P. Papamarinopoulos and P. Stefanopoulos purpose, as the desired depth of investigation was down to 20 m. In parallel, the method of electrical soundings was used to obtain resistivity measurements. The methodology and its implementation for this specific project is described in detail in Papamarinopoulos & Papaioannou (1994) and Papamarinopoulos et al, (1995).

concentrating rain water and sewage water etc. It had its source in the southern foothills of Lycabettus opposite the Gates of Diochares, that are placed today in the NW of Syntagma square. It crossed the ancient city and entered the walls by the Holy gate at Keramicos, turned the south and met with Ilissos. The maps of Figure 1, relate the topography of Athens in 3500-600 BC to the topography in 1833-1959 AC.

During geophysical investigation in the area, the ancient river Eridanos was discovered together with various other archaeological features, most of which were proved, through excavations, to be related to the presence of the river itself.

Other views about the course of Eridanos are given by Kordellas (1879), who describes the river as “a dry stream”. The most acceptable one by the author is that the source of Eridanos is in Mount Ymitos and it crosses Kesariani by the church of St. Thomas at Goudi. Strabo also places the source of Eridanos by the Gates of Diochares near the Lyceum. He also states that its once clean water is now so dirty that even animals do not drink it. Judeich (1905) describes only the part of the river that goes north of the church of Kapnikarea, between Hephestus and Adrian’s streets towards the Holy Gate.

Eridanos in the texts – ancient and modern Ancient Athens had three main rivers that were all connected to each other. Eridanos was the smallest, but with the most water. According to mythology, Eridanos was a river flowing in Helisia Fields, the place where as Homer describes in his poems, heroes lived. A river by the same name, also according to mythology, was located in the land of the Gauls near to the ocean. It was there that the burning Phaeton fell when Zeus struck him with lightning. Another version identifies Eridanos as the river Pados in Italy that Argo crossed to reach the Adriatic Sea during the Argonautical Expedition. It was Plato in one of his dialogues, Timeos and Critias, that connects Eridanos with a series of catastrophic events. The original text follows with the English translation by Bury (1989).

All the above descriptions and interpretations are based on what Pausanias states in his book “Attica” (Shilleto 1900) and Hesiod in his book “Theogony” (Wender, 1973) in which the first reports on Eridanos can be found. Roides (1896), in one of his short novels published in the newspaper “Estia”, mentioned that “it was in vain that I asked the contemporary archaeologists to show me where the plenty and good waters of Eridanos so much praised by Athenian virgins, were located”. “Estia”, June 11, 1931, also reports on the spring “Boubounistra”, the water of which comes from the mysterious Eridanos and is only used to water animals. The Gate of Boubounistra is at the cross-section of present day Othonos Str and Amalias Ave.

“…tÄ d’ ðstu jatemom Íd ¶m Ñν τ© τότε χρόνz. Πρ´τον μÑν τÄ τ±ς Çκροπόλεως ε²χε τότε οÌχ Ýς τÀ ν³ν ñχει. ν³ν μÁν γÀρ μία γενομένη νÅξ ÜγρÀ διαφερόντως γ±ς αÕτÂν ψιλÂν περιτήξασα πεποίηκε, σεισμ´ν çμα καà πρÄ τ±ς Ñπà Δευκαλίωνος φθορ°ς τρίτου πρότερον ìδατος Ñξαισίου γενομένου· τÄ δÁ πρÃν Ñν Ñτέρz χρόνz μέγεθος μέν Òν πρÄς τÄν ’GριδανÄν καà τÄν TλισÄν Ðποβεβηκυ²α καà περιειληφυ²α ÑντÛς τÂν Πύκνα καà τÄν ΛυκαβηττÄν ðρον Ñκ το³ καταντικρ³ τ±ς ΠυκνÄς ñχουσα, γεώδης δ’ ¶ν π°σα καà πλÂν Ðλίγον Ñπίπεδος ðνωθεν…”

Geological findings and ancient texts Athens’ bedrock lacks a large variety of rock formations. The well known undivided Athens Schists consisting mainly of sericitic-chloritic schists, sandstones with intercalation of shales, lenticular layers of sandy limestones, crystalline limestones and some effusions (Katsikatsos et al., 1976), is one of the most common formations that one sees in the centre of the city. This completely contrasting group of rock formations (Dounas, Kalergis, Morfis, 1976), is based not only on lithological aspects, but caused also by their reaction to external and internal factors resulting from the particularly heterogeneous and anisotropic behaviour of the group as a whole.

“…In the first place, the Acropolis, as it existed then, was different from what it is now. For as it is now, the action of a single night of extraordinary rain has crumbled it away and made it bare of soil, when earthquake occurred simultaneously with the third of the disastrous floods which preceded the destructive deluge in the time of Deucalion. But in its former extent, at an earlier period, it went down towards the Eridanus and the Ilissus, and embraced within the Pnyx, and had the Lecabettus as its boundary over against the Pnyx; and it was all rich soil and save for a small space, level on the top…”

Another common feature is a variety of deposits. Firstly, there are river deposits consisting of fine to medium grained material (clays, sands with isolated cobbles and pebbles) alternating locally and sometimes consisting of cohesive conglomerates; and secondly, fan deposits of coarse loose material, conglomerates and different

Travlos (1993) mentions that Eridanos was a small river with lots of water, not visible since the time of Themistocleus. A big part of it was turned into a pipe for 86

Geophysical Research at Syntagma Square, Athens

Figure 1.: Two maps produced by Travlos (1993), relate the topography of Athens in 3500-600 BC to the topography in 1833-1959 AC. The course of Eridanos River can be associated with today’s buildings.

material. The geophysical investigations in the area produced vertical sections immediately relating to the above description and were confirmed by the drillings at the area (Papamarinopoulos et. al. 1997). Georadar methods identified the Eridanos River on Filellinon Street, southeast of Syntagma Square. On the vertical georadar section, the deepest part of the river is about 6 metres and is about 17 metres wide. In parallel with the georadar, electrical soundings were recorded at Amalias Ave, which with the use of special software, were combined to give horizontal slices of 160m x 30 m up to a depth of 5 metres. These are presented together with the vertical georadar section of Eridanus in Figure 2. On the geoelectrical horizontal sections concentrations of high and low values of electrical resistivity can be seen. The high values (warm colours) correspond to archaeological findings of walls, baths, a cemetery and other features later excavated in the area. The low values are due to the moisture in soil reflecting River Eridanos, this anomally can be seen up to the depth of 5 metres that the electrical soundings reached and is pin- pointed by two black arrows on one of the horizontal slices. A v-shaped drawing explains the horizontal electrical sections up to five meters in depth.

Figure 2.: On the top right is the georadar vertical section with the image of the subterranean river Eridanos. Horizontal slices produced by electrical soundings on Amalias Ave follow with an explanatory v-shaped drawing showing the various consecutive layers of the river.

drillings during the construction of the Athens Metro, the Athenian Schist starts. Since it is a very disturbed and weathered formation it is most probable that it is a tectonic feature that allows the passage of water through it at depths greater than that of Eridanos.

This complex geological situation combined with the presence of Eridanos was topped by another geophysical discovery. This was the existence of a syncline on the south part of Syntagma Square shown on figure 3.


The georadar vertical section corresponds very well with the geology of the area. The depth of 5 metres at which the formation begins, is the depth that according to the

The works for the construction of the Athens Metro, provided a great opportunity to apply geophysical methods 87

M.G. Papaioannou. S. P. Papamarinopoulos and P. Stefanopoulos to a city environment, dealing with real problems and most importantly getting confirmation of the geophysical interpretation through excavations. Further geophysical investigation revealed the course of River Eridanos (figure 4) thus providing information about buildings and locations that ancient texts associate with the river. Figure 5, presents two photographs of Eridanos that can be seen today in the Syntagma Metro Station and Monastiraki Metro Station accordingly. Finally, extreme conditions described in ancient texts, again seem to find a plausible explanation through the geophysical discoveries.

Figure 3.: The syncline at the south part of Syntagma Square. The dotted line presents an estimated projection of the feature.

Acknowledgments Many people helped to realize this project and we take this opportunity to thank them. They are the people from the Olympic and Attiko Metro and especially, Markos Novak, Katerina Delouka, Jack Blanke and Tolis Exarchou, Chr. Kappopoulos and John Balatsas who performed the electrical resistivity measurements and Dr Chris De Wispelaere and the members of the Steering Committee of the NATO Science for Stability program who encouraged and supported us. References Figure 4.: The map of the centre of Athens showing the course of River Eridanos (dotted line) from the foothills of Mount Lycabettus towards Keramikos.

Bury, R.G., Plato IX, Timaeus-Critias-Cleitophon-Menexenos Epistles, Loeb Classical Library, Harvard Series, Eds, 1989, 111E-112B. Dounas, Α, Callergis, G., Morfis, Α., Tasios, Ν. and Gakis, Α., 1976, Hydrogeological investigation as part of the study for the Athens Metropolitan city train (METRO), ΙGME, Hydrological & Hydrogeological Investigations 19 (in Greek). Judeich, W., 1905, Topographig von Athen. Munchen, pp. 4446. Katsikatsos, G., Kounis, G., Antoniadis, P., Mettos, Α., Papadopoulos, P. and Gakis, Α. 1976, Geological Map of the Area of the Athens Metro (1:7.500). Kordellas, A., 1879, Athens examined from a Hydraulic point of view, Publication House of Philocalia, pp.156 and 60-64. Papamarinopoulos, St.P. and Papaioannou, M.G., 1994, Geophysical investigations with the georadar in the middle of Athens at Syntagma square and the discovery of the subterranean river Eridanos, GPR ‘94, Proceedings of the Fifth International Conference on Ground Penetrating Radar, vol. 2, Kitchener, Ontario, Canada, pp. 569-576. Papamarinopoulos, St. P., Papaioannou, M.G. and Stefanopoulos, P., 1997, New geological evidence in the center of Athens using the georadar, Proceedings of the International Symposium on Engineering Geology and the Environment, Athens, Greece, 23-27 June. Papamarinopoulos, St.P., Papaioannou, M.G., Kappopoulos, Ch. and Balatsas, Y., 1995, Multiple geophysical studies at the urban central environment of Athens in connection with the construction of the Metro’s main station, SAGEEP, April 23-26, Orlando, Florida. Plato, Timaios – Kritias, Publications Cactos, 1993, Original ancient text, in Greek.

Figure 5.: Two photographs of River Eridanos as it can be seen today at the Metro Station at Syntagma (left) and Monastiraki (right). Roidis, Ε., Stories, The Countryside of Athens. Nefeli publications (in Greek). Shilleto, A.R., 1900, Pausanias’ Description of Greece, Volume 1, Attica: George Bell & Sons. Strabo, The Geography of Strabo IV, Translated by H.L.Jones, The Loeb Classical Library, Book 9, Chapter Α, para 19. Travlos, J., 1993, City planning evolution of Athens. 2nd Publication, Kapon: Athens. Wender, D., 1973, Hesiod, Theogony-Works-Days, pp. 27 and 34, Penguin Classics, Reprinted 1976.


INTEGRATED GEOPHYSICAL STUDIES AT ANCIENT ITANOS (GREECE) A. Vafidis Applied Geophysics Laboratory, Mineral Resources Engineering, Technical University of Crete, Chania 73100, Greece, [email protected]

A. Sarris, Th. Kalpaxis Laboratory of Geophysical-Satellite Remote Sensing and Archaeo-Environment, Institute of Mediterranean Studies, Foundation of Research and Technology (F.O.R.T.H.), Rethymno, 74100 Greece Abstract: Since 1994, the Technical University of Crete and the Institute for Mediterranean Studies/F.O.R.T.H, in collaboration with the Ecole Francaise d’Athènes, have been involved in a geophysical research program at the archaeological site of Itanos, Ε.. Crete. The project integrates geophysical imaging techniques for the mapping of monuments of the archaeological site. Test excavations brought to light a number of monuments and finds related to the occupation of the site that spans from Classical to Early Christian/Hellenistic period. The site covers an area of about 400,000 square meters and it was soon realized that the assessment of the archaeological monuments of the site could be accomplished only through the utilization of geophysical and satellite remote sensing techniques. The purpose of the geophysical project was to map the buried archaeological relics. Seismic refraction and reflection techniques were used for detecting the ancient port of Itanos and mapping the bedrock of the area, covered by alluvium deposits. The location of the ancient port in the region south of the two acropolis was revealed by combining seismic, soil resistance, magnetic gradient and electromagnetic data with the digital elevation model. These prospection techniques were employed to map the subsurface architectural remnants and provide information on the extension of the settlement. Further geophysical prospection work was carried out through the use of nonconventional techniques such as ground penetrating radar and electrical tomography. The goal was to study in depth the dynamic potential of these methodologies for a more productive application of them in archaeological prospection. At Itanos, the non-conventional techniques refined the geophysical signature of subsurface targets and improved the interpretation process. In this way, excavations are only required to investigate regions of significant interest, providing a feedback to the process of data manipulation. The 3-dimensional model of the port’s basin was integrated by further seismic surveys such as S-wave refraction and inversion of surface waves. The resulting model was fed into the 3D elevation mode of the site to recreate the environmental settings of the particular area in question. Geophysical, geological and archaeological data from ancient Itanos are combined in order to achieve a valid interpretation of the geophysical data. Περιληψη: Η Ίτανoς, μία από τις σημαντικότερες πόλεις της Κρήτης κατά την αρχαιότητα, βρίσκεται στην επαρχία Σητείας του νομού Λασιθίου. Ο αρχαιολογικός χώρος της Ιτάνου προσέλκυσε το ενδιαφέρον του Εργαστηρίου Εφαρμοσμένης Γεωφυσικής του Πολυτεχνείου Κρήτης που άρχισε να τον μελετά από το 1994 σε συνεργασία με το Ινστιτούτο Μεσογειακών Ερευνών του Ιδρύματος Τεχνολογίας και Έρευνας και τη Γαλλική Αρχαιολογική Σχολή της Αθήνας. Ο αρχαιολογικός χώρος της Ιτάνου βρίσκεται 10 χιλιόμετρα βόρεια του Παλαίκαστρου, στην επαρχία Σητείας. Είναι παραθαλάσσια θέση και περιβάλλεται από τρεις λόφους. Στον ανατολικό και δυτικό λόφο έχουν χτιστεί δύο αρχαίες ακροπόλεις, ενώ ο τρίτος λόφος στο νότο, χωρίζει τη θέση από το φοινικόδασος του Βάι. Ο αρχαιολογικός χώρος καταλαμβάνει έκταση 16000 m2 και ανασκαφές έχουν γίνει μόνο στο 1% της έκτασής του. Ίχνη της μινωικής εποχής με πιο σημαντικό ένα λίθινο βάζο που φυλάσσεται στο μουσείο του Ηρακλείου της Κρήτης, κεραμικά της γεωμετρικής περιόδου (900 – 700 π.Χ.), η νεκρόπολη στο βόρειο τμήμα της περιοχής με σημαντικό αριθμό τάφων ελληνιστικής εποχής, μέρος του τείχους που έγινε για την προφύλαξη της αιγυπτιακής φρουράς, δύο ναοί βυζαντινής εποχής και υπολείμματα κατοικιών είναι τα πιο σημαντικά ευρήματα. H γεωφυσική διασκόπηση πραγματοποιήθηκε στο χώρο, με σκοπό την ανίχνευση ανωμαλιών που προκαλούνται στις γεωφυσικές παραμέτρους του εδάφους, όταν βρίσκονται μέσα σε αυτό θαμμένες αρχαιότητες. Η διασκόπηση περιελάμβανε χαρτογράφηση με μεθόδους μαγνητικής και ηλεκτρικής αντίστασης καθώς και την ηλεκτρομαγνητική μέθοδο, αλλά και μεθόδους υψηλής ανάλυσης όπως η ηλεκτρική τομογραφία και το γεωραντάρ. Η συνδυασμένη ερμηνεία των δεδομένων από τις παραπάνω μεθόδους οδήγησε στον εντοπισμό θαμμένων ερειπίων, προσφέροντας πολύτιμες πληροφορίες για την πόλη της Ιτάνου κατά την Ελληνιστική περίοδο καθώς και για το λιμάνι και τη Νεκρόπολη της. Στην περιοχή του λιμανιού όπου πραγματοποιήθηκε σεισμική διασκόπηση, δημιουργήθηκε τρισδιάστατο μοντέλο της οροφής του βραχώδους υποβάθρου της περιοχής. Το μοντέλο αυτό οριοθετεί την πιθανή περιοχή του αρχαίου λιμανιού, το οποίο σήμερα είναι καλυμμένο με προσχώσεις. Το βραχώδες υπόβαθρο εμφανίζεται στα νότια του λόφου. Προς βορρά βρίσκεται κάτω από την επιφάνεια της θάλασσας, ενώ ακόμη πιο βόρεια εμφανίζεται στη δυτική ακρόπολη του αρχαιολογικού χώρου. Αναπτύχθηκε βάση δεδομένων Γεωγραφικού Συστήματος Πληροφοριών για την διαχείριση και παρουσίαση των αποτελεσμάτων από την γεωφυσική έρευνα στον αρχαιολογικό χώρο της Ιτάνου. Η δυνατότητα του Γεωγραφικού Συστήματος Πληροφοριών για συνδυασμένη απεικόνιση σεισμικών, ηλεκτρικών, μαγνητικών και ηλεκτρομαγνητικών δεδομένων με γεωλογικά και αρχαιολογικά δεδομένα οδήγησε στην ασφαλέστερη ερμηνεία των γεωφυσικών δεδομένων στο αρχαιολογικό χώρο της Ιτάνου.


A. Vafidis, A. Sarris and Th. Kalpaxis Introduction

Itanos, an ancient coastal site in Eastern Crete (Fig. 1a), is located 10 km north of Palaikastro, Lasithi prefecture close to the unique Vai Palm Forest. The sea to the east, a mountain to the south and the provincial road to the west and north surround the archaeological site. There are two acropoleis of the ancient city. Most of the remains of the buildings have been located in the region between the two acropoleis.

During the last decade, conventional geophysical methods have been integrated with other techniques like ground penetrating radar, electrical tomography and seismic refraction to maximize the information context of the subsurface archaeological remains (Bevan 1991, Noel 1992, Tsokas et al. 1997). Also, aerial photography and satellite remote sensing have been combined with groundbased geophysical data (Cox 1992, Donoghue et. al. 1992, Sarris et. al. 1998).

A public building, probably a temple from the Hellenistic period, was found on the western acropolis. South of the two acropoleis, the existence of a gulf suggests the potential location of the port. On the hill south of the archaeological site, remnants of fortification walls are still present, indicating the establishment of Lagides’ (Ptolemeoi) army from Egypt in the fortification (Kalpaxis et al. 1995). The inhabited region within the fortification does not exceed 400,000 m2. The only known necropolis is extended to the north.

An integrated geophysical survey was conducted at the archaeological site of Itanos located in North- eastern Crete, Greece (Fig. 1a). The geophysical mapping was carried out in selected grids (Fig. 1b) covering an area of 16,000 m2. The grids were scanned with magnetic, the soil resistance and electromagnetic methods. Additional measurements were taken with electrical tomography, ground penetrating radar and seismic refraction. This paper presents the results of the five-year campaign of archaeological prospecting survey at the Hellenistic site of Itanos. The acquisition, processing and interpretation of the data are then presented and discussed in detail.

Itanos, according to Herodotus, was one of the most important cities in Eastern Crete in the middle of 7th century BC. It is among the first Cretan cities to mint coin (possibly at the beginning of the 4th century BC, Kalpaxis et al. 1995, Greco et al. 1999). The French archaeologist,

Figure 1. (a) Air photograph of Itanos archaeological site showing the surveyed grids and lines (white) as well as the areas of excavation (indicated by white arrows).


Integrated Geophysical Studies at Ancient Itanos ancient relics can produce remarkable resistivity contrast with respect to the surrounding soil (Carr 1882, Sarris and Jones 2000). At Itanos, we utilized the Geoscan RM15 twin probe array consisting of two remote electrodes (M, A) and two mobile ones (N, B), with electrode separation of 1 m to 2 m (effective depth of 1-2.5 m, Vafidis et al. 1996, Sarris et al. 1998). The electromagnetic (EM) method determines the electric and magnetic properties of the rocks from the observations of induced EM fields (Kaczor and Weymouth 1981). The EM method has been used in archaeological prospecting since the 1960s (Tabbagh 1984). The grids covering an area of 16,000 m2 were scanned with magnetic (vertical magnetic gradient), soil resistance and electromagnetic methods (Fig. 1b). During data acquisition, measurements were taken along common lines of adjacent grids. Measurements on adjacent grids may show deviations due to differences resulting from the calibration of the magnetic instrument or the position of remote electrodes in soil resistance techniques. Edge matching of adjacent grids is based on the calculation of the mean value of measurements along the common lines. For grids with common lines in one to four sides of the grid, a trial and error method, based on the average value of the whole grid, is applied in order to determine the correction factor for each grid (Sarris 1992). For the preparation of the geophysical maps krigging is performed on the data using the processing software Transform V3.4. Krigging involves the construction of a smooth surface from the data with a minimum curvature interpolation method. After obtaining an acceptable minimum curvature surface, the cell is subdivided and the same procedure is repeated until the cell size of the minimum curvature surface is equal to the grid spacing.

Figure 1. (b) The geophysical mapping techniques employed for each surveyed grid.

Demargne, conducted excavations at Itanos during the summer of 1899 (Spyridakis 1970). In 1950, French archaeologists started a second and more systematic archaeological project. Itanos is marked mainly by three periods: Geometric, Roman and Late Christian, while the periods of original occupation and abandonment are not known. A new archaeological collaborative campaign between the French School at Athens and the Institute of Mediterranean Studies FORTH was initiated in 1993. Within the context of archaeological investigations, a geophysical prospecting expedition was also carried out.

Processing of geophysical data Generally, most geophysical data sets do require processing but few of the available packages give the background of what the processing entails. Thus, most papers of geophysical surveys do not describe the mathematics of the processing. In this paper an explicit effort is made to present the processing methods. In the processing software Transform V3.4, the user can develop their own algorithms.

Magnetic, soil resistance and electromagnetic survey

The following differential operators were applied to soil resistance, magnetic gradient, quadrature and in phase data at Itanos (Blakely, 1996).

At Itanos, the vertical gradient of the earth’s magnetic field was measured (Kaczor and Weymouth, 1981) using the Geoscan Fluxgate gradiometer FM18 and the Geometrics Proton Precession Magnetometer G-856A. Diurnal variations of the magnetic field did not affect magnetic gradient data.

The resistivity method was also employed, since the 91

The gradient operator √{(∂xZ)2+ (∂yZ)2} where Z denotes the measurements and ∂x, ∂y the partial differential operators with respect to x and y, generates a grid of steepest slopes. The first derivative along the x-axis is described by

A. Vafidis, A. Sarris and Th. Kalpaxis

the operator Lx= kx i, where i=(-1)0.5 and kx denotes wavenumber in the x direction. The Laplacian operator provides a measure of discharge (negative values) or recharge (positive values) on a surface. It is described by the operator L2 = (ri)2 where r is the radial wavenumber r=√(kx2+ ky 2) and kx, ky denote wavenumbers in the x and y directions.

Upward continuation in wavenumber domain was performed by simply multiplying the transformed map with the operator: L = e-hr, where h denotes height. Apart from upward and downward continuation in the wavenumber domain other operators like the directional derivative Lx and the Laplacian L2 were also applied in the wavenumber domain. Additionally, a highpass Butterworth filter was used to enhance the local field. Its operator in the wavenumber domain is described by (1) where rc is the cutoff radial wavenumber, n the order of the Butterworth filter (here n=8) which controls its slope. If Gibbs phenomena are present, the order of the filter must be reduced. The cutoff wavenumber can take values from 0 to rnyquist ( = 0.5 / m for grid spacing 1 m).

Figure 2. Geophysical maps for grid H (a) high-pass filtered (Butterworth) soil resistance map, (b) magnetic gradient map, (c) quadrature map filtered with a gradient operator and (d) residual magnetic map which is generated by upward continuation of the original magnetic gradient data.

The application of a highpass Butterworth, (rc=0.2/m) filter in the soil resistance data of grid H (Fig. 2a) enhanced the anomalies H6 and H8. Parallel elongated anomalies H1, H2, H3, H4 and H5 are shown in the map of magnetic gradient (Fig. 2b). Features H8 and H6 are also present in the quadrature data filtered with the gradient operator (Fig. 2c). The above advocate the existence of more than one closed structure in grid H, which probably belong to the coastal facilities of the settlement. Most of the mentioned anomalies are shown on the magnetic map (Fig. 2d) which was generated by calculating the upward continued field at h=5 m and subtracting it from the measured magnetic field. Excavations that followed brought to light parallel walls at distances 4.5 m, 6.5-7 m and 9-9.5 m from the edge of the grid and at depths of 0.7 m, 0.25 m and 0.4 m respectively (Figs. 7a and 9). More specifically, the walls at distances 4.5 m, 6.5-7 m and 9-9.5 m are related to the observed anomalies H4, H3 and H2 respectively. In grid K, the quadrature map suggested an elongated anomaly K6 attributed to the sewage system (Fig. 3a). The quadrature map, filtered by a directional derivative operator (Fig. 3d), emphasized the elongated anomaly K4. The soil resistance gradient map (Fig. 3b) and the high pass filtered magnetic gradient map (rc=0.125/m) (Fig. 3c) indicated the same anomaly K5 which was attributed to the existence of a stone-built well (Fig. 10) and confirmed the anomaly K6.

Figure 3. Geophysical maps for grid Κ (a) quadrature map, (b) soil resistance gradient map, (c) high-pass filtered (Butterworth) magnetic gradient map and (d) quadrature y-derivative map.


Integrated Geophysical Studies at Ancient Itanos

Figure 4. Electrical tomography sections for lines (a) R5 and (b) R4 on grid H.

In conclusion, processing of the magnetic electromagnetic and electrical data was of extreme help in suggesting several potential targets. Apart from the excavated three walls (Fig. 9), grid H encountered a wealth of anthropogenic anomalies.

Figure 5. Geophysical survey at the port (a) Grid P was scanned with magnetic gradient, soil resistance and EM methods. The seismic lines are dotted and electrical tomography line R2 is continuous. (b) Residual quadrature map on subgrid Pb. Black corresponds to high resistivity. No data have been collected in region Pn.

Electrical tomography Electrical tomography surveys can be used to delineate areas with complex subsurface geology where conventional resistivity sounding or profiling is inadequate (Loke and Barker 1996). At Itanos, the Sting / Swift multielectrode system was used to collect data along specific traverses, which were contoured in the form of a pseudosection, giving an approximate picture of the subsurface resistivity. Accurate depth determination and interpretation was achieved through inversion of the measured apparent resistivity.

excavated trench. The high resistivity zone (Fig. 4b) near the surface at horizontal distance 6-7 m corresponds to the anomaly H9 on the high pass filtered (Butterworth) soil resistance map (Fig. 2a). The image also exhibits high resistivity anthropogenic anomalies at depths 1.5-2.5 m and horizontal distances 9-10 m, 12-14 m and 19-21 m. For grid Pb (Fig. 5a) the residual quadrature map (Fig.5b) indicates two high resistivity / low conductivity anomalies, namely P1 and P2. According to electrical tomography measurements carried out in grid P with electrode spacing 1.5 m using the dipole-dipole array (Fig. 6a), the anomalies P1 and P2 are attributed to shallow targets at depth less than 4 m.

2D and 3D electrical tomography measurements were collected at Itanos using the Wenner, the pole-pole or the dipole-dipole arrays with electrode spacing ranging from 0.5 m to 1.5 m. The data were contoured in the form of a pseudosection. To obtain a more accurate picture of the subsurface it is necessary to invert the apparent resistivity data. The inversion method was based on a smoothness constrained least squares method (Gauss - Newton method, Loke and Barker 1996).

Further experiments were carried out in the area of necropolis, located north- west of Itanos acropolis, on a small hill (grid N, Fig. 1a), where excavation had already taken place. The electrical tomography lines R7, R8 and R9 scanned a small portion of grid N (Fig. 1a). Α three dimensional image was constructed from electrical tomography data (Fig. 6b) showing a shallow linear anomaly EB. An additional high resistivity anomaly EA may be attributed to a wall, which may belong to a series of almost parallel ones in the east-west direction, partially revealed in the excavated area.

Figure 4a shows the resistivity section for line R5 next to the excavated trench on grid Η (Figs. 1a). This image shows five high resistivity zones, probably attributed to wall remains, although one should always keep in mind that these anomalies can originate from fallen rocks, present human activities or even poor electrode contact in the ground. The outline of the potential structures is also shown. The anomalies W1, W2 and W3 are probably caused by the extension of the walls revealed in the excavation trench (Figs. 7a and 9). The tomography line R4 (Figs. 1a) runs in a direction perpendicular to the

In summary, electrical tomography helped in confirming the anomalies present on soil resistance, magnetic or electromagnetic maps for grids H and Pb. 93

A. Vafidis, A. Sarris and Th. Kalpaxis

Figure 6. (a) Electrical tomography section at the port for line R2 (b) Three dimensional image from electrical tomography survey in the Necropolis seen from south-west (The dimensions of image are 12m x 1m x 1m, vertical exaggeration 3:1). Dark grey colour corresponds to high resistivity anomalies.

Ground penetrating radar

antennas, (Fig. 7b) exhibits features which correspond to the anomalies on soil resistance, magnetic gradient and conductivity maps discussed earlier. In particular the anomaly H6 is better resolved than in figure 2. The anomaly H9, which is also shown in Fig. 2a, is more detailed in Fig. 7b and seems to be connected with the anomaly H6. It is also worth mentioning the improved resolving power of GPR for anomalies H2 and H3.

Ground penetrating radar (GPR) employs radio waves, typically in the 1 to 1000 MHz frequency range, to provide information about the stratigraphy of a site (Vaughan 1986, Pipan et al. 1996). The most widely used mode of GPR surveying is common-offset, single fold reflection profiling. In such a reflection survey, a system with fixed antenna geometry is transported along a survey line. Radio energy transmitted into the subsurface, is reflected by the structural boundaries, and is returned to the receiver. The data can be viewed and interpreted as they are collected and this makes this technique very appealing and popular.

The seismic refraction survey A shallow seismic survey was carried out, in an effort to locate and map the ancient port. The target layer (basement) consists of Permian-Triassic phyllites covered by recently deposited sediments (Vafidis et al. 2003). Seismic refraction is especially suited for shallow investigationss (Tsokas Tsokas et al. 1995, Vafidis et al. 1995).. Eight profiles have been selected (Fig. 5a) with total length of 580 m and geophone spacing 2 m. The resonance frequency of the geophones was 14 Hz. The hammer and the seisgun (Betsy) produced seismic waves. The Geometrics ES2401 seismograph was used. Five shots were selected for most seismic lines: one in the middle, two near shots at the edges of the line and two far shots. For the seismic data sections processing, the software package SIP, supplied by Geometrics, was used. The velocity model along line S3 shows three layers (Fig. 8a). The third layer shows an

The 225 MHz and 450 MHz antennas of a pulse EKKO 1000 ground penetrating radar system were used with line spacing 0.5 m and station spacing 0.1 m and 0.05 m respectively. On grid Η,, fifty east-west 30 m long lines were surveyed with the 225 MHz antennas (Fig. 7a). A portion of the same grid (nine 12 m-long lines) was also scanned with the 450 MHz antennas (Fig. 7a). The radargrams along the line HR show diffraction patterns at x=4.4 m and 6 m (Fig. 7c). These diffractions were better resolved with the 450 MHz antennas (Fig. 7d) and are attributed to walls buried at a depth of 0.15 m (3 nanoseconds). The time slice at 2-3 nanoseconds, for the 225 MHz 94

Integrated Geophysical Studies at Ancient Itanos

Figure 7. Details (a) from survey grid H. (b) The time slice from the GPR survey grid H using the 225 ΜΗz antennas at 2-3 nanoseconds. Black corresponds to anomalies probably caused by walls. Ground penetrating radar sections for line HR using the 225 MHz (c) and 450 ΜΗz (d) antennas.


A. Vafidis, A. Sarris and Th. Kalpaxis

Figure 8. (a) Depth section from seismic refraction experiment for line S3. (b) Three-dimensional image of the top of the basement. (c) Elevation map of the top of the basement.

increased velocity of 3233 m/s, and is attributed to the basement or to an eroded layer on top of the basement. The top layer is unsaturated colluvium and the second layer is water saturated colluvium according to a nearby well. Seismic, geological and topographic data were utilized for the creation of a top-to-bedrock image (Figure 8b). The processing software for the creation of Figure 8 was Surfer 6. This image indicates that the ancient port, covered by colluvium, is located south of the acropolis (Vafidis et al. 2003). It extends to the west about 100 m from the present seashore. It covers a rectangular region of approximately 1500 m2, surrounded by the sea to the east, two acropolis to the north and a hill to the south. Figure 9. Photograph of the excavated trench on grid H. Arrows indicate wall remains.

The elevation map of the basement relief (Fig. 8c), correlates well with the results from soil resistance mapping, conductivity mapping and electrical tomography. The high resistance anthropogenic anomalies P1 and P2 on the geophysical maps (Figs. 5b and 6a) are located at places where the basement exhibits depths less 5 m from the surface (or absolute height –3 m).

Conclusions Magnetic gradient, soil resistance and electromagnetic measurements showed several potential targets. Grid H, apart from the three excavated walls, encounters a wealth of anthropogenic anomalies. The integration of geophysical mapping and electrical tomography delineated selected anomalies. In particular, ground probing radar and electrical tomography sections crossing the three parallel walls showed the vertical extent of these features. Time

The seismic refraction data, combined with topographic and geological data, helped in obtaining a more understandable image of the old port and contributed in the interpretation of the anomalies indicated by electromagnetic mapping and electrical tomography.


Integrated Geophysical Studies at Ancient Itanos data for archaeological prospecting using a Geographic Information System, in Proceedings of the 18th Annual Conference of the Remote Sensing Society (eds. A.P. Cracknell and R.A. Vaughan), University of Dundee, 197-207. Greco, E., Kalpaxis, Th., Schnapp, A., Viviers, D., 1999, Travaux menes en collaboration avec l’Ecole Francaise d’Athènes en 1998, Bulletin de Correspondance Hellènique, 123, 515-530. Kaczor, J. and Weymouth, J., 1981, Magnetic prospecting: preliminary results of the 1980 field season at the Toltee Site, Southeastern Archaeological Conference Bulletin, 24, 118123. Kalpaxis, Th., Schnapp, A, Viviers, D., Guy, M., Licoppe, C., Schnapp-Gourbeillon, A., Siard, H., Theodorescu, D., Tsigonaki, C., Vafidis, A. and Xanthopoulou, Μ., 1995, Rapport sur les travaux menes en collaboration avec l’Ecole Francaise d’Athènes en 1994, Bulletin de Correspondance Hellèniq, 119, 713-736. Loke, M.H. & Barker, R.D., 1996, Rapid least squares inversion of apparent resistivity pseudosections by a quasi Newton method, Geophysical Prospecting, 44, 131-152. Noel, M., 1992, Multielectrode resistivity tomography for imaging archaeology, in Geoprospection in the archaeological landscape (ed. P. Spoerry), Oxbow Monographs, 18, 89-99. Pipan, M., Finetti, I. and Ferigo F., 1996, Multi-fold GPR techniques with applications to high-resolution studies: two case histories, European Journal of Enviromental and Engineering Geophysics, 1, 83-103. Sarris, A., 1992, Shallow depth geophysical investigation through the application of magnetic and electric resistance techniques, Thesis (PhD), Lincoln, University of Nebrasca. Sarris, A. and Jones, R., 2000, Geophysical prospection of archaeological sites in the Mediterranean region, Journal of Mediterranean Archaeology, 13, 3-75. Sarris, A., Vafidis, A., Mertikas, S., Guy, M. and Kalpaxis, Th., 1998, Remote sensing techniques and computer applications for monuments and site assessment of Itanos (E. Crete), Abstracts, Computer Applications in Archaeology Conference (CAA 98), Barcelona. Spyridakis, 1970, Ptolemaic Itanos and Hellenistic Crete, Zurich. Tabbagh, A., 1984, On the comparison between magnetic and electromagnetic prospection methods for magnetic features detection, Archaeometry, 26, 171-182. Tsokas G.N., Papazachos, C.B., Vafidis, A., Loucoyannakis, M.Z., Vargemezis, G. and Tzimeas, K., 1995, The detection monumental tombs buried in tumuli by seismic refraction, Geophysics, 60, 1735-1742. Tsokas, G.N., Sarris, A., Pappa, M., Bessios, M., Papazachos, C., Tsourlos, P. and Giannopoulos, A., 1997, A large scale magnetic survey in Makrygialos (Pieria), Greece, Archaeological Prospection, 4, 123-137. Vafidis, A., Manakou, M., Kritikakis, G., Voganatsis, D., Sarris, A., and Kalpaxis, Th., 2003, Mapping the ancient port at the Archaeological site of Itanos (Greece) using shallow seismic methods, Archaeological Prospection, 10, 163-173. Vafidis, A., Sarris, A., Oikonomou N. and Kalpaxis, A., 1996, Geophysical survey in the archaeological site of Itanos, Lasithi Greece, 1st Balkan Geophysical Society Extended Abstracts, September 23-27, Athens, 54-55. Vafidis, A., Tsokas, G.N., Loucoyannakis, M.Z., Vasiliadis, K., Papazachos, C.B. and Vargemezis, G., 1995, Feasibility study on the use of seismic methods in detecting monumental tombs buried in tumuli, Archaeological Prospection, 2, 119-128. Vaughan, C.J., 1986, Ground penetrating radar surveys used in archaeological investigations, Geophysics, 51, 595-604.

Figure 10. Photograph of the stone-built well on grid K.

slices resolved better the anomaly H6, present on soil resistance and conductivity maps. In grid K, a small portion of the city plan is imaged. The soil resistance gradient map and the high pass filtered magnetic gradient map indicated the anomaly K5 where excavation revealed a stone-built well (Fig. 10) and burned soils. Additionally, the location of the ancient port at Itanos was confirmed by the threedimesional image of the bedrock. This integrated study showed that geomagnetic, resistivity and conductivity prospection techniques could be useful in locating and mapping anthropogenic anomalies at Itanos. The ground penetrating radar and electrical tomography techniques delineated many potential targets. These high-resolution geophysical methods are necessary for confirming anomalies on the geophysical maps attributed to buried remains, since they provide better images of the subsurface targets. Acknowlegments We thank the Department of Mineral Resources Engineering, Technical University of Crete for covering data acquisition cost. We also thank the French School of Archaeology, Athens for providing access to the archaeological site and Prof. S. Mertikas for providing GPS data. References Bevan, B., 1991, The search for graves, Geophysics, 56, 13101319. Blakely, R., 1996, Potential Theory in Gevity and Magnetic Applications, Cambridge University Press. Carr, C., 1982, Handbook on soil resistivity surveying, Evanston, Illinois: Center for American Archaeology Press. Cox, C., 1992, Satellite imagery, aerial photography and wetland archaeology- an interim report on an application of remote sensing to wetland archaeology: the pilot study in Cumbria, England, World Archaeology, 24, 249-267. Donoghue, D.N.M., Powlesland, D.J. and Pryor, C., 1992, Integration of remotely sensed and ground based geophysical


THE GEOPHYSICAL RESEARCH AT THE CHURCH OF ST. THEODORA AT VASTA MEGALOPOLIS, GREECE St. P. Papamarinopoulos, P. Stefanopoulos University of Patras, Department of Geology, Laboratory of Geophysics, 261 00 Rio-Patras, [email protected]

M. G. Papaioannou Sev. Kallisperi 32, 152 34 Chalandri, Athens, Greece

N. Charkiolakis Ministry of Culture, Directorate for the Restoration of Byzantine and Post-Byzantine Monuments, 56 Hermou str, Athens, Greece

Ch. Vachliotis Domos” Technical Company 38-40 Hellanikou str, Athens, Greece Abstract: Many stories have seen the light of publication for the small church of St. Theodora of particular religious importance, related mainly to its creation, the Saint that has given it her name and its story in general that also involves a “miracle”. The building of the small church, visited each year by many believers, is of the Byzantine era and besides the jurisdiction of the Church it is also under that of the Ministry of Culture as a Byzantine Monument. The geophysical team of the University of Patras tried to add to the restoration of this protected building. The problem was directly connected to the “miracle” that takes place there and the myth that is spread around involving the creation of the small church, which is as follows; seventeen large trees and several smaller have grown on the roof of the building. The problem that had to be dealt with concerning the trees on the roof was to find out where their roots were directed without provoking the religious feeling of the church people and the clergymen that were our “sleepless guardians” throughout the entire survey. After an onsite estimation of the situation it was decided that the high-frequency georadar would be used on the walls of the building together with electrical tomography using very small electrodes, so as to have a double check on the methodology. The results of the investigation threw ample light to the mystery and gave the Directorate for the Restoration of Byzantine and PostByzantine Monuments all the answers they needed to proceed safely with the restoration of the church of St. Theodora. It was proved that the roots follow the gaps existing inside the side stonemasonry walls creating repulsion stresses between the stones and thus they reach the ground. This leads to the creation of a net of roots that reinforces the building from a static point of view opposing the roof load but is also destroying the stone walls. It was also proved that there are fewer problems in the north wall while there seem to be many voids in the south wall. Περίληψη: Για το ιδιαίτερης θρησκευτικής σημασίας, εκκλησάκι της Αγίας Θεοδώρας στη Βάστα Μεγαλουπόλεως, έχουν κατά καιρούς δημοσιευθεί πολλά που έχουν κυρίως να κάνουν με τη δημιουργία του, την Αγία που του έχει δώσει και το όνομά της, και γενικά την ιστορία του η οποία συμπεριλαμβάνει και ένα «θαύμα». Το κτίσμα της μικρής εκκλησίας που επισκέπτονται χιλιάδες πιστοί κάθε χρόνο, χρονολογείται από τη βυζαντινή εποχή και εκτός από τη δικαιοδοσία της Εκκλησίας βρίσκεται και σε αυτή του Υπουργείου Πολιτισμού ως βυζαντινό μνημείο. Η ομάδα γεωφυσικής του Πανεπιστημίου της Πάτρας προσπάθησε να συμβάλει με την παρούσα έρευνα στη διάσωση του διατηρητέου κτιρίου. Το πρόβλημα για τη διάσωσή του συνδέεται άμεσα με το «θαύμα» που συντελείται εκεί και το μύθο που περιβάλλει τη δημιουργία της μικρής εκκλησίας και έχει ως εξής: στη σκεπή του κτιρίου έχουν φυτρώσει μεταξύ άλλων μικρότερων και δεκαεπτά μεγάλα δένδρα, οι ρίζες των οποίων δεν είναι ορατές κάτω από την μερικών εκατοστών σκεπή και μέσα ή έξω από την εκκλησία. Το όλο κτίριο βέβαια καταπονείται κάτω από το μεγάλο φορτίο και κατά καιρούς έχει επισκευαστεί με τον ένα ή τον άλλον τρόπο. Οι πρόχειρες αυτές επισκευές είχαν σαν συνέπεια παρεμβάσεις που ουσιαστικά καταστρέφουν την αρχιτεκτονική του κτιρίου. Το πρόβλημα που έπρεπε ωστόσο να επιλυθεί σχετικά με τα δέντρα της οροφής ήταν να βρεθεί που πηγαίνουν οι ρίζες τους, καταβάλλοντας ιδιαίτερη προσπάθεια να μη θιγεί το θρησκευτικό αίσθημα κληρικών και πιστών που καθ´ όλη τη διάρκεια της έρευνας υπήρξαν «ακοίμητοι φρουροί». Μετά από επιτόπια εκτίμηση της κατάστασης αποφασίστηκε η χρήση γεωραντάρ μεγάλης συχνότητας στους τοίχους της εκκλησίας καθώς και η πραγματοποίηση ηλεκτρικών τομογραφιών με πολύ μικρά ηλεκτρόδια, ώστε να υπάρχει έλεγχος της μεθοδολογίας. Τα αποτελέσματα της έρευνας διαλεύκαναν το μυστήριο και έδωσαν τις απαντήσεις που η Διεύθυνση Αναστήλωσης Βυζαντινών Αρχαιοτήτων χρειάζονταν για να προχωρήσει στις εργασίες διατήρησης και επισκευής της Εκκλησίας της Αγίας Θεοδώρας. Αποδείχθηκε ότι οι ρίζες ακολουθούν τα κενά που υπάρχουν στην λιθοδομή των πλευρικών τοίχων δημιουργώντας απωθητικές τάσεις ανάμεσα στις πέτρες και φτάνουν μέχρι το έδαφος. Αυτό έχει σαν συνέπεια την δημιουργία ενός πλέγματος από ρίζες το οποίο ενισχύει στατικά το κτίσμα σε αντιστάθμισμα του φορτίου της οροφής, καταστρέφει όμως σταδιακά τη λιθοδομή. Αποδείχθηκε ακόμα ότι ο νότιος τοίχος του κτίσματος είναι σχεδόν άθικτος ενώ βρέθηκε να υπάρχει κάποιο κενό στο βόρειο τοίχο.


the church of St. Theodora, following a request of the Directorate for the Restoration of Byzantine and PostByzantine Monuments of the Ministry of Culture. The 12th century church is located nearby the stream of Krya

During the summer of 1996, the team of the laboratory of Geophysics of the University of Patras undertook the difficult task of investigating the small building of 99

St. P. Papamarinopoulos, P. Stefanopoulos, M. G. Papaioannou, N. Charkiolakis and Ch. Vachliotis of the village of Vasta, 30 km of the city of Megalopolis on the mountains of Arcadia (Figure 1). The story of the creation of the church and the life of St. Theodora is tightly connected to the problem we were asked to deal with and goes like this: there was a poor family in the vicinity of ancient Melpia, the current area between Vasta and another small village, Isari. This was the family of young Theodora, a good looking, virtuous, strong and courageous girl. Her family, like all others, had to either give a young lad to become a soldier of the Monastery of Virgin Mary, or pay for a mercenary. Of course none of the above could happen so Theodora, disguised as a young man called Theodoros, joined the Army and was very well accepted, with her courage being recognized so that she became the captain of the soldiers. No one new her secret but a day came when, according to the legend, a girl that came to do the soldiers’ laundry, fell in love with the alleged captain who, from then on, prayed for her sins. The girl became pregnant by one of the soldiers and blamed Theodora for it, who to avoid dishonoring the regiment was sentenced by court-martial to death. Just before she died she asked god to turn her hair into trees, her body into a church and her blood into a river. So it was and there was also a star added in the sky shining brightly, the star of St. Theodora. This is how the legend goes about this holy pilgrimage.

Figure 2: Plan of the building (left) and the cross-section looking west view of the church (right). The thickness and shape of the walls can be seen together with the positions of the special wooden structure (black squared marks) constructed to help out the geophysical investigation (Drawings by N. Charkiolakis 1997).

The building of the church The church’s dimensions are about 4x7 m and its walls are made of materials from the river (stones and sand) while the roof is semi barrel covered with slates. On top there are 17 trees of variable diameters and heights together with many other plants (Charkiolakis, 1997). All the kinds of trees that grow on the roof have a very strong root system that lead to great depths in the soil or into cracks in the rocks in order for them to find moisture

Figure 1: Map of Peloponnese showing the location of the church of St. Theodora (spot) and a photo of the church at the time of the geophysical investigation.


The Geophysical Research at the Church of St. Theodora at Vasta Megalopolis



Depth (m)

0.5 Figure 3: Vertical georadar section depicting voids inside the south wall structure. The black arrows show the image of small voids.

and nutrients. It is estimated that twelve out of the total seventeen trees are fully-grown and rather large for their kind. Their mean height is about 8 meters and their mean diameter about 25 cm.

and Piccolo (1993) and taking into account for the survey design the optimum frequency versus desired depth of penetration (Annan & Cosway 1992). The data was processed with the help of special software (Radan III) to normalize time versus distance and only basic filters were used to enhance the data. An isolated georadar section created from the inside of the south wall 1.5 m above the floor, Figure 3, shows two to three voids inside the wall structure.

The estimated weight of all plants on the roof is about 5.5 tonnes. Their roots are not visible under the few centimeters of roofing, nor inside nor outside the church. The whole building is of course, under pressure due to this large load and there have been several reconstruction interventions during its lifetime. These non-professional interventions resulted in destroying the architecture of the church.

The same picture can be followed throughout the investigation with the georadar. The time slice sections created with special software for the south wall are shown in Figure 4. The method utilized to create the time slices is described by Goodman et. al., 1995 and the time interval for these sections is 4 ns that is about 10 cm depth wise.

The geophysical methodology Due to the sensitivity of the matter and the need to protect the building, the utilization of non-destructive geophysical methods seemed the only reasonable solution. It was decided that georadar and electrical methods be used in the form of tomography so that the construction of the walls could be investigated and a clear picture of their structure would be obtained. A special wooden construction was made to create leading lines for the measurements, also giving an accurate estimate of position. Figure 2 illustrates a horizontal and the front west view of the building, its dimensions, the thickness of the walls and the wooden construction with marks (small black squares) used for the geophysical investigation. The lines upon which the radar survey was conducted were 0.5 meters apart.

The round shaped formations with white to red colors represent voids. Another presentation of the same data is shown on Figure 5. The sections here are placed on top of each other creating the image of the wall with the top section being the inside and the last the outside of it. Figure 6 shows the parallel sections of the north wall, where similar voids can be identified. Electrical resistivity tomography was also utilized where applicable in the same lines as the georadar, to check the radar results. The method is described in detail in Papamarinopoulos et al, 1997. Special thin electrodes resembling large nails were used at an interval of 0.2 meters so as not to damage the frescoes (Figure 7). The electrical tomography sections showed also images of voids with particularly high values of electrical resistivity. Figure 8

Utilizing the 900 MHz frequency antenna, the walls were investigated both from the inside and the outside as described in Papaioannou et al (1996) based on the methodology applied on walls as described by Niccolini 101

St. P. Papamarinopoulos, P. Stefanopoulos, M. G. Papaioannou, N. Charkiolakis and Ch. Vachliotis

Figure 4: Five time slice sections of south wall. The red-white rounded anomalies represent voids.

Figure 5: The same data as in Figure 4 from a different point of view. The sections here are placed on top of each other creating the image of a wall with the top section being the “face” of it.


The Geophysical Research at the Church of St. Theodora at Vasta Megalopolis

Figure 6: Five time slice sections of the north wall. Once again black arrows show voids.

presents tomographical section of north and south walls respectively. Again the rounded anomalies of red (high) resistivity values are attributed to voids.

The roof was given a proper garret form after it was cleaned of the few centimeters of humus and an asphalt sheet was placed to prevent dampness. Then, slates were used to build it and also used to pave the immediate surrounding of the church. Cement constructions were brought down and a few arrangement works took place in the surrounding area. The result is shown in Figure 9.

The restoration works Based on the results of the geophysical investigation and the proved fact that the roots of the trees pass through the walls, the works approved by the Ministry of Culture aimed at the restoration of the building without disturbing the roots. So, the walls and joints were cleaned and repointed without the use of cement in order not to dry out the roots.

Conclusions The geophysical methodology was successfully applied in the particular problems encountered in the church of St. Theodora. Both the georadar and electrical tomography

Figure 7: The very thin electrodes placed in the wall ready for electrical tomography.


St. P. Papamarinopoulos, P. Stefanopoulos, M. G. Papaioannou, N. Charkiolakis and Ch. Vachliotis

Figure 8a: Electrical tomography pseudosection of the south wall.

Figure 8b: Electrical tomography pseudosection of the north wall. The black arrows point at images of small voids with great resistivity values.

Figure 9: The church after the restoration works.


The Geophysical Research at the Church of St. Theodora at Vasta Megalopolis Greece. The whole investigation proved once more that geophysics is the proper approach when there is a need for non-destructive, non-invasive and fast solutions. References Annan, A.P. and Cosway, S.W., 1992, Ground Penetrating Radar Survey Design, SAGEEP, Chicago. Charkiolakis, N., 1997, The small Church of St. Theodora at Vasta Megalopoleos (Greece), Programme and Abstracts, 17th Symposium of Byzantine and Post-Byzantine Art and Archaeology, 83-84 (in Greek), Christian Archaeological Society, Athens, 23-25 May 1997. Goodman, D., Nishimura, Y. and Rogers, J.D., 1995, GPR Time Slices in Archaeological Prospection, Archaeological Prospection, 2, 85-89. Niccoloni, G. and Piccolo, M., 1993, Use of G.P.R. Method to map buried archaeological structures and for detailed investigations of ancient walls and pavements. XVIII General Assembly of European Geophysical Sociey, Wiesbaden, May 3-7 1993. Papaioannou, M. G., Papamarinopoulos, St. P. and Stefanopoulos, P., 1996, Geophysical studies in Hermitage Museum, the Char’s former Winter Palace, in St. Petersburg, Proceedings, 6th International Conference on GPR, 101-106, Sept 30 – Oct 3 1996, Sendai, Japan. Papamarinopoulos, St. P., Stefanopoulos, P. and Papaioannou, M. G., 1997, Geophysical investigation in search of ancient Helike and the protection of the archeological site versus the rapid building expansion, Proceedings, Int’l Symposium of Engineering Geology and the Environment, 3229-3235, IAEG Athens, Greece 23-27 June 1997.

Figure 10: Part of roots creating a void within the rock structure of the church.

gave very good results, mapping the voids in the stone structure of the walls and thus providing the logical explanation of the “miracle”. The roots of the trees grow inside the walls creating small voids (Figure 10) and so reaching the ground. It is this very network of roots that reinforces the structure, helping it withstand the weight of the trees on the roof. This was the first time that georadar and electrical tomography were utilized successfully on a building in


ΙNVESTIGATION OF A MONUMENTAL MACEDONIAN TUMULUS BY THREE DIMENSIONAL SEISMIC TOMOGRAPHY L. Polymenakos, S. Papamarinopoulos, A. Liossis Department of Geology, Laboratory of Geophysics, University of Patra, Rio, Patra, Greece, [email protected]

Ch. Koukouli – Chryssanthaki Archaeological Society, Athens, Greece Abstract: Monumental tumuli are important monuments of past human activity, and may contain burial structures of high cultural and historical value. Seismic tomography is employed to investigate the internal structure of a monumental tumulus. Travel time data are analyzed and processed with three dimensional tomographic inversion in order to construct images of the distribution of seismic velocity in the interior of the tumulus. A case history from a Macedonian tumulus in Northern Greece is presented. The results are interpreted in terms of evidence for possible man-made buried structures, such as tombs, walls etc. Περίληψη: Οι μνημειακοί τύμβοι είναι σημαντικά μνημεία της ανθρώπινης δραστηριότητας στο παρελθόν και είναι δυνατόν να περιέχουν ταφικές δομές μεγάλης πολιτιστικής και ιστορικής αξίας. Στην παρούσα εργασία περιγράφεται η χρήση της σεισμικής τομογραφίας για τη διερεύνηση της εσωτερικής δομής ενός μνημειακού τύμβου. Γίνεται ανάλυση και επεξεργασία των χρόνων διαδρομής των σεισμικών κυμάτων με χρήση τρισδιάστατης τομογραφικής αντιστροφής, με σκοπό την απεικόνιση της κατανομής της σεισμικής ταχύτητας στο εσωτερικό του τύμβου. Παρουσιάζονται τα αποτελέσματα από μια εφαρμογή της μεθόδου σε έναν τύμβο μακεδονικού τύπου στην περιοχή της Βόρειας Ελλάδας. Γίνεται ερμηνεία των αποτελεσμάτων σε σχέση με ενδείξεις για πιθανές ανθρώπινες κατασκευές, όπως τάφους, τοίχους κ.α.


cases, tombs are reached from the periphery with a ramp (or tunnel, in case they are located at the centre), which was used during the construction of the tomb and thereafter filled with loose material from the vicinity. When several structures are contained in the tumulus, their location does not follow a specific trend, as depicted in the example in Figure 1b.

Tumuli are man-made hills erected to cover one or more tombs, that is, burial structures, as well as other structures (Lazaridis 1992, 1993). Their investigation needs to be at most non-destructive apart from the excavation at specific levels. According to the literature (Lazaridis 1993, Tsokas et al. 1995 and references therein) and direct observation (e.g. in Amfipoli and Vergina, N.Greece), tombs are mainly located close to the base (or a few meters lower) of the tumulus and usually not far from the periphery, (Figure 1a), or they may be located near or at the centre. In both

There have been various efforts to investigate the internal structure of tumuli with geophysical methods (Tsokas et al. 1995, provide extensive references thereof). Tsokas et al. (1995) exploited the effect of the material associated

Figure 1. Examples of tumuli and their internal structure in plan view- not to scale. (a) A tumulus containing a typical tomb (after Tsokas et al. 1995), (b) A tumulus containing various cultural and burial structures (after Lazaridis 1992).


L. Polymenakos, S. Papamarinopoulos, A. Liossis and Ch. Koukouli – Chryssanthaki A tomographic experiment was carried out at the tumulus of Kastas, located in the vicinity of the present village and the ancient city of Amfipolis in N.Greece (Figure 2). A large number of monumental burial structures, among them tumuli containing tombs of the “Macedonian” type, have been discovered and excavated in the surroundings of the fortified ancient city. The tumulus of Kastas is an artificial hill of a normal circular shape, with circumference of about 360 m, an average diameter of 160 m and height of 21m. It is made of alternating layers of sand, red soil and gravels. From scattered findings in the vicinity and within the material of its structure, the hill was thought to contain an important burial structure (Lazaridis 1993). Seismic tomography was one of several geophysical methods applied in an effort by the local archaeological service to investigate the tumulus by non-destructive methods. Experimental procedure Figure 2. Simplified location map of Kastas Tumulus, Northern Greece.

Seismic tomography is carried out by exciting a seismic energy source at several locations on the sidess of an investigated medium and using the arriving wave energy at all receivers located on the sidess of the medium. Measured characteristics of the received wave energy include amplitude and travel time, inversion of either type of data results in an image of velocity or attenuation, respectively, in the earth material.. Of the above, travel times / velocity will be considered hereafter, since it is easier to obtain from field records and is directly related to the physical character of the earth material.. Inversion can be performed in 2 or 3 dimensions,, considering data quality and angular coverage as well as time and cost. For details on inversion schemes the reader is referred to Nolet (1987) and Jackson and Tweeton (1994, 1997).

with the ramp on the arrival times of seismic waves, by exciting a source at the top of the tumulus and recording times around the base of it. Their simple field layout proved to be successful in locating a tomb. Seismic tomography was thought of as an attractive candidate for the investigation of the interior of a tumulus with a full coverage, non-destructive method in real three dimensions. Seismic tomography is a means of reconstructing the distribution of physical properties in a medium (e.g. earth material),, by using measurements of travel time or amplitude of wave energy propagated through it. it The fundamental concept in tomography is that of projection. Seismic measurements constitute a projection of the internal structure of the earth medium.. An image of the internal structure of the earth medium is therefore produced by combining information from a set of projections obtained at different viewing angles. Applications include lithologic characterization, fracture and void detection, fluid monitoring, stress evaluation, blast assessment and others (Nolet 1987, Friedel et al. 1992, Jackson and Tweeton 1994).

Considering the Kastass tumulus, the tomographic experiment was designed so as to provide the fullest possible angular coverage under the particular field conditions. Data recorders rs and sources were located along the periphery at the base of the tumulus.. Figure 3 presents a plan of the hill and location of sources and receivers.. The circular location of sources and receivers and the elevation difference between them along the base (about 12m in a NE-SW direction), provided for a real 3D tomographic application, aiming mainly at the hypothetical location of tombs near the base of the tumulus.. Along the periphery there here were 120 geophone locations at average distances of 4.5 m and 16 source locations at average distances of 34.5 m.. A total of 2000 records were made. The source was a falling 30 kg weight,, yielding a main frequency of about 70 Hz, while vertical geophones with a resonance frequency of 100 Hz were used as receivers. Data were recorded on a 24 channel EG&G Geometrics 2401X seismograph.

In the context of a tumulus, the fill material of the ramp,, possible disturbed earth material and the burial structures themselves are considered as constituting seismic velocity anomalies ies with their environment and capable of producing a change in the travel times and/or the amplitude of seismic waves. Seismic eismic tomography iss used in order to investigate the very existence, detectability and character of such anomalies. ies. The availability of 3D tomographic analysis algorithms permits an extensive coverage of a tumulus with seismic sources and receivers and allows for an n efficient investigation without a priori postulating the particular tomb locations, s,, thus greatly simplifying the overall investigation task while improving quality and robustness of interpretation.

The layout allowed for coverage of the internal structure of the tumulus in a volume containing the base, located at an average elevation of 90.0 m above sea level, and extending 10 m above and 13 m below the base, having a 108

Ιnvestigation of a Monumental Macedonian Tumulus by Three Dimensional Seismic Tomography

Figure 4. Plots from the tomographic dataset. (a) Time-Distance plot

Figure 3. Plan of Kastas tumulus with source/receiver locations of the tomographic experiment. Numbers near the isolines correspond to elevation a.s.l in meters.

total vertical length (thickness) of 23 m, between elevation of 77 m and 100 m above sea level. Data analysis and processing Figure 4. Plots from the tomographic dataset. (b) Histogram of estimated seismic velocities.

Data analysis Only first arrivals were used in the tomographic analysis. Initial quality control and processing of the field records was made by Tomtime software (Tweeton Tweeton 1999b). b). ). The error in time picking is 3 msec ec on average.. A time-distance plot for orr the first arrivals is presented in Figure 4a, a, while an average velocity of 0.9 km/s is calculated, as shown in the relevant velocity histogram in Figure 4b. With respect to the main frequency of 70 Hz, this yields an average wavelength of 12 m and a corresponding resolution of 3-6 m.

times, calculation of residuals and application of velocity corrections. Forward calculations may be carried out either by straight and/or or curved rays,, employing wavefront migration. Straight rays allow rapid calculation but are less physically realistic compared to curved rays, the latter however, being more time consuming,. The decision upon the use of curved rays is based on the velocity contrasts in the earth material.. The inversion algorithm allows for the use of constraints ts on the velocity values,, in order to reduce the nonuniqueness problem, which results from limited coverage of the investigated medium,, due to the geometry of source and receiver locations and the source character.

Independent information on the seismic velocities of the soil material of the hill and the underlain material provided a two-layer structure with the upper layer having an average velocity of 0.7 km/s and the lower one an average velocity of 1.2-1.4 km/s. The interface was estimated at an elevation of 90 m. Around the base of the tumulus, velocities have a range of 0.4-0.6 km/s, which is compatible with the existence of loose material as verified in situ.

The calculation grid was designed in 3D and matched the source-to-receiver geometry as closely as possible. Considering an average resolution of 4m and mathematical consistency, the grid has dimensions X*Y*Z = 5*5*2.5 m.. Inversion was performed with both straight and curved rays, to allow for handling of strong velocity contrasts, refraction and out-of-plane effects. The starting velocity model consisted of two layers, one upper assigned a velocity of 0.7 km/s m/s (at elevation 90-100 m), and one lower assigned a velocity of 1.2 .2 2 km/s (at at elevation 77-90 m). Minimum and maximum allowable velocities (global constraints) were applied with values of 0.3 3 km/s m/s and 2.0 km/s, respectively. Local (node) node)) constraintss were ere defined appropriately so as to allow for small modifications of velocities during the

Inversion Inversion of the travel time data was made using a variation of the SIRT algorithm (Lytle et al. 1978, Peterson et al. 1985; Um and Thurber 1987) with use of the Geotom software (Tweeton, 1999a).. This involves modification of an arbitrary initial velocity model by repeated cycles of three steps: forward computation of model travel 109

L. Polymenakos, S. Papamarinopoulos, A. Liossis and Ch. Koukouli – Chryssanthaki Modelling the tomb effect An attempt was made to simulate the effect of a tomb ‘buried’ in the soil material of the hill, in order to explore the imaging capability of the actual experimental configuration. The model consists of assigning ssigning a velocity of 1.5 km/s to nodes of the grid ‘occupied’ by the tomb, and 0.5 km/s to the ramp filling material.. After considering information from the literature and direct observation of some representative excavated tumuli umuli in N.Greece, the dimensions of the tomb and ramp were designed signed to be 20 m long, 5 m wide and 5 m high. Following on-site information, the ‘tomb’ was placed at various locations within the volume of the tumulus. Figure 6a shows a horizontal tomogram with the tomb at an arbitrary location and elevation. A tomogram is a slice of the tomographic volume in a preferred direction (horizontal, vertical or other) displaying the distribution of a specific quantity. Using the original experimental source/ receiver geometry, synthetic travel times were calculated and then were input to the inversion process with a uniform average velocity starting model. The calculation grid had similar dimensions than the original, and the same amount and type of iterations were performed. The combined result of straight and curved ray iterations for the arbitrary tomb location and elevation of Figure 5a is presented in a horizontal ontal ntal tomogram in Figure 5b.. The he result of the inversion for a tumulus volume without the tomb is presented for comparison in a horizontal tomogram in Figure 5c.. The original model is satisfactorily reconstructed, also taking into consideration that this result comes from synthetic travel time data which are limited in frequency and energy information with respect to real seismic data and that the initial velocity model is only an estimate with respect to real stratigraphy. In Figure 5b, the tomb itself is only marginally differentiating from the background velocity distribution. However, characteristic high velocity ‘tails’ (artifacts) are observed to expand from the ‘tomb’ location, while a low velocity concave lobe is formed as a result of the ‘ramp’ reconstruction. These features are not observed in the result without the tomb (Figure 5c) thus showing the clear effect of the tomb in the reconstruction. Moreover, the tails act as indirect indicators of the existence of a high velocity anomaly, located at their onset. The reconstructed velocity distribution is not homogeneous: it is characterized by high velocity ‘anomalies’ with a NE-SW trend, located mainly in the ENE sector of the tumulus. This is observed in both data with and without the tomb, and is an artifact attributed to the effect of the source/receiver geometry on the behaviour of curved rays: the elevation difference between the sources and receivers in the NE-SW and NW-SE direction and the velocity interface at 90m elevation produce an accumulation (focusing) of seismic rays in the NE portion, thus resulting in this velocity anomalies.

Figure 5. Modeling of the tomb effect. (a) Grid slice showing the ‘tomb’, (b) Results from inversion of synthetic data calculated for the ‘tomb’ model. (c) Results of synthetic data calculated without the ‘tomb’.

inversion process. The inversion process converged after 5 straight and 10 subsequent curved ray iterations, with an RMS error of 10%. Curved ray iterations were necessary as they did greatly improve the reconstructed velocity distribution, the coverage of the tomographic volume and also stabilized the inversion process. 110

Ιnvestigation of a Monumental Macedonian Tumulus by Three Dimensional Seismic Tomography The above results imply that the detection of even this modest anomaly is possible through an appropriately ppropriately designed experimental configuration and processing. Interpretation Horizontal tomograms of velocity and ray density are presented in Figure 6,, at the elevation of 87.5 m that is below the base of the tumulus which is at an average elevation of 90.0 m. Ray density (rays per unit model cell) is a simple global indicator of the inversion reliability and contributes to the interpretation of the velocity image. The major features of velocity and ray density tomograms are as follows (see Figure 6 a,b): a)

high velocity anomalies (velocities greater than 1.2km/s) appear in the central (H1), eastern (H2, H3), northern (H4) and southwestern (H5-H7) sector. Characteristic high velocity ‘tails’ are observed to radiate from anomalies H1 and H5, resembling the “tomb effect”, such as those observed in Figure 5b. These anomalies are interpreted as possible stone structures.

b) low velocity anomalies (velocities less than 1.2km/s) appear in the S and SW sectors (L1,L2) and along the periphery (L3). Anomaly L1 extends between the periphery and anomaly H1 while anomaly L2 is confined between anomalies H5-H7, producing an image of a highly disturbed material. Low velocity anomalies are interpreted as loose material, which is probably the fill of the trenches dug inside and around the tumulus. In the ray density tomogram, a sharp decrease in ray density is observed in the sector containing anomalies L1 and L2, implying the presence of loose material. c)

the resulting velocity distribution is considered as generally reliable since most parts of the tomogram locations are sampled by more than 100 rays on average and the model fit error (RMS value) is quite low (about 10%). Local increase/decrease in ray density such as at location of anomalies H1-H2/L1-L2 respectively is due to focusing/defocusing of rays in high/low velocity areas.

By comparing Figure 6b to the image of the synthetic data without the tomb (Figure 6c), it is clear that the above anomalies are real and not a result of layout geometry or errors in the initial data assessment.

Figure 6. Tomograms of the results from inversion of real experimental data at elevation 87.5m. (a) Velocity (b) Ray density. Results of synthetic data without the tomb model are presented for comparison (c)

It is proposed that various stone structures, some of which could be tombs, are buried in the tumulus, the overall image resembling that of the example in Figure 1b. Of the anomalies observed, anomaly H1 is the strongest one and, with anomaly L1, have the characteristics to be good candidates for a tomb and a ramp, respectively.

Conclusions The capability of very good angular coverage along the periphery of the tumulus base with a simple field 111

L. Polymenakos, S. Papamarinopoulos, A. Liossis and Ch. Koukouli – Chryssanthaki procedure and three dimensional inversion, makes seismic tomography an attractive exploration tool for the investigation of monumental tumuli. Synthetic tests provide evidence that anomalies produced by buried structures within the tumulus can be reliably detected. Tomograms from inversion of real data provide interesting images of the internal structure of the tumulus, with features which can be associated with man-made structures. Although these results remain to be ground-proofed, it is clearly shown that seismic tomography has the ability to provide important subsurface information for the exploration of artificial hills, in a non-destructive manner. A main advantage of the method is that it can be adapted to the specific site conditions. The study and improvement of the method is a subject of ongoing research at the Geophysical Laboratory of the University of Patra.

Tomography, In 62nd Annual Meeting and Exposition of the Society of Exploration Geophysicists, New Orleans, USA, 58-62.. Jackson, M.J. and Tweeton, D.R., 1994, MIGRATOM Geophysical Tomography Using Wavefront Migration and Fuzzy Constraints, US Bureau of Mines, RI-9497. Jackson, M.J. and Tweeton, D.R., 1997, 3DTOM: ThreeDimensional Geophysical Tomography, US Bureau of Mines, RI-9617. Lazaridis, D., Romiopoulou, K. and Touratsoglou, G., 1992, O tymvos tes Nikissianes (The tumulus of Nikissiani), Ellenomnemon Libr. Arch. Soc. Athens, 121, Athens (in Greek with english abstract). Lazaridis, D., 1993, Amfipolis, a guide to the antiquities, Ministry of Culture, Arch. Res. Restor. Fund (TAPA), Athens. Lytle, R.J., Dines, K.A., Laine, E.F. and Lager, P.L., 1978, Electromagnetic Cross-Borehole Survey of a Site proposed for an urban Transit Station, UCLR-52484, Lawrence Livermore Laboratory, University of California, USA. Nolet, G. (ed.), 1987, Seismic Tomography, with Applications in Global Seismology and Exploration Geophysics, Reidel, Boston, USA.. Peterson, J.E., Paulson, B.N.P. and McEvilly, T.V., 1985, Applications of Algebraic Reconstruction Techniques to Crosshole Seismic Data, Geophysics, 50(5),1566-1580. Tsokas, G.N., Papazachos, C.B., Vafidis, A., Loukoyiannakis, M.Z., Vargemezis. G. and Tzimeas, K., 1995, The detection of monumental tombs buried in tumuli by seismic refraction, Geophysics, 60(6), 6), ), 1735-1742. Tweeton, D.R., 1999(a), Geotomcg, Three dimensional geophysical tomography, Geotom LLC, Minneapolis, USA. Tweeton, D.R., 1999(b), Tomtime, Seismic data analysis and processing, Geotom LLC, Minneapolis, USA. Um, J. and Thurber, C., 1987, A fast algorithm for two-point ray tracing, Bulletin of the Seismological Society of America, 77, 972-986.

Acknowledgements We would like to thank Dr. Chris De Wispelaere the Science For Peace NATO programme director and all the members of the steering group of the former Science for Peace programme for financial assistance in the realization of this highly interesting project. Xenophon Bafitis who greatly assisted during the fieldwork and the members of the Archaeological Service of Kavala who participated and assisted in many ways, are gratefully acknowledged. References Friedel, M.J., Tweeton, D.R., Jackson, MJ.., Jessop, J.A., Cumerlato, C.L., 1992, Mining Applications of Seismic


GEOPHYSICAL INVESTIGATIONS IN PELASGIA & ANDRON A. Sarris, S. Topouzi, S. Soetens, A. Baba Laboratory of Geophysical - Satellite Remote Sensing & Archaeo-environment, Institute of Mediterranean Studies - Foundation of Research & Technology (F.O.R.T.H.), Melissinou & Nik. Foka 130, PO. Box 119, Rethymnon 74100, Crete, Greece, [email protected]

M-F. Papakonstantinou,, K. Psarogianni 14th Ephorate of Prehistoric and Classical Antiquities, Kastro Lamias, Lamia 35100, Greece, Abstract: Geophysical prospection techniques were applied in the regions of Pelasgia & Glyfa/Andron, in Central Greece. In Pelasgia, investigations were concentrated around a mud constructed Bronze Age tumulus. Previous excavations had revealed a couple of burials, wall remains and residues of burning 1m below the surface. The bulk of the tumulus has been destroyed by past interventions. In Andron the prospection survey was concentrated on the lower slopes of the Acropolis to the north and close to the coastline. In the surrounding region, excavations revealed a cemetery containing a large number of graves. High-resolution magnetic gradient and soil resistivity techniques were applied to both regions. In Pelasgia excavations verified some of the candidate targets, revealing concentrations of ceramic fragments, three burials and other finds, mainly of the Middle Helladic era. A number of architectural features were also identified by magnetic techniques in the region of Glyfa/Andron. Περίληψη: Γεωφυσικές διασκοπήσεις πραγματοποιήθηκαν σε διάφορες αρχαιολογικές θέσεις στην ευρύτερη περιοχή της Πελασγίας και Αντρώνα/Γλύφα, στη Φθιώτιδα. Στην Πελασγία, οι έρευνες επικεντρώθηκαν σε τύμβο Εποχής του Χαλκού, στον οποίο είχαν προηγηθεί ανασκαφές από την ΙΔ’ Ε.Π.Κ.Α. οι οποίες αποκάλυψαν δύο ταφές από μεταγενέστερη εποχή, πλιθιές, τοιχοδομή και στρώματα στάχτης σε ένα βάθος μικρότερο από 1μ από την επιφάνεια του εδάφους. Ο τύμβος βρίσκεται στην θέση «Κουμούλι» και είναι κατασκευασμένος από πηλό και χώμα, ενώ μεγάλο τμήμα αυτού έχει καταστραφεί από παλαιότερες επεμβάσεις. Στην Αντρώνα, οι έρευνες πραγματοποιήθηκαν σε δύο σημεία, γύρω από τον λόφο της ακρόπολης. Η μία θέση βρισκόταν πλησίον του δρόμου που οδηγεί προς την παραλία, περιμετρικά της ακρόπολης, ενώ η δεύτερη βρισκόταν πλησίον της ακτογραμμής. Ανασκαφές που είχαν προηγηθεί στην ευρύτερη περιοχή είχαν φέρει στο φως εκτεταμένα τμήματα περιβόλου νεκροταφείων. Η γεωφυσική χαρτογράφηση των παραπάνω περιοχών πραγματοποιήθηκε με την χρήση ηλεκτρικών και μαγνητικών διασκοπήσεων. Μετά από επεξεργασία και ερμηνεία των γεωφυσικών χαρτών και συσχέτισης των αποτελεσμάτων των διαφορετικών τεχνικών, πραγματοποιήθηκαν ανασκαφές σε συγκεκριμένους υποψήφιους στόχους. Στην Πελασγία, οι ανασκαφές έφεραν στο φως συγκεντρώσεις οστράκων κυρίως Μεσοελλαδικής περιόδου, τρεις ταφές που περιείχαν στο σύνολό τους 5 σκελετούς, τμήματα λιθοσωρών και άλλα ευρήματα. Από τις μαγνητικές μετρήσεις έχουν επίσης εντοπισθεί πιθανά αρχιτεκτονικά λείψανα στην περιοχή της Αντρώνας/Γλύφας.

Tumulus of Middle Helladic Date at Pelasgia (Phthiotis) On the summit of the “Koumouli” hill, north-east of the modern town of Pelasgia, the 14th Ephorate of Prehistoric and Classical Antiquities has been excavating (AD. AD. 1970, 1998, 1999) a tumulus of Middle Helladic date and the surrounding area since 1997 (Fig 1). The partially preserved tumulus (Fig 2 and Fig. 9, area I on the excavation plan) is 3

m in height and 5 m in diameter at the base and was built of layers of earth and mud-brick. A large part of its periphery was missing due to human activities in early seventies (Spyropoulos, 1972). Two beaked jugs with Matt-Painted decoration (Fig 3 & 4) were found in the course of these works. South of the tumulus area a Π-shaped -shaped dry-built construction covered with an ash layer was discovered (Fig. 5 – and Fig. 9, area II on the excavation plan), surrounded by a semi-circular artificial mound on the west side.

Figure 1. General view of the excavation.

Figure 2. The tumulus.


A. Sarris, S. Topouzi, S. Soetens, A. Baba, M-F. Papakonstantinou and K. Psarogianni

Figure 6. Details of the magnetic survey that was conducted in the vicinity of the MH tumulus at Pelasgia.

In response to a request from the Ephorate, the Institute of Mediterranean Studies carried out a geophysical survey in order to accelerate the archaeological research. Geophysical investigations employed magnetic and soil resistance techniques; the magnetic methods (Fig. 6) involved the measurement of vertical magnetic gradient using a Geoscan FM36 with high-resolution sampling (0.5m) and the soil resistance techniques used the Geoscan RM15 Twin probe configuration with sampling interval of 1m. Figure 3. Matt-painted beaked jug.

Figure 4. Grey Minyan goblet.

The results of the magnetic survey are characterized by a lack of extreme anomalies, with the exception of a metallic structure constructed above a recently excavated trench which masked the data taken in its vicinity (Fig. 7). Compression of the original dynamic range of the magnetic data was able to enhance the weaker anomalies, some of which were suggested as potential archaeological targets. The linear anomalies indicated in the southern part of the surveyed region were emphasized through the application of high-pass filters and the computation of directional derivatives. The area around the excavation trench indicated a number of isolated anomalies, some of which may have been caused by the disturbance of the anthropogenic layers and residues of the excavations.

Figure 5. The Π-shaped construction.

Similar conclusions were drawn from the measurement of the soil resistance, which indicate a trend of increasing values towards the north section of the region, suggesting that the tumulus was probably constructed just a small distance above the bedrock. Comparison of the magnetic and the soil resistance maps indicated a few anomalies of specific interest (Fig. 8). In the south section of the region, a highly magnetic linear anomaly extends for a length of 35m (x,y=464E, 454N to x,y=501E, 447N), which does not show in the soil resistance map. It is probable that this specific anomaly is related to an older path or property boundary. Another relatively linear feature with a S-N direction intersects the previous one (at A) and is probably related to erosional effects of the hill. The high intensity anomalies E and D are probably related to concentration 114

Geophysical Investigations in Pelasgia & Andron

Figure 7. Geophysical investigations at the MH tumulus of Pelasgia. Map of the measurements of the vertical magnetic gradient (left) and of the soil resistance (right).

of metal fragments. In contrast, the area (x,y=464E, 454N to x,y=501E, 447N) indicates a cluster of linear anomalies (W to E direction), the center of which (B) appears of particular interest in both datasets. Anomaly C was found to be caused by the existence of a terrace. In contrast, evidence for the extension of the tumulus originated by anomalies G, F & H.

have revealed evidence of cemeteries with a large number of tombs (Papakonstantinou, 1994, 1997). The magnetogram of the region exhibits a few strong dipolar anomalies most of which are caused by metal fragments which were distributed all over the surveyed area (Fig. 11). Anomalies D, B and I deserve some further investigation. The application of directional derivatives emphasized the linear features of the magnetic measurements, most of which are concentrated in the center, NE and NW sections of the region. The compression of the dynamic range enhanced anomaly A (x,y=8-15E, 13-18N), where there are indications of the existence of architectural features. Most of the anomalies at E and J extend in the same direction from south to north. Finally, a relatively well-preserved architectural structure of 5m x 5m is obvious on the NW side of the area.

The excavation program of the following periods 2000 – 2002 was based on the results of the above investigations (Fig. 9). Trial trenches have been opened on the six indicated points. No architectural remains were found on points A, B, C and H but at point G excavation revealed a burial. More interesting was area F, where the remains of a building of Middle Helladic date, perhaps a single farmstead, were discovered (Fig. 10). During the Early Christian times the area was used as a cemetery. Ancient Andron

Magnetic methods were also applied along the coast of Andron about 10-15m from the coastline, where the ancient port of the settlement is expected to be. A strong magnetic anomaly at the center of the magnetic map caused by an earth covered well, masked a wide area (about 3m in radius) around the target. Transect equalization techniques managed to smooth away the extreme anomalies and

The geophysical investigations were expanded in the region of ancient Andron, close to the coast of Glyfa. Two areas were surveyed, one at the foot of the Acropolis close to the road leading to the coast and another one in the vicinity of the coastline. Previous excavations in the nearby region 115

A. Sarris, S. Topouzi, S. Soetens, A. Baba, M-F. Papakonstantinou and K. Psarogianni

Figure 8. Diagrammatic representation and codification of geophysical anomalies.


Geophysical Investigations in Pelasgia & Andron

Figure 9. The excavation plan.

Figure 11. Magnetic prospection survey at Andron (Glyfa). Original measurements (left), application of high pass filtering (middle) and diagrammatic representation of the candidate targets (right).

AD.. 1999, Archaeologikon Deltion, 54, Chronika (in press). Papakonstantinou,, M.-F., .-F., F.,., 1994. To notion kai dytiko tmima tis Achaias Fthiotidas apo tous Klassikous mechri tous Romaikous Chronous, in Proceedings of the International Conference “Thessaly”: Fifteein years of Archaeological Research, 1975-1990. Results and Perspective, Lyon 1994, Β, Athens. Papakonstantinou,, M-F., -F., F.,., 1997, Andron: 10 chronia anaskafon - 2000 chronia istorias, Deltio tis Filarchaiou Etaireias Almyrou, Othrys, Period B´, 1, Almyros, 54-68. Spyropoulos,, Th., ., 1972, Dyo ramfostomoi ΜΕ prochoi ek Pelasgias Fthiotidos,, Archaeologika Analekta ex Athinon,, V(3), 470-474.

Figure 10. The remains of the MH building.

highlight the weaker features, a few of which may be related to architectural relics. References AD.. 1970, Archaeologikon Deltion, 25, Chronika 243. AD.. 1998, Archaeologikon Deltion, 53, Chronika (in press).


USE OF REMOTE SENSING AND GIS METHODS IN THE RECONSTRUCTION OF COASTAL PALAEOGEOGRAPHY OF ALEXANDRIA, EGYPT A. Chalari, D. Christodoulou, G. Papatheodorou, M. Geraga, A. Stefatos and G. Ferentinos Laboratory of Marine Geology and Physical Oceanography, Department of Geology, University of Patras, 26110 Rio, Greece, [email protected] Abstract: Submerged monuments and ancient shipwrecks, surficial or buried on the seafloor, constitute significant evidence of the evolution of human culture over time. Remote sensing techniques consisting of profilers and sonars, have become a common practice in underwater archaeology. The treatment of multidisciplinary selecting data can be managed by the applications of GIS. In this paper we propose a simple methodological scheme for the processing of analogue remote sensing data in a GIS environment and we present the results of this application in the coastal zone of Alexandria in Egypt, one of the richest areas in terms of abundance of archaeological remains. Detailed bathymetric and geomorphological maps created in GIS environment, have permitted us to reconstruct the coastal palaeogeography of Alexandria during its glorious past around 2300years B.P. The reconstruction of coastal palaeogeography showed the extent of the submerged ancient Cape Lochias and the existence of a natural breakwater zone protecting the city of ancient Alexandria from heavy weather. The latter is the shallowest member of a sequence of parallel rocky ridges running along the seafloor off the coast of Alexandria. The ridges probably correspond to submerged fossil shorelines. Περιληψη: Η Μεσόγειος θάλασσα, μαρτυρία αέναων συγκρούσεων και εξέλιξης του ανθρώπινου πολιτισμού, αποτελεί πρόκληση για κάθε ερευνητή. Στη ενάλια αρχαιολογία είναι πλέον συνήθης τακτική η εξερεύνηση καταβυθισμένων μνημείων και αρχαίων ναυαγίων με τη βοήθεια σύγχρονων γεωφυσικών μεθόδων. Η πολυσυλλεκτική πληροφορία που προκύπτει από γεωφυσικές και αρχαιολογικές έρευνες για την εξερεύνηση καταβυθισμένων μαρτυριών ανθρώπινης δραστηριότητας, μπορεί να διαχειριστεί σε περιβάλλον Γεωγραφικών Συστημάτων Πληροφοριών (GIS). Σκοπός της παρούσας εργασίας είναι να προτείνει ένα απλό μεθοδολογικό σχήμα αποτύπωσης και διαχείρισης αναλογικών γεωφυσικών δεδομένων σε περιβάλλον GIS και να παρουσιάσει τα αποτελέσματα που προέκυψαν από την εφαρμογή αυτής της μεθοδολογίας, σε δεδομένα που συλλέχθηκαν στην παράκτια ζώνη της Αλεξάνδρειας στην Αίγυπτο. Η παράκτια ζώνη της Αλεξάνδρειας, περιοχή με ιδιαίτερο αρχαιολογικό ενδιαφέρον, χαρακτηρίζεται από πληθώρα αρχαιολογικών ευρημάτων. Από γεωφυσικά και αρχαιολογικά δεδομένα κατασκευάστηκαν σε περιβάλλον GIS βυθομετρικός και μορφολογικός χάρτης μεγάλης ακρίβειας, από τους οποίος προέκυψε η ανάπλαση της παλαιογεωγραφίας της περιοχής με έμφαση στην περίοδο των ελληνιστικών χρόνων γύρω στα 2300 BP. Η έρευνα αποκάλυψε την συνολική έκταση της καταβυθισμένης Άκρας Λοχιάδος καθώς και την ύπαρξη ενός φυσικού κυματοθραύστη που φαίνεται να προστάτευε την αρχαία πόλη. Ο κυματοθραύστης αποτελεί το αβαθές μέλος μιας σειράς παράλληλων καταβυθισμένων βραχωδών ράχεων, που φαίνεται να αντιστοιχούν σε απολιθωμένες παλαιοακτές.


carried out from that year up until October 2002. The earlier three missions took place in partnership with the Hellenic Institute of Ancient and Medieval Alexandrian Studies (H.I.A.M.A.S) and were focused on an area located between the El-Silsila promontory and Stanley Bay (fig.1). The latter three were carried out in partnership with the French Centre d’Etudes Alexandrines (CEA) under the direction of Jean-Yves Empereur and were focused offshore at the Qait Bey where the ancient Pharos used to stand (fig. 1).

Alexandria was founded by Alexander the Great in 330 BC and was one of the most renowned ports of the ancient world controlling the trade of Eastern Mediterranean during Hellenistic and Roman times. The coastal zone of Alexandria hosts the ruins of one of the seven wonders of the ancient world, the famous Pharos Lighthouse, as well as many other ancient and historical monuments and shipwrecks. Archaeological and oceanographic data based on long and short-term observations, suggest that the coastal zone of Alexandria has been under subsidence for the last 2,300 years. This suggests that part of Ptolemaic Alexandria is currently below the present sea-level and thus the coastal zone of Alexandria is one of the most promising areas of the world for archaeological studies.

Modern remote sensing techniques introduce many advantages in underwater archaeology. There are two general approaches (Papatheodorou et al. 2002) regarding the application of these techniques in underwater archaeology: They are being increasingly used to identify, locate and map (i) ancient shipwrecks lying on the seafloor or partly buried in it (Ballard et al. 2000, Phaneuf et al. 2002), and (ii) submerged sites of archaeological interest (submerged ancient cities, port facilities and man-made structures). The deductions of the latter approach leads to reconstruction of coastal paleogeography whose evolution is controlled by eustatic and isostatic sea-level changes

In October 1999, the Laboratory of Marine Geology and Physical Oceanography (E.THA.GE.F.O) of the University of Patras initiated a marine remote sensing survey in the coastal zone of Alexandria. Six scientific missions were 119

A. Chalari, D. Christodoulou, G. Papatheodorou, M. Geraga, A. Stefatos and G. Ferentinos

Figure. 1: Map of the two ports of Alexandria. The grey rectangles indicate the three underwater archaeological survey areas that are taking place in the coastal zone of Alexandria: (1) CEA under the direction of J-Y Empereur, (2) IFAO under the direction of F. Goddio, (3) HIAMAS under the direction of H. Tzalas.

and/or local tectonic movements (Van Andel & Lianos 1984, Lambeck 1996, Papatheodorou et al. 2004 in this volume).

submerged shorelines, which are now buried under loose marine sediments. GIS technologies are currently undergoing a valid expansion into the field of archaeology. The Geographical Information System (ArcView 3.2) presents a useful tool to create flexible databases, which contain and combine spatial data from multidisciplinary remote sensing techniques including data from different surveying periods. The rapid development of GIS methods has permitted archaeologists to link human and environmental factors with spatial information and create maps in space and time. (Brewster et al. 2003, Bal et al. 2003). In the field of Underwater Archaeology, remote sensing technology when used in concert with other common sources of archeological information, generate enormous amounts of data to be handled by underwater or marine GIS. For example, Breman (2003) used marine GIS to incorporate coastal geomorphology, bathymetric data and

The most common remote sensing techniques which are used for the second approach are: (i) side scan sonar and (ii) sub-bottom profiling system. Side scan sonar is an acoustic device used to provide two-dimensional pictures – called sonographs – of the seafloor. The system provides useful information on the morphology and the texture of the seafloor. A sub-bottom profiling system provides an acoustic profile of a section of the sub-bottom. The main advantages of side scan sonar regarding the reconstruction of the coastal palaeogeography are: (i) the ability to rapidly survey large seafloor areas, (ii) the ability to detect seafloor morphological features indicative of submerged archaeological sites or presently submerged shorelines. A shallow-penetration, high-resolution sub-bottom profiling system permits the identification and mapping of 120

Use of Remote Sensing and GIS methods in the Reconstruction of Coastal Palaeogeography of Alexandria Mameluke Sultan Qait Bey built a fort (Qait Bey Fort) in 1480. This area has undergone three underwater archaeological surveys in the last decade. J-Y Empereur (2000) uncovered 2500 pieces of stonework of archaeological interest, scattered over 2.5 hectares off the Qait Bey Fort (fig. 1). These include columns, bases, capitals, sphinxes, statues and some blocks of granite, which were probably parts of the great Pharos Lighthouse. The Institut Europeén d’Archaeologie Sous-marine under the direction of F.Goddio detected, located and mapped submerged harbours and dock works inside the Eastern Harbour of Alexandria, close to the western side of El-Silsila promontory (Goddio 2000) (fig. 1). Tzalas (2000) discovered three architectural elements weighing 4, 10 and 12 tons, which are located on the seafloor adjacent to the eastern side of El-Silsila promontory at a water depth of 7m (fig. 1). These pieces are probably remnants of the Temple of Isis Lochias, which was located at the tip of ancient Cape Lochias. Figure. 2: General map showing the studied area.


archaeological findings in order to support the maritime use of the archaeological site of Tel Shiqmona during the Persian Period.

Instrumentation A marine remote sensing survey was conducted utilizing a 3.5kHz sub-bottom profiling system and a dual-frequency EG&G 260 side scan sonar. Survey navigation was provided by a single Magellan NAV 6500 GPS with an accuracy of ±15m. The survey was carried out onboard two zodiacs, which were suitably modified to meet the specific needs of the present research.

The aim of this paper is twofold: (i) the establishment of a simple methodological scheme for the processing of analogue bathymetrical and marine geomorphological data in a GIS environment and (ii) the applicability of this methodological scheme to the data collected at the coastal zone of Alexandria in order to reconstruct the coastal palaeogeography of the area.


Survey area – archaeological background

The following software programs were used: (a) the navigation software Touratech QV 2.5, (b) the CALDERA GRAPHICS “Cameleo” Vers. 3 software, applied to SEISCANEX images and databases, which provided the facilities to scan the analogue remote sensing records and (c) the Arc View 3.2, used to store data and create digital maps, using several scripts in Avenue language.

The study area extends for 32.5km2 from the Qait Bey site where the ancient Pharos of Alexandria used to stand, to Stanley Bay about 4.6km to the east of the El-Silsila promontory (fig. 2). Strabo (XVII) left a picturesque description of the survey area as it was during the period of Alexandria’s prosperity under the Ptolemaic kings: “On entering the great port (Eastern Port), the island and lighthouse of Pharos lie to the right while on the left are seen a cluster of rocks and Cape Lochias, with a palace standing on its summit. As the ship approaches the shore, the palaces behind Cape Lochias astonish one because of the number of dwellings they contain, the variety of constructions and the extent of their gardens…” (Morcos 2000).

Results and discussion (a) Data flow and processing The positioning, bathymetric and morphological data, which were collected during the marine remote sensing survey, were input into ArcView GIS in order to construct bathymetric and geomorphological maps of the coastal zone of Alexandria. For this purpose two ArcView Projects (R1 and R2) were created; the R1 to construct a bathymetric and the R2 a geomorphologic map (fig. 3).

In the course of time, the Pharos Lighthouse was destroyed whilst Cape Lochias has been lost to the sea, due to subsidence, and only the El-Silsila promontory remains. The Pharos lighthouse was completely destroyed between 1303 and 1349 AD by an earthquake and at its site the

The first step was to obtain an accurate base map with links to positioning database. The positioning data given by the GPS was recorded in UTM projection using the 27o 121

A. Chalari, D. Christodoulou, G. Papatheodorou, M. Geraga, A. Stefatos and G. Ferentinos



Sub-bottom Profiler System Unit

Side Scan Sonar System Unit

GPS Navigation Software

Sub-bottom Seismic Profiles

Side Scan Sonar Records

Tracklines Tables T1, T2 (coordinates, vessel velocity, direction)

Admiralty Chart



GIS (Arc View) Bathymetrical MAP

Geomorphological MAP

Figure 3: Database Flowchart

central meridian and the WGS84 datum. During the survey, a navigational software that supported the GPS operation permitted storing of the positioning data every 20 seconds. This time interval was chosen as the optimum on the basis of vessel velocity (~4knots) and the GPS accuracy (±10m). Two tables of positioning data were created; T1 and T2, consisting of sub-bottom profiler and side scan sonar tracklines positioning data respectively (fig. 3). The data tables were imported to the ArcView GIS as separate “event themes” files. A nautical map (Admiralty chart, 1:37500) of the studied area was imported to the ArcView as “theme” in JPEG (.*jpg) format and was used to digitise the shorelines, the breakwaters and the port facilities.

facilitate the procedure that matches the positioning data to the corresponding depth information, an Avenue script was developed. Initially, the distance between successive positioning data was calculated. Then, by using image processing techniques, this information was exploited to match the positioning data to the corresponding recorded depths. A kriging method was chosen to present the bathymetry of the coastal zone of Alexandria. The selection of method was based on statistical criteria and tests (see below). Scanned side scan sonar records were imported to R2 project and were georeferenced at the UTM projection and the WGS84 datum (fig. 5). A sonograph mosaic was produced by locating georeferenced side scan sonar records from adjacent survey tracklines. The interpretation of the side scan sonar mosaic was based on two main acoustic criteria: (i) the acoustic reflectivity, which is determined by the texture of the seafloor; rock and gravel are better reflectors than sand or mud and therefore show up darker on the record, and (ii) the acoustic shadow that occurs near features relieved from or depressed into the seafloor. High and low-reflectivity seafloor areas and seafloor morphological features were digitised from side scan sonar mosaic. All the digitised textural and morphological data was imported to the ArcView GIS as “themes” in shape (.*shp) file format and seven seafloor types were defined (fig. 6). Through the use of ArcView 3.2 and ArcView Spatial Analyst, the areal extent of the seven seafloor types was calculated.

The second step was to scan the analogue seismic and side scan sonar data using the CALDERA GRAPHICS Cameleo Vers.3 software. This program was developed within the framework of the European SEISCANEX project and the main objective was to create a digital database of analogue remote sensing data (Miles 2001). SEISCANEX also includes digital conversion of analogue data and seismic processing, but for the needs of this survey the digital translation to SEG-Y file format was not used. The scanned remote sensing data was imported to ArcView GIS, as “themes” in JPEG (*.jpg) file format and then were converted to shape (*.shp) file format. Within R1 project, a bathymetric map was constructed using sub-bottom profiles (theme *.shp) and positioning data (table *.dbf) (fig. 4). The scanned records were imported into the ArcView environment. In order to 122

Use of Remote Sensing and GIS methods in the Reconstruction of Coastal Palaeogeography of Alexandria

Sub-bottom Profiler echograms


Themes (*shp)

GIS Arc View

Tables (*dbf)

script (Avenue Language)

Kriging Interpolation

Table (x,y,z)

Bathymetrical MAP

Figure 4: Bathymetrical Map Flowchart (R1 Project)

(b) Selection of interpolation method

Side Scan Sonar records

There are numerous mathematical techniques available for griding irregularly spaced data. The four most common methods are: (i) C1 Spline, (ii) Kriging, (iii) Inverse Distance Weighted (IDW) and (iv) Polynomial trend. Kriging is the most widely used interpolation method, which is applied in two main types: Ordinary and Universal. Universal kriging takes into account the global trend of the data and ordinary kriging assumes that there is no trend in the data and thus the data have a constant mean value.


Comparisons between kriging (universal and ordinary) and inverse squared-distance weighted (ISDW) methods based on the analysis of synthetic data from a computational experiment, revealed that the two kriging methods were substantially superior to the ISDW over different surface types, sampling patterns, noise and strength of spatial correlation among error terms (Zimmerman et al. 1999). Furthermore many studies based on real geostatistical data suggest that the kriging interpolation method performs better (Phillips et al. 1997). On the other hand, some scientists (Gallichand & Marcotte 1993) suggest that spline or IDW fits on their data set as well, or even better, than kriging.

GIS Arc View

georeference scanned sonographs


Sonograph Mosaic



Geomorphological Map

Figure 5: Geomorphological Map Flowchart (R2 Project)

kriging,]. The main use of trend surface analysis is not to interpolate, but mainly to remove broad trends from the data prior to the use of other interpolation methods. This method is highly affected by extreme values and uneven distribution of observational data points, therefore is inappropriate in interpolating topographic surfaces. Triangulation is a very flexible and accurate interpolation method, but gives a jagged appearance and it is not

In order to choose the most appropriate interpolation method for our data we performed a lot of tests and criteria using the five most common methods that exist in the GIS environment [polynomial trend (TREND), triangulation (CREATETIN), inverse distance weighted (IDW), spline, 123

A. Chalari, D. Christodoulou, G. Papatheodorou, M. Geraga, A. Stefatos and G. Ferentinos Ordinary kriging was used in order to interpolate our data and create the bathymetric map (fig. 8). (c) Application: Coastal area of Alexandria The remote sensing survey identified seven main seafloor types that extend into the study area, on the basis of the acoustic reflectivity and morphological configuration (fig. 6). The calculation of the areal extent of each seafloor type shows that the coastal zone of Alexandria mainly consists of rocky seafloor (fig. 6) (table 1). There are extensive areas (8.3 km2) of highly irregular rocky relief and small areas of smooth rocky relief (0.6 km2) as indicated by the presence and the lack of acoustic shadow respectively, on the sonographs (fig. 6). Loose sediments cover a significant part (5.8 km2) of the surveyed area and consist of sand, sandy mud and mud as indicated by the high, medium and low reflectivity pattern on the sonographs (fig. 6). The sandy cover of the seafloor is often well rippled by the waves, whilst the sandy-mud and mud cover is fairly smooth. Within the loose sediments and close to rocky seafloor, small areas of scattered rocky outcrops were detected (fig. 6). Among the above described seafloor types, two are the most significant. The first is the extensive sandy/muddy seafloor areas, which are considered of archaeological importance for the detection of well-preserved ancient shipwrecks and the second type is the rocky ridges, which are the most striking morphological features on the seafloor and are considered significant in the reconstruction of coastal palaeogeography.

Figure 6: Geomorphological map created in an ArcView3.2 environment based on side scan sonar data. A-F: the location of the six rocky ridges.

suitable for extrapolation beyond the observed data points. Although IDW is a popular interpolation method due to its simplicity and speed in calculation, it has not been chosen since it can be easily affected by an uneven distribution which characterizes the observational data points of the survey area. The Spline interpolation method is best for gently varying surfaces, and therefore was not considered suitable for our data set, which is characterized by abrupt changes in depth within short horizontal distance.

The sandy/muddy seafloor is the most promising environment for the preservation of ancient shipwrecks. So far, the most well-preserved remnants of shipwrecks have been discovered on the surface or buried under soft seafloor rather than on a rocky seabed (Ballard et al. 2000). Furthermore, the detection and identification of ancient shipwrecks, using remote sensing techniques, is much easier on a flat sandy/muddy seafloor than on a rocky one.

The Kriging interpolation method seems to be the most suitable geostatistical method for our data set. Kriging was applied in the (x,y,z) table (fig. 4) in order to create a detailed bathymetric map. For this purpose, the variogram of the data set was constructed. The variogram values increase with distance and reach an upper level or sill, which indicates that there is not an obvious global trend in our data and therefore universal kriging is not chosen.

Based on the above geomorphological map, (fig. 6) Chalari et al. (2003) carried out a small-scale side scan sonar and a diver cross validated survey focused on the sandy/muddy seafloor areas. The side scan sonar survey resulted in the identification of 26 targets of potential archaeological interest, on the basis of acoustic signatures. The visual inspection of the 26 targets showed that most of them were natural features (predominantly rocky outcrops) and man-

High relief 20.2%

Rocky Seafloor

Seafloor Covered with Loose Sediments



Ridges 29.9%

Scattered Rocks 6.45%

Low relief 4.2% Table 1


Sand 26.7%

Mud 9%

Muddy Sand 3.5%

Use of Remote Sensing and GIS methods in the Reconstruction of Coastal Palaeogeography of Alexandria





Figure 7: Sub-bottom profiling data showing: (1) ridge A, (2) ridge C, (3) ridge D and (4) ridge E.

made objects with no archaeological interest and only one target was a well-preserved shipwreck.

ridge, with a width of about 680m and height of about 614m above the surrounding seafloor. The minimum water depths of this ridge obtained at 12m forming a narrow planar strip (fig. 8a).

Six individual ridges were detected which are between 20m and 680m wide and about 40m to 800m apart (fig. 6). The ridges, which are located between the 12m and 45m isobaths, run almost parallel to the contours and presentday shoreline, having a dominant strike direction of about ~45o (fig. 6 and fig. 7).

The fact that the ridges discussed above aligned almost parallel to the present-day coastline, lead us to suggest that these ridges are a “submerged barrier reef system” which must have developed along with the advancing sea-level and therefore is considered a “fossil” shoreline system. Fossil shorelines, which include submerged terraces and shelf edge ridges being considered as reefs or beach barriers, are known from the shelf of India (Vora et al. 1996), outer shelf of Bengal (Wiedicke et al. 1999), and Madagascar, Tahiti, New Guinea and the Fiji islands (Guilcher 1988). Wiedicke et al. (1999) interpreted the outer-shelf rocky ridges of Bengal as “drowned” beach barriers based on the geomorphology (parallel to contours), shallow water lithology (oolites) and their Late Pleistocene age.

Sonographs display the ridges as highly reflective stripes, sharply contrasting against the low reflectivity of the surrounding seafloor, which is covered by loose sediments (fig. 8). Examination of the sonographs showed that the individual ridges are often composite features made of patches or smaller strips of high reflectivity and thus have irregular and undulating boundaries (fig. 8). It should be mentioned that the boundaries configuration is also controlled by the wave-induced transportation of loose sediments. The southern ridge (ridge A) is the best-shaped 125

A. Chalari, D. Christodoulou, G. Papatheodorou, M. Geraga, A. Stefatos and G. Ferentinos There is no doubt that to verify the above hypothesis further study with lithological and dating criteria is required. Within the above concept so far three samples, which were collected from rocky seafloor in the vicinity of the rocky ridges, were microscopically examined. All the samples consist of spheroid and to a lesser extent, ellipsoid oolites cemented with micritic calcite. Ooid nuclei include calcite and less frequently quartz, chert and Fe-Ti oxides. These microfacies characteristics argue for a very shallow highenergy nearshore environment, similar to that of a coast. The age determination of the ridges require radiometric dating measurements since age estimation at coastal environments based on a global eustatic sea-level curve (see Fairnbanks 1989), as is widely used for similar cases, is not recommended here due to the rate of subsidence of the studied area. The subsidence of the coastal zone of Alexandria is well documented by short-term (oceanographical data) and long-term (archaeological data) observations. Tide-gauge data suggests that Alexandria has been subsiding during the last 60 years at a rate of 2mm/year (Frihy 1992). Greek and Roman settlements along the coastline are 5 to 8m below the present sea level suggesting an average rate of subsidence of about 2.6mm/year for the last 2300 years (Frihy 1992). In addition Jondet (1912, 1916, 1921), the Chief Engineer of the Department of Port and Lighthouse, discovered a line of submerged ruins of an ancient breakwater of the Western port of Alexandria at a water depth ranging between 6.5 and 8.5m below the seasurface at 1915. Furthermore submerged ancient harbour facilities were also discovered in the Eastern Harbour at a water depth of about 6-7m (Goddio 2000). All the above suggest that the coastal morphology of Alexandria is now markedly different from what it was 2300 BP. If we assume a subsidence of 8m for the last 2300 years, a new coastal configuration is obtained using the bathymetric map (fig. 9). Based on the above-mentioned scenario, some remarkable conclusions can be reached regarding the coastal palaeogeography of Alexandria: a)

Figure 8: Side scan sonar records showing: (a) ridge A, (b) ridge B and (c) ridge D.

An extended part of ancient Cape Lochias where the Royal Palaces stood during Ptolemaic times, is now under the sea-level surrounding the presentday El-Silsila promontory (fig. 9 and 10) .

is no doubt that for the detailed reconstruction of ancient Cape Lochias during its glorious past, more underwater archaeological and small scale remote sensing surveys are required.

The areal extent of ancient Cape Lochias, defined by the 8m isobath, covered an area of about 0.854km2, of which 0.79km2 is now submerged under the present sea-level (fig. 9 and 10). This realistic estimation is well supported by recent archaeological findings. Goddio (2000) discovered sunken port structures west of El-Silsila at a water depth of 6-7 meters, and Tzallas (2003) found architectural elements of large dimensions on the submerged Cape Lochias east of El-Silsila promontory at a water depth of 7m. There

b) The entrance to the Eastern Harbour was very narrow in the past and was located very close to Pharos site (fig. 10). In the course of time due to subsidence, it became so wide that a breakwater was constructed to protect the piers. c)


The southern ridge (ridge A), which is located about 1km to the north of the Qait Bey, was 4m below the sea-level 2300years ago (fig.9 and 10). Taking into consideration that the maximum swell is about

Use of Remote Sensing and GIS methods in the Reconstruction of Coastal Palaeogeography of Alexandria 5m, the southern ridge was occasionally exposed at the sea-surface and therefore can be considered as a reef. That ridge was probably a natural barrier, a breakwater zone protecting the Eastern harbour from heavy weather. On the other hand, that reef could also have a disadvantage; the rapid and unexpected rise of the rocky seafloor, during heavy weather, was a dangerous zone for ancient ships approaching the entrance of the Eastern Harbour of Alexandria . The above scenario is well demonstrated by ancient literature and recent archaeological findings. Strabo (XVII) pointed out that the danger of the Alexandrian coast for ancient mariners arises from reef at the seasurface or just below it. J-Y Empereur discovered remains of 40 Greek and Roman shipwrecks 350m to the north of Qait Bey. All the wrecks are located on a rocky bar, which is now at a water depth of 12m. Although the area is nearby the Pharos Lighthouse site, which was operating at that time, the existence of the wrecks suggests that the ancient ships crashed onto the exposed rocky bar during heavy weather which, based on a subsidence of 8m in the last 2300 years, was 4m below the sea-level . The geomorphological characteristics of the southern ridge (A) match well J-Y Empereur’s hypothesis, suggesting that this is a site of potential interest and a detailed underwater archaeological survey must be carried out in the near future.

Figure 9: Bathymetric map created into an ArcView 3.2 environment based on sub-bottom profiling data.

Figure 10: 3D representation of the coastal palaeogeography of Alexandria created in an ArcView environment. The arrows indicate: (1) the submerged ancient Cape Lochias, (2) the entrance of the Great Eastern Harbour of ancient Alexandria and (3) the ridge A, which is discussed in the text.


A. Chalari, D. Christodoulou, G. Papatheodorou, M. Geraga, A. Stefatos and G. Ferentinos Acknowledgments

Pharos, Mémoires présentés à l’ Institut Égyptien, IX, pp. 101, planches I-IV (Cairo). Jondet, G., 1921, Atlas Historique de la Ville et des Ports d’ Alexandrie, Mémoires présentés à la Société Sultanieh de Géographie, II. XVII, pp. 13, planches I-L IV (Cairo). Lambeck, K., 1996. Sea-level changes and shore-line evolution in Aegean Greece Upper Palaeolithic time, Antiquity, 70, 588610. Miles, P.R. et al., 2001, Developing a European facility to Reuse Seismic Data, EC FP5 project EVR1-CT-2001-40016, European Commission. Morcos, S.A., 2000, Early discoveries of submarine archaeological sites in Alexandria, Coastal management sourcebooks 2. Underwater archaeology and coastal management. Focus on Alexandria, pp: 33-45, Unesco Publishing. Papatheodorou, G., Geraga, M., Chalari, A., Stefatos, A., Cristodoulou, D., and Ferentinos, G., 2002. Remote sensing in Underwater archaeology: Case studies from Greece and Alexandria (Egypt). 8th International Symposium on ship construction in Antiquity, Hydra 28-30 August 2002 (in press). Papatheodorou, G., Geraga, M., and Ferentinos, G., 2004, The reconstruction of prehistoric shorelines in Dokos island, Aegean sea, using remote sensing techniques, 4th Symposium on Archaeometry organised by the Hellenic Society of Archaeometry (28-31/5, 2003, Athens, Greece) (in this volume). Phaneuf ,B.A., Ciavola, P., Papatheodorou, G., and Ferentinos, G., 2002, Ionian Sea Study 2001, I.N.A Quarterly, 29(3/4), 28-29. Phillips, D.L., Lee, H.E., Herstrom, A.A., Hogsett, W.E., and Tingey T., 1997, Use of auxiliary data for spatial interpolation of ozone exposure in southeastern forests, Envirometrics, 8, 43-61. Stabo (XVII). Geography, 17.1.6 and 17.1.9, Thesaurus Linguae Graecae (TLG), University of California Irvine. Tzalas, H.E., 2000, The two ports of Alexandria. Plans and maps from the 14th century to the time of Mohamed Ali, Coastal management sourcebooks 2. Underwater archaeology and coastal management. Focus on Alexandria, pp: 22-32. Unesco Publishing. Tzalas, H.E., 2003. Greco-Egyptian underwater archaeological survey near Alexandria. Towards integrated management of Alexandria’s coastal heritage, pp: 74-79. Unesco Publishing. Van Andel, T.H., and Lianos, N., 1984. High-resolution seismic reflection profiles for the reconstruction of postglacial transgressive shorelines: an example from Greece, Quaternary Research, 22, 31-45. Vora, K.H., Wagle, B.G., Veerayya, M., Almeida, F. and, Karisiddaiah, S.M., 1996, 1300 km long late PleistoceneHolocene shelf edge barrier reef system along the western continental shelf of India: occurrence and significance, Marine Geology, 134, 145-162. Wiedicke, M., Kudrass, H.-R., and Hübscher, Ch., 1999, Oolitic beach barriers of the last Glacial sea-level lowstand at the outer Bengal shelf, Marine Geology, 157, 7-18. Zimmerman, D., Pavlik, C., Ruggles, A., and Armstrong M.P., 1999, An experimental comparison of Ordinary and Universal Kriging and Inverse Distance Weighting. Mathematical Geology, 31(4), 375-390.

The authors thank Dr. J-Y Empereur, Director of Céntre D’Etudes Alexandrines (CEA) for his keen interest and encouragement. Thanks are also due to Mr. H. Tzalas, President of Hellenic Institute of Ancient and Medieval Alexandrian Studies (H.I.A.M.A.S). We also thank Dr. P.Karaivazoglou who helped with the development of the ArcView script that we used for the bathymetry. This survey was funded by the CEA and the “Karatheodory Programme” of University of Patras (grant: 2460). References Bal, Y., Kelling, G., Kapur, S., Akça, E., Çetin, H., and Erol, O., 2003, An improved method for determination of Holocene coastline changes around two ancient settlements in southern Anatolia: a geoarchaeological approach to historical land degradation studies, Land Degradation & Development, 14, 363-376. Ballard, R.D., McCann, A.M., Yoerger, D., Whitcomb, L., Mindell, D., Oleson, J., Singh, H., Foley, B., Adams, J., Piechota, D., and Giangrande, C., 2000, The discovery of ancient history in the deep sea using advanced deep submergence technology, Deep-Sea Research I, 47, 1591-1620. Breman, J., 2003, Marine archaeology goes underwater with GIS, Journal of GIS in Archaeology, I, 24-32. Brewster, A., Byrd, B.F., and Reddly, S.N., 2003, Cultural landscapes of coastal forages: An example of GIS and drainage catchment analysis from Southern California, Journal of GIS in Archaeology, I, 49-60. Chalari, Α.,., Christodoulou, D., Papatheodorou, G., Geraga, M., and Ferentinos, G., 2003, Marine Geophysical Investigation Around the Site of the Famous Pharos of Alexandria (Egypt) for the Detection of Ancient Shipwrecks. (Preliminary Results), 2nd World Congress. Ancient Greece “The Modern World”, Ancient Olympia (International Olympic Academy), 12-17 July 2002, pp: 62-73. Fairbanks, R.G., 1989. A 17,000 year glacio-eustatic sea level record: influence of glacial melting rates on the Younger Dryas event and deep-ocean circulation, Nature, 342, 637-642. Frihy, O.E., 1992, Sea level rise and shoreline retreat of the Nile Delta promontories, Egypt, Natural Hazards, 5: 65-81. Gallichand, J., and, Marcotte, D., 1993, Mapping clay content for subsurface drainage in the Nile Delta, Geoderma, 58(3-4), 165-179. Goddio, F., 2000, Underwater archaeological survey of Alexandria’s Eastern Harbour. Coastal management sourcebooks 2. Underwater archaeology and coastal management. Focus on Alexandria, pp: 60-63. Unesco Publishing. Guilcher, A., 1988, Coral Reef Geomorphology, Wiley, New York. Jondet, G., 1912, Les ports antiques de Pharos, Bulletin de la Société Archéologique d’ Alexandrie, 14, 252-266 (Alexandria). Jondet, G., 1916, Les ports submergés de l’ ancienne île de


PROSPECTING OF THE CITY WALL AT THE ASEA PREHISTORIC SITE (GREECE) USING GEOELECTRIC AND GPR TECHNIQUES M. Dogan, St.P. Papamarinopoulos and P. Stefanopoulos University of Patras, Department of Geology, Laboratory of Geophysics, 26110, Rio, Patras, Greece Abstract: The Multi-electrode Resistivity Tomography and GPR geophysical techniques were applied at multi-cultural site of Asea as collaboration between the Swedish Institute of the Athens and University of Patras. This paper presents the geophysical results of the second field season 2002. The survey intended to explore especially the buried extension of the city wall to contribute progressive research of responses of methods to subsurface intact archaeological features (wall, void, tunnel etc.). The survey was performed by means of the Geopulse imager system with 25 electrodes of 1m electrode spacing, Geoscan RM4 and SIR-10 GPR system. The surveys accomplished both in profile and grid modes and Wenner-Schlumberger hybrid array was selected for resistivity measurements. The city wall was imaged. The results revealed that resistivity method can provide a clear pseudo-image of medium to large size wall in high accuracy for archaeological sites. The interpretation of properly processed 3-D GPR data allowed identifying the position, the dimension and the orientation of subsurface city wall. Μέθοδοι ηλεκτρικής τομογραφίας και του γεωραντάρ εφαρμόστηκαν σε ερευνητικό πρόγραμμα συνεργασίας μεταξύ της Σουηδικής Αρχαιολογικής Σχολής και του Πανεπιστημίου της Πάτρας στον αρχαιολογικό χώρο της Ασέας στην Αρκαδία. Η παρούσα μελέτη αφορά τα αποτελέσματα της δεύτερης φάσης των ερευνών που διεξήχθησαν το 2002. Οι γεωφυσικές διασκοπήσεις εστιάστηκαν στην διερεύνηση της προέκτασης του τείχους της πόλης, ενώ ταυτόχρονα είχαν και ένα πειραματικό χαρακτήρα σε σχέση με την καταγραφή των σημάτων από διαφορετικούς αρχαιολογικούς στόχους. Οι μετρήσεις της ηλεκτρικής τομογραφίας διεξήχθησαν με το Geopulse imager και 25 ηλεκτρόδια με απόσταση 1m m μεταξύ τους και σε διάταξη Wenner-Schlumberger. -Schlumberger. Schlumberger.. Το Geoscan RM44 και το SIR-10 -10 GPR χρησιμοποιήθηκαν επιπρόσθετα στην ευρύτερη χαρτογράφηση της περιοχής. Η ηλεκτρική τομογραφία είχε ως αποτέλεσμα την αποτύπωση τμημάτων του τείχους μέσου και μεγάλου μεγέθους. Η τρισδιάστατη απεικόνιση των αποτελεσμάτων του γεωραντάρ επέτρεψε την καλύτερη οριοθέτηση του τείχους, τον προσδιορισμό των διαστάσεων αυτού και του προσανατολισμού του. Λέξεις-Κλειδιά:Aσέα, οχυρωματικό τείχος, ηλεκτρική τομογραφία, γεωραντάρ, Wenner-Schlumberger array, SIR-10 Keywords: Asea, City wall, Multi-electrode resistivity tomography, Geopulse resistivity meter, and Wenner-Schlumberger array, Pseudo-image, SIR-10..


the early 1990s, the first reconnoitering tour of the valley was directed by J. Forsen and the Asea valley again came into focus of archaeological research as the Asea Valley Survey 1994-1996 and the excavation of the Doric temple on Agios Elias 1997.

The application of geophysical methods changed the starting point of the scientific orientation in archaeology in the 20th century (Weymouth, 1986). The modern developments in geophysical related digital electronic technology have increased the ability of geophysical capabilities to provide an extremely rapid, three-dimensional reconnaissance of a site (Scollar et al., 1990). In addition to the use of multiple techniques on a given site can vastly expand our understanding of its geophysical characteristics and archaeological structure (Clay, 2001). In this respect, combination of a various information obtained from the different methodologies and their 3-D image presentation from different processing can help give a clear picture of the studied areas. Concordance between the results of two or more different techniques in signaling a specific target would increase the reliability and confidence of the target detection itself.

Geographically, the site is a restricted area between the larger alluvial areas in the Megalopolis basin and the plains of Tripolis (Fig.1). The valley bottom is the source area of two large Peloponnesian Rivers, the Alpheios and the Eurotas. Asea (Paleokastro) is a large (250x100 m) evidently flat-topped hill with steep sides that is detached from the lower-most southeast slope of Agios Elias (Forsen, 1996). Immediately to the southwest of the hill lies the railway station of Asea, around which geophysical surveys were carried out (Fig.1). Geological characteristic for the Paleokastro is the bedrock, which consists of limestone lenses in flysch (Jacobshagen, 1986). This evidently separates this area from the alluvium deposits of the Alpheios River at the southeast down hill (Fig.1) on where the grids of interest were laid out.

The site The Asea valley is located between Tegea and Megalopolis and measures 8x7 km in size (Forsen & Forsen, 2002). It has always been of great strategic importance, as it is providing the only easy way of communication between the prosperous plain of Tripolis, with the strong ancient cities of Tegea and Mantinea, and the plain of Megalopolis. In

The city wall The Lower city fortification walls of Asea were built using the rustic polygonal technique and belong to the 3rd century B.C, namely the Hellenistic period (Forsen & Forsen, 129

M. Dogan, St.P. Papamarinopoulos and P. Stefanopoulos Early studies at The Asea Prehistoric Site The site was first excavated in the 1930s by a Swedish archaeologist, Erik J. Holmberg. He found a prehistoric settlement with abundant remains from the later part of the Neolithic period to the Middle Hellenistic period (Holmberg, 1944). He excavated the Paleokastro close to hill. In the early1990s, the first reconnoitering tour of the valley was made by Forsen and archaeological fieldwork started in 1994. Concurrently with the archaeological survey, a geological team led by M. Levanto (2002) studied the geomorphology of the valley. He analyzed of total 31 hand auger drilling core samples collected from the alluvium deposit at the east and the south of the Paleokastro. He evidenced that cultural layers dating from the Archaic to Roman periods are buried at a depth of 2-3 m below the surface, whereas the deeper layers were 14C dated to the Early Bronze Age respectively the Late Bronze Age. 2002 Season survey details and methodology The results achieved in 2001 survey (Dogan & Papamarinopoulos, 2003) season were encouraging to shift survey location through downhill side of the Paleokastro region on the alluvial deposits of the river Alpheios to carry on searching buried archaeological features and the southern extension of the subsurface lower city wall and its turning points inside the site. Total number of 16 grids in 20m x 20m, and 1 grid 10m x 10m in dimensions were set up on the downhill side of the site (Fig.2). Two of them were measured at upper side of Megalopolis-Tripolis highway and the rests were surveyed at the lower side of the road.

Figure 1: Location maps of the research area. The location of the Asea is indicated by white circle.

2002). The only visible parts of the wall that ringed the lower city are the two spurs which run down slopes of the Acropolis (Fig.2) in the north and the south The Northern wall, which is the subject of the resistivity survey in Asea in 2001 research season, downhill from the northeastern corner of the Acropolis and runs straight down to the valley in a northeasterly direction (Fig.2). The distance from the Acropolis walls to the tower is 40 m. It is visible in all its length. The width of wall is 3.3 m. The wall blocks are very large and frequently measuring 1-1.5 m in length and 0.5-1.2 m in heights. The two wall faces are built with large blocks in the polygonal technique of the same kind as the southern lower city wall. The blocks have straight sides and were placed on the bedrock without a footing course. The blocks of the two wall faces, seemingly stepping up the hillside like staircase, have been visible through the centuries. A substantial tower having an inner room is located 40 m from the Acropolis on the wall. It measures 6.6 m in projection and 6.45 m in width. It was built with very large blocks up to 1.5 and even 2 m length, and about 0.7 m in height (Forsen & Forsen, 2002). On the wall, the towers are located at intervals of 100 Greek feet (33 m).

Electrical imaging Initially, a site-reconnaissance survey was set up using Geoscan RM4 resistance meter that was configured as twin probe array. Data were acquired at 1 m interval along the traverse and with 1 m separation between traverses. For the duration of the survey, it was beard in mind that one consequence in relevance to the twin-probe array is the probe separation or depth criterion becomes only an approximation in complex archaeological deposits. The G15 and G16 displayed prominent resistance anomalies (Fig.2) with RM4 survey and were selected to be surveyed using multi-electrode electrical tomography technique. The method is more time-consuming than the twinprobe resistance measurements, other than it specifically investigates the stratigraphy and plan views of subsurface at various depths. By the tomography technique the same piece of the ground at different sensing depths yielded profiles of data along any transect. Campus Geopulse resistivity meter with 25 electrodes was opt for tomography 130

Prospecting of the City Wall at the Asea Prehistoric Site (Greece) Using Geoelectric and GPR Techniques

Figure 2: Topographic map of the Asea site illustrating locations of measured sections (lines-2001) and grids (squares) (adapted from Holmberg, 1944).

survey at the Asea site. Considering that the site consists of subsurface cultural deposits up to 4 m depth (Levanto, 2002), 1m electrode spacing for Grid-15 and 0.8 m for Grid-16 were selected to scan the site up to the aforesaid depth. Wenner-Schlumberger electrode configuration was employed for the tomography survey considering its high horizontal and vertical depth resolution. Throughout the resistivity survey at the Asea site, problems of instrumental internal temperature increase due to hot weather (45oC) that restrained the measurements and insufficiency of batteries were experienced.

12 grids (Fig.2), 20m x 20m in size, were scanned using a SIR-10 digital pulse radar system produced by GSSI at the prehistoric site of Asea. The radar sections are in a traverse configuration at 2 m intervals. Radar reflections along the transects were recorded continuously across the ground at 16 scans per meter. All radar data reflections within 80ns (500 MHz) time windows were recorded digitally in the field as 8 bit data on tapes and 512 data per scan. The useful results were extracted from the data by background filtering. The results from the survey were presented using time-slices techniques (Goodman et al., 1995).

Ground penetrating radar

An attempt to convert the time slices to depth has been investigated (Fig.3) in which a hyperbola was fitted to faint hyperbolic reflections seen in the raw data. Hyperbola fitting is a standard process available in GPR-Slice software (Goodman, 2003). A hyperbola is fit to real data in which a dielectric of 25 was found. Figure (3) shows the hyperbola obtained with GSSI 500 MHz antenna at the Asea site. The horizontal axis represents distance in meters. The range setting is 80 ns two-way travel time (TWTT). The ground surface reflection is approximately at 14 ns, the peak of the hyperbolic reflection is at approximately 33.3 ns, and a good choice for the “bottom” of the hyperbola is

Ground-penetrating radar (GPR) survey was carried out at the prehistoric site of Asea in addition to the tomography survey. 2001 season resistivity results (Dogan & Papamarinopoulos, 2003) suggested that the site may be amenable to GPR survey up to 2 meter depth in which is limited by water table. The purpose of the GPR survey was initially to test the value of radar in respect to depth of penetration and to be carried out as a complementary method to provide additional information on the vertical extent of the two-dimensional horizontal structures imaged with electrical techniques. 131

M. Dogan, St.P. Papamarinopoulos and P. Stefanopoulos

Figure 3: Process of hyperbola fitting in an attempt to convert the time windows to depth.

at about 40 ns TWTT. The distance (2 xs) of the hyperbola is approximately 1.07 m. •

2x = (12.15-13.22) =1.07 m; x=0.535 m.

t1 = (40 –14)/2= 13 ns {divided by 2 to get one-way travel time}

K=C/V K= 0.2998 m/ns / 0.06 m/ns

K = 25

t2 = (33.3 –14)/2= 9.65 ns

Soil Velocity: V = x / V = 0.535 /

where C = speed of light in a vacuum (0.2998 m/ns) and V = velocity of radar energy (in m/ns). All radar travel times were converted into depth slices using predicted average velocity 6 cm/ns for saturated soil (Figs.10 & 12)

(t12 - t22)

(9.652 - 132)=0.06 m/ns Geophysical prospection at the asea archaeological site

In this case a velocity of 6 cm/ns is found for the for saturated soil (Annan & Cosway, 1992) matrix material of Grid16. Given the velocity, the depth can be assigned to the radargram. With the peak of the hyperbola at t2 (ns) below the surface, the depth to the anomaly may be estimated by following equation:


The area of grid-15 The Grid-15 is located next to the western side of TripolisMegalopolis highway and it was extremely in dreadful situation for the applications of geophysical techniques due to thorny plants and intense vegetation. Initially, the undesirable vegetations were removed by means of a rented bulldozer so that geophysical applications would be facilitated. Figure (2) illustrates the layout of the Grid-15. The photo (Fig. 4d) shows the grid area after the surface cleaning and flattening courses.

Vt2 2

d = 0.06m / ns * 9.65ns = 0.6m with an estimate of soil velocity the dielectric (K) can be computed by: 132

Prospecting of the City Wall at the Asea Prehistoric Site (Greece) Using Geoelectric and GPR Techniques 25 20








5 m





a )


20 15





Resistance Ohms










Resistivity Ohms-m


Asea Prehistoric Site Resistivity Survey RM-4 & Electrical Tomography Twin probe & Wen-Schl arrays Sampling interval 1 m


Figure 4: Results of geophysical application in the Grid15 at Asea site. a) Color plot of resistance data b) Electrical tomography depth-slice 0.45m c) 3D wire-net map of depth slice 0.45 m, white arrows indicate the building walls d) photo of the Grid15.

The same features were detected by multi-electrode resistivity tomography technique at the site. Figure (4b) shows enlarged 0.45 depth slice of the Grid-15. The results were acquired with 1m electrode spacing, and the Grid15 was extended 4 meters through west direction in order to be able to insert 25 electrodes with 1 m separation. 20 parallel profiles with 1 m transverse separation were measured allowing %1 error range. Acquired resistivity data were inverted using Res2Dinv software (Loke & Barker 1996) and pseudo-sections were created for each measured profile. All pseudo-sections were combined into three-dimensional form so that to be sliced in depths. Depth slices (Fig.6) were presented using Geosoft software.

The RM4 resistance survey was conducted in Grid-15 (20m x 20m) and 400 data were recorded. Due to high accuracy of data it was not attempted to apply any correction factor upon them and they were gridded by means of Geosoft software so that could be mapped. The color image map of the soil resistance of Grid-15 is shown in Figure (4a) at top-left side of the map. The reddish-purple colors indicate the areas of high resistance whereas green-blue colors represent the low resistance. Resistance values for Geid-15 vary 178 Ohms to 147 Ohms in range, which display about 30 Ohms variation in readings. The map shows a clear picture of a collapsed settlement probably a building. White arrows indicate the collapsed southwestern corner of the building. The likely tower complex of the Southern lower city wall at the southeastern section of the color map is also indicated. It is exactly in the same direction with the visible part of the Southern city wall tower complex that measures 6.6 m in projection and 6.45 m in width. It was built with very large blocks up to 1.5 and even 2 m length, and about 0.7 m in height (Forsen & Forsen, 2002). The width of city wall is 3.3 m and the towers are located at intervals of 100 Greek feet (33 m) on it. The low resistive area at the southeastern section of the map is most probably indicating the filled internal room of the tower complex.

All geophysical responses of the Grid-15 are plotted. Figure (4) shows the geoelectric maps and 3D resistivity distribution of the Grid-15. From the obtained data, it can be seen that there are striking consistencies between RM4 and multi-electrode resistivity tomography (MRT) methods. Turning to the tomography method, one can say that it revealed the clearer image than the twin-probe technique for Grid-15. The predicted building shape is more visible on the color image map (Fig.4b) and its resistivity values range between 167 to 6 Ohms-m (Fig. 5). It sources a 133

M. Dogan, St.P. Papamarinopoulos and P. Stefanopoulos

Figure 5: Line01-Grid15 Inverse model resistivity pseudo-sections for the Wenner-Schlumberger array indicating resistivity values and true depths. Electrode interval is 1m.

Figure 6: Illustration of multi-electrode resistivity tomography depth slices in meters from Grid15 at Asea archaeological site

rounding image of the building. This also can be observed in 3D wire-net image of 0.35 m horizontal depth slice (Fig.4c) indicated by white arrows.

whereas the 15 m long E-W trending linear part has not accurate width estimation due to restriction of the grid. Since, these dimensions greatly support the hypothesis of existence of the city wall and its tower multipart, it was concluded that the Southern part of the Lower city wall prolongs throughout the location of Grid-15.

The linear feature (Fig.4b) was interpreted to be the Hellenistic Period Southern lower city wall and the rectangular one as its tower complex with an inner room 0.8-0.9 m in length (Forsen & Forsen, 2002). The rectangular part of the feature is about 6 m in width,

The created depth slices (Fig.6) provided depth estimation for the demarcated features within the Grid-15. The large 134

Prospecting of the City Wall at the Asea Prehistoric Site (Greece) Using Geoelectric and GPR Techniques correction factor was applied to its data and ambiguity of them could not be entirely eliminated. The resistance values of Grids-16-17 are ranging about between 297 to 314 Ohms. The three-dimensional wirenet image of resistance distribution of Grids 16-17 (Fig.7, bottom) provides a better understanding of the location and trending of the city wall (white arrow) and a scattered fallen block. In the resistance map, the S-N trending city wall sizes 1 to 3 meters in width including a shallow depression in the company of it. The examination of comprehensible linear trend of the subsurface features in Grid-17 was concluded that they are the walls of buildings, which most probably belong to the dwellings outside of the surrounding lower city wall (J. Forsen personal communication). Due to the restricted time, only Grid-16 was covered by multi-electrode resistivity tomography method. The measurements were obtained by way of 0.8m electrode interval and Wenner-Schlumberger array. The contact resistance with the ground was slightly reduced by saturating the region around the electrodes with water. The produced depth slices of Grid-16 were illustrated in Figure (8). From the obtained resistivity responses, the striking consistency of RM4 results and tomography results was yet again discovered (Figs. 7 & 8).

Figure 7: Geoscan RM4 Twin probe resistance survey of area Grids 16-17. Color plot map and 3D wire-net plot of acquired resistance data, white arrow points the city wall.

multifaceted building and its southern tangential wall to the tower can be well seen at depth 0.65 m. The depth slice of 0.75 m provides a clear image of the subsurface of the Grid-15. At this depth, the lower city wall, the half image of the probable tower complex, the building walls and scattered blocks (about 1 m in length) can be clearly distinguished. It seems that the remains of building starts at 20 cm and disappears at about 1.30 m depth below the subsurface. From this point, it can be concluded that it has approximately 1.1 m thickness, whilst the city wall has 1.2 m and the tower displays more than 1.5 m thickness.

Although, the city wall and its scattered blocks are only visible at depth 0.20 m, the tower can be seen at deeper levels in the 1:800 scaled depth slices color map (Fig.8). This is due to the data normalization to reveal the best image of the subsurface. Otherwise, the rectangular-shape tower could be seen at the same depth with the city wall (Fig. 9). The scattered stones, which are about 1 m in length and 0.30 m in height, are most probably the construction blocks of later dwellings located outside of the city wall. The inverse model resistivity pseudo-section of line06 provides an idea about the dimensions of the city wall constructing blocks in addition to their burial depth. The blocks size approximately 1 to 1.5 m in length and 1 to 1.5 m in height (Fig. 9) and their resistivity ranges about 148 to 7 Ohm-m. These are well consistence with the referred dimensions (Forsen & Forsen, 2002) for the Southern segment of the lower city wall. The tower blocks give the impression that is more weathered and disintegrated in comparing with the city wall’s blocks. The inner room of the tower complex can be very well seen between depth of 0.68 and 1.34 m (Fig.8) below the subsurface.

The area of grids 16-17 The grids of interest are located about 2 meters away from eastern side of Megalopolis-Tripolis highway (Fig.2). This survey area cleaned from long dry grasses by means of the bulldozer and was laid down for further geophysical applications. Grid-16 is 20x20 m in dimension, whereas Grid-17 is 10x10 m due to the difficulty of intensive vegetation in the survey field. Grid-16 and Grid-17 were combined for the purpose of electrical resistance survey using Geoscan RM4 resistance meter. The result acquired from this study is shown in Figure (7). The color image map of the subsurface resistance distribution of Grid-16 exposed an exciting reflection of the city wall, which is clearly outlined by hot colors (reddish-purple) at the left side of the map (It also is indicated by white arrow), whilst Grid-17 (Fig.7) displayed an ambiguous image to be interpreted even if a trend

Further, the Grid-16 GPR sections were measured. The results from the survey were presented using time-slice techniques (Fig.10). These time-slices can be actually used as archaeological site plans based on mapping of radar reflection anomalies across the site (Nishimura & Goodman, 2000). The measured GPR data were subject to background filtering in order to cut low frequency anomalies created 135

M. Dogan, St.P. Papamarinopoulos and P. Stefanopoulos

Figure 8: Demonstration of Grid16’s electrical tomography depth slices in meters.

Figure 9: Inverse model resistivity section of Line06 from the survey conducted at the Grid16. Sampling interval is 0.8 m.

by subsurface geology. In addition to a hyperbola was matched with diffraction patterns throughout the profiles using the processing GPR-Slice software, so that a 2D velocity field can be created (Fig.3). 12 time-slices (Fig.10) were generated at 5.7 ns (or approximately 17.28 cm depth windows) selecting the best color distribution for the plotting.

Despite of the relatively high water table causing attenuation of the radar signal, which proved to be more problematic than was thought originally, useful results were obtained for the Grid-16. An advance technique of three-dimensional visualization of depth-slices allowed fine details of the archaeological remains to be seen. A number of points may be made from those images (Fig. 10). Interpretation 136

Prospecting of the City Wall at the Asea Prehistoric Site (Greece) Using Geoelectric and GPR Techniques

Figure 10: a) produced depth slices for Grid16

of the GPR data revealed two sets of structures locating at W and E sides of the grid, which essentially exist within the top 2 m of stratigraphy separated from by a natural depression run across between them. The western main feature standing out is obviously the city wall regarding its blocks size. It may be also inferred that the northern part of the wall constitutes the tower complex comparing the stone block magnitude with the rest of it. The Eastern one is ascribed to fit in later dwellings in the site concerning its shallower depth. GPR was better resolved for this feature than the resistivity tomography. The constructing blocks size (about 1 m) and location of this later period wall can be well seen between depths of 20 cm and 75 cm (Fig.10a, first 4 depth slices). The results from this survey provided subsurface evidence of the city wall preserved within the topsoil between 20 cm and 2.10 m (5.76ns and 69.3 ns). Especially, S-N trending wall and its well-preserved blocks (Fig.10b) in rectangular shape can be well discerned at depths 52.08 and 75.36 cm. The wall blocks are approximately1 and 3 m in size.

Figure 10: b) magnified depth 52.08-75.36 cm, white arrow shows the city wall and its tower complex constructing block, red shows high amplitude.

A careful visualization and interpretation of the data allowed a detailed description of the structures the 137

M. Dogan, St.P. Papamarinopoulos and P. Stefanopoulos different and changing depths with which each wall was constructed. Additionally, from the height of a single stone layer and the total height of the wall, the changing number of stone layers preserved over the length of the wall can be determined. Taking into account some archaeological constraints, this might show if more than the base of wall is conserved and therefore if there are still some archaeological layers buried in the surrounding subsurface. In the example of the east wall the mean layer thickness is less than 75 cm, showing that it was constructed of rather small stones of the structure. From an archaeological point of view, critical problems regarding the shape of the city wall has been resolved. Previously unknown direction of the wall was discovered. GPR data not only provided the ground plan, but also an estimate of the thickness and depth of the buried walls. In this case, the results of the geophysical survey may suggest that there is no need for excavation providing a clear answer to the southern section fortification wall. However, it is strongly suggested that the grid should be extended towards East direction for further geophysical investigations. The area of grids 3-4-5-6-7-8-9 In this region 7 grids in 20 x 20 m dimensions (Fig.2) were studied using the Geoscan RM4 resistance meter. Sampling interval was determined as 1 m. 7 individually measured grids were combined in order to facilitate the interpretation of the data and have a broad picture of the subsurface resistance distribution of this region. Figure (11) shows the results obtained for Grids 3-4-5-6-7-8-9. Hot colored areas characterize the high resistance, whilst blue-green areas signify low resistance in the map produced of 1:400 scale. Three-dimensional net image of the resistance data was also formed to smooth the progress of recognition of the subsurface buried archaeological features within the Grids 3-4-5-6-7-8-9. The resistance values of the data range between 218 and 187 Ohms.

Figure 11: The color plot map and 3D wire-net plot of resistance data measured within Grids 3456789; white arrow indicates the E-W trending low resistive linear feature. The photo shows the survey area and details.

create low resistance values grasping high moisture in the soils matrix by reason of irrigation in the field. As a part of survey conducted in Asea during 2002 field season, Ground-penetrating radar was used to scan in the area of Grids 4-5-6-7-8-9 (Fig.2). Grid-3 was eliminated to make GPR data collection more comfortable. The SIR-10 GPR system was used with the frequency of the antennas 500 MHz and the survey-sampling interval was set to 2m. Keeping profiles interval 2 m, 158 GPR sections were measured. The acquisition range was 80 ns (500 MHz). The sampling rate was 512 sample/scan. Based on recorded GPR data, every radargram was reworked and background filtered before slicing them, since there is many gain problems between grid sets as well as within the grids. The background removing filtering is expected to remove all the reflection events that occur at the same time such as due to operator, antenna ringing effect (Sternberg & McGill, 1995), uneven soils and vegetation in the same grid etc., leaving only those that are more random (Conyers & Goodman, 1997) similar to those significant archaeological or geological reflections in the subsurface. However, the quality of the data was found poor and noisy even after data processing.

Although, this region of the site consists of largest surveyed area in the 2002 season, any significant archaeological element could not be discovered. This area consists of fine silts and clays (Levanto, 2002) and has been heavily ploughed. A massive low resistive linear feature in the map (near x range of 64-67m and y range 0-40m) may well be a water arc or an ancient street. It seems that the area was subjected mankind work, which distorts the soil allocation within the region. Three-dimensional net-view of subsurface clearly pictures this feature (Fig.11). The linear distribution of the high resistance areas in x range 65-75 m and y range 30-40 m (Fig.11, black arrow) were supposedly attributed to be archaeological. While, high rank piled, compact silts and clays due to plough may provide higher resistance values, low level ones may 138

Prospecting of the City Wall at the Asea Prehistoric Site (Greece) Using Geoelectric and GPR Techniques

Figure 12: a) Illustration of converted GPR time-slices into depth (in cm) as depth slices for Grids456789. 500 MHz antennas, time window is 80 ns, offset is 10.64 ns.

Further, color plot time-slices were created for the GPR data collected within Grids456789. The very last time slice is the assumed 69.36 ns, where an offset of 10.64 ns was found to the ground surface. Figure (12) illustrates the converted color plot transform for time-slice levels into depth slices applying 6 cm/ns average velocity factor for the Asea site.

A high amplitude massive square feature in the slices at SE corner of the depth slices may be interpreted as a well concerning its remote and isolated location and its dimensions from the features located at northern part of the map. The linear trending feature, observed in the resistance map (Fig.11), can also be perceived in the depth slices below 17.28 cm (Fig.12b). White arrows indicate the location of this feature. This good consistency of resistance data and GPR results for the E-W trending linear depressionlike feature may further support the assumption that the observed features can be archaeological and the linear feature may be a street which runs between settlements.

The GPR depth slices of Grids 4-5-6-7-8-9 have not clearly revealed any significant archaeological structure, however, the features having geometric shape and linear trend in the slices below 17.28 cm (near x range of 0-40 m and y 40-60 m range) may well be archaeological, although interpretation of such features are not possible at this time. The thickness of these features was guessed as approximately 40 cm to 2 m in subsurface.

GPR was used as the site reconnaissance tool and data were acquired with 2 m interval. In future, it is strongly recommended that the area of Grids 4-5-6-7-8-9 should be extended to East direction and the records should be done in a more appropriate way for archaeology with 0.5 m interval in order to answer research problems in the site.

Another aspect related to Grids 4-5-6-7-8-9 to be stated is the linear trending of high amplitude features in S-N direction. These features can be well seen in all depth slices and they are about 2 m in thickness below the subsurface. 139

M. Dogan, St.P. Papamarinopoulos and P. Stefanopoulos survey confirmed the existence of the city wall and gave information regarding its dimensions and depth using timeslices technique at the Asea. There is a strong correlation between GPR and MRT data. Thus, it can be concluded that together with MRT, the GPR can be used as an alternative method in the geophysical prospection of Prehistoric sites in Greece. References Annan, A.P. and Cosway, S.W. (1992) Ground Penetrating Radar survey design. Paper presented at the Annual Meeting of SAGEEP, April 1992, Chicago. Barker, R.D., Venkateswararao, T and Thangarajan, M. (2001) electrical imaging for borehole siting in India. Research Note 3. University of Birmingham, UK. Clay, B. 2001 Complementary geophysical survey techniques: why two ways are always better than one. Southeastern Archaeology,   : 31-43. Conyers, L.B. and Goodman, D. (1997) Ground-penetrating Radar: An introduction for Archaeologists. Alta Mira Press, Walnut Creek, CA. Dogan, M. and Papamarinopoulos, S. (2003) Geoelectric prospection of a city wall by multi-electrode resistivity image survey at the prehistoric site of Asea (southern Greece). Archaeological Prospection, (): 1-8. Forsen, J. (1996) “ Prehistoric Asea revisited”. Opuscula Atneniensia, (): 41-72. Forsen, J. and Forsen, B. (2002). The Asea valley survey. An Arcadian mountain valley from the Paleolithic period until modern times. ActaAthens-4, (), Stockholm. Goodman, D., Nishimura, Y., Uno, R. and Yamamoto, T. (1995) GPR time-slices in archaeological prospection. Archaeological Prospection, (): 85-89. Goodman, D. (2003) GPR-SLICE software for Windows, (c.1993-2003). Geophysical Archaeometry Laboratory, University of Miami Japan Division, Otsubu 39-1, Nakajima Machi, Ishikawa Ken 929-22, Japan. Holmberg, E. J. (1944) The Swedish excavations at Asea in Arcadia. ActaRom-4: 11, Lund & Leipzig. Lavento, M. (2002) The geoarchaeological investigation. In Forsen, J. and Forsen, B., The Asea valley survey. An Arcadian mountain valley from the Paleolithic period until modern times. (2002), ActaAthens-4, (), Stockholm. Loke, M.H. and Barker, R. (1996) Rapid least-squares inversion of apparent resistivity pseudosections using a quasi-Newton method. Geophysical Prospecting, (): 131-152 Nishimura, Y. and Goodman, D. (2000) Ground-penetrating radar survey at Wroxeter. Archaeological Prospection, (): 101-105. Scollar, I., Tabbag, A., Hesse, A., Herzog, I. (1990) Archaeological Prospecting and Remote sensing: Topics in Remote Sensing 2. Cambridge University Press, Cambridge, England. Sternberg, B.K and McGill, J.W. (1995) Archaeology studies in southern Arizona using ground penetrating radar. Journal of Applied Geophysics, (): 209-225. Weymouth, J.W. (1986b) Geophysical methods of archaeological site surveying. In M.B. Schiffer (ed.), Advances in Archaeological Method and Theory, New York: Academic Press, (): 311-395.

Figure 12: b) Depth level la4-92 was magnified on purpose. Red signifies high amplitudes

Discussion and conclusions The geophysical prospection project of the prehistoric site of Asea can be considered to be one of the most successful experiments to explore a buried city wall including its segments. One of the proceeds of the above project was that the information concerning location, dimensions and depth of the buried city wall has been improved. The subsurface city wall characterized by strong resistivity was better resolved with the MRT than any other method. The inverse model pseudo-sections and produced depth slices of acquired resistivity data delineated with a remarkable accuracy the dimensions and burial depths of the fortification wall in the site. Thus, it can be stated that the tomography technique provides a more detailed view of the subsurface structure than can be obtained using other geophysical techniques and may therefore lead to better understanding of depth. On the other hand, it is important to remember that electrical imaging is not a reconnaissance tool (Barker et al., 2001). It is too slow and more suitable for the detailed study of problem areas that have been selected using other techniques such as GPR. Similarly, the use of SIR-10 GPR has shown the advantages and disadvantages of the system. Leading the correction of its malfunctions and keeping the sampling interval appropriated regarding to the buried archaeological feature of interest to record much denser data, the instrument can be proved valuable in geophysical surveys. The GPR


THE MOST FAMOUS ERUPTION OF THE THERA VOLCANO: A REVIEW OF MORE THAN SIXTY YEARS OF STUDIES V. M. Francaviglia CNR - Istituto per le Tecnologie Applicate ai Beni Culturali, I-00016 Monterotondo S., Roma, Italy [email protected] Abstract: More than sixty years have passed since the formulation of the Spyridon Marinatos catastrophic hypothesis describing the end of the Minoan civilization in Crete. Now, at the beginning of the third millennium the scientific debate, initiated in 1939, does not seem to have diminished, given the quantity of work still being produced on the subject. It has been recently demonstrated, beyond any doubt and contrary to the preceding model, that a wide caldera, invaded by the sea, already existed on the volcanic island of Thera (known also as Thira, Santorini, Santorino, and Santorin) during the Minoan epoch. It follows that the famous Late Minoan (LM) eruption of this volcano only brought minor changes to the morphology of the pre-Minoan caldera, which was very similar to the existing one. This in turn implies that if powerful tsunamis are only generated from caldera collapses (in addition to earthquakes), they (as in the case of the Thera volcano) must have been of modest size. If, on the other hand, tsunamis are also generated by huge pyroclastic flows entering the sea, we believe that the intensity of this kind of sea-quakes could not have been comparable to those created by caldera collapses – also implying that those starting from Thera were of modest size. It has since been accepted that the layer of ash that covered at least half of Crete must have been of very modest thickness and therefore had a negligible effect on Minoan agriculture. All the radiocarbon datings on specimens found under the ashes of the famous eruption give contradictory results. If we base the dating of the Thera eruption on the acidity anomalies of the ancient ice of Greenland and on the dendrochronological anomalies of old trees in the western hemisphere, the Thera event must be placed back in the 17th century BC. These data questions the commonly accepted Aegean chronology, which places the eruption at a later date. However, more recent experimental evidence supports the older date. Curiously, archaeometric studies conducted by various researchers worldwide haven not led to a common point of view regarding the dynamic of the eruption, the impact on the local and distant environments, or even the accurate dating of the eruption itself. At this point, we have been able to assert that the only victim of the Thera eruption was the Minoan settlement of Akrotiri located on the island itself. Περιληψη: Έχουν περάσει περισσότερα από 60 χρόνια από τότε που ο Σπυρίδων Μαρινάτος διατύπωσε τη «θεωρία καταστροφής» που περιέγραφε το τέλος του Μινωικού πολιτισμού στην Κρήτη. Σήμερα, στο ξεκίνημα της τρίτης χιλιετίας, η επιστημονική συζήτηση που ξεκίνησε το 1939, δεν φαίνεται να μειώνεται, αν αναλογιστεί κανείς τον όγκο των εργασιών που ακόμα παράγεται στο θέμα αυτό. Έχει αποδειχθεί πρόσφατα, χωρίς καμία αμφιβολία και ενάντια στο προηγούμενο μοντέλο, ότι μια μεγάλη καλντέρα, που καλύφθηκε από τη θάλασσα, ήδη υπήρχε στο ηφαιστειογενές νησί της Θήρας (γνωστό επίσης και ως (Thira, Santorini, Santorino, και Santorin) κατά τη διάρκεια της Μινωικής εποχής. Συνεπώς, η διάσημη Υστερομινωική έκρηξη του ηφαιστείου αυτού, μόνο δευτερεύουσες αλλαγές επέφερε στη μορφολογία της προ-Μινωικής καλντέρας, η οποία ήταν όμοια με την υπάρχουσα. Το συμπέρασμα αυτό υποδεικνύει ότι αν ένα ισχυρό τσουνάμι δημιουργείται μόνο από την κατάρρευση μιας καλντέρας (εκτός από την περίπτωση σεισμού), τα τσουνάμι στην περίπτωσή μας πρέπει να ήταν μεσαίου μεγέθους. Αν, από την άλλη μεριά, ένα τσουνάμι μπορεί να δημιουργηθεί και από την τεράστια πυροκλαστική ροή υλικού που εισέρχεται στη θάλασσα, πιστεύουμε ότι η ένταση θαλασσίων σεισμών τέτοιου τύπου δεν μπορεί να είναι συγκρίσιμη με την ένταση αυτών που δημιουργήθηκαν από κατάρρευση καλντέρας – υποδεικνύοντας ότι τα τσουνάμι της Θήρας ήταν μεσαίας έντασης. Από τα παραπάνω, γίνεται αποδεκτό ότι το στρώμα στάχτης που κάλυψε τουλάχιστον τη μισή Κρήτη θα πρέπει να είχε μέτριο πάχος και επομένως αμελητέα επίδραση στη Μινωική γεωργία. Όλες οι ραδιοχρονολογήσεις σε δείγματα που βρέθηκαν κάτω από τις στάχτες της διάσημης έκρηξης δίνουν αντιφατικά αποτελέσματα. Αν βασίσουμε τη χρονολόγηση της έκρηξης της Θήρας στις ανωμαλίες οξύτητας που παρατηρούνται στους αρχαίους πάγους της Γροινλανδίας και στις ανωμαλίες δενδροχρονολόγησης δέντρων στο δυτικό ημισφαίριο, το γεγονός της Θήρας πρέπει να τοποθετηθεί πίσω στο 17ο αιώνα π.Χ. Τα δεδομένα αυτά θέτουν υπό αμφισβήτηση την κοινώς αποδεκτή Αιγαιακή χρονολογία, η οποία τοποθετεί την έκρηξη σε μεταγενέστερη ημερομηνία. Όμως, πιο πρόσφατες πειραματικές αποδείξεις υποστηρίζουν την παλαιότερη χρονολογία. Κατά περίεργο τρόπο, οι αρχαιομετρικές μελέτες από διάφορους ερευνητές ανά τον κόσμο, δεν έχουν οδηγήσει σε κάποιο κοινό σημείο όσον αφορά στη δυναμική της έκρηξης, την επίδρασή της στο εγγύς και μακρινό περιβάλλον, αλλά και την ακριβή χρονολόγηση της ίδιας της έκρηξης. Στο σημείο αυτό, είμαστε σε θέση να επιβεβαιώσουμε ότι το μοναδικό θύμα της έκρηξης της Θήρας ήταν η Μινωική εγκατάσταση του Ακρωτηρίου που εντοπίζεται στο ίδιο το νησί.


the Minoan civilization in Crete had been destroyed by the nearby Thera/Santorini volcano (Fig. 1).

The objective of the present study is to critically evaluate the more than sixty years of scholarship which appeared after S. Marinatos brought forth his 1939 hypothesis that

We are not only interested in the separation of the catastrophic and non-catastrophic researchers, but also in 141

V. M. Francaviglia selecting among the mass of data specimens the reliable and pertinent ones, and in evaluating the legitimacy of the interpretations and conclusions. This author has been occupied with this famous eruption for 25 years as a geologist. For a first synthesis in Italian see Francaviglia (1999).

wave started from Thera could have destroyed interior Cretan sites, he was forced to admit, then, that such destruction had to be caused by the seismic shocks that precede volcanic eruptions. The catastrophe must have been unexpected because in the Minoan palace of Kato Zakros (located on the eastern extremity of Crete and excavated by the Greek archaeologist Nikolaos Platon (1971)), a capsized oven was found with a pot containing food still on the stove. The Minoan civilization of Crete seems to have never recovered from that devastating blow. Soon after the Achaean invasions, Crete became a Greek province.

The catastrophic theory of Marinatos In the presence of a series of destructive events, more or less happening at the same time in the Minoan Crete of the 15th century BC, the Greek archeologist Spyridon Marinatos (1939) believed that he had enough evidence to validate his hypothesis that an eruption of the Thera (Santorini) volcano (Fig. 1), which actually took place in the Cycladic archipelago, could be responsible for the destruction and therefore, indirectly, the decline of the Minoan civilization. Such hypothesis was subsequently called the “catastrophic theory.”

The difficulty in demonstrating the simultaneous destruction of Cretan villas and palaces in a time so distant from ours is evident, given that in those epochs changes in style (which are needed for identification) happen very slowly. In our time, on the other hand, it is very easy to accurately identify the age of an old film or photograph by examining the car or the clothes worn by the actors, being that fashions change rapidly.

The archeological analysis reveals, in fact, that a series of destructive events affected villas and palaces of Crete in a period that, according the Aegean chronology of Sir Arthur Evans, is called LM-IB (Late Minoan) datable at around the middle of the 15th century BC. This destruction affected both the interior and coastal areas on the northern and southern parts of the island, as well as isolated villas. The devastation of the coastal settlements was, according to Marinatos, caused by great seaquake waves that had developed during the eruption. Given that no seaquake

Marinatos was aware of the semi-archaeological excavations on the island of Thera, 110 km north of Crete, executed in the 1860’s (Fouqué 1879). Such excavations had given light to Minoan style villa ruins (which were generally contemporary with those of Crete), buried under meters and meters of pumice and volcanic ashes. He then extrapolated from these excavations an analogy

Figure 1: The volcano of Thera/Santorini to the North of Crete.


The Most Famous Eruption of the Thera Volcano: a Review of More Than Sixty Years of Studies testify to what really happened during the paroxystic phase of the eruption in the summer of 1883 (kein Menschenauge hat den Akt der Kalderawerdung gesehen). In fact, a thick cloud of smoke and ashes wrapped the entire island for days which, incidentally, was deserted. To this day, 120 years after the event and despite the progresses made in volcanology, there still isn’t an agreement on the exact eruptive dynamics of this episode. Though it is obvious that three quarters of the island was under the sea (precise Dutch topographical maps were available to show this) with radical modification to the bathymetry, it is still unclear whether the subsequent tsunamis were caused by the sinking of the volcanic structure or by pyroclastic flows (of pumice) entering the sea (Self & Rampino 1981). Pyroclastic flows, when launched into the sea, cover a distance of tens of kilometers on the surface of the water, staying afloat because of calefaction caused by extreme heat, and generating enormous waves by entering the sea (Fig. 3), as is the case when launching a ship into the water. In the case of Krakatoa, we now know that such waves were formed. The stories of Sumatra survivors now have been correctly interpreted: they say that scorching ash filtered through the unconnected tables of wooden house floors burning occupants’ feet (Furneaux 1967). Others reported a strange phenomenon appearing at night on the beach: something appeared, with an intermittent brightness, similar to an immense swarm of fireflies coming from the sea and landing on the beach. Unknown to the survivors, the phenomenon was actually caused by pyroclastic flows that were launched into the sea from the Krakatoa, and landed on Sumatra on the evening of August 25 1883. The flows were still burning red inside, under the colder rind. Agitated by the wave motion, it would crack open and close again here and there, revealing the burning, luminescent nature underneath (Furneaux 1967). Recently, some dredgings near the coast of Sumatra have confirmed that some pyroclastic material actually did arrive on the island (Haraldur Sigurdsson, oral communication). A question remains, however: is this volcanic material truly the product or the 1883 eruption or of a previous one?

Figure 2: Location of one of the relic layers of Minoan pumice-fall found on the walls of the pre-Minoan caldera. In the background, the lava formed Cape Skaros.

Figure 3: Genesis of tsunamis due to pumice-flows entering the sea.

with the Krakatoa (or Krakatau) eruption of 1883, where three quarters of the island were sunk under the sea, and enormous seaquake waves, tsunamis, had devastated the adjacent coasts of Java and Sumatra, with more than 36,000 casualties (Verbeek 1884, 1885, van Doorn 1884, Furneaux 1967). The development of the catastrophic theory

One must further ask himself if this great explosion, heard up to 4000 km away, truly had a relation with the great seaquake wave. The French researchers Camus and Vincent (1983) think not, and claim that the causes of the seaquake are found elsewhere. These various interpretations for such a recent event must make us reflect on the difficulty and uncertainty involved in reconstructing the eruptive dynamics of a prehistoric event of more than 35 centuries ago, such as that of Thera.

The catastrophic theory of Marinatos was based on three presuppositions which, at that time (1939), there was no reason to doubt: •

The Minoan eruption of the Thera volcano had the same dynamic as the Krakatoan eruption of 1883

The Minoan eruption of Thera was contemporary with the destruction of Cretan villas and palaces

The morphology of ancient Thera, before the eruption, was conical, similar to present day Stromboli (Liparian Islands)

It is obvious that Marinatos could not have predicted the developments to come in volcanology. Moreover, the Second World War prevented any further archaeological survey in Greece until ten years after the war. However, those who later on enthusiastically accepted Marinatos’ hypothesis had all the means to carry out some form of verification.

Regarding the uncertainties pertaining to the effective dynamics of the Krakatoan eruption, Hans Reck (1936), more cautiously pointed out that nobody had been able to 143

V. M. Francaviglia The issue of tephra fall out

a nuisance. Moreover, he calculated that the layer of ash on the other islands, Rhodes and Anatolia, added up to a maximum 40 cm.

In the course of plinian eruptions, which take their name from the eruption of Vesuvius on August 24-25, AD 79, described to Tacitus by Plinius the Young in the Epistulae, a great amount of volcanic ash was also emitted along with a denser material (tephra) that, carried by winds, were widely deposited on land and also, logically, on the bottom of the sea. The ash from the Minoan eruption of Thera was punctually identified amongst the ash from different eruptions and different volcanoes on the marine floor around Crete and in all of the eastern Mediterranean in the course of continued oceanographic campaigns (Ninkovich & Heezen 1965, 1967). The hypothesis regarding tephra from Santorini was thus set forth, and has since been speculated on, in our opinion, somewhat excessively (Keller & Ninkovich 1972, Keller 1971, 1980a). We are skeptical of the use of the extrapolations derived from marine tephra thickness to estimate what the thickness of tephra on dry land could have been. These extrapolations do not take into account the phenomena of erosion, transport, and re-deposition of marine and earth sediments, or the work of marine currents and atmospheric agents, all of which eventually alter the original thickness. Above all, the following axiom seems unjustified: the greater the thickness of marine ash, the greater the permanent devastation on land. The fact that ashes from the Minoan eruption of Santorini are recoverable, not only from the eastern Mediterranean (Watkins et al. 1978, Federman & Carey 1980), but also from Cretan archaeological layers (Smith 1975, Cadogan & Harrison 1980, Soles, Taylor & Vitaliano 1995), Rhodes (Doumas & Papazoglou 1980), Cyprus (Åström 1980), the delta of the Nile (Stanley & Sheng 1986), the Anatolian peninsula (Sullivan 1990) and the Black Sea (Guichard et al. 1993), strengthens the convictions of the catastrophists. For them, it is more than obvious that the Theran ash had obscured the sun even in Egypt, making the Cretan fields barren for years, that earthquakes had destroyed commercial warehouses, and that the seaquakes had sunk the Minoan fleet (Luce 1969, 1976).

Most recently, Soles, Taylor and Vitaliano (1995) identified Minoan tephra found in Cretan archaeological sites to be datable, with certainty, to the late LM-IA period, the same period of the eruption in question. This excludes the possibility that the famous Santorini eruption had any relationship whatsoever with the decline of the Minoan Civilization on Crete, which would have occurred in the successive period, LM-IB. Various scholars have analyzed the eruptions of famous volcanoes - from the Vesuvius eruption of 79 AD to those of Merapi, Toba (Acharyya & Basu 1993) and Tambora and Krakatoa in Indonesia - and claimed that the Minoan eruption of Thera must have had the same tragic consequences (Neumann van Padang 1971, van Bemmelen 1971). It was shown that the amount of ash emitted during a given volcanic eruption does not depend solely on the energy given off, but also on other factors, such as the interaction of marine or phreatic water with magma. Eventually, this ash can be scattered throughout an area whose size depends, more or less, on the force and the direction of the winds, with varying consequences. Very recently, Eastwood et al. (2002) report that the tephra shard concentration (TSC) around the Gölhisar lake in southwest Turkey, due to the Minoan Santorini/Thera eruption, did not effect significantly terrestrial pollen, nonsiliceous micro-fossils or diatom assemblages. However, an enhanced lake productivity due to accelerated input of silica and other nutrients following tephra dissolution, is measurable. Finally, to show how dangerous it is to trust overly simplistic models, one need only cite Vinci’s work (1984). He surprisingly demonstrates that while the tephra from the Late Minoan eruption of Thera is easily recovered even on the western side of the island (where the predominant winds would not have deposited the tephra, according to maps published by several authors), the origin of the tephra inside the Santorini caldera itself is still uncertain. In fact, if there is any Minoan tephra at the bottom of the caldera, this implies that it was pre-existing at the time of the famous eruption. If, instead, the tephra at the bottom is not Minoan, it cannot be other than post-Minoan.

Platon (1971) mistakenly identified the slag found near some kilns for making bricks as a volcanic scoria, thus claiming to have found a proof of the volcanic destruction of the Kato Zakros palace. Apart from the fact that it is demonstrable that no volcanic eruption on earth could have enough energy to hurl a distance of 140 km masses of several tens of grams, over-fired material identical to those found by Platon is quite frequently present in archaeological sites worldwide.

The issue of the pre-minoan caldera of Thera As it has been said previously, until a few years ago it was a given that the present caldera of Santorini had formed at the end of the eruption of this volcano during the Minoan age, also because the sinking of the caldera cut some Minoan constructions in two. Only recently have Pichler & Schiering (1980), Pichler & Friedrich (1980) and then Heiken & McCoy (1984) seen the possibility

But not all are of the opinion that the amount of tephra deposited from the Thera volcano had been great, or even constituted any threat whatsoever, i.e. Pichler (1977), Sparks et al. (1978) and Keller (1980b). McCoy (1980) maintains that the insufficient amount of ash deposited in Crete following the Minoan eruption created, at most, 144

The Most Famous Eruption of the Thera Volcano: a Review of More Than Sixty Years of Studies and the pumice came from a central cone that could have been more or less isolated at the center of the caldera itself. Francaviglia & Di Sabatino (1990) demonstrated that the presumed Minoan pumice of McCoy & Heiken actually belongs to an eruption more ancient than the Minoan one (Günther & Pichler 1973, Druitt et al. 1989). Even during his first visit to Santorini in 1978, Francaviglia had had the sensation, particularly because of its morphology, that the caldera of Thera had to be, at least in some parts, older than that traditionally believed. It was obvious that some cuts of the caldera were fresher than others. However, there wasn’t any supporting evidence. One day in 1988, while using binoculars to explore the rugged walls of the Thera caldera, he noticed, by chance, some adherent whitish material at a point of Cape Apanofirá, at half-height of the caldera (fig. 2, 4, 5). Having reached this point, he immediately recognized it as a plinian (from the first phase of the eruption) and primary (not reworked, in other words) pumice-fall, very similar to that of the Minoan eruption. The comparative analyses, carried out in Rome, confirmed the hypothesis made by Francaviglia; that the materials were the remains of a layer of Minoan pumice that had covered all of the island, including the pre-existing caldera and that later had been almost completely eroded on the caldera’s upwardly sloping areas (Francaviglia 1989).

Figure 4: Geomorphological map of the caldera wall on Thera

Druitt and Francaviglia (1990, 1992) continued exploring the walls of the Thera caldera and discovered other deposits of the same type, among which was one at sea level at the foot of the lava Cape of Skaros, on a small wave-cut platform. Strangely, these deposits were overlooked by all those who had completed detailed studies of the Santorini volcano: Fouqué (1879), Neumann van Padang (1936), Pichler & Kussmaul (1972), Günther & Pichler (1973), Bond & Sparks (1976), Pichler & Friedrich (1980), Heiken & McCoy (1984), Druitt et al. (1989), and Barberi & Vecci (1989). In synthesis, Druitt & Francaviglia (1990, 1992) demonstrated that the pre-Minoan caldera morphology of Thera was not very different from the present-day one; that the Minoan eruption had only produced local collapses that widened the caldera; and that at the center there was a volcanic cone (or cones) which started the famous eruption. Therefore, being that there was not a huge collapse, as assumed by catastrophic theory, great seaquakes could not have resulted.

Figure 5: The first relic layer of Minoan pumice-fall in its geological context.

that the inner bay at Thera could have existed even before the Minoan eruption. However, the small pre-Minoan bay of Pichler & Schiering does not excessively modify the appearance of that which must have been a round shaped island. More importantly, the two authors do not deliver any convincing tests for their assertions. Actually, Heiken & McCoy are the ones who most closely approach the palaeogeographic reconstruction of Thera, made later by Druitt & Francaviglia (1990, 1992). Unfortunately, the basic assumptions of the two Americans were wrong. They had interpreted as Minoan the plastered pumice found on the walls of the caldera of Thera, above the phyllite rocks of Athiniós (Fig. 4), in the southern part of the bay. Based on these assumptions, the caldera pre-existed the eruption,

This author, on the basis of simple qualitative comparisons of various caldera morphologies, is of the opinion that the present Santorini morphology is the result of - not counting its age - a combination of powerful phreatic explosions and collapses. Furthermore, an illustrious student, V. Seebach (1867), had previously made observations of the same sort: “Die Kaldera von santorin ist durch Explosion entstanden und durch Denudation nur venig vergrößert 145

V. M. Francaviglia worden” (the caldera of Santorini formed as a result of an explosion and widened only because of erosion). Judging from the slope of the caldera walls, we can distinguish the two phenomena. Slopes of 60-70 degrees, particularly those observable in the northern part of Santorini bay, are attributed to phreatic explosions at shallow depths, which, like explosions from bombs, produce walls of various inclinations (though never vertical), lying on the surface of a cone. The clean cut of 60degrees on the cape of Skaros, consisting of 25 horizontal layers of lava and pyroclastic material (see map), cannot be explained by the caldera collapse. As points of comparison we can consider the craters of Albano and Nemi lakes (Latium Volcano, Rome), which were certainly created by phreatic explosions. Or, on the other hand, we can consider the complex caldera containing vertical walls in Erta Alé, Ethiopia, which was surely caused by collapses. True caldera collapses (that is, those caused by the decrease of underground support to the magma chamber’s vault, created immediately after an eruption empties the magma chamber itself) produce vertical walls. A proof of this was given by an eruption that occurred under ice on October 1996 (Bourseiller & Durieux 1997) at Vatnajökull, Iceland. The great heat emitted during the eruption melted the layers of ice immediately above the craters, forming an underground lake that was then emptied. Lacking this support, the superior ice collapsed along circular and radial, but rigorously vertical, fractures.

Figure 6: The wave-cut platform in the Skaros lavas covered by Minoan pyroclastics.

of rounded pumice were recovered, one of which was from the Thera Minoan eruption (Francaviglia 1990). As an exercise, Yokoyama (1980) tried to calculate the velocity and direction of propagation of a hypothetical tsunami that would have originated in Thera: the wave, diffracted by several islands and straits, could have reached even the most sheltered areas of the eastern Mediterranean basin. His paper reveals that a seaquake wave centered in Thera would form a circular wave front which would diffract when encountering an island or strait. Some of the newly formed wave fronts, especially the one corresponding to the strait between eastern Crete and Karpathos island, could even tangentially hit the southern coast of Crete.

In the case of the Santorini caldera, the observable subvertical cuts in the southern part of the bay of Asprónisi island can surely be attributed to a collapse due to a lack of underground support. In the case of the Krakatau eruption in 1883 (which is known for causing the loss of 2/3 of the island), the morphology of the cracked cone of Rakata remains ambiguous. Its cut is not vertical, suggesting, therefore, an explosion, which did in fact occur (Camus & Vincent 1983).

While Druitt, Francaviglia, Heiken and McCoy were carrying out their research, the German Walter Friedrich and his group were identifying some of the calcareous algae colonies (stromatolites) which were mixed in with the ash-fall (the third phase or Bo3) from the Minoan eruption of Thera. They uncovered a material that nobody before them, including this writer, had ever noticed in more than 110 years of studies and surveys in Santorini. Friedrich’s discovery, probably the most decisive among all those of the last years, not only demonstrated unequivocally that a pre-Minoan caldera existed in Thera, as Druitt and Francaviglia had done in another way, but also that it had been invaded by the sea. Evidently the once-living algae, preferring warmer waters, had attached themselves to the central cone of the volcano (Fig. 7) that had emerged in the bay, and were launched into the air along with pumice and other rocks during the dismantling of the cones in the first phase of the eruption (Friedrich et al. 1988, Eriksen et al. 1990, Friedrich 2000). Nevertheless, it is very difficult to ascertain whether these algae were hurled from the central cone ante or post mortem!

Marinatos (1932) believed to have found the traces of ancient seaquake at Amnisós (the port of Knossos in Crete) solely because some large orthostates had been displaced. We should remember that the sea storm that occurred at Genoa in 1952 was able to dislocate the larger blocks of the outer dam of the port. Melidonis (1963) and Marinos & Melidonis (1971) thought themselves to have recovered volcanic pumice from the Thera eruption brought to Anáfi island (32 km to East of Thera) (Fig. 1) by a sea-quake at 250 m a.s.l. Francaviglia, however, demonstrated that the material is much more ancient and was shot ballistically onto the island. The only proof of an ancient seaquake in Crete would be the one from Amnisós, on a terrace formed by marine abrasion behind a Minoan villa uncovered by Marinatos. Two types


The Most Famous Eruption of the Thera Volcano: a Review of More Than Sixty Years of Studies

Figure 7: The best-fit reconstruction of Thera immediately before and after the Minoan eruption. The shoreline is dotted where unconstrained by field evidence. The presence of calcareous algae imbedded in the pumice-fall and that of relic pumice-fall layers on the wall of the present-day caldera are a proof of the existence of a pre-Minoan water-filled caldera.

As it is easy to be wise after the event, Francaviglia found tens of stromatolites embedded in the pumice-fall (the first phase or Bo1) of the Minoan eruption of Thera. One of them, analysed in Rome by Gilberto Calderoni, has been 14 C dated to 19,100±150 (1σ) yr BP, in good agreement with those published by Walter Friedrich and his collaborators.

Is this result due to endogenous carbon dioxide polluting the material, during and after the eruption, and therefore making it poorer in radiocarbon? Or is it that, contrarily, more recent carbon dioxide polluted the material making it richer in radiocarbon? Calderoni & Turi (1998) have clearly demonstrated how juvenile CO2 can affect organisms living in the proximity of active volcanic areas. Biddle & Ralph (1980), realizing these two possibilities, propose a different approach. Aitken (1988) - the author of a critical review of all the available dating methods - also tries, unconvincingly, to explain this strange phenomenon.

The issue of dating the late minoan eruption of Thera Even though the ceramics discovered during the excavation of Akrotiri (the Minoan Pompeii) are clearly of Minoan type, they stylistically precede the ceramics recovered amongst the ruins of Cretan villas and palaces. The last Theran ceramics are said to have geometric decoration and also came from Crete. The ceramic recovered more recently amongst the Cretan ruins was said to be marine but had never been found at Thera. This circumstance has been interpreted as the proof that Thera had already been lost - or at least taken over - when the marine ceramics were manufactured in Crete. The time lapse between the two styles is estimated to be two generations at most, though both were made in the 15th BC.

The moment of Greenlandic ice Contemporarily, Hammer (1980) and Hammer et al. (1987), while analyzing the stratification of ancient deep ice in Greenland, discovered an anomalous acidity level which they at first dated at 1390±50 BC, and then revised to 1645±20 BC. They maintain that this high acidity could be due to the Minoan eruption on Santorini which could have actually emitted sulfur. Clearly as a hypothesis it is as only valid as any other, because other eruptions were not known from that period. Other eruptions, like those in the New World, could have occurred and passed unnoticed at the archaeological level, or simply have yet to be studied. This is also the opinion of Renfrew (1996).

It was inevitable that radiocarbon dating was applied to the abundant organic material (woods, seeds, etc.) recovered at the Theran excavation. Given the identical and narrow archaeological horizon examined, all within the ancient pre-Minoan palaeosoil, one could legitimately expect the dates to more or less coincide (Pichler & Friedrich 1976, Betancourt 1987, Betancourt et al. 1978, Friedrich et al. 1980, Warren 1984, Manning 1990, Nelson et al. 1990). Strangely, the analyses give a time lapse of more than 400 years between the earlier and later dates (Pomerance 1971).

The decisive evidence would seem to have come from the same Greenlandic ice cores, thanks to Zielinski et al. (1994). They found that amongst the many anomalies in the annual layers of Greenlandic ice, there is above all an anomalous level of sulfates corresponding to the interval between 1627-1623 BC, which is arbitrarily connected to 147

V. M. Francaviglia the second dating of the famous volcanic Minoan eruption of Thera given by Hammer et al.

suggest that the date of 1620 BC is not, by any means definitive, for this eruption.

La Marche & Hirschboeck (1984) and Baillie & Munro (1988) had observed anomalies in the growth rings of an ancient Irish oak, indicating that the tree suffered an unusually cold spring corresponding to the years 1628-1626 BC. The aforesaid authors had connected the phenomenon to the Minoan eruption at Santorini, which would have launched into air so much ash as to disturb the climate in the succeeding years, which is what actually happened with Krakatoa (Nature, XXX, 12 June 1884). The objections made for the acidity of the Greenlandic ice also hold for this hypothesis.

In July 1999 a study by Eastwood et al. (1999) appeared, concerning the tephra found between the shoreline sediments of Gölhisar Gölü lake in South-Western Turkey, immediately above a layer of peat, which was dated 3330±70 and 3225±45 years BP (calibrated interval of time 1749-1406 BC). Without doubt, the volcanic material is younger than the peat on which it settled, but is not possible to say how much, without another datable level of peat immediately above of it. Concerning the origin of this tephra there should not be any doubt; it was produced during the Late Minoan eruption of the Santorini volcano. The interpretation given by the authors of the results on the radiocarbon dosage contained in the peat - without foundation, in our opinion - lean towards the notion that this eruption occurred ~3300 years ago. In fact, they give intervals (2σ) of 1749-1434 or 1604-1406 years BC, that is between 17th-15th century BC, an interval too wide to either satisfy archaeologists or to definitively date the most famous volcanic event of Thera/Santorini.

Undoubtedly, in some part of the world between the years 1627-1623 BC there must have been a great volcanic eruption that influenced the global climate, producing the anomalies observed in the growth rings of American pines, Irish oaks, and not yet well identified Anatolian trees (Kuniholm et al. 1996), as well as depositing an anomalous quantity of sulfates on Greenlandic ice. Rather curiously, the same Theran event, producing a dust mantle which drastically reduced solar radiation reaching the Earth’s surface, did impair tree growth in Californian pines but did enhance the normal growth of Turkish juniper trees! It is evident, however, that there is not yet any evidence that these anomalies are due to the Late Minoan eruption of the Thera volcano. Moreover, it should be noted that the tables published by Zielinski also reveal previously unknown anomalies, but of minor intensity, that corresponded to the traditional dating of the eruption (ca. 1450 BC).

Recently (2000), Italian excavations of Iasos (Western Turkey) have revealed an extremely thick layer of volcanic tephra. Analyses carried out by Jörg Keller showed that this volcanic material belongs to the Theran Minoan event. We must keep in mind that it is not evident that the Theran eruption produced a great amount of sulphur dioxide, just as it is equally true that it is not evident that the same eruption necessarily left tangible signs in Greenlandic ice or in the rings in the trees of the northern hemisphere. In the same way, it is not evident that the measure of ice anomalies in any part of the world is directly proportional to the magnitude of the volcanic eruptions that generated them. Finally, although it is easy to find signs of historically known volcanic events in ancient glaciers, it becomes difficult to estimate ages on the bases of other criteria. Hammer himself (1980) reminds us that the famous eruption of Thera is dated between the 1700 and 1300 BC. In our opinion, isotopic distribution analyses of the sulfur emitted by Santorini volcano during the eruption of Late Minoan age are necessary and should be compared with the ones on the anomalies present in Greenlandic ice.

The last study on the Greenlandic ice cores (Zielinski & Germani 1998) revealed, not only the SO4-2 peak, but also the presence of volcanic glass at a level corresponding to 1628-1627 BC. However, SEM analyses carried out on this tephra (GISP2) concur in excluding the possibility that it comes from the eruption of the Santorini volcano. In his inevitable reply, Manning (1998) maintained that, based on magma considerations, the tephra present in GISP2 does not exclude a dating of 1628 BC for the Minoan eruption of the Santorini volcano. In Zielinski & Germani’s counter-reply (1998), the two authors confirm that the glass included in GISP2 does not originate from Santorini, and that they do not see any reason to change their initial conclusions that the glass found in the layer pertaining to 1623±36 BC of GISP2 is not comparable to that which was formed during the Santorini eruption. This suggests that the Santorini Minoan eruption might not have taken place around 1620 BC and that another eruption might be responsible for the volcanic traces observed in several recordings dating to that period. However, as they declared in their initial article, Santorini could have erupted in 1620 BC. In fact, they cannot exclude this possibility, but are convinced that their results strongly show that this possibility is questionable. Therefore, they

The latest work of Hammer et al. (2003), based on SEM analysis of volcanic glass from new Greenlandic ice cores, tries, in a dubious way (the question mark in the title of their paper is symptomatic), to conclude that the Minoan Thera eruption very likely took place in 1645 BC ± 4 yr. But, if we compare MnO content in Hammer’s glass shards with that of Eastwood’s (1999) volcanic material found in lake sediments from Southwest Turkey, we can see a significant difference. In our opinion the Turkish volcanic material belongs to the Minoan eruption, the Greenlandic one does not. Moreover, these authors find


The Most Famous Eruption of the Thera Volcano: a Review of More Than Sixty Years of Studies it hard to envisage another strong volcanic eruption with features similar to that of Thera (Francaviglia (1990).

profoundly changes all the eruption’s believed dynamics and consequences. Some localized caldera collapses have taken place, causing only minor changes, mostly in the Southern sector. If tsunamis are directly proportional to the intensity of caldera collapses, it follows that the tsunamis that resulted from Thera were of minor intensity. If, on the other hand, tsunamis are also generated when great pyroclastic flows enter the sea, we believe that they are nonetheless smaller than those created by grandiose caldera collapses. In either case, the tsunamis resulting from the Theran event were of minor intensity.

Strange sea-borne pumice On almost every beach of the Eastern Mediterranean (continental Greece and its islands, Turkey, Israel, Egypt, and more to come as findings continue) it is possible to find great quantities of white, biotite-bearing pumice, also present in funerary trousseau in a Middle-Minoan tomb at Malia on Crete and therefore more ancient than the Theran event (Pichler & Schiering 1980). The volcano of origin of the aforementioned pumice is still unknown (Francaviglia 1990). Being that the diffusion of this mysterious pumice, which is often mixed with the Late Minoan pyroclastics of Santorini, is of much greater quantities than the pumice of the Thera eruption, we can suppose that its origin is the great event of the 17th century BC, as recorded in ices of Greenland. The Theran event, on the other hand, could be represented by the minor anomaly of Zielinski et al., and in doing so would put the archaeological debate to rest.

The ash layer that covered the eastern part of Crete was very thin, and therefore incapable of damaging Minoan agriculture. The various radiocarbon datings of the material lying below the pumice and above the Minoan palaeosoil are unusable, at least until a correction factor is found. If the datings made using the acidity anomalies in Greenland ice are to be believed, the date of the famous eruption at Thera would be taken back to 1625 BC, forcing us towards a difficult review of the Aegean chronology. The most recent study on volcanic glass particles found in the Greenlandic ice seem to exclude the Minoan eruption of Santorini as the cause of the acidity peaks and of the tephra itself. In theory at least, the acidity anomalies in ancient ice could be linked (through isotope analysis?) to the volcanic eruptions which caused them. But it is certain that a direct connection between a given eruption and growth anomalies of an ancient tree will never be possible.

This biotite-bearing pumice, identified for the first time by Fornaseri et al. (1975) and found again by Pichler & Schiering (1980), later being studied in depth by Francaviglia (1990), has not been fated to receive the attention it deserves (not even by the authoritative magazine Nature), despite its widespread diffusion and semi-coexistence with the Late Minoan pumice of Thera/ Santorini. Palaeomagnetism studies applied to Theran pumice and archaeological material from Crete have given surprising results that are difficult to explain by any possible dynamics of the Minoan volcanic eruption (Downey & Tarling 1984). Such studies lead to different phases of the eruption separated by several decades. But it is known that a plinian eruption lasts only a few hours, or days, at the most.

The very recent - and important - work by Eastwood et al., on the tephra of Santorini, which is dispersed throughout South-Western Turkey, cannot unfortunately support one of the datings, early or late, for the eruption of this volcano. Currently we can state that the only certain victim of the famous Late Minoan eruption of the Thera-Santorini volcano was the village of Akrotiri, on the extreme South of the island.

Dating based on the cores of ancient ice from Greenland, independently from the way they were carried out, places the Minoan eruption of Thera at a much earlier date, thus throwing the archaeologists in understandable consternation. Betancourt (1987), on the other hand, proposed an extension of the Aegean chronology. If the event of Thera occurred much earlier than what is commonly believed, we would need to assign a much slower evolution to the arts in the Aegean area. If, instead, we maintain the Theran event current dating (ca. 1625 BC) and fix the time interval separating it with the Cretan destructions (ca. 50 years to it), a review of the entirety of Aegean chronology becomes necessary.

Renfrew (1996) reports: “Recently the advocates of the conventional (lower) chronology have taken great comfort from the discoveries of pumices, presumably derived from the Minoan eruption of Thera, have been discovered in strata which follows those of the late Hyksos palace at Tell Dab’a in Lower Egypt. Deposits associated with that palace contained fragments of fresco paintings of Minoan character closely resembling some of those found in Thera and dating from the period there immediately before the great eruption.” But, what does presumably mean? Have they been analyzed? Minoan Theran pumices can be easily identified by many means.

Conclusions It seems evident beyond doubt that the present caldera of the Theran volcano partially existed in pre-Minoan times and had been already invaded by the sea. This circumstance

We could say, jokingly, that until some ashes of the late Minoan eruption of Santorini are found in, for example, the 149

V. M. Francaviglia interior of the sarcophagus of the pharaoh Tutankhamun (which is historically dated at the half of the 14th century BC or 1350 BC), it will not be possible to assign an accurate dating to this eruption.

correlation of tephra layers from Eastern Mediterranean abyssal sediments and the island of Santorini, Quaternary Research, , 160-171. Fornaseri, M., Malpieri, L., and Tolomeo, L., 1975, Provenance of pumices in the North coast of Cyprus, Archaeometry, , 112-116. Fouqué, F., 1879. Santorin et ses éruptions, Masson, Paris. Francaviglia, V., 1986, Provenance of pumices from the Kastelli excavations (Chania, West-Crete), PACT, , 67-70. Francaviglia, V., 1989, Post-caldera pumice deposits on Santorini (Cyclades, Greece), Neues Jahrbuch für Mineralogie Abhandlungen, H6, 275-288. Francaviglia, V., 1990, Sea-borne pumice deposits of archaeological interest on Aegean and Eastern Mediterranean beaches, Thera and the Aegean World III, Acta III, London, : 127-134. Francaviglia, V.M., 1999, Thera: una catastrofe ridimensionata, in: L’eruzione vesuviana delle “Pomici di Avellino” e la facies di Palma Campana, 381-389, Centro Universitario Europeo per i Beni Culturali di Ravello, Claude Albore Livadie Ed., Edipuglia, Bari. Francaviglia, V., and Di Sabatino, B., 1990, Statistical study on Santorini pumice-falls, Thera and the Aegean World III, Acta III, London, , 29-57. Francis, P., and Self, S., 1987, L’eruzione del Krakatoa, in: Le Scienze - Quaderni, , 66-77. Friedrich, W.L., 2000, Fire in the sea, Cambridge University Press, ISBN 0 521 65290 1. Friedrich, W.L., Eriksen, U., Tauber, H., Heinemeier, J., Rud, N., Thomsen, M.S., and Buchardt, B., 1988, Existence of a water-filled caldera prior to the Minoan eruption of Santorini, Greece, Naturwissenschaften, , 567-569. Friedrich, W.L., Friborg, R., and Tauber, H., 1980, Two radiocarbon dates of the Minoan eruption on Santorini (Greece), Thera and the Aegean World II, Athens, Acta II, London, : 241-243. Furneaux, R., 1967. Krakatoa, Sugar Ed., Milano. Guichard, F., Carey, S., Arthur, M.A., Sigurdsson, H., and Arnold, M., 1993, Tephra from the Minoan eruption of Santorini in sediments of the Black Sea, Nature, , 610-612. Günther, D., and Pichler, H., 1973, Die Obere und Untere Bimsstein Folge auf Santorin, Neues Jahrbuch für Geologie und Paläontologie - Abhandlungen, H7, 394-415. Hammer, C.U., 1980, Acidity of polar ice cores in relation to absolute dating, past volcanism and radio-echoes, Journal of Glaciology, 25, , 359-372. Hammer, C.U., Clausen, H.B., Friedrich, W.L., and Tauber, H., 1987, The Minoan eruption of Santorini in Greece dated to 1645 BC?, Nature, , 517-519. Hammer, C.U., Kurat, G., Hoppe, P., Grum, W., and Clausen, B., 2003, Thera eruption date 1645 BC confirmed by new ice core data?, SCIEM, (in press). Heiken, G., and McCoy, F. jr, 1984. Caldera development during the Minoan eruption, Thira, Cyclades, Greece, Journal of Geophysical Research, , B10: 8441-8462. Keller, J., 1971, The major volcanic events in recent eastern Mediterranean volcanism and their bearing on the problem of Santorini ash layers, Thera and the Aegean World I, Athens, 1969. Acta I: 152-169, Athens. Keller, J., 1980a, Prehistoric pumice tephra on Aegean islands, Thera and the Aegean World II, Athens, 1978, Acta II: 49-56. London, 2. Keller, J., 1980b, Did the Santorini eruption destroy the Minoan world?, Nature, , 779. Keller, J., and Ninkovich, D., 1972, Tephra-Lagen in der Ägäis, Zeitschrift der Deutschen Geologischen Gesellschaft, , 579-587. Kuniholm, P.I, Kromer, B., Manning, S.W., Newton, M., Latini, C., and Bruce, M.J., 1996, Anatolian tree rings and the

References Acharyya, S.K., and Basu, P.K., 1993, Toba ash on the Indian Subcontinent and its implications for correlation of Late Pleistocene Alluvium, Quaternary Research, , 10-19. Aitken, M.J., 1988, The Thera eruption: continuing discussion of the dating, Archaeometry,   , 165-182. Baillie, M.G.L., and Munro, M.A.R., 1988, Irish tree rings, Santorini and volcanic dust veils, Nature, , 344-346. Barberi, F., and Vecci, R., 1989, Le fasi freatomagmatiche nell’eruzione Minoica di Santorino (Grecia), Bollettino GNV, , 649-667, Roma. Betancourt, P.P., 1987, Dating the Aegean Late Bronze Age with radiocarbon, Archaeometry,   , 45-49. Betancourt, P.P., Michael, H.N., and Weinstein, G.A., 1978, Calibration and the radiocarbon chronology of Late Minoan 1B, Archaeometry,   , 200-209. Biddle, M., and Ralph, E.K., 1978, Radiocarbon dates from Akrotiri: problems and a strategy, Thera and the Aegean World II, Acta II: 247-252. London. Bond, A., and Sparks, R.S.J., 1976, The Minoan eruption of Santorini, Greece, Journal of the Geological Society, London, , 1-16. Bourseiller, P., and Durieux, J., 1997, Éruption en Islande, GEO, 215, Janvier 1997: 10-24. Cadogan, G., and Harrison, R.K., 1980, Evidence of tephra in soil samples from Pyrgos, Crete, Thera and the Aegean World II, Acta II, I: 235-256. London. Calderoni, G., and Turi, B., 1998, Major constraints on the use of radiocarbon dating for tephrochronology, Quaternary International, , 153-159. Camus, G., and Vincent, P.M., 1983, Un siècle pour comprendre l’éruption du Krakatoa, La Recherche, 14, 1452-1457. Doumas, C., and Papazoglou, L., 1980, Santorini tephra from Rhodes, Nature, , 322-324. Downey, W.S., and Tarling, D.H., 1984, Archaeomagnetic dating of Santorini volcanic eruptions and fired destruction levels of Late Minoan civilization, Nature, , 519-523. Druitt, T.H., and Francaviglia, V., 1990, An ancient caldera cliff line at Phira, and its significance for the topography and geology of pre-Minoan Santorini, Thera and the Aegean World III, Acta III, London, : 362-369. Druitt, T.H., and Francaviglia, V., 1992, Caldera formation on Santorini and the physiography of the islands in the Late Bronze Age, Bulletin of Volcanology, , 484-493. Druitt, T.H., Mellors, R.A., Pyle, D.M., and Sparks, R.S.J., 1989, Explosive volcanism on Santorini, Greece, Geological Magazine, , 2, 95-126. Eastwood, W.J., Pearce, N.J.G., Westgate, J.A., Perkins, W.T., Lamb, H.F., and Roberts, N., 1999, Geochemistry of Santorini tephra in lake sediments from Southwest Turkey, Global and Planetary Change, , 17-29. Eastwood, W.J., Tibby, J., Roberts, N., Birks, H.J.B., and Lamb, H.F., 2002, The environmental impact of the Minoan eruption of Santorini (Thera): statistical analysis of palaeoecological data from Gölhisar, southwest Turkey, The Holocene   , 431-444. Eriksen, U., Friedrich, W.L., Buchardt, B., Tauber, H., and Thomsen, M.S., 1990, The Stronghyle caldera: geological, palaeontological and stable isotope evidence from radiocarbon dated stromatolites from Santorini, Thera and the Aegean World III, Acta III, London, : 139-150. Federman, A.N., and Carey, S.N., 1980, Electron microprobe


The Most Famous Eruption of the Thera Volcano: a Review of More Than Sixty Years of Studies rocks of the Santorini group (Aegean Sea, Greece), Neues Jahrbuch für Mineralogie - Abhandlungen,   , 268-307. Pichler, H., and Kussmaul, S., 1972a, Inselbildung und MagmenGenese im Santorin-Archipel, Naturwissenschaften, , 188197. Pichler, H., and Schiering, W., 1980, Der spätbronzezeitliche Ausbruch des Thera-Vulkans und seine Auswirkungen auf Kreta, in Archäologischer Anzeiger, 1-37. Platon N., 1971, La destruction volcanique du centre palatial de Zakros, in Thera and the Aegean World I, Acta I, Athens, pp. 395-402. Plinius Caecilius Secundus, 80, Epistulae, VI, 16. Pomerance, L., 1971, The final collapse of Santorini (Thera), 1400 B.C. or 1200 B.C.?, Thera and the Aegean World I, Acta I: 384-394. Rapp, G.Jr, Cooke, S.R.B., and Herickson, E., 1973, Pumice from Thera (Santorini) identified from a Greek Mainland archaeological excavation, Science, , 471-473. Reck, H., 1936, Santorin. 3 vol, Dietrich Reimer, Berlin. Renfrew C., 1996, Kings, tree rings and the Old World, Nature, 381, 733-734. Seebach, K., 1867, Der vulkan von Santorin, in: Samlung gemeinverständlicher Vorträge, , 38, Berlin. Seebach, K., 1867, Ueber den Vulkan von Santorin und die eruption von 1866. Göttingen. Self, S., and Rampino, M.R., 1981, The 1883 eruption of Krakatau, Nature, , 699-704. Smith, P.J., 1975, Volcanic tephra on Crete, Nature, , 9-10. Soles, J.S., Taylor, S.R., and Vitaliano, C.J., 1995, Tephra samples from Mochlos and their chronological implication for neopalatial Crete, Archaeometry, 37(2), 385-393. Sparks, R.S.J., Sigurdsson, H., and Watkins, N.D., 1978, The Thera eruption and Late Minoan-IB destruction on Crete, Nature, , 91. Stanley, D.J., and Sheng, H., 1986, Volcanic shards from Santorini (Upper Minoan ash) in the Nile Delta, Egypt, Nature, , 733-735. Sullivan, D.G., 1990, Minoan tephra in lake sediments in Western Turkey: dating the eruption and assembling the atmospheric dispersal of the ash, Thera and the Aegean World III, Acta III: 114-119. London, 3. Van Bemmelen, R.W., 1971, Four volcanic oubursts that influenced human history: Toba, Sunda, Merapi and Thera, in: Thera and the Aegean World I, Acta I: 6-50. Van Doorn, M.C., 1884, The eruption of Krakatau, Nature, , 268-269. Jan. 1884. Verbeek, R.D.M., 1884. The Krakatau eruption, Nature, 10-15. 1 May 1884. Verbeek, R.D.M., 1885. Krakatau. Batavia. Warren, P., 1984. Absolute dating of the Bronze Age eruption of Thera (Santorini), Nature, , 492-493. Warren, P., and Puchelt, H., 1990. Stratified pumice from Bronze Age Knossos, Thera and the Aegean World III, Acta III: 7181. London, 3. Watkins, N.D., Sparks, R.S.J., Sigurdsson, H., Huang, T.C., Federman, A., Carey, S., and Ninkovich, D., 1978, Volume and extent of the Minoan tephra from Santorini volcano: new evidence from deep-sea sediment cores, Nature, :, 122126. Vinci, A., 1984, Chemical differences between the island and submarine pumice layers of Thera, Marine Geology, , 487491. Vitaliano, D.B., and Vitaliano, C.J., 1980, Tephrochronological evidence for the time of the Bronze Age eruption of Thera, Thera and the Aegean World II, Acta II: 217-220. London, I. Yokoyama, I., 1980, The tsunami caused by the prehistoric eruption of Thera, Thera and the Aegean World II, Acta II: 277-283. London, I. Zielinski, G.A., Mayewski, P.A., Meeker, L.D., Whitlow, S.,

absolute chronology of the Eastern Mediterranean, 2220-718 BC, Nature, , 780-783. La Marche, V.C. jr & Hirscboeck K.K., 1984. Frost rings in trees as records of major volcanic eruptions, Nature, , 105150. Luce, J.V., 1969, The End of Atlantis. New Light on an Old Legend, Thames & Hudson. Luce, J.V., 1976, Thera and the devastation of Minoan Crete: a new interpretation of the evidence, American Journal of Archaeology, , 9-18. Manning, S.W., 1990, The Thera eruption: The Third Congress and the problem of the date, Archaeometry,   , 91-100. Manning, S.W., 1998. Correction. New GISP2 ice-core evidence supports 17th century BC date for the Santorini (Minoan) eruption: response to Zielinski & Germani (1998), Journal of Archaeological Science, , 1039-104. Marinatos, S., 1932, Anaskafi Amnisou Kritis, Praktika, 76-94. Marinatos, S., 1939, The volcanic destruction of Minoan Crete, Antiquity, , 425-439. Marinos, G., and Melidonis, N.G., 1971, On the strength of seaquakes (tsunamis) during the prehistoric eruptions of Santorin, in: Thera and the Aegean World I, Acta I: 277-282, Athens. McCoy, F. jr, 1980. The Upper Thera (Minoan) Ash in deep-sea sediments: distribution and comparison with other ash layers, Thera and The Aegean World II, Athens, 1978, Acta II: 5778. London. McCoy, F.W., 1981, Areal distribution, redeposition and mixing of tephra within deep-sea sediments of the Eastern Mediterranean Sea, in: S.Self & R.S.J. Sparks, Tephra Studies: 245-254, D. Reidel Publishing Company. Melidonis, N.G., 1963, Die Geologie der Insel Anaphi, IGME, Athens, VIII, 3: 308 (in Greek with German abstract). Nelson, D.E., Vogel, J.S., and Southon, J.R., 1990, Another suite of confusing radiocarbon dates for the destruction of Akrotiri, Thera and the Aegean World III, Acta III: 197-206. London, 3. Neumann van Padang, M., 1936, Die Geschichte des Vulkanismus Santorins von ihren Anfängen bis zum zerstörenden Bimssteinausbruch um die Mitte des 2. Jahrtausends vor Christus, in: Reck H., 1936: Santorin - Der Werdegang eines Inselvulkans und sein Ausbruch 1925-1928, I: 1-72. Berlin: Dietrich Reimer. Neumann van Padang, M., 1971. Two catastrophic eruptions in Indonesia, comparable with the plinian outburst of the volcano of Thera (Santorini) in Minoan time, Thera and the Aegean World I, Acta I: 51-63, Athens. Niemeier, W.D., 1980, Die Katastrophe von Thera und die spätminoische Chronologie, JdI, , 1-76. Ninkovich, D., and Heezen, B.C., 1965, Santorini tephra. Lamont Geological Observatory (Palisades, New York), contribution , 413-454. Ninkovich, D., and Heezen, B.C., 1967, Physical and chemical properties of volcanic glass shards from pozzuolana ash, Thera Island, and from Upper and Lower ash layers in Eastern Mediterranean deep sea sediments, Nature, 213, 582-584. Page, D., 1971, The volcano at Santorini and the devastation of Minoan Crete: an introduction to the historical and archaeological problem, Thera and the Aegean World I, Acta I: 371-376, Athens. Pichler, H., 1977, The Thera eruption and Late Minoan-IB destruction on Crete, Nature, , 819-822. Pichler, H., and Friedrich, W., 1976, Radiocarbon dates of Santorini volcanics, Nature, , 373-374. Pichler, H., and Friedrich, W.L., 1980, Mechanism of the Minoan eruption of Santorini. Thera and the Aegean World II, Acta II, 15-30, London, 2. Pichler, H., and Kussmaul, S., 1972, The calc-alkaline volcanic


V. M. Francaviglia Zielinski, G.A., and Germani, M.S., 1998, Reply to: Correction. New GISP2 ice-core evidence supports 17th century BC date for the Santorini (Minoan) eruption: response to Zielinski & Germani (1998), Journal of Archaeological Science, , 1043-1045. Åström, P., 1980, Traces of the eruption of Thera in Cyprus?, Thera and the Aegean World II, Acta II: 221-234. London.

Twickler, M.S., Morrison, M., Meese, D.A., Gow, A.J., and Alley, R.B., 1994, Record of volcanism since 7000 B.C. from the GISP2 Greenland ice core and implications for the volcano-climate system, Science, , 948-952. Zielinski, G.A., and Germani, M.S., 1998, New ice-core evidence challenges the 1620s BC age for the Santorini (Minoan) eruption, Journal of Archaeological Science, , 279-289.


BETWEEN PEAK AND PALACE. REINTERPRETATION OF THE MINOAN CULTURAL LANDSCAPE IN SPACE AND TIME S. Soetens Département d’Histoire de l’Art et d’Archéologie, Université catholique de Louvain, Place B. Pascal 1, 1348 Louvain-la-Neuve, Belgium, [email protected] and Laboratory of Geophysical and Satellite Remote Sensing & Archaeo-environment, Institute for Mediterranean Studies, Foundation of Research and Technology, Hellas (F.O.R.T.H.), Melissinou & N. Foka 130, P.O. Box 119, Rethymno, GR- 74100, Crete, Greece,

A. Sarris Laboratory of Geophysical and Satellite Remote Sensing & Archaeo-environment, Institute for Mediterranean Studies, Foundation of Research and Technology, Hellas (F.O.R.T.H.), Melissinou & N. Foka 130, P.O. Box 119, Rethymno, GR- 74100, Crete, Greece

K. Vansteenhuyse Département d’Histoire de l’Art et d’Archéologie, Université catholique de Louvain, Place B. Pascal 1, 1348 Louvain-la-Neuve, Belgium and Department of Oriental Studies, Katholieke Universiteit Leuven, Blijde Inkomststraat 21, 3000 Leuven, Belgium Abstract: Crete is the spatial entity where the history of the Cretan Bronze Age was enacted. It is obvious that the landscape as a whole and the temporal dynamics within it contain many distinctive properties for the better understanding of its culture. One of the key parameters for capturing the dynamics of the Minoan landscape is the specific social, religious and topographic character of the Minoan peak sanctuary. In the past, it has been argued that the location of the peak sanctuaries depends on the topography of the landscape. Their position would also be determined by their relation with the court complexes. In order to test the validity of the above argument, Differential Global Positioning System (DGPS) was employed for the accurate positioning of the relevant archaeological sites. The topographical and environmental parameters of peak sanctuaries and palaces were extracted from digitized maps and from SPOT stereoscopic satellite images. GIS (Geographical Information System) analysis was systematically applied to investigate the spatial characteristics of these sites and their spatial relations. Among other techniques, intervisibility between peak sanctuaries and other sites was simulated through viewshed analysis and line of sight. Results of the least-cost distance computed from the peak sanctuaries to the near-by environmental and archaeological features were subjected to statistical analysis in order to define the weight of importance of these features as an indication of their relevance to the location of the peak sanctuaries. Hypothetical territories were suggested for the court complexes, through the application of the most commonly used models, namely Thiessen polygons, Cost Surface Analysis and the Xtent model. The location of peak sanctuaries within these territories suggests that they played a neutral role in the early stages of their existence. This picture evolves through time and geographical regions. The origin, acme and decline of peak sanctuaries seem to be strongly related to the political development within the island. Περιληψη: Η Κρήτη είναι η χωρική οντότητα στην οποία πραγματοποιήθηκε η ιστορία της εποχής του χαλκού της νήσου. Προφανώς το συνολικό περιβάλλον και η χρονολογική δυναμική της νήσου περιέχουν πολλά συστατικά για την καλύτερη κατανόηση του Μινωϊκού πολιτισμού. Ο ιδιαίτερος κοινωνικός, θρησκευτικός και τοπογραφικός χαρακτήρας του Μινωικού ιερού κορυφής αποτελούν από τις σημαντικότερες παραμέτρους για την κατανόηση του Μινωικού περιβάλλοντος. Στο παρελθόν προτάθηκε ότι η τοποθεσία των ιερών κορυφής βασιζόταν στην τοπογραφία του περιβάλλοντος. Επίσης η τοποθεσία τους θα μπορούσε να είχε καθοριστεί από την σχέση τους με τα λεγόμενα ‘court complexes’ ή συγκροτήματα αυλής. Διαφορικά Συστήματα Παγκόσμιου Εντοπισμού (DGPS) χρησιμοποιήθηκαν για την μέτρηση συντεταγμένων ακριβείας των σχετικών αρχαιολογικών θέσεων με στόχο την εξακρίβωση του προαναφερομένου επιχειρήματος. Τοπογραφικά και περιβαλλοντικά στοιχεία των ιερών κορυφής και των ανακτόρων αντλήθηκαν από ψηφιοποιημένους χάρτες και στερεοσκοπικές δορυφορικές εικόνες του SPOT. Τα χωρικά χαρακτηριστικά των θέσεων και των αμοιβαίων σχέσεών τους διερευνήθηκαν συστηματικά μέσω των αναλυτικών εργαλείων των Γεωγραφικών Συστημάτων Πληροφοριών (GIS). Η προσομοίωση της ορατότητας μεταξύ των ιερών και άλλων θέσεων πραγματοποιήθηκε μέσω αναλύσεων οπτικού πανοράματος (viewshed analysis) και της ευθύγραμμης οπτική επαφής (line of sight). Τα αποτελέσματα της ανάλυσης του ελαχίστου κόστους απόστασης (least-cost distance) από τα ιερά κορυφής στα εγγύτερα γεωμορφολογικά χαρακτηριστικά και αρχαιολογικά μνημεία υποβλήθηκαν σε στατιστική ανάλυση για τον καθορισμό του συντελεστή βαρύτητας αυτών ως ένδειξη προσδιορισμού της θέσης των ιερών κορυφής. Η μοντελοποίηση των περιοχών επικράτειας (hypothetical hypothetical territories)) για τα συγκροτήματα αυλής πραγματοποιήθηκε με την εφαρμογή διαφορετικών διαδικασιών όπως τα πολύγωνα Thiessen,, η ανάλυση επιφάνειας κόστους (Cost Cost Surface Analysis)) και το μοντέλο Xtent.. Τα αποτελέσματα των ερευνών προτείνουν έναν σχετικά ουδέτερο ρόλο για τα ιερά στις αρχές της ύπαρξής τους. Η εικόνα αυτή εξελίσσεται με διαφορετικό τρόπο τόσο χρονικά όσο και ανά γεωγραφική περιοχή. Η προέλευση, η ακμή και παρακμή των ιερών κορυφής φαίνονται να σχετίζονται άμεσα με την εξέλιξη της πολιτικής ανάπτυξης στη Κρήτη.


S. Soetens, A. Sarris and K. Vansteenhuyse Introduction Before we jump into the pool of GIS spatial analysis and archaeological interpretation, a short introduction to the theme and its particular approach is in order. Cretan peak sanctuaries from the Bronze Age (c. 3000 – 1070 BC) have been the focus of attention almost since the birth of Minoan archaeology at the beginning of the 20th century AD. Nowicki (1994) defined the main features of a Minoan peak sanctuary as: positioned on a mountain top, with one or more sides formed by a cliff, the presence of pottery/figurines, and the presence of pebbles. The peak sanctuaries emerged in Early Minoan III – Middle Minoan IA, or perhaps as early as Early Minoan II where Iuktas is concerned (Karetsou 1981). Iuktas has basically survived the millennia as a sacred mountain (fig. 1). The site is still known as the tomb of Zeus and an orthodox church dedicated to the Metamorphosis, Agios Georgios, Agii Pantes and Agia Zoni is still located at the peak. Its remarkable shape, as seen from the area south - west of the mountain still triggers our imagination. Its profile, as seen from this angle, resembles a resting human head and it is right on its forehead, the human metaphor of the mountain’s spirit, that the Minoan peak sanctuary is located. The major religions of the world today and our perception of mountains through the millennia show that mountains were always highly humanised. The western appreciation tends to be more competitive, in a sense that we conquer the mountain (read: ‘stand on its peak’), while the oriental world has a more humble tradition in which the mountain must be respected as the abode of the immortals, the axial pillars that support the sky.

Figure 1: Iuktas from Giofyro, Iraklion

Figure 2: Iraklio from Iuktas, Knossos should be somewhere in the plain on the right

‘Palaces’, namely Knossos, Mallia, Phaistos, and Zakros, we opted here to present the analysis of a larger group of sites: the court complexes (Agia Triada, Kommos, Galatas, Makrygialos and Petras, including Knossos, Mallia, Phaistos and Zakros), together with some sites where a court complex can be expected or sites which can be understood as centres, based on architectural quality of structures, size of site, and presence of administrative documents (Archanes, Chania, Palaikastro, Stavromenos-Chamalevri) (Driessen et al. 2002). The identification of central places in the Protopalatial (c. 2000 – 1640 BC) period is often obscured by the Neopalatial structures constructed on top of them. The variability - central place or not central place - remains mostly unknown for the Protopalatial period. Therefore the Protopalatial dataset should be seen more as a test case. The burial sites were not included here, because we are interested in the interaction of the ‘living Minoan’ with his/her landscape. It is possible, however, that the tholos tombs in the Mesara (Early Minoan period) were the foregoers of the peak sanctuaries (Branigan 1998). A future analysis is being prepared to compare the relation tomb/settlement and peak sanctuary/settlement.

Peatfield (1983), one of the first to make a comprehensive study of the concept of the peak sanctuary, defined peak sanctuaries foremost on their topographical characteristics. He stated that “The sanctuary should be seen from the region it served” and “it should ‘see’ that region”. From below we see “…the most prominent mountain” and therefore “the best landmark for worshippers to travel to” (Peatfield 1983, 274-276). Intervisibility between the peak sanctuaries was understood as “the expression of ritual unity that may have transcended political boundaries” (Peatfield 1994, 25). The proximity of peak sanctuaries to nearby settlements, the intervisibility of the sanctuaries and their visual quality as landmarks from both land and sea, and the diachronic changes of distribution of both Minoan ‘Palaces’ and the peak sanctuaries hold many clues to a better understanding of the history of the Cretan Bronze Age. The relationship of peak sanctuaries with central places of power, presumably the so-called ‘Palaces’, seems intensified in the Neopalatial period (c. 1640 – 1550/1425 BC), as Cherry (1978, 429431) and Peatfield (1987) remarked a long time ago.

How ‘real’ is the observation that the relations between the central places and the peak sanctuaries intensified? First of all, the chronological coincidence between the construction of the first court complexes and the emergence of the nearby peak sanctuaries is not as clear as was previously assumed. The chronology of peak sanctuaries itself remains dubious, due to the limited publications and rescue character of the

The classification of a site as a central place of power needs some further clarification. Instead of using the canonical 154

Between Peak and Palace. Reinterpretation of the Minoan Cultural Landscape in Space and Time excavations. A reviewed chronology dates twenty peak sanctuaries to the Protopalatial period (Soetens 2004), but the most intensively investigated sites emerged earlier. At the beginning of the Neopalatial period, most go out of use while the foundation of two new sanctuaries at the same moment, Kofinas peak (Karetsou and Rethemiotakis 1990) and Liliano Kefala (Rethemiotakis 2002: 62, 65) is rather unusual. Secondly, the common cult apparatus and presence of Linear A may indeed be present at both site types, but are not exclusive to these sites. Thirdly, the presence of iconographic representations of peak sanctuaries only found at the court complexes is a rather risky argument to argue for an exclusive relationship. The representations are extremely few and fragmentary, while some other representations from non-palatial sites may also be interpreted as peak sanctuary representations.

unexpected results: not a single peak sanctuary is visible from the central sites of Mallia, Zakros, Monastiraki, Myrtos, Kommos, Agia Triada or Gournia at any moment. In the Protopalatial period even Phaistos does not have a visible peak sanctuary. As a matter of fact, of the original canonical ‘Palaces’, only Knossos can see a peak sanctuary (Iuktas). Even if some peak sanctuaries remain to be discovered, the number of visually unrelated sites makes us wonder whether the ‘Palace’-peak nexus is a real one. The exceptions are the peak sanctuaries Iuktas, Kofinas, and Liliano Kefala. Those are indeed visible from respectively: Knossos - Galatas, Phaistos – Protoria (if indeed a central place) and Galatas. Could this connection have been an exclusively Central Cretan phenomenon? Possibly, but the richness of the Vrysinas peak sanctuary in West Crete is highly suspicious and suggests the presence of a visible important place. The location of Stavromenos – Chamalevri seems promising. In the far East of Crete, the location of Petsofas and Prinias is clearly related to Palaikastro and Petras respectively. Palaikastro, however, has no court complex.

Investigated here is not so much the common finds and architectural characteristics of peak sanctuaries vis-à-vis central places of power, but rather their spatial interaction. Visibility and distance between the peak sanctuaries and the contemporary settlements are not so much landscape characteristics, but rather indicate the human experience of that landscape.

When these analyses were compared to the line of sights between peak sanctuaries and all sites (excluding peak sanctuaries and burial sites), only very few of the peak sanctuaries were not visible at all, and it is especially indicative that the sanctuaries with the highest visibility are those close to intensively surveyed areas. In both periods, the lack of any visible sites from peak sanctuaries seems due to a bias in the archaeological dataset.

Data collection and organization The sites, including the most relevant contemporary sites of the peak sanctuaries (only published material) were either visited with differential GPS receivers or they were digitized. The peak sanctuaries were all located by DGPS. Most of the survey sites were digitized, and the level of accuracy here, depends entirely on the accuracy of the published map. The archaeological data was organized in a database where all sites were related to their typology, chronology and bibliography, peak sanctuary finds, landscape characteristics of the peak sanctuaries and more (see Soetens 2004).

It is important to note that the analogy of peak sanctuaries and visible settlements switches from settlements with a low profile in the Protopalatial period to important, larger sized settlements in the Neopalatial period. The intervisibility amongst peak sanctuaries is much more intensive (figs. 5 & 6). Three networks appear, responding amazingly well to large Cretan eco-zones, which are divided basically by two main mountain chains, namely Dikti and Idi. No peak sanctuaries have been found on those mountains. The eastern and western networks are almost completely disconnected in the Neopalatial period, a gradual process, indicative of increasing hierarchy and lower connectivity (cf. Haggis 2002). The central network, however, changes more dramatically. The number of peak sanctuaries declines and their intervisibility increases. Especially the addition of Kofinas to the Neopalatial peak sanctuaries increased the connectivity and intervisibility of north and south central Crete in a visual system.

The background maps were all geo-referenced to the same ΕΓΣΑ ’87 projection (the Greek Geodetic Reference System of 1987), and include a DEM (digital elevation model), based on a SPOT stereoscopic satellite image (50x50m pixel), slope, aspect and hill shade grids (through analysis of the DEM), digitized topographical, geological, land use and land capability maps (on 1:50000 scale). Visibility Empirically, the peak sanctuaries are far more visible from the settlement than vice versa. The court complex of Knossos can hardly be located from the Iuktas peak sanctuary (fig. 2).

Environmental characteristics Instead of viewsheds, line of sight analysis was chosen to better visualise a series of visibility networks (Soetens et al. 2002; Soetens et al. 2004). The analysis of the visibility of peak sanctuaries from central places (figs. 3 & 4) provided

The peak sanctuaries are almost all (87%) located in the category of phrygana or in severe vegetation, with very little human interference in the natural scenery. This 155

S. Soetens, A. Sarris and K. Vansteenhuyse

Figure 3: Lines of sight between central places and peak sanctuaries in the Protopalatial period

Figure 4: Lines of sight between central places and peak sanctuaries in the Neopalatial period


Between Peak and Palace. Reinterpretation of the Minoan Cultural Landscape in Space and Time vegetation type forms about 42% of the entire Cretan landscape, and it excludes all agricultural land. One could define this, as has been proposed in past archaeological research as pasture land (Rutkowski 1986, 73; 1988, 75). In terms of land region, almost all are located in the broadleaved evergreen zone. Two exceptions, Karfi and Kofinas, belong respectively to the oak-cypress-maple zone, and to the Pinus bruttia-cypress zone. These three zones cover almost the entire island, except for the pseudo alpine zones and the urban zones, so these are not relevant for the peak sanctuary identification.

sites of Apodoulou, Monastiraki (Rethymnon prefecture) and Chamezi (close to Siteia) were also abandoned, while we can observe the growth and/or emergence of a number of other sites: Gournia, Mochlos, Makrygialos, Plati (Lasithi plain), Galatas and possibly Protoria. At the very final phase of the peak sanctuaries, in Late Minoan IA, the ‘villas’ emerge. Although the archaeological data are incomplete, it seems that the regional differences in settlement patterns are bigger than the temporal differences. This means that the settlement pattern is more closely dependent on its environment than on the temporal dynamics of human interaction. The relation between peak sanctuaries and central places must therefore first be studied on a regional scale. We have attempted to model these regional territories based on the list of presumed central places and the list of known peak sanctuaries. Five methods have been tested: Nearest neighbour analysis, Cost distance, Thiessen polygons, Euclidean distance and the Xtent model.

When we turn towards the basic geological characteristics, all peak sanctuaries belong to the sedimentary-metamorphic group of limestone and/or dolomites, a group which forms about 34% of the entire island (more specifically: hard limestone 19 sites, and one site on schist, peridotite, mixed flysch and tertiary deposits). Of the Neopalatial sanctuaries, all are on hard limestone, except Gonies Filiorimos, which can be found on a peridotite mountain. This is a remarkable feature, especially since serpentine, one of the elite materials in Minoan palatial material culture, is a peridotite mineral, which is very often used for the shaping of stone vases, stone offering tables, and seals.

Nearest neighbour analysis from peak sanctuaries to settlements The Nearest neighbour analysis looks for the closest neighbour of a given site. Most obvious and important here is that Iuktas is closer to Archanes than to Knossos. Every archaeologist acknowledges this but the relationship Knossos – Iuktas is often so overemphasized that we tend to forget that a very important settlement is right at the foot of Iuktas. Iuktas remains the closest peak sanctuary for Knossos. As confirmed by the line of sight analysis, Pyrgos is closest to Tylissos, Gonies Filiorimos to Sklavokambos and Modi to the archaeologically rich but not well investigated area of Magasa. An even more promising research concept would be to relate a cluster of settlements to each of the peak sanctuaries and not simply one presumed central place. One peak sanctuary probably served a cluster of settlements rather than one particular site.

When the same analysis is made for the central places (including identified court complexes and hypothetical ‘palatial’ sites in both the Proto- and Neopalatial period), alluvium and tertiary deposits are the main geological categories, whereas the exceptions are indeed those sites of which the identification as ‘palatial’ remains in doubt. As can be expected the artificial vegetation of almost all of these sites results from agricultural exploitation. It is possible that the gradual disappearance of peak sanctuaries and the emergence of the ‘villas’ in Late Minoan IA is an indication of an economic change of interest from a more husbandry focused society to a more agricultural landscape. The distance of the coastline from the Neopalatial peak sanctuaries is between 295m and 14.487m, streams can be found between 832m and 17.629m away, caves between 253m and 8.125m, and springs between 73 and 3.891m. These data only become more meaningful after statistical evaluation which shows that ‘normal peak sanctuaries’ (those that fall within the standard deviation distance) are 5.951m ±3.863 from the coastline, 2.108m ±1.697 from a cave, 1.382m ±975 from a spring, 6.714m ±4.069 from a stream.

The East Cretan area is highly interlinked, and West Crete shows a bias in the available data, as can be understood from the long distances in this area. Here, statistical analysis would locate the closest central places of power at a distance of 7.353m ±4.417 from the peak sanctuaries. The closest settlements are located at an average of only half the distance or 3.215m ±2.411 in the Neopalatial period and 4.750m ±4.256 in the Protopalatial period from the peak sanctuaries.

Territorial modelling

Nearest neighbour analysis between peak sanctuaries

On the diachronic changes of the general distribution of central places, only hypothetical comments can be made, because the Protopalatial central places are mostly hypothetical (see comments in Data Collection and Organization). At the end of the Protopalatial period, when most of the peak sanctuaries have been abandoned, the

The thicker density of East Cretan sanctuaries shows a regional divergence, where sanctuaries are between three and seven km apart, a situation similar to the cluster of Keria, Pyrgos and Filiorimos in the valley between Tylissos and Anogeia. In Central Crete the distances are much larger, between twelve and over thirty km apart. 157

S. Soetens, A. Sarris and K. Vansteenhuyse

Figure 5: Lines of sight between the Protopalatial peak sanctuaries

Figure 6: Lines of sight between the Neopalatial peak sanctuaries


Between Peak and Palace. Reinterpretation of the Minoan Cultural Landscape in Space and Time The sanctuaries in the Rethymnon area are between eight and eleven km apart. These differences have already been interpreted as related to the different topographical and political trajectories of these areas.

construction of the central places is no longer a certainty, and neither is the coexistence of all peak sanctuaries in the Protopalatial period. This fact alone suggests a dynamic political, economic and religious power game, although these modern concepts were probably not clearly distinguishable in Bronze Age Crete. It has further been shown that some older hypotheses on the relationship of peak sanctuaries with ‘Palaces’/court complexes need further consideration.

The averages for the whole of Crete show that peak sanctuaries are located 8.870m ±7.003 from each other. In general, near larger valleys, one encounters fewer and higher peak sanctuaries. Cost distance analysis

Analysis of the visibility has shown that the peak sanctuaries are the most important landmarks in the landscape, and not so much the court complexes or the settlements. Many important court complexes (e.g. Mallia and Kato Zakros) do not have one visible peak sanctuary. In the Protopalatial period peak sanctuaries are mainly visible from the ‘rural’ settlements. In the Neopalatial period the more important settlements have better visibility to the peak sanctuaries, but there are still few court complexes with such visibility. Peak sanctuaries that are totally invisible are located within archaeologically poorly investigated areas.

The cost distance territory is a model based on economic values. It shows how much energy is spent by crossing the landscape from any given point. This is presented graphically by irregular bands each representing one hour of walking (buffer zones). For the Protopalatial period (fig. 7), it is remarkable that within the two-hour buffer zone of some central places, we can find only half the number of peak sanctuaries. Exactly the same group of sanctuaries lies within the one-hour buffer zone of all settlements. Within the two-hour buffer zone of all Protopalatial settlements, we can find 16 out of 20 of the peak sanctuaries. The exceptions are Vrysinas, Keria, Filiorimos and Pyrgos which is probably the result of a bias in the archaeological dataset.

Modern land use and basic geological formations seem to distinguish peak sanctuaries clearly from central places of power. Past observations were confirmed: peak sanctuaries are part of the pasture land, while court complexes and alike are mainly located in areas with good farming possibilities. While the visitor could be rich or poor, from close by or further way, the relation of the type of site to the resources of the land surrounding it in human terms seems clear.

For the Neopalatial period (fig. 8), almost all peak sanctuaries are within 1 ½ hour walking from all settlements, and it is important to note that here we encounter sites such as: Azokeramos, Palaikastro, Achladia, Zou, Kastelli Pediados, Archanes, Tylissos, Sklavokambos, and Zominthos. This means that the peak sanctuaries in the Neopalatial period were indeed closer to the larger and more important sites.

The Cost distance analysis between peak sanctuaries and settlements shows a divergent pattern in largely three zones (West, Central and East Crete), very similar to the visibility networks, with the same empty regions near the Dikti and Idi mountain chains. The changes in the number of peak sanctuaries from the Proto- to the Neopalatial period are not impressive, but specific cases, such as the late chronology for Kofinas excluding Phaistos of the Protopalatial peak sanctuary landscape, require a different interpretation of the Minoan cultural landscape. In the Neopalatial period, peak sanctuaries tend to be situated closer to more ‘important’ sites.

Thiessen polygons and Euclidean distance Thiessen polygons and Euclidean distance are basically simplified versions of the Cost distance analysis, and were added for comparative reasons but are not presented in this paper. Xtent model The Xtent model, originally developed by Renfrew and Level (1984), presupposes that the (political) influence a site exercises is dependent on its size and the distance to other sites of the same hierarchical level. As such, a sphere of influence can be created, presented graphically as a cone. Applied to the list of central places, one observes that the link Knossos – Iuktas is reinstalled. This model is the only one that can support the concept of Knossian hegemony during the Neopalatial period (Vansteenhuyse 2004).

Territories, in terms of catchment areas based on Cost distance analysis, for the central places of power and for all sites, not only confirm the assumptions of a divergent spatial pattern, they also reinforce the idea that peak sanctuaries have closer connections to sites that have no court complexes. East and West Cretan settlement patterns loose connectivity with the peak sanctuaries, except maybe for the major sites, while in Central Crete the opposite trend can be observed, with a tendency towards hierarchy. The hegemony of Knossos on a large part of the island cannot be supported by any model, except for the Xtent model. The ideological powers of memory, the ancestry of the site of Knossos and of its focal point from the central

Conclusion The synchronism of the peak sanctuaries with the 159

S. Soetens, A. Sarris and K. Vansteenhuyse

Figure 7: Modelling of Protopalatial territories, and overlay with peak sanctuaries: Cost Distance, Thiessen polygons, Euclidean distance buffers

Figure 8: Modelling of Neopalatial territories, and overlay with peak sanctuaries: Cost Distance


Between Peak and Palace. Reinterpretation of the Minoan Cultural Landscape in Space and Time court toward Iuktas remain intriguing. If we do not take into account the sizes of the court complexes, as does the Xtent model, but apply Nearest neighbour analysis and Cost distance analysis to these central places, the results tend to verify the loose relation between peak sanctuaries and court complexes. The different results between Cost distance analysis and the Xtent model should not surprise or disappoint us. While the territories or spheres of influence of most central places tended to be regionally focused, the ideological influence exerted from, for example, Knossos – Iuktas may be more difficult to grasp archaeologically.

Karetsou, A., 1981, The Peak Sanctuary of Mt. Juktas, in Sanctuaries and Cults in the Aegean Bronze Age (eds. R. Hägg & N. Marinatos), 137-153 (Skrifter Utgivna av Svenska Institutet I Athen 4, XXVIII), Paul Aströms Förlag: Stockholm. Karetsou, A. & Rethemiotakis, G. 1990. ΑΔ : 429-430. Nowicki, K., 1994, Some Remarks on the Pre- and Protopalatial Peak Sanctuaries in Crete, Aegean Archaeology, , 31-48. Peatfield, A. A. D., 1983, The Topography of Minoan Peak Sanctuaries, Annual of the British School of Archaeology, 78, 274-276. Peatfield, A., Palace and Peak: The Political and Religious Relationship between Palaces and Peak Sanctuaries, in The Function of the Minoan Palace, Proceedings of the Fourth International Symposium at the Swedish Institute in Athens, (eds. R. Hägg & N. Marinatos), 10-16 June, 1984, (Skrifter Utgivna Av Svenska Institutet I Athen, 4, XXXV), Paul Aströms Forlag: Stockholm, 89-93. Peatfield, A. A. D., 1994, After the ‘Big Bang’ – What? Or Minoan Symbols and Shrines beyond Palatial Collapse, in Placing the Gods, Sanctuaries and Sacred Space in Greece, (eds. S. E. Alcock & R. Osborne), 19-36. Oxford University Press: Oxford. Rethemiotakis, G., 2002, Evidence on Social and Economic Changes at Galatas and Pediada in the New-Palace Period, in Monuments of Minos. Rethinking the Minoan Palaces. Proceedings of the International Workshop “Crete of the hundred Palaces ?” (eds. J. Driessen, I. Schoep & R. Laffineur), Université Catholique de Louvain, Louvain-laNeuve, 14-15 December 2001 (Aegaeum 23), Ulg-PASP, Liège–Austin, 55-70. Renfrew, C. & Level, E.V., 1984, Exploring Dominance: Predicting Polities from Centres, in Approaches to Social Archaeology, (ed. C. Renfrew), 54-78, Edinburgh University Press: Edinburgh. Rutkowski, B., 1986, The Cult Places of The Aegean, Yale University, New Haven (Conn.) & London. Rutkowski, B. 1988, Minoan Peak Sanctuaries: The Topography and Architecture, in Aegaeum (ed. R. Laffineur), 2, 71-100. Ulg-PASP, Liège-Austin. Soetens, S. 2004, Minoan Peak Sanctuaries. Building a Cultural Landscape using GIS. Unpublished doctoral dissertation, Université catholique de Louvain, Louvain-la-Neuve. Soetens S., Driessen J, Sarris A. & Topouzi S., 2002, The Minoan Peak Sanctuary Landscape through a GIS approach, Archeologia e Calcolatori, , 161-170. Soetens S., Sarris A. & Topouzi S., 2004, Peak Sanctuaries in the Minoan Cultural Landscape. Proceedings of the 9th International Congress of Cretan Studies. Vansteenhuyse, K., 2004, The Archaeology of Political Organisation. Conceptual and Methodological Issues and the Case of Late Minoan IA Crete, Unpublished doctoral dissertation, Université catholique de Louvain, Louvain-laNeuve.

Acknowledgments The authors would like to thank the Institute of Aegean Prehistory for their continued support of this project. K. Vansteenhuyse would also like to thank the Fonds National de Recherche Scientifique (B) for a travel grant. This poster is the combined effort of two individual research projects: ‘Building a cultural landscape model of Minoan peak sanctuaries through a GIS approach’ and ‘The Archaeology of Political Organisation. Conceptual and Methodological Issues and the Case of Late Minoan IA Crete’, and forms part of the wider framework project ‘A Topography of Power. Studies on the Political Structures of Minoan Crete’ based at the Université Catholique de Louvain. ( /Arka/ accueil/fsr.html) References Branigan, K., 1998, The Nearness of You: Proximity and Distance in Early Minoan Funerary Landscapes, in Cemetery and Society in the Aegean Bronze Age (ed. K. Branigan), 1326, Sheffield University Press: Sheffield. Cherry, J., 1978, Generalisation and the archaeology of state, in Social organisation and settlement (eds. D. Green, C. Haselgrove & M. Springs), 411-437 (BAR International Series 47) BAR Press: Oxford. Driessen, J., Schoep, I. & Laffineur, R. (eds), 2002, Monuments of Minos. Rethinking the Minoan Palaces. Proceedings of the International Workshop “Crete of the hundred Palaces?” held at the Université Catholique de Louvain, Louvain-laNeuve, 14-15 December 2001 (Aegaeum 23), Ulg-PASP, Liège–Austin. Haggis, D.C., 2002, Integration and Complexity in the Late PrePalatial Period. A View from the Countryside in Eastern Crete, in Labyrinth Revisited. Rethinking Minoan Archaeology, (ed. Hamilakis), 120-142, Oxbow Books: Oxford.


GEOLOGICAL CONTRIBUTION TO THE STRATIGRAPHIC STUDY OF THE PALAEOLITHIC SEDIMENTS AT THE LAKONIS BEACH, GYTHION, GREECE E. Chiotis Institute of Geology and Mineral Exploration, 70 Messogheion Str., Athens 115 27, [email protected] Abstract: The primary objectives of this paper are the stratigraphic correlation of the anthropogenic outcrops over the various sites at the Lakonis beach, the clarification of geological points mentioned in recent publications and finally the proposal of geological guidelines for the support of the archaeological excavations. Special emphasis is given to observations of erosion and transportation of the anthropogenic sediments at Lakonis, in order to distinguish primary in situ sediments from secondary redeposited strata. Among the transportation mechanisms observed, the cryogenic mass flow is the most impressive; it occurred most likely during the last glacial interval, in the period before 25 to 15 ka. The palaeolithic outcrops are classified into five formations, the oldest of which, formation A, is locally covered by the formation B, composed of conglomerates of marine transgression deposited during the last interglacial interval. The overlying formation C encompasses the most extensive outcrops and is characterized by a Levallois-Mousterian industry. The formation D forms a cone that was deposited as a result of a cryogenic mass flow and transported sediments from the proximity of the cave 2. The younger anthropogenic formation E, presumably assigned to the Upper Palaeolithic, occurs as an outcrop of a very limited extent; it is inferred though that it is maintained intact over a large area below the marble blocks of the collapsed cave 2. It is also noted that flowing water was available periodically at the various sites and this constituted an additional advantage of the settlement at the Lakonis beach. This is either directly witnessed by the presence of flowstones and travertine or implicitly inferred from the erosion and transportation of palaeolithic sediments as a result of water action. Περιληψη: Οι αντικειμενικοί στόχοι της εργασίας αυτής είναι ο στρωματογραφικός συσχετισμός των εμφανίσεων ανθρωπογενών ιζημάτων στις διάφορες παλαιολιθικές θέσεις κατοίκησης στην ακτή Λακωνίδος Γυθείου, η διευκρίνιση γεωλογικών αναφορών σε πρόσφατες δημοσιεύσεις και η διατύπωση γεωλογικών κατευθυντήριων γραμμών για την υποστήριξη των αρχαιολογικών ανασκαφών. Ιδιαίτερη έμφαση δίδεται σε παρατηρήσεις σχετικές με την διάβρωση και μεταφορά ανθρωπογενών ιζημάτων στη Λακωνίδα, ώστε να διαφοροποιηθούν τα πρωτογενή ιζήματα κατά χώραν από τα μεταφερμένα δευτερογενή. Μεταξύ των μηχανισμών μεταφοράς ο πλέον εντυπωσιακός είναι η κρυογενής ροή μάζας που έλαβε χώραν πιθανότατα κατά το τελευταίο παγετώδες διάστημα πριν από 25 έως 15ka. Οι παλαιολιθικές εμφανίσεις ταξινομούνται σε πέντε σχηματισμούς, ο παλαιότερος των οποίων, σχηματισμός A, καλύπτεται τοπικά από τον σχηματισμό B που συνίσταται από κροκαλοπαγή θαλάσσιας επίκλυσης και απετέθησαν στο τελευταίο μεσοπαγετώδες διάστημα. Ο υπερκείμενος σχηματισμός C περιλαμβάνει τις πλέον εκτεταμένες εμφανίσεις και χαρακτηρίζεται από τεχνολογία Λεβαλλουά-Μουστέρια. Ο σχηματισμός D απετέθη υπό μορφή κώνου, συνεπεία κρυογενούς ροής μάζας και μετέφερε ιζήματα από την περιοχή του σπηλαίου 2. Ο νεότερος ανθρωπογενής σχηματισμός E αποδίδεται πιθανότατα στο Ανώτερο Παλαιολιθικό και παρόλο που ως ανάγλυφο έχει μικρές διαστάσεις, θεωρείται ότι διατηρείται άθικτος κάτω από τα τεμάχη μαρμάρου του κατακρημνισμένου σπηλαίου 2. Σημειώνεται επίσης ότι στις παλαιολιθικές θέσεις υπήρχε περιοδικά φυσική ροή νερού, και αυτό θεωρείται ως πρόσθετο πλεονέκτημα της ακτής Λακωνίδος. Τούτο μαρτυρείται άμεσα από την παρουσία ροομορφών ή τραβερτίνη και έμμεσα από τις παρατηρήσεις διάβρωσης και μεταφοράς ιζημάτων με μηχανισμούς ροής ύδατος.

Important Palaeolithic sites in western Mani, near the Limeni port, have been studied at Kalamakia (Lumley and Darlas 1994, Darlas 1999) and at Apidima (Pitsios 2002). The cranial bone findings at Apidima include two Lower Palaeolithic archaic Neanderthal skulls. The high density of sites with indirect palaeoanthropological evidence in the western Mani is impressive; Bassiakos (1993) has recorded more than one hundred such sites in the area around Diros.

,QWURGXFWLRQ The Palaeolithic settlement at the Lakonis beach lies in Peloponnese, three kilometres NE of Gythion in the inner part of the Lakonian gulf (Chiotis 1997a). Significant Palaeolithic sites occur within a distance of one hundred and fifty meters along the coast (Fig. 1). Site No. 1 is the most significant, due to the greater depth and stratigraphic diversity of the Palaeolithic sediments which are rich in stone artifacts, burnt bones broken by human beings and other anthropogenic remains.

A preliminary report on the evidence for the, then unknown, Palaeolithic settlement at Lakonis was submitted to both the Greek Institute of Geology and the Ephoreia of Palaeoanthropology and Speleology (Chiotis 1996a). A working hypothesis was proposed for the period of settlement, on the basis of both the geological evidence and the typology of the artifacts (Chiotis 1996b). It was then suggested that the time span of human occupation at the Lakonis settlement ranges from at least 100 ka to 35

Middle Palaeolithic findings were also reported along the eastern side of the Lakonian gulf (Kowalczyk et al 1992). They were located during the field work of a geological study of the Tyrrhenian formations along the coastline from Elea to Neapolis. Two samples of fossilized molluscs from these coastal marine terraces were dated to 104 and 117 ka respectively (Zötl et al 1999). 163

E. Chiotis


b Figure 1. General view of the Palaeolithic sites, 1 to 6, at the Lakonis beach as seen from S-SW (Fig. 1a) and from SE (Fig. 1b); the dashed rectangle in Fig. 1a delineates the area shown in Fig. 2.

Figure 2. The Palaeolithic site No. 1; the dashed ellipses approximately delineate the collapsed caves No. 1 and 2 respectively. The dashed rectangles define the locations shown in the Fig. 3 and 8.

Levallois-Mousterian industry of the Middle Palaeolithic, whereas Upper Palaeolithic features were recognized in the microlithic artifacts of black chert found at the outcrop C3 (Fig. 8).

ka, with a very likely continuation into the Early Upper Palaeolithic. The geomorphology of the Palaeolithic sites and the lithology of the artifacts were described on the basis of the abundant outcrops (Chiotis 1997a). The typology and the lithology of the artifacts at the Lakonis settlement were further described by Chiotis (1997b); the lithic artifacts of outcrop C2 (Fig. 2) were attributed to an evolved

Since 1999, archaeological excavations undertaken by an interdisciplinary team, have been in progress at site No. 1 (Fig. 2) mainly in the formations A2 and C2. Among other findings, a Neanderthal tooth was discovered which has 164

Geological Contribution to the Stratigraphic Study of the Palaeolithic Sediments at the Lakonis Beach been described by Harvati et al (2003). In the framework of this project, Karkanas (2002) reported that “evidence of major hiatuses or naturally deposited sediments” and “indications of freeze-thaw activity in the sediments” were not observed. Both of these points are reconsidered in the present paper.

rise of the sea level, on top of the early cemented mass of A1. A similar deposition of conglomerate trapped in a small cave also occurs at site No. 6 (Fig. 1b). A1 is the oldest anthropogenic outcrop at site No. 1, as implied by its stratigraphic position below the conglomerate B. The significant thickness of the conglomerate bank, ca. one metre, suggests that the bank initially extended seawards. Its present truncation is due to tectonic subsidence along a coastal fault. This trend of tectonic movement is in accordance with the subsidence of ancient sites at a depth of five meters in the bay since 2000 B.P. (Zötl et al 1999), as well as with the trend of tectonic subsidence in the western Laconian bay since the Pliocene (Kelletat et al 1976).

Five successive stratigraphic units are distinguished by Harvati et al (2003). Their unit Ia is identified as Initial Upper Palaeolithic (IUP), dated to between ca. 44 to 38 ka based on two 14C AMS dates taken from charcoal. Typologically, the tool assemblage is considered as a mixture of Middle and Upper Palaeolithic morphotypes, with the latter dominating the tool inventory. The Lakonis IUP is considered to be characterized by the same raw materials as those found in the Middle Palaeolithic assemblages of the site, but with an emphasis on the higher quality varieties (Harvati et al 2003).

This typical conglomerate, formed on the sea bed in front of a sea cliff as a result of an interglacial marine transgression, is composed of pebbles larger than five centimeters, cemented with fine-grained sandstone; strangely enough, it is described by Harvati et al as beachrock (their unit V).

The primary objectives of this paper are the stratigraphic correlation of the anthropogenic formations over the various sites at Lakonis, the clarification of geological points mentioned in recent publications (Karkanas 2002, Harvati et al 2003) and finally the proposal of geological guidelines for the support of the archaeological excavations.

Another type of marine sediment also occurs at site No.1 dispersed in minor outcrops at various points. It has been preserved in small masses enclosed either in joints or in small cavities of the marble. It is a well classified marine sandstone or even coarser in grain size; it has been deposited closer to the coast along the sea cliff. It is found in the border of the outcrop A1, as well as a few meters higher.

/LWKRVWDWLJUDSKLFREVHUYDWLRQVRQWKHSDODHROLWKLF RXWFURSVDWWKHVLWHQR The main volume of the Palaeolithic sediments at the Lakonis settlement has been preserved at site No.1, shown in detail in Fig. 2. A number of various formations can be distinguished, taking local variations into account. Therefore, the stratigraphic observations are first described for the individual outcrops and finally a correlation and synthesis is presented for the identified formations.

Two different anthropogenic formations, C1 and D, rest directly in places on the conglomerate bank (Fig. 3 and 4). The conglomerate is intercalated between Palaeolithic anthropogenic sediments; therefore it constitutes a stratigraphic hiatus in the Palaeolithic sequence. The return of settlers after the regression of the sea indicates that the Lakonis site possessed some advantages attractive to Palaeolithic people.

The outcrop A1 (Fig. 3) rests on marbles, M. It is well cemented and composed of red terrigenous matrix enclosing angular fragments of marble and sparse bones and stone flakes. A transitional zone exists, A1trn, at the base of A1 where the inclusion of loose marble fragments gradually increases towards the underlying massive marbles M, as a result of progressive dissolution of the terrigenous matrix. The remarkable outcrop a1 corresponds to a small karstic cavity filled with a very well cemented red sediment rich in stone flakes and bones.

The outcrops C1 and D rest on the conglomerate bank, as shown in Fig. 3 and 4. C1 is covered at the top by marbles that apparently belonged to the border of the collapsed cave 1. Flowstones, F, formed on the floor of the cave are maintained, encircled by cryogenic breccia of the outcrop D (Fig. 4). C1 closely resembles the outcrop C2 described below. The outcrop D (Fig. 3 and 4) is distinct in both its genetic origin and the geomorphological setting; it has the form of a cone that moved as a result of a cryogenic breccia flow that eroded and truncated the underlying sediments. D is most extensive at the base of the cone where it is placed directly on the conglomerate bank. It narrows at higher elevations and overlies the outcrop C1 along a dipping contact. D is a very well cemented rock composed of unclassified angular pieces of marble, stone flakes and bones that vary in size from less than one millimeter to some centimeters (Fig. 5). It is clear in thin section microscopy that the matrix

Both outcrops A1 and a1 have been deposited in a cave, named cave 1 in the following and shown approximately in Fig. 2. This cave has been destroyed due to both seawards tectonic subsidence and marine erosion, so that only a residual mass of sediments, A1, deposited in the rear of the cave is preserved. The outcrop A1 underlies isolated blocks of conglomerate, B. The conglomerate blocks have been undermined due to the preferential erosion of the underlying softer sediment A1. The conglomerate has been deposited and trapped in the cave 1 during an interglacial 165

E. Chiotis cryogenic sediments subject to freeze-thaw action of water. The cryogenic action in outcrop D is further confirmed by the cluster of subparallel microfractures that penetrate the mineral grains (Fig. 6b). The outcrop A2, shown in Fig. 2, has been deposited in karstic cavities of the underlying marble. Its components were transported there by gravity along the geomorphological slope below the floor of the cave 2. The transported material was also subject to some water action, as a result of which the earthy argillaceous component was partially washed out; thus, both the stone flakes and the pieces of bones were enriched. A2 clearly differs from the overlying sediments of the outcrop C2; however, the contact between A2 and C2 is not very clear since collapsed blocks of marble and cave flowstones occur in between.

Figure 3. Delineation of the outcrops described in the text; frame width ca. 15 m. The location is shown in Fig. 2.

C2 is the most extensive outcrop at site No. 1 (Fig. 2). Its gray-reddish colour is the result of the admixture of ash and charcoal with red terrestrial sediment. It is inferred that C2 was originally deposited in the front part of cave 2 and was further transported by gravity and accumulated to form a ramp in the open air in front of the cave. C2 was loose at the time of deposition and subject to creeping as witnessed by steep surfaces of discontinuity penetrating the outcrop C2. These discontinuities (Fig. 7) are presently healed through the deposition of porous travertine similar to that of the fossilized roots which pierce the mass of C2. This dense net of fossilized roots confirms the accumulation of outcrop C2 was in the open air. The fossilization of the roots was associated with a short event that is also observed elsewhere at Lakonis, for example at the site 3a, shown in Fig. 1a and 9a. During this event a blanket of travertine, a couple of centimeters thick, covered the surface of the outcrops C2 and C3; parts of this blanket are still maintained and shown in Fig. 8. The outcrop C3 (Fig. 8), close to the entrance of the collapsed cave No. 2 is of special interest because of its horizontal layering, which implies primary deposition in situ. Therefore, underneath C3 the stratigraphy of the Palaeolithic sediments is expected to be undisturbed.

Figure 4. Closer view of the marine conglomerate B, the cryogenic breccia D and the flowstone F; tape length 50 cm.

The outcrop E (Fig. 8) represents the youngest accessible cave sediments at site 1. It is found at the face of the collapsed cave 2, on top of the travertine blanket T, enclosed by fallen blocks of marble from the roof. E is composed of angular fragments of marble surrounded by red terrigenous material which is locally cemented and partially maintained loose as a fine powder. Figure 5. Close-up of the cryogenic breccia D.


cementing the breccia is composed of isolated rounded particles of argillaceous material surrounded by crystalline calcite (Fig. 6a). This micromorphological texture is similar to that described in deposits of the Theopetra cave (Karkanas 2002; Fig. 2) and is considered to be typical of

Although the Palaeolithic sediments at the cave of site 3 have been most extensively eroded, the few preserved remains retain significant information (Fig. 9). One of them, outcrop A3a, lies at the present entrance of the cave 166

Geological Contribution to the Stratigraphic Study of the Palaeolithic Sediments at the Lakonis Beach a


Figure 7. Travertine sheets deposited in subvertical surfaces of the formation C2; the rim of the travertine is underlined by dashed lines (tape length 10 cm).

Figure 6. Photo-micrographs of the cryogenic breccia in polarized light (||N). (a): rounded argillaceous particles formed by repeated freeze-thaw action; frame width 4 mm. (b): cryogenic microfractures; frame width 1.5 mm.

(Fig. 9a) and the other one, A3b, at the innermost part of the cave (Fig. 9b). Both of them are red and very well cemented but look quite different.

Figure 8. The thin travertine crust T, as well as the outcrops C3 and E at the face of the collapsed cave No. 2; tape length 30 cm. The location is delineated in Figure 2.

A3b is protected below a large column of flowstones (Fig. 9b). The outcrop A3b is composed of rounded pieces of marble and sparse bones and stone flakes cemented with red terrigenous material. Apparently, the slow water flow that formed the flowstones could not shift the marble fragments, which were only subject to chemical erosion and rounding. However, the earthy red terrigenous material was partially washed out so that the residual mass was enriched in the pieces of marble. In addition, the thickness of the sediments was reduced by at least one meter. This is inferred from the flat bottom of the projected flowstone F shown in Fig. 9b, which is considered to mark a previous upper level of the cave sediments at the initiation of the flowstone formation. Sampling of flowstones from cave 3 should take into account that the base of the column just above outcrop A3b is younger than the flowstone F shown in Fig. 9b. It is noted that the marble fragments fell into the cave through a fault opening of the roof that was later sealed with flowstones.

belong to the same Palaeolithic formation. Both of them were covered by a flat flowstone layer a few centimeters thick, which was deposited on the floor of the cave at this time, as inferred by traces of it on the walls of the cave. Obviously, these differences between outcrops A3a and A3b reflect the different conditions prevailing at various points within the same cave. This is plausible, since A3a was situated closer to the entrance of the cave, whereas A3b was deposited at the less frequented back side which was open to the air through the fault opening. The sediments of the outcrop at site 3a are attached to the marbles and have therefore been deposited at the rear of an eroded cave. They resemble the layers of outcrop C2, are clearly bedded and dip about 15o towards the marbles. This setting indicates transportation of the sediments from the central part of the cave towards the rear. (URVLRQDQGWUDQVSRUWDWLRQPHFKDQLVPVRIWKH DQWKURSRJHQLFVHGLPHQWV

Outcrop A3a is rich in bones of big mammals and stone flakes but devoid of marble pieces, an indication that it was originally deposited inside the cave. A3a and A3b look quite different in composition, even though they clearly

Before attempting the stratigraphic correlation of the outcrops described above, special emphasis is given to 167

E. Chiotis 3.

Accumulation of sediments in the form of a ramp in front of a cave, a typical case of which is represented by outcrop C2 at site 1 (Fig. 2); this mechanism of accumulation results in a stratigraphic structure quite distinct from that of nearly horizontal layering inside a cave. Younger sediments are deposited over the older ones along dipping boundaries, so that isochronous sediments are spread at different altitudes. The outcrop A2 (Fig. 2) was deposited at site No. 1 under similar conditions in front of cave 2, so that the sediments were easily transported by gravity over a short distance to be finally trapped into karstic cavities.


Cryogenic mass flow represented by the outcrop D (Fig. 4), whereby the slow mass movement was due to the freeze-thaw action of the water and the transported material was deposited in the form of a cone, which during its placement eroded the underlying sediments. The source of the cryogenic cone is considered to be in the locality of the presently collapsed cave No. 2.

6WUDWLJUDSKLFV\QWKHVLVDQGFODVVLILFDWLRQRIWKHRXWFURS LQWRIRUPDWLRQV Based on the above stratigraphic observations five formations can be distinguished, namely A to E. The outcrops A1, a1, A2, A3a and A3b are considered to be stratigraphically equivalent, constituting the formation A. Facial differences between the neighbouring outcrops A3a and A3b or A1 and a1 are ascribed to local differentiation in the prevailing conditions within the same cave. This reasoning is also extended to the outcrops A1 and A2, formed in the neighbouring caves 1 and 2 respectively. The small residual pockets of red sediments attached to the marbles between sites 3a and 4 are tentatively considered to also belong to formation A. Formation A is the older one at sites 1 and 3 and perhaps in the whole settlement. The latter reservation is expressed because the sediments of cave No. 4 (Fig. 1b), which cannot be reliably correlated with the rest, might be even older.

Figure 9. (a): Anthropogenic sediments A3a at the entrance of the cave at site No. 3; the dashed rectangle indicates the inner part of the cave shown in Fig. 9b. (b): The lower part of the flowstone column and the underlying outcrop A3b.

observations regarding the conditions of erosion and transportation of the anthropogenic sediments at Lakonis, in order to distinguish between primary in situ and secondary redeposited strata. The following cases can be classified at the Lakonis sites. 1.


The formation A was cemented early and therefore it was only partially eroded by the transgression that deposited the overlying conglomerate of the formation B during the last interglacial interval.

Erosion in situ in a low energy environment, with a typical case being the formation A3b preserved below the flowstones in the cave at site 3 (Fig. 9b).

The anthropogenic formation C is represented by the outcrops C1, C2 and C3 at site No. 2 (Fig. 3 and 8), as well as the outcrop at site 3a (Fig. 9a). The deposition of C was terminated by the formation of the thin travertine mantle and the associated formation of the fossilized roots. C1 and C2 are considered equivalent, whereas C3 constitutes the terminal stage of formation C. By contrast to C2, C3 was deposited inside cave 2 and remained in situ. It is suggested that the outcrop C3 (Fig. 8) is a better site for the study of the stratigraphy. Despite the practical difficulties, the most challenging area for the stratigraphic study is that

Erosion of cave sediments in situ, as described above, and subsequent transportation and redeposition through the temporary action of higher energy water flow (Fig. 10). The rounding of the marble pebbles of the “petrified stream” S in Fig. 10 cannot be explained by the mechanical abrasion in the short distance of transportation. This is also confirmed by the abundant oblong pieces of thin bones. 168

Geological Contribution to the Stratigraphic Study of the Palaeolithic Sediments at the Lakonis Beach covered presently by the large blocks of the collapsed cave 2 (Fig. 2). Drilling a borehole there might be a reasonable starting approach. It is emphasized that the cryogenic formation D incorporated sediments from the proximity of cave No. 2 before its collapse. As explained previously, formation D was formed as a result of cryogenic mass flow, after the deposition of formation C, most likely during the last glacial interval. Formation E was similarly deposited after formation C and before the collapse of cave 2 and it is here where the beautiful leaf point was found (Chiotis 1977b). The correlation in time between the formations D and E cannot be inferred from the available evidence. &RQFOXVLRQV A stratigraphic correlation of the anthropogenic sediments at the Lakonis settlement is presented. Severe events of erosion and redeposition of sediments are recognized and this allows the distinction between sediments in situ (A1, A3a, A3b, C1, C3, E, small pockets of the type of a1) and redeposited sediments (A2, C2, D). Transportation and redeposition cause mixing of sediments from different stages of settlement and this seems to be the case particularly at the outcrop C2. A further implication is that charcoal used for dating might be subject to the above misleading effects. The deposition of C2 over a ramp results in a structure of convex dipping layers; therefore, excavation trenches at different altitudes might encounter the same or contiguous stages.

Figure 10. Reworked anthropogenic sediments, S, transported by water flow; tape length 50 cm. (a): general view; (b): close-up of the outcrop delineated in Fig. 10a.


The outcrop C3 (Fig. 1) is considered as a reasonable candidate site for testing the possible existence of Initial Upper Palaeolithic in situ. However, the greatest challenge is the study of the area covered by the large blocks formed from the collapse of cave 2; this can be initially attempted by drilling. It is anticipated that the collapsed blocks of cave 2 cover the most complete and intact Palaeolithic stratigraphy.

Bassiakos, Y., 1993, Dating of fossils from caves and speleothemes: evidence from electron spin resonance (E.S.R.) technique, the study of underground karst morphology and the relevant radiometric and geological conditions in speleoenvironments of Dyros, Mani. Ph. D. thesis, Univ. of Athens, 380p. Chiotis, E., 1996a, Report on the discovery of a new centre of Palaeolithic settlement at Gythion, Institute for Geological and Mineralofical Exploration, 24p. (in Greek, unpublished). Chiotis, E., 1996b, The discovery of a Palaeolithic settlement centre at Gythion, Peloponnese, Greece, Third Symposium on Archaeometry, Greek Society for Archaeometry, ProgrammeAbstracts, pp.68-69. Chiotis, E., 1997a, Geomorphology and lithology in a new centre of Palaeolithic settlement at Gythion, Lakonia: a preliminary study, Bulletin of the Geological Society of Greece, , 49-62 (in Greek). Chiotis, E., 1997b, Description of a Middle Palaeolithic industry at Gythion, Lakonia, University of Athens, Anthropology, , 53-68. Darlas, A., 1999, The Palaeolithic Mani: the excavation at Kalamakia, Gythion 92 p. (in Greek). Harvati, K., Panagopoulou, E. and Karkanas, P., 2003, First Neanderthal remains from Greece: the evidence from Lakonis, Journal of Human Evolution, , 465-473. Karkanas, P., 2002, Micromorphological studies of Greek

Finally, is noted that flowing water was available periodically at the various sites and this constituted an additional advantage of the settlement. This is either confirmed directly by the presence of flowstones and travertine or implicitly inferred by the erosion and transportation of Palaeolithic sediments as a result of water action. $FNQRZOHGJPHQWV Sincere thanks are expressed to the referee for the constructive comments thanks to which the final text was significantly improved.


E. Chiotis Lumley, H. de and Darlas, A., 1994, Grotte de Kalamakia (Areopolis Péloponnèse), Bulletin de Correspondance Hellenique, , 535-559. Pitsios, T., 2002, The fossil hominids from Apidima, South Peloponnese, Greece, Lakonian Studies, 16, 163-190. Zötl, J.G., Geyh, M.A., Riepler, F., Mettos, A. and Georgiou, Ch., 1999, Klimaepochen, eustatische Meeresspiegelschwankungen und Strandterrassen im östlichen Mittelmeer (Griechenland), Beiträge zur Hydrogeologie, , 5-66.

prehistoric sites: new insights in the interpretation of the archaeological record, Geoarchaeology,   , 237-259. Kelletat, D., Kowalczyk, G., Schröder, B. and Winter, K.-P., 1976, A synoptic view on the neotectonic development of the Peloponnesian coastal regions, Zeitschrift der Deutschen Geologischen Gesellschaft, , 447-465. Kowalczyk, G., Winter, K.-P., Steinich, G. und Reisch, L., 1992, Jungpleistozäne Strandterrassen in Südost-Lakonien (Peloponnes, Griechenland), Schriftenr. f. Geowiss., , 1-72.


TROY – GEOMORPHOLOGIC CHANGES IN THE VALLEY OF THE RIVER SCAMANDER P. Malfas Lemessou 44, Papagou 156 69, Attiki Greece e-mail: [email protected] Abstract: The lower valley of the river Scamander has its exit into the Hellespont and encloses the plain of Troy. The ancient ruins are found on a hillock at the eastern side. According to topographical indications in Homer’s Iliad, the part of the plain between the ruins and the coast of the Hellespont -5 kilometers north- can be identified with the battlefield between the Achaeans and the Trojans, because the former had camped on this coast. However, the issue is whether the above alluvial field existed in the era of the Trojan War. Theoretically, considering the alluvium of 33 centuries, one concludes that the sea would have surrounded Troy in that era. To clarify the real history of the plain, recent geological investigations have been undertaken. The findings of the boreholes contradict the theoretical estimations. The dating of the sediments reveal that the valley was already entirely alluviated in prehistoric times. At the time of the Great War the plain extended beyond Troy, like today. Nevertheless, an irregular evolution is indicated in the stratigraphy. Among the very ancient terrestrial deposits, there are also younger marine ones. The existence of a sea bay during the Hellenic Dark Ages and historic times can be ascertained. Based on geological and other indications the transformation must be attributed to subversive physical phenomena which occurred in the area towards the end of the Bronze Age. They caused an invasion of the sea into the valley and a great change in the environment of Troy for many centuries, until it was gradually restored by new alluviums. Περίληψη: Η κάτω κοιλάδα του ποταμού Σκαμάνδρου, έχει έξοδο στον Ελλήσποντο και περικλείει την πεδιάδα της Τροίας. Τα αρχαία ερείπια βρίσκονται σε ένα ύψωμα στην ανατολική πλευρά. Σύμφωνα με τοπογραφικές ενδείξεις της Ομήρου Ιλιάδας, το τμήμα της πεδιάδας μεταξύ των ερειπίων και της ακτής του Ελλησπόντου –5 χλμ. βορειότερα- μπορεί να αναγνωριστεί ως το πεδίο μάχης μεταξύ Αχαιών και Τρώων, καθώς οι πρώτοι είχαν στρατοπεδεύσει σε αυτή την ακτή. Όμως, το ζήτημα ήταν εάν το ανωτέρω προσχωματικό πεδίο υπήρχε και στην εποχή του Τρωικού Πολέμου. Διότι, υπολογίζοντας κανείς θεωρητικά τις προσχώσεις 33 αιώνων του ποταμού, εκτιμά ότι εκείνη την εποχή η Τροία θα περιβαλλόταν από θάλασσα. Για να διευκρινισθεί το πραγματικό ιστορικό της πεδιάδας έγιναν πρόσφατα γεωλογικές έρευνες. Τα στοιχεία των γεωτρήσεων διαψεύδουν τις θεωρητικές εκτιμήσεις. Η χρονολόγηση των καθιζημάτων αποκαλύπτει ότι η κοιλάδα ήταν ήδη ολόκληρη προσχωμένη στους προϊστορικούς χρόνους. Τον καιρό του Μεγάλου Πολέμου η πεδιάδα εκτεινόταν πέρα από την Τροία, όσο και σήμερα. Ωστόσο, στη στρωματογραφία διαγράφεται μία ιδιόμορφη εξέλιξη. Ανάμεσα στις πολύ αρχαίες χερσαίες αποθέσεις υπάρχουν και νεότερες θαλάσσιες. Διαπιστώνεται η ύπαρξη θαλάσσιου κόλπου κατά τους Σκοτεινούς Αιώνες και τους ιστορικούς χρόνους. Με βάση γεωλογικές και άλλες ενδείξεις η μεταβολή πρέπει να αποδοθεί σε ανατρεπτικά φυσικά φαινόμενα, που έλαβαν χώρα στην περιοχή προς το τέλος της Χαλκοκρατίας. Προκάλεσαν εισβολή της θάλασσας στην κοιλάδα και αλλαγή του τοπίου της Τροίας για πολλούς αιώνες, μέχρι να αποκατασταθεί βαθμιαία από νέες προσχώσεις.


been identified as the river Scamander of the Iliad. It is approximately 90 km long and its source is at the top of mount Ides (Kaz Dag, 1774m), which dominates the peninsula. It has many tributaries and at its beginning, it flows through a closed plateau – its upper valley – from which it accumulates large quantities of alluvial matter.

Our topic refers to the background of the creation of the alluvial plain of Troy and the relation of this plain with the “Trojan or Scamandrean Field” of Homer’s Iliad. The Peninsula of Troas, a broader domain of the homeric Troy, occupies the northwest part of Asia Minor (fig. 1 and 2). On the west, it is soaked by the Aegean Sea and on the north by the deep current of the Hellespont, the Strait of Dardanelles, which separates it from eastern Thrace (European Turkey). The extreme part of this peninsula towards the Hellespont constitutes a long, narrow valley, which contains the alluvial Trojan plain. It is 14 km long and 4 km wide and is crossed by one of the largest rivers of Troas, which is named today Karamanderes.

The ruins of Troy are on the eastern side of the plain at an altitude of 30m, at the edge of one of the low ridges, which surround the valley. The hill with the ancient castle is 5 km away from the coast of the Hellespont and on its northern foothills the tributary Dumrek flows, which is identified with Simoeis of the Iliad. Generally speaking, the topography of the area as it is today, coincides with the one portrayed in the Iliad’s descriptions, a fact which contributed to the discovery of Troy. Thus, the part of the plain between Troy and the coast

This river, which flows out into the Hellespont, has 171

P. Malfas

Figure 1. The Homeric Geography of Greece and western Asia Minor (Leaf, 1912).

of the Hellespont can be identified as the battleground between the Achaeans, who were camped by the coast, and the Trojans, who entrenched themselves in the city. This was due to the fact that, according to the epic, “along the strong stream of the Hellespont” (M 30: «… … παρ’αγαρροον ’ αγαρροον Ελλησποντον» »)*, *,, there was the Achaean camp camp. Whereas, “in the mid space between the ships and the river and the high wall” (Π Π 397:: «μεσηγυ μεσηγυ νηων και ποταμου και τειχεος υψηλοιο»)* * battles were conducted (fig. 3)..

Therefore, one may conclude that the poetic descriptions bear no relation to reality. Nevertheless, if we take into account the text of the Iliad, we are entitled to suppose that things cannot have been so simple in the case of the plain of the Scamander river. Thus, we are inclined to examine another possibility: that Strabo’s bay likely constituted a later state of the valley in the place of a former plain. In other words, the difference between the “Homeric” topography of Troy and Strabo’s topography might have arisen due to the return of the sea, which was caused by extraordinary natural phenomena during the period between the prehistoric and historic eras.

However, we know from the geographer Strabo of the 1st century B.C., who describes the area in detail in one of his books, that in the Hellenistic and the Roman times at the same place there was a sea bay, swamps and lagoons (fig. 4). Taking into consideration the silting, we estimate that the situation must have been much worse thirteen centuries before, at the time of the Trojan War. The sea bay might have widely covered the area around Troy and there might have been no ground for tactics and the deployment of troops – meaning flat ground of any description.

Improbable as this view may seem, it can be supported today by the findings of the geological investigations, which have been conducted over the last 25 years within the framework of the contemporary archaeological expeditions in the area of Troy. 172

Troy – Geomorphologic Changes in the Valley of the River Scamander

Figure 2. The peninsula of Troas, according to the description of the geographer Strabo (Malfas, 1998).

Figure 4. The vicinity of Troy in the first century B.C. according to Strabo’s description (Malfas 1998). Regarding the site of “Stomalimne”, elements of geological researches were also taken into account.

the sea during the rise of its level and its retreat after the establishment of alluviums. The process described above was believed to have taken effect in a single and uninterrupted act, also in the case of this considerably idiomorphic area (a strong and continuous current in front of a delta). But, there was a contradiction in their interpretations, first based on the whole length of the valley (1982, fig. 5a) and then only on its northern half (1995, fig. 5b) for the same period of time (last 6000 years). As a result, their palaeo-geographic reconstructions depict the prehistoric Troy (3000 – 1000 B.C.), either as a coastal city, or surrounded, always, by the deltas of rivers, swamps and lagoons. The above interpretations are not only inconsistent with Iliad’s requirements, but also instead of solving the socalled Trojan topographic problem they complicate it more, to such an extent that Schliemann’s very discovery is put in doubt.

Figure 3. The plain of Troy according to the Iliad (Leaf 1912).

However, the geologists working in Troy, affected by the prevalent prejudice against the Iliad’s credibility, have not yet considered this possibility. After being led initially to plausible conclusions by basing themselves on inadequate elements, later, even though they obtained a more complete -and complex- picture of the subsoil, they became reserved and ever since hesitate to accept such an irregular evolution of the valley.

The same interpretations accompanied by slightly modified reconstructions were presented at the last conference about Troy in Heidelberg, Germany in 2001(Kraft et al. 2001). Study of the stratigraphy – first realizations We have at our disposal enough diagrams of the stratigraphy of the Scamandrean plain –vertical sections, sideways and along the axis- as well as some good topographic maps with the drilling points. These were published partially (Kraft et al. 1982; Kayan 1991; 1995; 1996; 1997; and 2002)

Therefore, until today they produce interpretations, which are compatible with the classic model of evolution of the valleys during the Holocene, meaning the extension of 173

P. Malfas

Figure 5. Palaeographic reconstructions of the Scamandrean valley for several eras from 6000 B.P. until today according to the geologists: a. John Kraft et al. (1982). b. Ilhan Kayan (1995).

during the progress of the investigations together with the corresponding expositions of the participating geologists. In the present text, we attach copies of the most important ones (fig. 6, 8, 9). As we can see in the diagrams, the nature of the depositions is described and the dating is indicated, a vertical scale is given in meters and a horizontal scale in kilometers as well as any additional information.

evaluate the speed at which the alluviums must have developed between 7000 and 5500 B.P. The stratigraphy of the flat indentation of Kumtepe (fig. 8a) at the foot of the ancient Sigeion where the river Scamander flowed 2000 years ago, according to Strabo’s description (distances in stadiums)1 speaks more eloquently. In the vertical section of this small semi-circular field, we can see the neogene bedrock (shallow and horizontal), a few meters of sea-mud followed by a thick layer of deltaic coarse sand, which starts at four meters below sea level and is interrupted at one meter above sea level and is dated between 7000 and 5500 B.P.

If we examine the stratigraphy carefully, we observe some interesting characteristics. Firstly, we can see that the deltaic coarse sand, which the river had deposited all along the west side of the plain (fig. 8a), between Yenikoy and Kumtepe, at a depth that coincides with the sea level, does not escalate in time from the South to the North, i.e. the course of the delta, but is approximately of the same age (5000 B.P. ±500).

We realize, in other words, that at that time the delta of the river was close to this point, as in Strabo’s era. By assuming that it is not possible for a delta to remain on the same spot for so many millennia, we conclude that another alluvium deposit must have existed previously in the main basin of the valley, and a respective regression of the sea. A process, which must have been much more ancient than the one Strabo recorded 2000 years before (fig. 4 and 9).

Something similar happens with the more ancient depositions, dating to approximately 7000 B.P., mostly at a depth of 4-5 meters below sea level (fig. 7). We believe that the above characteristics reveal a homogeneous distribution and increase of the depositions on the bottom of the initial estuary, covering the whole stretch of the northern part of the valley as far as its mouth, while the rising of the sea level continued. We estimate that this fact must be attributed to the effect of surface sea current of the Hellespont, which most likely penetrated the valley and was mixed with the river currents.

Knowing what happened from 2000 B.P. until today, we can also estimate what could have formed within an equal spell of time in the same area after 7000 B.P. This allows us to draw the first possible conclusion: that the regression of the sea in the Scamander valley had been completed very early (5000 B.P.) in the Holocene and that a plain extended several kilometers north of Troy during prehistoric times.

Apart from this fact, the minimal depth of the depositional surface of 7000 B.P. gives us a means with which to

Strabo: Geography, 13th Book - «Τρωάς-Αιολική Τρωάς-Αιολική -Αιολική Αιολική Γη», », 36 (C.598). C.598). Εκδ. ΚΑΚΤΟΣ, Αθήνα 1993. 1


Troy – Geomorphologic Changes in the Valley of the River Scamander

Figure 6. A north-south cross section along the flood plain of the Scamandrean river with boreholes (vertical lines), dates of sediments (numbers in years B.P.) and the deposits: marine (muddy or sandy) and fluvials (different colors). The exact points of the included boreholes are marked on the above map. They were drawn by the author of the article based on respective diagrams, side cross sections and topographical maps, which are attached to the geologists’ reports (Kraft et al. 1982, Kayan 1995, 1996, 1997, Göbel et al. 2002).

time (2000 B.P.), the stratigraphy of which is closely dated (fig. 8b and 9). In a zone of 1-2 km radius around the hill, at a depth of 4-5 meters below sea level, marine and fluvial depositions accumulated and increased significantly during a period of 4000 years (7200-3100 B.P.) If we also consider the coastal zones, which would have been present

In this case, the supposition that Strabo’s bay was formed afterwards due to the return of the sea into the valley starts to look plausible. We observe something similar on the eastern side of the plain between the hill of Troy and the coastline of Strabo’s 175

P. Malfas in Strabo’s time (2000 B.P.) at the fringes of the zone, they would have increased in total within 5000 years. It is obvious that this condensed deposition cannot be considered as a product of a single and unified regression of the sea. This would be equivalent to the immobilization of alluviums in front of Troy and stagnation in the development of the plain over millennia. We must consider that the accumulation arises because of a new recurrence of the sea and its subsequent retreat from the valley, a new regression, which leaves its traces in the same place as the initial one completed long before. As we can see the correlation of Strabo’s bay to the stratigraphy of the eastern side, this leads us to similar evaluations such as the correlation of the stratigraphy in the west (Kumtepe). Consequently, the conclusion, that the bay in question is “uninvited” and was formed by a younger penetration of the sea into the valley, is inevitable. The difference at Kumtepe is that apart from the very ancient either marine or fluvial depositions between 7000 and 5000 B.P., on the eastern side there are also the much younger deltaic depositions of 3500-3150 B.P. As newly formed as much of Strabo’s coastline, they developed between much older, terrestrial depositions of the perimetric zone.

Figure 7. A ground plain of the depositional surface 8000-7000 B.P. in the basin of the Scamandrean valley. The numbers are dates of sediments in years before present and their depth in meters under the present sea level. The approximate limits of the extent of the Holocene sea in the valley are also marked. It was drawn by the writer, based on data from the boreholes

However, it is precisely these latest alluviums that we need in order to define more accurately the new penetration of the sea, along with its cause. We know that the landscape at Strabo’s time was similar to

Figure 8. Side cross sections of the Scamandrean valley: a. Through its western zone (Yenikoy, Kesik and Kumtepe plains) on the left. b. Through its eastern zone (Dumrek and Ηiplak valleys) on the right. The locations of the corresponding drillings are indicated on the inserted map. They were composed by the author as simplified copies of the respective drawings, which were drawn and published by the scientific team of the geological investigations (Kayan 1995, 1996 and 1997). The above representations do not feature certain details of the originals, but contain all necessary information here.


Troy – Geomorphologic Changes in the Valley of the River Scamander

Figure 9. The stratigraphy of the northern zone of the valley from the Kumtepe inlet to the Dumrek valley (up). A map of the main valley (Scamander) with the points of the included drillings, the coastlines of Strabo’s time and at present, as well as the hypsometrical lines of 5 and 7 meters of the flood plain (down). They were drawn by the writer based on the respective geological diagrams and topographical maps.

the one existing several centuries before his time, before the 3rd century of Istiaia, whom he invokes, like Dimitrios of the 2nd century B.C.

the dating of samples of deltaic deposits to between 3500 & 3150 B.P. is of interest because it represents the time of the new marine penetration.

Because there are indications that the landscape was similar throughout all historic Antiquity (Herodotus’ implications about the alluvial field of Troy)2 but also before Homer’s time, we locate the transformation which caused the difference between the Homeric topography and Strabo’s topography towards the end of the Bronze Age. Therefore,

The contested ground Of course, there are no specific elements in the Iliad giving the exact location of the coast in the Hellespont, or the distance between the Achaean’s camp and the hill of Troy. The dimensions of the camp, with its protective wall and the defensive trench, is considered to be very large (hundreds

Herodotos: History, B’ Book - «Ευτέρπη», Ευτέρπη», », 10. Eκδ. κδ. ΠΑΠΥΡΟΣ, Aθήνα θήνα 1975.



P. Malfas Evaluation of the findings – correlations and estimations

of meters), considering on the one hand the number of the encamped force (thousands of men, hundreds of ships drawn ashore) and on the other hand the military activities within the camp. We also only have indications about the extent of the terrestrial space between Troy and the camp. There should be some kilometers of free flat ground to justify the tactical movements, the charioteers’ action and the pitched battles, but everything is hypothetical and therefore disputable.

On the western side of the plain, along the ridge of Sigeion and specifically in the three inlets of Yenikoy, Kesik and Kumtepe (fig. 8a), we can say that the entire history of the creation of Troy’s prehistoric plain is recorded and preserved. Each of these marine bays, on the side of the estuary, dated to around 7000 B.P. underwent a special evolution. They were transformed from marine bays into terrestrial areas either directly or gradually, after turning into lagoons, then lakes and later into swamps. This evolution, temporally independent but consistent with the structure and the position of each one, is exclusively affected by the course of alluvium in the main plain, since on the west side -that of the ridge which covers them- there is no river or gorge, which could contribute to their development.

However, there is a specific topographic feature in the Homeric text, which allows for more substantial evaluations. It is to this fact that the present research owes its persistence. It is about “the rising ground of the plain” (K 160: «θρωσμός θρωσμός πεδίοιο» »)*, *, with Ilos’ tomb thereon– the founder of Priamos dynasty (fig. 3). It was a point of reference for the tactical moves during attacks and retreats alternately; that is to say approximately in the middle of the field. It used to be, at times, the base of Trojan operations, thus closer to Troy than the Achaean camp.

The first one (Yenikoy) is cut off from the main plain and transformed into a lagoon (or lake) in 6000-5500 B.P., when its entrance silted up, and after the flooded field had been adequately developed in the main valley. This delay must have been due to the eccentric position of the inlet.

There is even today a low and wide, sandy hill, which is located approximately 1.5 km north of Troy (fig. 9). It might have developed during an older geological period (coastal sand dunes) or it could be due to similar factors existing during the initial retreat of the sea and the subsequent retreat (marine and fluvial currents, strong winds etc).

The second bay towards the north (Kesik), is almost completely surrounded by the extensions of the ridge except for a narrow opening from the east, and is likewise unable to keep up with the development of the main plain, because it is bypassed by the river currents. Therefore, its entrance is blocked even later (5000-4500 B.P.) and is covered by alluvium at its mid point between 4500 and 3400 B.P., the rest being covered many centuries later (A.D. and Strabo).

Based on the above indications, we estimate that the walls of the Achaean camp, the charging line of the Achaeans must have been more than twice the distance of the «θρωσμός» θρωσμός»» from Troy, approximately 3-4 km, and the coast of the Hellespont would have been more than triple the same distance (given the great width of the camp), approximately 4-5 km from Troy. In other words, it is approximately where it is today.

We have already referred to the northern embayment (Kumtepe) and what the transitional layer from marine to terrestrial environment (7000 -5500 B.P.) proves here.

Besides this, the currents of the Hellespont, which delay the progress of alluviums in the outer part of the valley, define in some way the northern borderline of the Trojan plain.

All these changes in the areas mentioned above on the western side of the plain are taking place in accordance with the rise in the sea level (fig. 10 and 11). If we consider also the shallow depth and the slight inclination of the neogene bedrock -elements which would suggest that the initial stratigraphy remained intact throughout the subsequent millennia until today-, we can acknowledge that these vertical sections are more reliable compared to the deep stratigraphy of the central axis of the valley.

So, lacking any additional arguments, we can say that there is no room for compromise, neither for the orientation of the Achaeans base, nor for the size of the field north of Troy, even if we admit that Homer’s descriptions are fictitious. Especially since according to all data, at the poet’s time (8th century B.C.) the area in front of the ruins of Troy was most likely dominated by deltas, swamps and lagoons. It would thus be inaccessible and he would not have been able to provide topographically such a detailed “scenario”. While, on the contrary, the facts clearly speak of a real, extensive Trojan field, accessible and crossable, at the time of the War (13th century B.C.)

In the vertical section diagram alongside the Dumrek (Simoeis) valley (fig. 8b) and in the similar diagram of the smaller southern plain, formed between the secondary two arms of the ridge of Troy and crossed by the small gorge of Çiplak, we note to the east, the largest extension of the Holocene marine intrusion at 8000-7000 B.P.

Taking into account everything mentioned above, we must apply a more comprehensive and bolder explanation of the findings of the geological investigations.

According to the dating of the sediments, which are 178

Troy – Geomorphologic Changes in the Valley of the River Scamander marked on the inner side of the first valley and on the exit of the second one, we can also observe on this side an homogeneous rising of the bottom of the estuary in 81007000 B.P. Here also, this elevation of the deposits is, as on the west side, in accordance with the local curve of the rise in sea level. In the vertical section of Dumrek valley (Simoeis) and in its inner part -between drillings “9” and “33”- we can see the steady elevation of the sandy marine deposits dating between 8100 and 7600 B.P., followed by a thick (3-4 meters) and long (1.5 km) layer of coarse deltaic sand, which constitutes a transitional zone from marine to terrestrial environment. It begins, as in Kumtepe, from a depth of four meters and is interrupted at almost one meter below present sea level. It is dated to 7000 B.P. and thus proves the early transformation of the area from a marine to a terrestrial environment. Subsequent alluvial depositions follow that are earlier than in the opposite northern inlet of Kumtepe (5500 B.P.) and much earlier than in Kesik (4500 B.P.), thus indicating the direction of alluvial deposition in the estuary from the east to the west.

Figure 10. A sea level curve from 17000 B.P until today (Bintliff, 1991).

The same cannot be said for the part of Simoeis valley closest to Troy and the area where it meets the main valley of Scamander (Karamanderes) between drillings “16” and “36” (fig. 9). We note that the transitional level here is not synchronous to the transitional zone 7000-5500 B.P., which is aligned sideways at the two ends (Kumtepe – Dumrek), but belongs to an earlier era, younger by 3000 years (3500-3150 B.P.), which constitutes, as we have also gathered, a posterior aspect of the valley.

Figure 11. Sea level changes on the Aegean coast of Troy during the last 7000 years (Kayan 1991).

However, independently of what occurred in this intermediate area and based on the more ancient stratigraphy of the small valley, we can say that very early in the Holocene the active and productive river Simoeis not only created its own plain as far as Troy, but also considerably influenced the evolution of the northern part of the main Scamandrean valley between 7000 and 5500 B.P.

During the next phase of the decelerated elevation of the sea level, the inundation by the sea is suspended and starts retreating from the valley, due to the accelerated establishment of alluviums after the sea level stabilized 6000 years ago at approximately level it is today. So between 7000 and 3500 B.P., the shallow estuary is gradually restricted until it is obliterated from the valley, while at the same time the northern plain of prehistoric Troy expands continuously until it is fully formed according to the reconstructions of fig. 12 approximately.

Retreat of the sea – development of the Trojan Plain 7000-3500 b.P. Between 10000 and 7000 B.P., a period of rapid sea level rise and a gradual penetration of the sea into the valley of river Scamander took place. The marine penetration, narrow and initially brief, in the bottom of the basin (4030m ?), occupied the wider extent of the -mostly- shallow basin of the valley, while the rise sea level covered the last 10 meters.

It can be seen that the marine penetration regresses from all of the southern part of the plain even before the rise in sea level begins to decelerate (7000 B.P.), seemingly because the penetration in this area was too shallow and too narrow compared to the width of the valley. This may be attributed to a shallower bottom of the basin here or to the Pleistocene fill maintaining high levels.

At the same time, with the rise in sea level, marine and fluvial deposits rapidly fill the deeper parts of the flooded basin, raising the level in such a way that towards the end of the era (7000 B.P) a shallow (1mm sieve, as the fine-flots likely to contain processing by-products have not yet been examined. The seed evidence indicates the presence of various pure or less pure concentrations of grains in various parts of the houses, either in relation to constructions or on the floor. Based on the observed concentrations of grain the species list can be translated as follows (Table 1): a. concentrations of at least 4 cereal species including: einkorn, emmer, spelt and barley, b. concentrations of 4 species of pulses including: peas, lentils, grass pea, Celtic bean, c. concentrations of two oil-rich plants: flax and cf. Lallemantia sp. (Figure 4), d. concentrations of acorns and d. one concentration of a cf. Leguminosae plant (Figure 5).

Figure 3. Complex clay facility from Archondiko

Cooking ingredients The next step towards our goal, i.e. to approach cooking ingredients, is to distinguish among the crops and harvests from the wild identified at the site, those that represent food intended to be consumed by people, from those intended to be used as fodder. This is a rather difficult task as the distinction between food and fodder is culturally defined and may be flexible even within the same culture as modern ethnographic evidence suggests (Halstead 1990, Halstead & Jones 1989, Jones & Halstead 1995, Jones 1998 Megas 1988). If we use species to approach the issue we are faced with biases posed by modern perceptions and lifestyles, we are faced with classifications culturally defined and modified through time within each society. Let us take the two major cereal categories both identified at Archondiko, and at many other sites: wheat and barley. Although barley in recent times is described as low status food (Halstead 1993: 64) and was used in bread making only in times of wheat crop failure (Voutsina 2000) there are no inherent properties in barley that would render it unsuitable for human consumption. In fact it is used on the island of Crete for the preparation of gallettes that shepherds take with them up on the mountains (Voutsina 200: 15). This is also reflected in classical Athens where barley was the ingredient of ‘maza’, a staple foodstuff for farmers, soldiers and citizens alike, a food for the masses, highly valued but at the same time considered ‘inferior’ to wheat preparations and cheaper than wheaten bread (Micha Lambaki 1984:

This list of deliberately accumulated seeds/nuts and therefore used by the inhabitants of Archondiko is impressive, more numerous by far compared to the crops identified from EBA Kastanas (Kroll 1983), Sitagroi (Renfrew J. 2003) and Mandalo (Valamoti & Jones in press), the only sites in northern Greece from which a fully published archaeobotanical study is available. At Kastanas only 5 species represent concentrations of crops/ harvests from the wild and these include einkorn, emmer, barley, flax and acorns, while at Mandalo these include the following 7 species: einkorn, emmer, barley, grass pea, lentil, bitter vetch and Celtic bean (Table 1). At the other sites, the narrower range of crops/harvests represented may be, at least partly, due to the small excavated area and/or small number of samples. Thus, this exceptional position of Archondiko among these other Early Bronze Age sites in terms of number of crops/harvests represented, is an outcome of the extensive horizontal excavation and the large scale, intensive sampling (over 1000 samples).


S.M. Valamoti, A. Papanthimou and A. Pilali

Figure 4: Charred seeds of cf. Lallemantia sp.

Figure 5: Charred seeds of Leguminosae indet.

24-33); moreover, during Roman times, barley was a foodstuff used to distinguish Greeks, consuming it, from Romans consuming emmer wheat (Braun 1995). Thus both crops could have been used as food at Early Bronze Age Archondiko. The same applies to the pulse species. Grass-pea for example has occasionally been considered a fodder crop, dangerous for human consumption (Cohn & Kislev 1987, Halstead 1990, Halstead & Jones 1989), yet it is sold in Greek supermarkets and consumed in restaurants and households as a delicacy under the name fava. Acorns although usually thought of as fodder they were considered an old fashioned or barbaric food by Greeks; they were nevertheless considered a food of honourable men by Hesiod and comprised the basic foodstuff for certain tribes of American Indians (Mason 1995, Renfrew J. 1973). Given the potentially dual use of the crops/harvests identified at Archondiko as either food for humans or fodder, are we then to conclude that it is basically impossible to approach food beyond a general assumption that edible species featuring in pure and large concentrations in the archaeobotanical record represent potential food? Although at many sites this may indeed be the case, at Archondiko we are presented with a rare opportunity to move beyond this stage of speculation and positively state that some of the available grain concentrations were indeed intended for human consumption.

involved charring of fragmented einkorn grain soaked and unsoaked and commercial fragmented cereal grain, Greek ‘pligouri’ and Arabian ‘taboulé’. Charring was carried out at temperatures of 2000C and 3500C under both oxydizing and anaerobic conditions. The aim of the experiment was to investigate the effect of a) charring conditions and b) treatment with water on the cut surface of the grain. The experiment indicated that the concentration of the fragments found in House E have been generated from the disintegration of the conglomerations of cereal grain fragments, probably barley and that the fragments had been generated prior to charring. It was not possible to tell whether the conglomerations represent some durable form of food in the form of cakes or whether they are the result of high charring temperatures. Those from House B have only recently been found and represent fragments broken prior to charring as they show a characteristic bulging. To our knowledge, both finds from Archondiko, together with a similar find from the Early Bronze Age levels at Mesimeriani Toumba (Valamoti 2002) are so far among the earliest archaeobotanical indications for cereal grinding from Greek settlements; the other one reported is dated to the Late Bronze Age and comes from Santorini (Sarpaki 2001). Similar finds occur in much earlier deposits dated to the Neolithic in Bulgaria (Marinova, E., June 2001, pers. comm.) and it may be a matter of time and systematic recovery before we find them in Greek Neolithic contexts. Due to high charring temperatures it is not possible to distinguish whether the Archondiko fragments had been treated with water; had this been the case one would expect to see the characteristic shiny surface observed on the experimental material (Valamoti 2002). This shiny surface is a result of the gelatinization of starch caused when grains are boiled or soaked in hot water, a practice involved in the making of bulgur (Bayram 2000).

Food remains from Archondiko Samples from the interior of house B, from construction A, and samples from house E, from a construction complex, contained concentrations of fragmented cereal grain either loose or in conglomerations (Figure 6). The origin of those fragments was investigated experimentally in order to determine whether: a) the fragments had been generated prior to charring and b) whether the fragments had received treatment with water analogous to that involved in bulgur making. The experiment, the results of which have already been published elsewhere (Valamoti 2002),

The evidence for grain fragmentation in large and dense concentrations in a storage/food preparation context, as is indicated by the large number of constructions and whole 190

Cooking Ingredients from Bronze Age Archondiko: the Archaeobotanical Evidence

Archondiko Einkorn Emmer Spelt Barley Peas Lentils Grass pea Celtic bean Flax Cf. Lallemantia sp Acorns Leguminosae indet.

Mandalo Einkorn Emmer Barley Grass pea Lentil Bitter vetch Celtic Bean

Kastanas Einkorn Emmer Barley Flax Acorns

Sitagroi Einkorn Bread/macaroni wheat Bitter vetch Celtic Bean

Table 1: Preliminary crop/harvest list from the Early Bronze Age1 levels from Archondiko, Mandalo, Kastanas and Sitagroi (crops have been identified.on the basis of more or less pure or mixed but rich concentrations of a species; cf. Halstead 1994).

pots found in the houses, suggests that the ground cereal grain was intended to be used as food, stored in a form that could be cooked easily, especially if it had been treated with water prior to charring. It could have been stored in the form of loose fragments, like modern bulgur sold nowadays in Greek grocery-shops and supermarkets. Or, it could have been prepared in some form of dried cakes like ‘chondros’ used in Crete nowadays or like ‘kishk’ prepared in Jordan nowadays with bulgur and strained full fat yogurt as a means of storing cereal and milk products for a long time (Palmer 2002). In support of this possibility comes another find, at times associated with the cracked barley finds: besides the loose fragments, conglomerations of what appears to be fragmented grain but which could also represent some form of food (Figure 6b/c). Such finds are known from archaeological sites in France (Lannoy et al. 2003) and Austria (Kohler-Schneider 2001, pp 153154). Although it is tempting to think that the ‘masses’ from Archondiko represent the remains of some kind of food preparation, it is important to explore first the contribution of charring conditions in such formations, a possibility raised during the course of experimentation (Valamoti 2002). We hope that future analyses of the pattern and texture of the surface and the interior of these masses along the lines suggested by Lannoy et al. (2003) as well as using scanning electron microscopy, might shed some light on the issue. In this way we may be able to identify the ingredients of some Early Bronze Age food preparations.

Vakirtzoglou 1990, Kiziridou 2002). The Archondiko find suggests that this was a widely used processed form of cereal food as it has been found in two separate houses. And so, to return to the question wheat or barley, it seems that barley was almost certainly used for human food, and was not reserved for animals (or animals only). We cannot tell whether it represented an emergency food, used when wheat cultivation was unsuccessful, or whether it represented an ‘honourable’ food, like the ‘maza’ of ancient Greeks. The next thing to ask is whether the different houses differed in their preferences concerning cooking ingredients, judging by the hoards of food they contained. The level of analysis of the archaeobotanical samples does not allow a detailed representation on a sample by sample basis of the individual concentrations of crops/harvests contained within each house. A general impression is, however, that most houses hoarded different cereal and pulse species, as well as an oil plant, and/or acorns. A combination of a wide range of crop species, including cereals and pulses, is seen as a means adopted by Neolithic and Bronze Age farmers to spread labour throughout the year, to exploit different soils and microenvironments in the vicinity of the settlement and thus secure at least some harvests in cases of unfavourable weather (Halstead 1981, 1990, Sarpaki 1992). This variety and combination of cereals and pulses in ‘economic’ terms is advantageous not only from an agricultural point of view, namely reducing risks in crop production, but also from a nutritional point of view, providing carbohydrates (cereals) and proteins (pulses) (Halstead 1981, Renfrew1972,). This variety at the ‘house’ (household?) level may suggest equal access to a wide range of culinary ingredients by each unit inhabiting the post-framed houses. The full analysis of the archaeobotanical data set will definitely bring to light more information as regards shared characteristics and differences between houses/household units (‘shared and specialised ingredients’ in our approach here).

This form of preparing cereals recognized at Archondiko has a long tradition in the Mediterranean region, the Middle and Near East (Rivera-Nuñez Rivera-Nuñez & Obon de Castro 1989).. Nowadays it is still used by Greeks originating from the Black Sea and the Caucasus region who distinguish two types of coarsely ground cereal grain, ‘korkoto’, i.e. unboiled cracked wheat used in soups, and ‘pligouri’, i.e. boiled cracked wheat used as rice (Gregoriadou1 The Early Bronze Age time span (second half of 4th millennium B.C. to end of 3rd millennium B.C.) represented at each site varies considerably, with Archondiko covering the end of the period.


S.M. Valamoti, A. Papanthimou and A. Pilali with lentils, fava beans or peas, days when acorns were being soaked or roasted to remove tannins, when flax or Lallemantia seeds were crushed and soaked to remove their edible oil. Archondiko, a settlement where food was being cooked in elaborate clay constructions, where different plant species and different foods may have signified different days of the year, different origins or different status, presents a challenge to future archaeobotanical work which will hopefully reveal more secrets of the Early Bronze Age cuisine of northern Greece. Acknowledgements Figure 6a: Charred fragmented cereal grain; loose fragments

We would like to thank the numerous archaeology students of the Aristotle University of Thessaloniki and in particular Olga Terzidou, Tasoula Dimoula and Fotis Ifandidis for their help in processing of the soil samples by flotation. We are also grateful to Georgia Kotzamani, Alexandra Livarda, Paschalis Zafiriadis, Irini Delidaki and Ismini Ninou who helped in sorting the flots and to Dr Anaya Sarpaki for useful comments to an earlier draft of the manuscript. References Albanakis K.S., and Syridis, G.E., unpublished report. Παλαιογεωγραφία και φυσικό περιβάλλον των προϊστορικών και ιστορικών πληθυσμών της ευρύτερης περιοχής του Θερμαϊκού κόλπου. Bayram, M., 2000, Bulgur around the world, Cereal Food World, , 80-82. Bintliff, J., 1976, The Plain of Western Macedonia and the Neolithic site of Nea Nikomedeia, Proceedings of the Prehistoric Society, , 241-262. Bourdieu, P., 1977, Outline of a Theory of Practice, Cambridge: Cambridge University Press. Braun, T., 1995, Barley cakes and emmer bread, In Food in Antiquity (J. Wilkins, D. Harvey and M. Dobson eds.), 25-37, Exeter: University of Exeter Press. Butzer, K.W., 1982, Archaeology as Human Ecology, Cambridge: Cambridge University Press. Chrysostomou, A. and Chrysostomou, P., 1999, Αρχοντικό Γιαννιτσών: Τράπεζα. Η έρευνα των ετών 1996-1997 (Archondiko Giannitson: The 1996-1997 research), ΑΕΜTH, , 179-188. Dennell, R., 1976, The economic importance of plant resources represented on archaeological sites. Journal of Archaeological Science, , 229-247. Douglas, M., 1984, (ed.) Food in the Social Order, New York: Russel Sage Foundation. French, D. H., 1971, An experiment in water-sieving, Anatolian Studies, , 59-64. Gregoriadou-Vakirtzoglou, E., 1990, Παραδοσιακή και σύγχρονη Ποντιακή Κουζίνα (Traditional and Modern Cuisine of the Pontioi), Thessaloniki: Melissa Asprovaltas. Halstead, P., 1981, Counting sheep in Neolithic and Bronze Age Greece, In Pattern of the Past. Studies in Honour of David Clarke. (I. Hodder, G. Isaac and N. Hammond, eds), 307-339, Cambridge: Cambridge University Press. Halstead, P., 1990, Waste not, Want not: Traditional Responses to Crop Failure in Greece, Rural History,   , 147-164. Halstead, P., 1992, Agriculture in the Bronze Age Aegean. Towards a model of Palatial economy, In Agriculture in Ancient Greece (B. Wells ed.), 105-117, Stockholm: Swedish Institute at Athens. Halstead, P., 1993, Banking on Livestock: Indirect Storage in

Figure 6b: Charred fragmented cereal grain; conglomerations of fragments

Figure 6c: Charred fragmented cereal grain

Conclusions The destruction layer of the post-framed houses has revealed a long list of potential cooking ingredients which reveals a wide variety primarily of cereals, but also of pulses, oil plants and nuts. We could suppose that their edible properties were recognised and that different species may have been intended for different recipes. Also that the variety in the ingredients could have been related to a variety in culinary practices is also underlined by the different types of cooking constructions unearthed at the site. We can also imagine busy days during which cereals were ground to be stored for later use, gruels being cooked with the addition of a few drops of oil (either from linseed or cf. Lallemantia), sprinkled with flax seeds or mixed 192

Cooking Ingredients from Bronze Age Archondiko: the Archaeobotanical Evidence Papaefthymiou-Papanthimou, A., Pilali-Papasteriou, A., Basogianni, D., Papadopoulou, E., Tsagaraki, E. and Fappas, I., 2002, Αρχοντικό 2000. Τυπολογική παρουσίαση και ερμηνευτικά προβλήματα των πηλόκτιστων κατασκευών (Archondiko 2000. Typological presentation and interpretive problems concerning the clay structures), AEMTH, , 421433. Papaefthymiou-Papanthimou, A. and Pilali-Papasteriou, A., 1995, Ανασκαφή στο Αρχοντικό Γιαννιτσών (Excavations at Archondiko Giannitson), ΑΕΜTH, , 151-156. Papaefthymiou-Papanthimou, A. and Pilali-Papasteriou, A., 1997, Οι προϊστορικοί οικισμοί στο Μάνδαλο και στο Αρχοντικό Πέλλας, ΑΕΜTH, 10α, 143-158. Papaefthymiou-Papanthimou, Α. and Pilali-Papasteriou, Α., 1999, Ανασκαφή στον προϊστορικό οικισμό του Αρχοντικού κατά το 1997 (Excavations at Archondiko during 1997), ΑΕΜTH, , 165-177. Papaefthymiou-Papanthimou, A., Pilali-Papasteriou, A., Basogianni, D., Papadopoulou, E., Tsagaraki, E. and Fappas, I., 2002, Αρχοντικό 2000. Τυπολογική παρουσίαση και ερμηνευτικά προβλήματα των πηλόκτιστων κατασκευών (Archondiko 2000. Typological presentation and interpretive problems concerning the clay structures, AEMTH, , 421433. Palmer, C., 2002, Milk and Cereals: Identifying Food and Food Identity among Fallahin and Bedouin in Jordan, Levant, , 173-195. Parker-Pearson, M., 2003, Food, identity and culture: an introduction and overview, In Food, Culture and Identity in the Neolithic and Early Bronze Age (Parker-Pearson M. ed.), British Archaeological Reports, 1117, 1-30. Renfrew, C., 1972, The Emergence of Civilisation: The Cyclades and the Aegean in the Third Millenium BC, London: Methuen. Renfrew J. M., 1973, Palaeoethnobotany. The Prehistoric Food Plants of the Near East and Europe, New York: Columbia University Press. Renfrew, J. 2003, Grains, seeds, and fruits from prehistoric Sitagroi, In Prehistoric Sitagroi: Excavations in Northeast Greece, 1968-1970 (E.S. Elster and C. Renfrew eds.), Volume 2: The Final Report, 1-29. (Monumenta Archaeologica 20), Los Angeles: Cotsen Institute of Archaeology. Rivera-Nuñez, D. and Obon de Castro, C., 1989, La dieta cereal prehistorica y su supevivencia en el area Mediterranea, Trabajos de Prehistoria, , 247-254. Sarpaki, A., 2001, Processed cereals and pulses from the Late Bronze Age site of Akrotiri, Thera; preparations prior to consumption: a preliminary approach to their study, Annual of the British School in Athens, , 27-40. Sarpaki, A., 1992, The palaeoethnobotanical approach: the Mediterranean triad or is it a quartet? In Agriculture in Ancient Greece (B. Wells ed.), 62-76, Stockholm: Swedish Institute at Athens. Sherratt, A., 1987, Cups That Cheered, In Bell Beakers of the Western Mediterranean (W. H. Waldren & R.C. Kennard eds.), British Archaeological Reports, S331, 81-114. Sherratt, A., 1991, Palaeoethnobotany: from Crops to Cuisine, Paleoecologia & Arquelogia, , 221-236. Sherratt, A., 1995, Sacred and Profane Substances: the ritual use of narcotics in later neolithic Europe, In Sacred and Profane: Proceedings of a Conference on Archaeology, Ritual and Religion (P. Garwood, D. Jennings, R. Skeates and J. Toms eds.), 50-64, Oxford: University Committee for Archaeology. Triantafyllou, S., 1999, A bioarchaeological approach to prehistoric cemetery populations from western and central Greek Macedonia, British Archaeological reports, S976. Urem-Kotsos, D., Kotsakis, K., and Stern, B., 2002, Definining function in Neolithic ceramics: the example of Makriyalos, Greece, Documenta Praehistorica, , 109-118.

Greek Agriculture, Bulletin on Sumerian Agriculture, , 6275. Halstead, P., 1994, The north-south divide: Regional paths to complexity in prehistoric Greece, In Development and Decline in the Mediterranean Bronze Age (C. Mathers and S. Stoddart eds.), 195-219, (Sheffield Archaeological Monographs 8). Sheffield: J.R. Collis Publications. Halstead, P. and Jones, G., 1989, Agrarian ecology in the Greek islands: time stress, scale and risk, Journal of Hellenic Studies, , 41-55. Halstead, P. and Barrett, J. (eds.) (in press). Food, Cuisine and Society in Prehistoric Greece. (Sheffield Studies in Aegean Archaeology 5), Oxford: Oxbow. Hamilakis, Y., 1996, Wine, oil and the dialectics of power in Bronze Age Crete: a review of the evidence, Oxford Journal of Archaeology, , 1-32. Hamilakis, Y., 1999, Food technologies/technologies of the body: the social context of wine and oil production and consumption in Bronze Age Crete, World Archaeology, , 38-54. Hamilakis, Υ., 2000, The anthropology of food and drink consumption and Aegean archaeology, In Palaeodiet in the Aegean (S.J. Vaughn and W.D.E.Coulson eds.), 55-63, Oxford: Oxbow Books. Hastorf, C., 1991, Gender, Space and Food in Prehistory, In Engendering Archaeology (J.M. Gero and M.W.Conkey eds.), 132-159, Oxford: Blackwell. Jones, G., 1998, Distinguishing food from fodder in the archaeobotanical record, In Fodder: Archaeological, Historical and Ethnographic Studies (M. Charles, P. Halstead and G. Jones eds.), Environmental Archaeology, , 95-98. Jones, G. and Halstead, P., 1995, Maslins, mixtures and monocrops: on the Interpretation of archaeobotanical crop samples of heterogeneous composition, Journal of Archaeological Science, ,103-114. Jones, G. and Valamoti, S.M., 2005, Lallemantia, an imported or introduced oil plant in Bronze Age northern Greece, Vegetation History and Archaeobotany,   , 571-577. Kiziridou, Th., 2002, Ποντίων Εδέσματα (Delicacies of Pontioi), Thessaloniki: Kyriakidis Publications. Kohler-Schneider, M., 2001, Verkohlte Kultur- und Wildpflanzenreste aus Stillfried an der March. Wien, Verlag der Österreichischen Akademie der Wissenschaften. Kroll, H. J., 1983, Die Pflanzenfunde. In Kastanas. Ausgrabungen in einem Siedlungshugel der Bronze und Eisenzeit Makedoniens 1975-1979 (B. Hänsel ed.), Berlin: Spiess. Kroll, H.J., 1991, Suedosteuropa, In Progress in Old World Palaeoethnobotany (W. van Zeist, K. Wasylikowa and K.E. Behre eds.), 161-177, Rotterdam: Balkema. Lannoy, S., Marinval, Ph., Buléton, A., Chiron, H., Méjanelle, Ph., Rech, J., and Tchapla, A., 2003, Etude de “pains/galettes” archéologiques français, In Bread, Ovens and Hearths of the past (K Fechner and M. Mesnil eds.), Civilisations, , 119160. Levi-Stauss, Cl., 1962, Totemisme: Le Totemisme Aujourd’hui, Paris. Mason, S., 1995, Acornutopia? Determining the role of acorns in past human subsistence, In Food in Antiquity (J. Wilkins, D. Harvey and M. Dobson eds.), 12-23, Exeter: University of Exeter Press. Megas, G.A., 1988, Ελληνικές Γιορτές και Έθιμα της Λαϊκής Λατρείας (Greek Festivals and Customs of Folk Worship), Athens: Odysseas Publications (1992 reprint). Micha-Lampaki, A., 1984, Η διατροφή των Αρχαίων Ελλήνων κατά τους Αρχαίους Κωμωδιογράφους (The Diet of Ancient Greeks According to Ancient Comic Poets), PhD Thesis, University of Athens, Athens. Papaefthymiou-Papanthimou, A. and Pilali-Papasteriou, A., 1995, Ανασκαφή στο Αρχοντικό Γιαννιτσών (Excavations at Archondiko Giannitson), ΑΕΜTH, , 151-156.


S.M. Valamoti, A. Papanthimou and A. Pilali Valamoti, S. M., 2001, Archaeobotanical investigation of Late Neolithic and Early Bronze Age agriculture and plant exploitation in northern Greece, PhD Thesis, University of Sheffield, Sheffild. Valamoti, S.M., 2002, Food remains from Bronze Age Archondiko and Mesimeriani Toumba in northern Greece?, Vegetation History and Archaeobotany, , 17-22. Valamoti, S. M., 2003, Neolithic and early Bronze Age ‘food’ from northern Greece: the archaeobotanical evidence, In Food, Culture and Identity in the Neolithic and Early Bronze

Age (M. Parker-Pearson ed.), British Archaeological reports, 1117, 97-112. Valamoti, S.M. and Jones, G., 2003. Plant diversity and storage at Mandalo, Macedonia, Greece: archaeobotanical evidence from the Final Neolithic and Early Bronze Age, Annual of the British School at Athens, , 1-35. Vaughan, S.J. and Coulson, W.D.E., 2000, Palaeodiet in the Aegean, Oxford: Oxbow Books. Voutsina, E., 2000, Το Ψωμί (Bread), Athens: Kastaniotis Publications.


SEASONALITY AND RADIOLOGY: A PILOT APPLICATION ON RED DEER (CERVUS ELAPHUS) DENTARIES FROM THE UPPER PALAEOLITHIC CAVE OF KASTRITSA, NW GREECE E. Kotjabopoulou IB’ Ephorate of Prehistoric and Classical Antiquities, Plateia 25th March 6, Ioannina 45 221, Greece, [email protected]

C. N. Kaftantzis Serres General Hospital, Ayias Sofias 3, Serres 62100, Greece Abstract: Seasonality studies are especially crucial in interpreting mobile hunter-gatherer settlement patterns and their correspondence to the socio-economic regional landscapes. Under the term seasonality modern research encompasses at least two concepts. One refers to the number of season(s) over which humans made use of a particular site/locus (i.e. duration of occupation). The other considers whether prey was seasonally targeted. However, from an analytical point of view tackling the issue of age/season at death of ungulate species, the most common dietary staple of human foraging communities, has not proven an easy task. In this paper we present the application of a relatively recent, but still not widely known, non-destructive method, which uses radiological data and criteria for estimating age/season at death. The sample analysed comprises partial semi-mandibles of red deer (Cervus elaphus) from the upper palaeolithic cave of Kastritsa, Epirus, where this herbivore was the most abundant species in the faunal assemblage. To date, this is the first such study made on archaeological material from the Greek territory. The particular radiological technique applicable to osseous structures employed by the authors is firstly delineated. Subsequently, the results obtained are evaluated against other analytical means (e.g. eruption-wear-replacement sequence) by which age at death was also estimated for these same specimens (Kotjabopoulou 2001). This comparison clearly demonstrates the higher resolution achievable via the radiological data. In this light, the significance and the potential of this method in rigorously probing into the mode of occupation at Kastritsa, a wetland intermountain site situated close to the high Pindus, are addressed. Finally, a discussion is undertaken about the Late Glacial landscape and the mobility patterns practiced in the rugged and dissected topography of NW Greece. Περιληψη: Ο προσδιορισμός αφενός της εποχής – στο πλαίσιο του ενιαύσιου κύκλου- χρήσης μιας αρχαιολογικής θέσης και αφετέρου του εάν η θηρευτική διαδικασία είχε χαρακτήρα εποχικό αποτελούν δύο κρίσιμες συνιστώσες για τη διατύπωση ερμηνευτικών προτάσεων που αφορούν στα οικιστικά δίκτυα και το κοινωνικο-οικονομικό υπόβαθρο της Παλαιολιθικής. Στην εργασία αυτή παρουσιάζεται η εφαρμογή μίας μη καταστρεπτικής μεθόδου (η οποία χρησιμοποιεί ακτινολογικά δεδομένα και κριτήρια για την εκτίμηση της ηλικίας/εποχής θανάτου οπληφόρων) σε κάτω γνάθους από ερυθρά ελάφια που προέρχονται από το ανώτερο παλαιολιθικό σπήλαιο της Καστρίτσας στην Ήπειρο. Τα αποτελέσματα αξιολογούνται σε σύγκριση με τα πορίσματα συμβατικών τρόπων εκτίμησης της «εποχικότητας». Ιδιαίτερη μνεία γίνεται για τη σημασία και τις δυνατότητες που η συγκεκριμένη μέθοδος συνεπάγεται για την ερμηνεία του χαρακτήρα της κατοίκησης στην Καστρίτσα, μία παραλίμνια θέση στην ορεινή ενδοχώρα της Πίνδου. Τέλος, επιχειρείται μία συνοπτική συζήτηση για τα δίκτυα μετακίνησης των παλαιολιθικών ομάδων στο ανώμαλο και κατακερματισμένο γεωγραφικό ανάγλυφο της ΒΔ Ελλάδας κατά τις ύστερες χιλιετίες του Πλειστοκαίνου.


as modern palaeoanthropological research starts to reconsider, historical trajectories are fundamental shaping forces of regional subsistence objectives and organization and settlement pattern variability.

Under the impact of ethnographically documented resource- and land-use patterns the “bi-seasonal configured annual-round cycle” has been for a long time the favored interpretative framework for Pleistocene mobile foraging systems (e.g. Audouze 1987, Bolomey 1973, Gordon 1988, Higgs et al. 1967, Straus 1977, Sturdy 1975, White 1971). Increasingly, though, regional palaeolithic studies have raised caution against the simplicity of the concept and, in a number of cases, have already demonstrated the entailed neutralizing effects of such direct analogical model-building (e.g. Jochim 1991, Keeley 1991, Miracle & O’ Brien 1998). Undoubtedly, confronting environmental asymmetry – diachronically and/or spatially- has been a critical selective pressure on hominid adaptation and radiation from an evolutionary perspective. However, context specific ecological and,

Seasonality studies – i.e. methodologies applied to zooarchaeological material for determining number of season(s) over which humans made use of a site/locus, alias duration of occupation, and whether prey was seasonally targeted - are proving especially crucial in providing insights into camp selection and habitation mode and in interpreting multi-layered socio-economic regional landscapes. Moving, however, from season of site-use (or resource-use) to regional mobility networks and land-use patterns is not only a demanding task, requiring a diverse suit of on- and off-site data, but also one that depends on the standpoint adopted as regards the links between physical attributes of space, including range, availability, 195

E. Kotjabopoulou and C.N. Kaftantzis quality of resources, and the social landscape (cf. Gosden & Head 1994, Ingold 1993, 1996).

Asprochaliko rockshelter (c. 200m a.s.l.), further to the south, roughly at the mid point of the Louros river valley, was assigned the status of the wintering territory.

From an analytical point of view, Brown & Chapman (1991a,b) have recently elaborated, in the context of wildlife management studies, a non-destructive method, which uses radiological data and criteria, for estimating age and by implication season at death for two species of cervids, namely fallow and red deer. Carter (1997) has expanded the method and showed that it can also work on roe deer mandibular dentition, without alterations.

External environmental data (e.g. geological, geomorphological, palynological), often conflated with present day prevailing conditions (e.g. wind direction), formed by and large the basis for assessing the economic potential, including seasonal availability of resources, of what was designated the “exploitation territory”, an arbitrary 10kms circle/contour around the site-location (ibid). The central tenet of the model, i.e. the complementarity of inland (cold) - coastal (warm) territories, was mainly built around the migratory behaviour of red deer, the presumed most abundant prey in the collections. For Kastritsa, this has been recently confirmed by a comprehensive faunal analysis, which, moreover, sheds light on various diachronic subtleties as regards taphonomic variables, range and emphasis of resource exploitation, procurement and consumption strategies exerted on animal resources and discard practices along the c. 10,000 years that the site formed part of a wider socio-economic landscape (Kotjabopoulou 2001, 2003). For Asprochaliko, however, no dominance of red deer can be postulated for the Upper Palaeolithic deposits (Bailey et al. 1983a, Kotjabopoulou 2001); and this is besides the inadequate evidence to demonstrate, at a finer level of resolution, contemporaneity of use with Kastritsa (e.g. Adam 1989). In the initial model, a local ethnohistoric analogy with transhumant small-ruminant pastoralists (cf. Campbell 1964) added “verification”, albeit of a uniformitarian nature. Also, site types and settlement distribution models documented in modern African hunter-gatherers were extrapolated wholesome onto the prehistoric past. In short, the urge for capturing the “big” picture, the need to develop the new, at the time, and insightful off-site methodologies, the purely ecological orientation of the research design and not least, contingency, had had the counter effect of leaving unexplored (or unfinished) what potential evidence, including seasonality, could be extracted from the “bare bones” themselves.

In this paper we present a pilot application of this analytical route on red deer (Cervus elaphus) material from the Upper Palaeolithic cave of Kastritsa, Epirus, NW Greece. This site’s archaeological record is for the first time used for probing into the issue of seasonality, which was hitherto addressed exclusively via reconstructions of (palaeo)landscape dynamics. Our first aim, in this study, is to evaluate the resolutionpotential of this analytical avenue with respect to highly fragmented collections, as is the case of Kastritsa (see Kotjabopoulou 2001). Secondly, we show that wellestablished medical radiological techniques for osseous structures can be used for obtaining radiographs of archaeological specimens of appropriate quality and resolution to suit the specific method at issue. Furthermore, and in light of the results obtained, we discuss certain options raised as regards the regional rhythm of huntergatherer mobility in the ragged and fragmented physical space of Epirus during Upper Palaeolithic times. In this research, besides functional, economic and economical inferences that can be drawn from the available record, we posit an equal emphasis on delineating a framework for tracing the way(s) foragers engaged with the prehistoric mountainscape(s) (cf. Kotjabopoulou 2001) in this peninsula of the Northern Mediterranean belt. Palaeogeographic models Pioneering research in the 1960’s, inspired and led by E. Higgs, in the region of Epirus, NW Greece, provides an early, come to be classic, attempt of an integrative approach, which combined archaeological survey and excavation with multifaceted palaeoenvironmental investigations. In the case of Epirus, which served as the rehearsal for what was later crystallized into the “site-catchment analysis” and the “palaeoeconomy approach” (e.g. Higgs & VitaFinzi 1972), the scope was to document bi-annually configured regional mobility patterns, i.e. complementary winter (lowland/coastal) and summer (upland/hinterland) exploitation territories (Higgs et al. 1967, Higgs & Webley 1971). In the initial model, Kastritsa cave, a lake-shore site - presently at an elevation of c. 460 m a.s.l. - situated in the largest inland basin below the high and precipitous mountain peaks of the Pindus massive, (Figure 1) provided the summer-end “home-base” of the functionalist equation.

Subsequent, productive questioning and modification of the seasonality model has debunked Kastritsa and Asprochaliko from their earlier “properties” as homologous home base sites (Bailey et al. 1983a, b, Bailey & Gamble 1990). Moreover, a particular fieldwork strategy has concentrated on refining the external, palaeogeographic reconstruction in order to assess foraging potential and opportunities with respect to quantified densities of ungulate communities (Sturdy & Webley 1988, Sturdy et al. 1997). The revised model grades landscape units according to a variety of edaphic (e.g. water retentiveness, nutrient content, vegetation cover) and palaeoecological (e.g. geology, tectonics, snow and coastline positions) factors (see ibid. for details). Aspects of species-specific ethology, i.e. tolerance, preference or avoidance of landscape attributes, constitute a key variable in the attempt to place animals on 196

Seasonality and Radiology: a Pilot Application on Red Deer (Cervus Elaphus) Dentaries

Figure 1: Map of Northwest Greece showing inferred principal routes of seasonal migration of large herbivores. Triangles denote Upper Palaeolithic excavated rockshelter sites (after Sturdy et al. 1997, with additions). the ‘prehistoric maps’. The territory comprising north-west Greece, including the present island of Corfu and the now submerged continental shelf between that and the present coastline, is thus partitioned, following a NNW-SSE axis, into a three-belt model: a western coastal winter-attractive lowland, a middle flysch “bad” (i.e. of low productivity) country and an eastern composite summer productivity area (Figure 1). The cave of Kastritsa falls within the latter band, and its location is interpreted as being adjacent but offset from major ungulate migration routes. Close-up carrying capacity reconstructions are also offered for the areas in the vicinity of excavated Palaeolithic sites, but this time these are more freely delineated, for instance on account of natural features/barriers. For Kastritsa’s exploitation territory, the model predicts summer as the optimum productivity period of the annual cycle, but

spring/summer and autumn/winter cannot be excluded from offering viable opportunities of exploitation, always on the basis of off-site factors (Sturdy et al. 1997: Tab. 3.10). Interestingly, Higgs et al. (1967: 14) were keen in noting these same alternatives, a point often neglected by later critical reviews. The model attempts a gross two-tier temporal resolution (at c. 20/16 kyr BP and c. 13 kyr BP respectively) within the frame of “Late Upper Palaeolithic Times” (Sturdy et al. 1997). In relation to the Kastritsa setting, however, the discussion of possible temporal variability is confined to the puzzling theme of the lake’s extent in relation to the waxing and waning effects of climatic conditions, as these were registered in the pollen sequences from lake Pamvotis available at the time (e.g. Bottema 1974, Tzedakis 1993). In all, while seasonal mobility remains the central component of the 197

E. Kotjabopoulou and C.N. Kaftantzis revised palaeolandscape dynamics, a more complex model of subsistence strategies is advocated. These involve selectivity between densities of available resources and those actually exploited and micro-regional diversity in exploitation goals. This re-assessment was influenced by the caprine-rich faunal records recovered from three Late Glacial hinterland rockshelters, namely Klithi, Megalakkos and Boila, in the Voidomatis river catchment (Bailey 1997, Gamble 1997, Kotjabopoulou et al. 1997, 1999, Sinclair 1997), and echoes recent anthropological research in the region which challenges the reduction of cultural traits to environmental limitations (e.g. Green 1997, see also Nitsiakos & Kassimis 2000).

prime adult cohort is evident, but both juveniles and old adults are not uncommon. The above features of the red deer component at Kastritsa underscore the restrictions and difficulties imposed on tackling seasonality. By using tooth eruption data on isolated immature unworn premolars of very young red deer animals, usually the most reliable seasonality age sub-set, it has been shown that groups of hunter-gatherers made use of the Ioannina basin, sensu lato, in late spring/early summer (Kotjabopoulou 2001). The question, however, remained, whether this pattern could be extrapolated to other age-segments of the procured red deer population. Given that such assumptions are not any more warranted, other methods for pinpointing season at death had to be explored. Crown height measurements of red deer permanent molars as well as of the P4 have been shown to be inconclusive for the question at issue (Klein et al. 1983, Legge & Rowley-Conwy 1988). Incremental structure or dental annuli analysis, while it constitutes a potentially powerful, but destructive, speciesspecific seasonality tool (e.g. Burke 2000, Lieberman & Meadow 1992, Pike-Tay 1991, 2000), should be sought only after other options are firstly examined. This applies especially to a site like Kastritsa, where fragmentation and post-mortem dislocation of teeth from the encasing jawbone is high, thus adding to the problems of compiling a suitable and meaningful sample, at least until the entire collection is sampled for such an analysis. Meanwhile, then, the application of the radiological technique worked out by Brown & Chapman (1991b) was judged to be worth pursuing as its efficacy with archaeological material had already been tested with encouraging results at the famous Mesolithic site of Star Carr (Carter 1998), where, though, in contrast to Kastritsa, preservation of organic remains was excellent.

In spite of theoretical and/or methodological reservations that can be raised (for discussion see Kotjabopoulou 2001), the above briefly sketched palaeogeographic model provides an invaluable framework for furthering research on the operational aspect of land use patterns through onsite seasonality indicators. Materials and method Hominid agency is the principal signature of accumulation and modification of faunal remains at Kastritsa. However, and largely owing to this factor, the organic material is extremely fragmented. Out of a sample of c. 62,000 remains, which represents approximately half of the material that can be now referred to a minimum contextual (excavation) provenance, only a mere 1,1% survived intact. Concomitantly, identifiability is low, e.g. 7,5 % at the site-assemblage level (Kotjabopoulou 2001). In the identified fraction, non-fragmented skeletal elements comprise 14,5%. Carnivore remains are extremely low key (2,2%), while ungulates form the bulk of the collection (c. 85%). In the latter sample, species relative frequency, using Number of Identified Specimens (NISP) as an analytical currency, reveals a clear red deer dominance (c. 76%). A 3,4 bone:tooth ratio characterizes the red deer component. Body-part relative abundance, as measured by CNISP (Corrected NISP) – a method designed to compensate for differences in intra-skeletal element frequency (see Gamble 1997, Kotjabopoulou 2001), shows a clear bias against red deer head transportation to the site. For the stratigraphic packages, which have yielded faunal assemblages relatively unaffected from major postdepositional taphonomic attrition, namely strata 5, 3 and 1, carcass unit parcels are statistically positively correlated with food utility ranking. In other words, a pattern of hoof and head abandonment at kill-locations is depicted (Kotjabopoulou 2003). In terms of preservation, loose teeth dominate the dental red deer component (74%). At the site level, a comprehensive analysis of dental mortality, based on eruption-wear-replacement sequences of lacteal and permanent elements, demonstrates that red deer culling at Kastritsa falls in the range of a living structure model (sensu Stiner 1990, 1994). Also, a slight emphasis on the

The strong asset of the radiological method is that it concentrates on tooth development, i.e. from the formation of the crypt inside the jawbone to dental maturity. In modern populations of red deer the latter is achieved by the age of 40 months, a benchmark, which constitutes the method’s upper limit of application. In the original study the development of lacteal premolars was already completed in the jaws of the control sample. In effect, then, the method is applicable on the permanent molariform dentition (premolars and molars) and particularly useful for sub-adult animals. A scheme of ten distinct developmental stages based on radiological data and criteria has been devised (Brown & Chapman 1991b: 90). Each tooth is assigned a scoring value pending to which stage it corresponds. The rationale of the model is founded on the sequential nature of the developmental stages for each tooth, which are, in turn, considered to be relatively immune from nutritional and environmental factors. However, the time-increment, i.e. the time-span represented by each stage for any particular tooth, is variable. Thus, the more teeth present on a jaw198

Seasonality and Radiology: a Pilot Application on Red Deer (Cervus Elaphus) Dentaries mandibular fragments also have the deciduous premolar four in place and fully formed. The results readily demonstrate that MAND2 represents an animal older than the one to which MAND3 belonged. Using May/June as a conventional peak birthing timing for European red deer (cf. Clutton-Brock et al. 1982) the interpretation of age/season at death is discussed below.

Dental Unit

Scores of MAND2

Scores of MAND3

Lower Premolar 4



Lower Molar 1



Lower Molar 2



Table 1: Scores of permanent dental elements of red deer mandibles MAND2 and MAND3 from Kastritsa cave using radiological data and criteria.

For MAND2, if the scores are taken at face value, the combination of developmental stage 6 (i.e. root less than half formed) for P4 and stage 9 (i.e. root fully formed, but with apex open) for M1 suggests an animal aged 18 months at the time of death. This can be translated to a November/December death, i.e. a late autumn kill. It is of importance to point out that in the modern control sample stage 6 of P4 is consistently correlated with either stage 9 or 10 of M1, and that these two combinations were always observed on animals aged 18 months. Still, and given the fragmentation bias, we should not fail to note – in adopting a rather conservative approach to the method- that the range of possible procurement could theoretically fall between 13 and 19 months, i.e. June/ July and December/January (cf. Brown & Chapman 1991b: Tab. IV).

For MAND3, the particular order of developmental stages for the three permanent teeth present on this mandible cannot be found in the published scheme of modern known-aged individuals. Animals of certain ages were missing from the original comparative collection. Carter (1998: Tab.1) has attempted to provide scoring estimates for the missing age/timesegments. If the Kastritsa MAND3 tooth row is set against Carter’s likely score-assignment, then the widest age-range interpretation for MAND3 falls between >6 months and 50 % = unacceptable agreement

Magnesium. Magnesium shows a poor paleonutritional significance for all samples. Although the average amount decreases from herbivores to carnivores as expected (see e.g. Antoine et al. 1988; Subira & Malgosa 1992) and increases with consumption of marine products (see e.g. Magou et al., 1990) the difference is not sufficient to provide a good separation among different samples. The concentration of Mg for the investigated samples is 640 and 1560 mg/kg.

Table 3. Analytical data obtained for IAEA-SRM-H-5 [mg/kg].

Reconstruction Of Paleodiet On the basis of analytical data, bivariate plots and results from Cluster and Factor analysis, a hypothesis for the paleodiet of Thracian people may be proposed. The interpretation of the bivariate plots: [Ba]:[Ca]/[Sr]:[Ca], Cu/Zn, Al/Ba, lg ([Ba]:[Sr])/lg ([Sr]:[Ca]) and lg ([Ba]:[Sr])/lg ([Sr]:[Ca] were discussed in detail in Zlateva et al. (2003). As an example the bivariate plots lg ([Ba]:[Sr])/lg ([Sr]:[Ca]) are presented (Fig. 4).

Figure 2. Bivariate plot Ba versus Al for archaeological bones from Apollonia Pontica

According to the data of Burton & Price (1990) the ratio lg (Ba/Sr) can be used for the differentiation between consumption of food from terrestrial and marine origins. For most of the investigated individuals from Apollonia Pontica the ratio lg (Ba/Sr) is lower than –1, which is an indication of predominantly fish and mollusc consumption. This result, in combination with the established indicators (correlation of Ba to Al, Zn, Cu and Mg), confirms the elevated consumption of marine food and indicates the low consumption of vegetable food. It is very important to mention that there is some heterogeneity of the diet among the investigated individuals from Apollonia Pontica. For a

Figure 3. Bivariate plot Cu versus Zn


B. Zlateva, I. Kuleff, R. Djingova, M. Kuzmanov and D. Gergova significant, smaller part of the community the ratio lg (Ba/ Sr) indicates that these individuals have not consumed any food of a marine origin. In fact, the population of Apollonia Pontica during the 4th - 3rd century is mixed - Thracians and ancient Greek, and the proposed hypothesis for daily diet of Greek people at the same time, is confirmed by results obtained from Magou et al.(1990) that the predominant food in Greece (along with marine) was meat. For the rest of the individuals the ratio lg (Ba/Sr) indicate that they have consumed terrestrial food. Figure 4. Bivariate plot Log (Ba/Sr) versus Log (Sr/Ca)

Using the ratio log ([Sr]:[Ca]) for the investigated bones from Malak Porovets it appears that people have consumed vegetarian food. The relatively high Ba concentrations are an indication for the presence of nuts (hazelnuts, walnuts), cereals and vegetables in the diet, although the concentrations for Sr are not very high.

for all individuals, excluding sucklings and babies, was determined. Using cluster analysis only on bone samples from Malak Porovets additional information has been obtained. Concentrations of elements Ba, Cu, Mg, Sr and Zn were used as variables and results are presented in Fig. 6. Six clusters and 2 outliers (sample 042.MAP-women, about 40 years old and 041.MAP – child, 7-7.5 years, see Table 1) are established. The results show that all bones of babies and suckling infants from both periods form one cluster - clearly children have consumed milk. A posteriori this provides good evidence that diagenesis has been minimal. It is very interesting that female bone samples from both periods are mixed and form 2 clusters, while male Thracians bones form an isolated cluster. On the basis of the clustering data some interesting conclusions may be drawn about the life-style of people inhabiting Thracia during the 4th - 3rd century BC. It is probable that the diet of the men from the Thracian tribe gety included more meat and animal products than the diet of the women, however the diet of women for both periods is similarly based on vegetarian food. An indirect confirmation for the influence of sex on the diet of Thracians can be found in the words of ancient authors. Athenei (VII: 7, 8) mentions that the Thracian followers of one family used food of different origin or cooked in a different way.

Investigated individuals from Gledachevo are vegetarians too, with nuts and corn dominating the diet. This assumption is confirmed by the very high values of lg ([Ba]:[Sr]), this means that the basic food is vegetables, cereals and other plants. Simultaneously the high Ba concentrations are evidence for nuts (hazelnut, walnut) and berries. The data results for the investigated individuals from Glavanak, Pchelari, Dolno Lukovo and Veliko Tarnovo show that they have consumed mixed food – meat and as well as vegetables. Multivariate statistics 1. Cluster analysis. The Cluster option of the SSPS statistical package was used, and concentrations of the elements Ba, Cu, Mg, Sr and Zn were used as variables. Distance calculation was done using Ward’s method, and clustering by Joining method (Tree Clustering) was chosen, the results are presented in Fig. 5.

2. Factor analysis - Malak Porovets.

As a result, compact groups are formed, mostly depending on the investigated region, i.e. type of diet of the individuals. For this reason, the second method for clustering was Kmeans clustering. Analysis of variance indicates Ba and Sr to be the best indicators of paleodiet for the investigated individuals from both periods (Hellenistic period and the end of the 18th c. AD), the level of significance of these elements is about 41-44%. Secondly is Zn, with levels of significance of 34% and thirdly - Cu (P=27%). In this case Mg is an inefficient discriminative element (see Table 4).

Using the same statistical program, after a Z-transformation three factors were extracted from the correlation matrix with varimax rotation. These factors are presented in decreasing importance, measured in terms of the amount of variance explained individually by each one of them (eigenvalue). Only factors with eigenvalue >1 were considered. The variables included in this analysis were the concentrations of Ba, Ca, Cu, Mg, Sr and Zn. In Table 5 the eigenvalues for each factor and the percentage of variance explained by each one of them are presented.

As mentioned above, there were possibilities that the analyzed bone samples belong to individuals from Malak Porovets from different periods – the Hellenistic period and people from the end of the 18th c AD. On the basis of the results obtained using bivariate plots, a vegetarian diet

These factors explained 80.44 % of the variance. The rotated factor matrix (Table 6) indicate that Ba and Sr show the highest correlation with first factor; Mg and Zn with second factor and Cu with third. 220

Elemental Analysis of Bones and Diet Reconstruction of the Inhabitants of Thracia in the Hellenistic Period

Elements Елементи

Between clusters

Inside clusters


6.08 31.1 0.63 34.6 14.9

0.91 7.75 1.19 5.61 2.91

Cu Mg Sr Zn

Level of significant P (%) 44.60 26.82 3.57 40.85 33.77

Table 4. Analysis of variation between and inside of clusters

Factor 1 2 3

Eigenvalue 2.02 1.25 0.96

Variation, % 40.48 23.95 16.01

Accumulated, % 40.48 64.43 80.44

Figure 5. Results from Tree Clustering

Table 5. Percentage of variance explained by each factor and accumulated Variables

Ba Cu Mg Sr Zn Ca

Factor 1 0.82*

Factor 2 -0.09

Factor 3 0.004
















Table 6. Rotated factorial matrix with the weights of each variable in the three principal factors Factor 1

2 3

Eigenvalue 3.07 1.61

Variation, % 38.43 20.21

Accumulated, % 38.43 58.64




Figure 6. Results from Tree Clustering for samples from Malak Porovets

give information about vegetarian food and nuts in the daily menu. The closed distance between Ba, Sr and Ca is connected by the structure of bones and for the possibilities for ionic changes during fossilization. The high correlation of Mg and Zn in the second factor seems due to consumption of meat and animal proteins, while the third factor can be explain by Cu – “milk” factor.

Table 7. Percentage of variance explained by each factor and accumulated variance Variables

Al Ba Cu Mg Sr Zn

Factor 1 -0.92*

Factor 2 0.12

Factor 3 -0.05
















The results from Factor analysis of bones from Malak Porovets confirm the basic hypothesis of paleodiet nutrition of the community made on the basis of the bivariate plots. - Apollonia Pontica.

Table 8. Rotated factorial matrix with the weights of each variable in the three principal factors

The correlation with the factors is presented in Fig. 7 where the normalized eigenvectors have been multiplied by the square roots of the associated eigenvalues.

In this case classification was performed using concentration of Ba, Cu, Mg, Sr, Zn and Al. (Obviously the concentration of Ca in bones is a structural element). From the correlation matrix only factors with eigenvalue > 1 were considered. In Table 7 the eigenvalues for each factor and the percentage of variance explained by each one of them are presented.

The high value of Ba and Sr in the first factor indicates that these elements behave in a similar manner and

These factors explained 77.10 % of the variance. The rotated factor matrix (Table 8) indicate that Ba and Al


B. Zlateva, I. Kuleff, R. Djingova, M. Kuzmanov and D. Gergova show the highest negative correlation with the first factor; Cu and Zn with the second factor and positive correlation of Sr and negative of Mg with the third. The correlation with the factors is presented in Fig. 8 where the normalized eigenvectors have been multiplied again by the square roots of the associated eigenvalues. The high values of Ba and Al in the first factor confirm the hypothesis for the dietary origin of Al in the archaeological samples from Apollonia Pontica and may indicate that these elements give information about a fish diet. For modern bones there is evidence that concentration of Al increases when individuals have consumed more corn and food from a marine origin (see e.g. Greene 1997). The obtained results support the possibility of using the concentration of Al as paleodietary indicator for a fish diet when the preservation of bone is very good. In the remaining cases the concentration of Al could be used as indicator for diagenesis of the bone. The high correlation of Cu and Zn in the second factor together with other published data (see e.g. Rheingold 1983; Subira & Malgosa 1992) is evidence that the content of Cu and Zn in bone is connected to consumption of animal proteins.

Figure 7. Three-dimensional principal component analysis of six elements in the population of Malak Porovets

The covariation of Mg and Sr cannot be explained exactly. The reason is of different signs of correlation coefficients. The foods rich in Mg are vegetable, as well as some animal bi-products (Greene 1997), but assimilation of Mg in the human body is very different from assimilation of alkaline earth metals – closer to Zn than others (Silva and Williams 2001). This correlation seems due to consumption of mixed food - meat and vegetables. Conclusion Although the number of investigated individuals may be considered to be rather limited (53 skeletons) they were excavated in settlements situated in different regions of Bulgaria, thus the results may be considered to be representative of the diet of the Thracian population in the Hellenistic period. The interpretation of the analytical data indicates that the population consumed predominantly vegetarian food, and along the sea coast marine food as well. During the same time, the predominant food in Greece (along with marine food) was meat. Whether this difference in the dietary habits is due to religious and ethical reasons or to the social status of the population is a question still to be answered.

Figure 8. Three-dimensional principal component analysis of six elements in the population of Apollonia Pontica

of the population consumed milk to a considerable extent may be found in Homer, who wrote about the “famous Hypemolgians, -who are galactophagoi (milkeaters)” (Homer, Hias, XXII, 1-6) and Xenophon mentioned the tribe Melinophagoi (‘’mi lief-eaters”) (Xenophon VII: 5, 12).

The conclusion made in the present paper that Thracians were vegetarians is confirmed by ancient authors. Euripides (Euripides 952-955) mentions that the followers of Orpheus used “non-breathing food” and Poisidonius writes that Moesians did not use food of an animal origin (Strabo VII: 1).

Using methods of multivariate statistics new information about sex and age related dietary habits could be done: •

Evidence for the assumption that at the least some groups

• 222

the consumption of milk and diary products is connected to very low levels of Cu in the archaeological bones the diet of men from Thracian tribe gety is different to

Elemental Analysis of Bones and Diet Reconstruction of the Inhabitants of Thracia in the Hellenistic Period that of the diet of women and included more meat and animal products.

Greene, J., 1997, Grynpas, M., Pritzker, K.P.H., Hancock, R.G.V., 1987, NAA of bulk and selected trace elements in bones using a low flux SLOWPOKE reactor, Biological Trace Element Research, 13, 333-344. Iyengar, G.V., Kollmer, W.E., Bowen, H.J.M., 1978, The Elemental Composition of Human Tissues and Body Fluids, 28 Weinheim: Verlag Chemie. Kosugi, H., Hanihara, K., Suzuzki, T., Himeno, S., Kawabe, T., Hongo, T., Morita M., 1986, Elemental composition of ancient Japanese bones, The Science of the Total Environment, 52, 93-107. Kuleff I., Djingova, R., Zlateva, B., 2000, Archaeometric investigations of bones from Thrace, Archaeologia Bulgarica, IV, 7-12. Lambert, J., Weydert-Homeyer, J., 1993, The Fundamental Relationship between Ancient Diet and Inorganic Constituents of Bone as Derived from Feeding Experiments, Archaeometry, 35, 279-294. Lambert, J., Vlasak, S.M., Thometz, A., Buikstra, J., 1982, A comparative study of the chemical analysis of ribs and femurs in Woodland populations, American Journal of Physical Anthropology, 9, 289-294. Lambert, J., Simpson, S., Buikstra, J., Hanson, D., 1983, Electron Microprobe Analysis of Elemental Distribution in Excavated Human Femurs, American Journal of Physical Anthropology, 62, 409-423. Lambert, J., Simpson, S., Szpunar, C., Buikstra, J., 1984, Ancient Human Diet from Inorganic Analysis of Bone, Acc Chem Res, 17, 298-305. Magou, H., Panagiaris, G., Manalis, S., Zafeiratos, C., 1990, Identification of chemical elements in excavated human bones of ancient cemeteries from Greece, PACT, 45, 97-110. McFarren, E., Lishka, R., Parker, J., 1970, Criterion for judging the acceptability of analytical methods, Analytical Chemistry, 42, 358-362. Nie, N.H., Hull, C.H., Jenkins, J.C., Steinbrennerk, G., Bent, C.H., 1991. Statistical package for social science. New York: McGraw Hill. Pollard, A.M., Antoine, S.E., Dresser, P.Q., Whittle, A.W.R., 1989, Methodological Study of the analysis of bone, in Proceedings of the Conference Application of scientific techniques to archaeology (eds. Budd, P., Chapman, B., Jackson, C., Janaway, R., Ottaway, B.), Bradford, Great Britain, Oxbow, Monograph 9, 363-372. Price, T., 1989, Multi-element studies of diagenesis in prehistoric bone, in The Chemistry of Prehistoric Human Bone (ed. T. Price), 126-154. Cambridge: Cambridge University Press, Price, T., Kavanagh, M., 1982, Bone composition and the reconstruction of diet: examples from the midwestern United States, Midcontinental Journal of Archaeology, 7, 61-79. Rheingold, A., Hues, S., Ohen., M., 1983, Strontium and zinc content in bones as an indication of diet, Journal of Chemical Education, 60, 233-234. Safont, S., Malgosa, A., Subira, M., Gibert, J., 1998, Can Trace Elements in Fossils Provide Information about Paleodiet?, International Journal of Osteoarchaeology, 8, 23-37. Subira, M.E., Malgosa, A., 1991, Alkaline earth metal content of human bones at the site of “S’lllot des Porros” (Iron age, Mallorca, Spain), International Journal of Anthropology, 6, 225231. Subira, M.E., Malgosa, A., 1992, Multi-element analysis for Dietary Reconstruction at Balearic Iron Age Site, International Journal of Osteoarchaeology, 2, 199-204. Zlateva, B., Kuzmanov, M., Kuleff, I., Djingova., R., Gergova, D., 2003, Multi-Element Analysis of Bones for Dietary Reconstruction of the Inhabitants of Thracia at Hellenistic Times (Bulgaria), Berliner Beiträge zur Archäometrie, 20, 117-131.

Obviously it is necessary to determine the concentration of elements in archaeological bones not only by the level of Ba and Sr but as many elements as possible. Finally, in order to more fully understand the inhabitants of Thracia during Hellenistic times, future fieldwork is essential, particularly in Bulgaria, but also in Greece to locate and sample for elemental analysis as much archaeological bone as possible. Acknowledgements This work was supported financial by the “Open Society” Foundation in Bulgaria. The authors would like to thank Mrs. Kr. Panayotova (Archaeological Museum, Sofia), Mrs. E. Penkova (National Historical Museum, Sofia), Dr. M. Tonkova (Archaeological Institute, Bulgarian Academy of Sciences, Sofia), and Mr. D. Nedev (Archaeological Museum, Sozopol) for kindly supplying the bone material for this investigation. References Antoine,S.E., Dresser, R.Q., Pollard, A.M., Whittle, A.W.R, l998, Bone chemistry and dietary reconstruction in prehistoric Britain: examples from Wessex, in Oxford: British Archaeological Report, British Series (eds. E. Slater E, J. Tite), 196, 10, 369380. Baraybar, J.P., 1999, Diet and death in a Fog Oasis Site in Central Coastal Peru: A trace element study of tomb 1 in Malanche 22, Journal of Archaeological Science, 26, 471-482. Baraybar, J.P., de la Rua, C., 1997, Reconstruction of diet with trace element of bone at the Chalcolithic site of Paco Ramos, Basque Country, Spain, Journal of Archaeological Science, 24, 355-364. Burton, J., Price, T., 1990, The ratio of Ba to Sr as a paleodietary indicator of consumption of marine resources, Journal of Archaeological Science, 17, 547-557. Burton, J., Price, T., 2000, The use abuse of trace-elements for paleodietary, in Biochemical Approaches to Paleodietary Analysis (S.H. Ambrose and M.A. Katzenberg), 159-171. Berlin: Kluwer Academic/Plenum Publishers. Edward, J., Fossey, J., Yafe, L., 1984, Analysis by neutron activation of human bone from the Hellenistic cemetery at Asine, Greece, Journal of Field Archaeology, 11, 37-46. Euripides, Ippolit. Ezzo, J., 1994a, Zinc as paleodietary indicator: Two decades of spurious logic and unscientific critique in archaeological bone chemistry analysis, American Antiquity, 59(4), 606621. Ezzo, J., 1994b, Putting the “chemistry” back into archaeological bone chemistry analysis: Modeling potential paleodietary indicators, Journal of Anthropology and Archaeology, 13, 1-34. Francalacci, P., 1989, Dietary reconstruction at Arene Candide Cave (Liguria, Italy) by means of trace element analysis, Journal of Archaeological Science, 16, 109- 124. Francalacci, P., Subira, M., Borgognini Tarli, S., Macchiareli, R., Malgosa, A., Gilbert, C., Sealy, J., Sillen, A., 1994, An investigation of barium, calcium and strontium as paleodietary indicators in the Southwestern Cape, South Africa, Journal of Archaeological Science, 21, 173-184.


IN SEARCH OF MOTHERS AND DAUGHTERS IN 3000 YEAR OLD BURIALS IN NORTHERN GREECE L. Kovatsi Laboratory of Forensic Medicine and Toxicology, School of Medicine, Aristotle University of Thessaloniki

S. Triantafyllou Wiener Lab, American School of Classical Studies in Athens

J. Eliopoulos Institute of Agrobiotechnology, The Centre for Research and Technology, Hellas

M. Besios Archaeological Service of Thessaloniki

S. Andreou Department of Archaeology, Aristotle University of Thessaloniki

A. S. Tsaftaris Archaeological Service of Thessaloniki Department of Genetics and Plant Breeding, Aristotle University of Thessaloniki

A. Dimitriadou and S. Kouidou Laboratory of Biological Chemistry, School of Medicine, Aristotle University of Thessaloniki, [email protected] Abstract: Multiple burials in Early Iron Age communities of Northern Greece are probably indicative of the initiation of social emphasis on small family groups (matrilineal lineages) eages) ages) as basic organizational units. Analysis of ancient mitochondrial DNA (mtDNA) obtained from burial remains, mainly bones and teeth, could facilitate the establishment of such linkages. We presently report our preliminary results on the analysis of aDNA from teeth obtained from a Late Bronze Age to Roman Period cemetery in Makrigialos, Southern Macedonia, Greece (ancient Pydna), to investigate this archaeological interpretation. DNA was extracted from teeth using organic solvents. Two segments egments of the control region of the mtDNA corresponding to the Hypervariable Regions I and II were amplified by polymerase chain reaction (PCR) in two overlapping fragments and sequenced, using all the suggested precautions to avoid contamination. In parallel, as a control, we also analyzed, in a separate working space, contemporary DNA from peripheral blood obtained from females from three generations of the same family. Four ancient teeth were studied, from which one was successfully analyzed for the two overlapping regions, which were subsequently sequenced. Fragment HV2b could not be amplified from ancient DNA. Matrilinear lineage was confirmed by analysis of the contemporary samples. The differences in the ancient and contemporary samples are discussed. Περιληψη: Οι πολλαπλές ταφές σε κοινότητες Πρώιμης Εποχής του Σιδήρου της Βορείου Ελλάδας πιθανόν να παρέχουν ενδείξεις για την έναρξη κοινωνικής έμφασης σε μικρές οικογενειακές ομάδες (γραμμές μητρικής διαδοχής) ως βασικές μονάδες οργάνωσης. Η ανάλυση αρχαίου μιτοχονδριακού DNA (mtDNA) mtDNA)) από ταφικά υπολείμματα, κυρίως δόντια και οστά, θα διευκόλυνε την απόδειξη επικράτησης τέτοιων γραμμών διαδοχής. Παρουσιάζουμε εδώ τα πρώτα αποτελέσματα της ανάλυσης aDNA δοντιών που βρέθηκαν σε νεκροταφείο από την Ύστερη Εποχή του Χαλκού έως τη Ρωμαϊκή Εποχή, στο Μακρύγιαλο (Βόρεια Μακεδονία, Ελλάδα – αρχαία Πύδνα), για να διερευνήσουμε την παραπάνω αρχαιολογική ερμηνεία. Η εξαγωγή του DNA των δοντιών έγινε με οργανικούς διαλύτες. Δύο τμήματα της περιοχής ελέγχου του mtDNA που αντιστοιχούν στις Περιοχές Υπερ-μεταβλητότητας Ι και ΙΙ (Hypervariable Hypervariable Regions I , II)) επαυξήθηκαν με αλυσιδωτή αντίδραση πολυμεράσης (PCR) PCR)) και έγινε ανάλυση αλληλουχίας, με τη χρήση όλων των προτεινόμενων προφυλάξεων για την αποφυγή επιμόλυνσης. Παράλληλα, για λόγους σύγκρισης και ελέγχου, αναλύσαμε επίσης, σε διαφορετικό χώρο εργασίας, σύγχρονο DNA περιφερειακού αίματος από θηλυκά άτομα τριών γενεών της ίδιας οικογένειας. Μελετήθηκαν τέσσερα αρχαία δόντια, από τα οποία το ένα αναλύθηκε επιτυχώς στις δύο επικαλυπτόμενες περιοχές, στις οποίες έγινε κατόπιν ανάλυση αλληλουχίας. Το τμήμα HV2b 2bb δεν έγινε δυνατό να επαυξηθεί από το αρχαίο DNA.. Η μητρική διαδοχή επιβεβαιώθηκε από την ανάλυση των σύγχρονων δειγμάτων. Γίνεται τέλος συζήτηση για τις διαφορές μεταξύ αρχαίων και σύγχρονων δειγμάτων.


also focused on major research questions regarding primarily the differences in the physical characteristics of human populations and thus developed a series of methodological tools in this direction.

Archaeologists have long had an interest in identifying geographically distinct clusters of material culture, which they assigned to distinct human groups of the past. In an intensive attempt to trace the “other”, physical anthropology, especially in the 19th and early 20th centuries,

The measuring of skull shapes (craniometry) together with the systematic recording of metric and non metric traits in 225

L. Kovatsi, S. Triantafyllou, J. Eliopoulos, M. Besios, S. Andreou, A. S. Tsaftaris, A. Dimitriadou and S. Kouidou the cranial and post cranial skeleton were considered to be strictly biologically determined and thus their detailed study and statistical analysis would offer potential answers to the identification of different peoples. Skeletal traits however are largely affected by both local and general epigenetic and environmental factors and thus mechanisms regarding skeletal and dental heritability are poorly understood. Besides, the measurement and interpretation of relatedness and divergence between population groups have undergone a striking decline during previous years due to their misuse in the construction of “race” and “ethnicity”. More recently however, both archaeology and physical anthropology moving within modern approaches of social anthropology, addressed their interests towards the investigation of different aspects of the relationships between population groups and subgroups.

Figure 1. Multiple burial

New methodological advances involving ancient DNA analysis may provide a new boost to crucial issues on the identification of population boundaries, post marital residence patterns, familial and kin groupings (Buikstra 1990, Mirza 1995). Molecular archaeology, the field associated with ancient DNA analysis, offers an historical perspective to molecular investigation and a molecular insight into the natural history of the living species. At the same time, it has to face a formidable array of technical problems which arise due to the fact that ancient DNA is chemically modified and is refractory to most of the procedures of nucleic acid analysis. Ancient DNA analysis became the subject of molecular analysis in the late 1980s (Paabo 1985, Paabo 1989, Stringer 1988) and generated an autonomous field defined as ‘molecular archaeology’. Starting from the simple idea of trying to extract and characterize deoxyribonucleic acid from archaeological and fossilized samples, molecular archaeology has found an innovative way to tackle ambitious goals such as studying the genetic structure of extinct species and their relationship to contemporary species (Greenwood 1999, Leonard 2000) and solving phylogenetic relationships within the genus Homo, throwing light on the origin of modern humans (Krings 2000, Stringer 1988).

Figure 2. Single burial

Because of the degraded nature of ancient DNA (aDNA) extracts, mitochondrial (mt)DNA has proven to be the molecule of preference for genetically characterizing prehistoric samples. This is because mtDNA is present in several hundreds of copies per cell, in contrast to the singlecopy nuclear genome. Thus, target sequences of mtDNA are more likely to be present in any single extract, and accessible for amplification, than are nuclear sequences.

Figure 3. Ancient tooth sample

Actually, some regions of the mtDNA genome appear to be evolving at a rate of 5 to 10 times higher than single copy nuclear genes. It is these regions that are of interest for human identity testing. It has been estimated that among unrelated Caucasians, there is an average of eight nucleotide difference between individuals in specific sequences.

Mitochondrial DNA lacks repair mechanisms, and thus the rate of changes in its sequence (mutations) is high. 226

In Search of Mothers and Daughters in 3000 Year Old Burials in Northern Greece Materials CLAN MOTHER Ursula Xenia Velda Tara Katrine Jasmine

POSITION OF CHANGE 16270 16223 16298 16126 16294 16224 16311 16069 16126


Our samples come from the cemetery of Makrigialos (ancient Pydna), which is located in Southern Macedonia, Greece, close to the geographical borders of the region with Thessaly. The cemetery was used from the Late Bronze Age to the Roman Period and its excavation has revealed several thousands graves. Approximately 5% of the total number of graves belongs to the Early Iron Age and more specifically to the 9th century BC. This segment of the cemetery has undergone a thorough study of the archaeological and the osteological data (Bessios 1989, Triantaphyllou 1998).

Table 1. Characteristic deviations from the CRS for each of our ancient ancestral mothers-clan mothers.

The mortuary picture of the 9th century burials presents the following characteristics:

The CRS (Cambridge Reference Sequence) is a standard mtDNA sequence, against which we compare all corresponding sequences; it is believed that this sequence is most common among native Europeans. Our common ancestral heritage is verified by the deviations from the CRS (specific DNA sequences). Seven of these sequences are considered to be more ancient and to characterize corresponding lineages that are named after 7 females, the ‘Seven Daughters of Eve’ (ancient ancestral mothers – clan mothers). Since modern day descendents of any single clan mother will be from many different countries, some ethnic backgrounds and races can also be determined, based on other characteristic sequences, provided that sufficient comparison data is available. However, as a general rule, for individuals whose ancestors lived in the geographical area now called Europe, absence of the characteristic sequences listed in table 1, signifies descent from the Helena lineage.



3. 4.

The highlighted letters show where one’s mtDNA sequence varies from the Cambridge Reference Sequence (CRS). One’s mtDNA changes overtime and only certain changes at specific positions are characteristic of one’s clan. If one’s sequence contains one of these characteristic changes, then one’s clan mother is identified appropriately. If one’s sequence does not contain one of these changes, then one’s clan mother was Helena.


A prevalence of single inhumations. Nevertheless few graves display multiple burials. This differentiation of burial practices is not associated with either sex or age differences. A large variety of grave types. Pit graves are the most frequent followed by cist graves. There is however a group of five chamber tombs, a type which is very rare in Macedonia and appears for the first time during the Early Iron Age. However, chamber tombs were the most common type of grave in Southern Greece during the Mycenaean period. There is no differentiation in sex and age groups in the use of the different grave types. A variety of burial orientations, which as a rule were associated with the sex of the deceased. A variety of grave goods (pottery, jewellery, weapons etc) some of which are also associated with the sex of the deceased. A spatial clustering of burials appearing either as clusters of single burials or multiple burials in cist or chamber tombs has been observed.

The samples selected were teeth. Teeth were preferred to bones because of the better preservation of DNA (Kurosaki 1993, Oota 1995). Method

There is little work to our knowledge concerning aDNA analysis in Greece (Brown 2000, Evison 2001). In the present paper we report the preliminary results of a project that employs aDNA analysis to confirm the archaeological interpretation of the commencement of the cultural practice of multiple burial as an indication for the initiation of social emphasis on small family groups (matrilineal lineages) as the basic organizational unit in the Early Iron Age communities of Northern Greece.

Maximum precautions were taken during the collection of samples from the excavation site in order to avoid contemporary contaminations. Gloves and masks were worn and all tools and containers were sterile and DNAfree. The selection of the tooth was based on the health of the specimen, since teeth with caries allow contaminating DNA to enter the pulp cavity. Therefore, we chose to use unerupted teeth and we avoided sectioning the tooth and removing the pulp cavity. Instead, we chose to use the whole tooth.

For this reason, we analyzed segments of the control region of the mtDNA from selected ancient tooth samples. For reasons of methodological comparison we also analyzed modern mtDNA from members of a selected family.

The selection of the samples from the human skeletal remains was determined: 227

L. Kovatsi, S. Triantafyllou, J. Eliopoulos, M. Besios, S. Andreou, A. S. Tsaftaris, A. Dimitriadou and S. Kouidou To remove dirt and other contaminants from previous handlings, and therefore avoid contamination and false positive results, the surface layer of the teeth was removed and the teeth were then exposed to UV light (254 nm) at 2.5 cm from the light source for at least 10 minutes on each side. The teeth were then pulverized and the pelleted bone powder was decalcified by incubation (at 55oC overnight and at 37oC for 24 hours) in a solution containing EDTA 0.5M, SDS 20% and Proteinase K. For the isolation of mitochondrial DNA, we used two different approaches. The first protocol, designated the Centricon approach (Hagelberg 1991) involved proteinase digestion to solubilize the DNA. The DNA was then separated from the cellular debris through a series of phenol-chloroform extractions and concentrated by passage through a Centricon microconcentrator. The second protocol involved the use of silica particles that have a high binding capacity for DNA molecules (Hoss 1993). We compared the two protocols by testing them on contemporary teeth, before deciding which one was to be employed on our ancient samples.

Figure 4. Successful amplification of all 4 target fragments in DNA coming from contemporary tooth samples.

Following isolation and purification of mtDNA, each of the two Hypervariable Regions I and II were amplified in two overlapping fragments. The target fragments were:

Primer Sequence

Target fragments






Figure 5. Successful amplification of target fragment HV1a in DNA coming from ancient tooth sample and lack of contamination in the extraction reamplification blank.




1. 2. 3.

By the archaeological objectives of the research. By the limitations set by the type of information provided by the mitochondrial aDNA (genetic characteristics inherited solely through the mother). By the state of preservation of the skeletal material.




For the amplification, the reaction mixture was briefly UV-irradiated prior to the addition of DNA and enzyme. Amplification reactions were carried out in a final volume of 50μl. μl. l. The reaction mixture contained 20-40 pmol of each primer (depending on the primer pair), 200 μM M each of dNTPs, 1.5-5.0 mM MgCl2 (depending on the primer pair), 20 mM Tris-HCl pH 8.4, 50 mM KCl and 2 mg/ml of BSA. Cycling conditions were optimized for each of the individual target fragments and were as follows:

It needs to be pointed out that these constraints considerably limited the potential population from which samples could be selected. Only ten samples could be found that fulfilled all the aims and limitations of the project. Four of them belong to the same cluster of single inhumations, another four of them belong to pairs of burials laid in two cist graves and the remaining two belong to burials laid in the same chamber tomb. 228

In Search of Mothers and Daughters in 3000 Year Old Burials in Northern Greece included. The sample extract and the extraction blank, plus one negative reagent control (PCR blank) were subjected to PCR amplification. When a second round of PCR was performed, the product of the first round PCR from the relevant extraction blank was reamplified alongside the samples subjected to the second round of PCR (extraction reamplification blank).

For target fragment HV1a 5 min at 94oC 20 s at 94oC 20 s at 62oC

40 cycles

30 s at 72oC 5 min at 72oC

Another problem aDNA analysis has to cope with, are PCR-introduced errors due to both DNA damage and inappropriate enzyme activity (Krings 1997). In order to assure the authenticity of our results, our target sequences were amplified and sequenced twice and in some cases the whole procedure was applied to 2 different DNA extracts of the same specimen.

For target fragments HV1b, HV2a, HV2b 5 min at 94oC 20 s at 94oC 10 s at 56oC

40 cycles

30 s at 72 C o

5 min at 72oC

Results Out of 4 ancient teeth we studied, we only succeeded to extract and amplify DNA from one of them.

Reamplification of the first PCR product was routinely used to produce sufficient amounts for sequencing. For the second round of PCR, 10 μl of the first amplification product were used. The amplified DNA was sequenced using Dye Terminator Cycle Sequencing. The primer sequences are shown below: Primer Sequence 5’-TTA ACT CCA CCA TTA GCA CC-3’ 5’-CTT TGG AGT TGC AGT TGA TG-3’

We were not able to amplify fragment HV2b when using ancient DNA, while all 4 fragments were successfully amplified when using contemporary DNA specimens. Contemporary DNA specimens were analyzed in order to be able to compare the ancient with the contemporary sequences.

Region sequenced HV1a



Comparison of the two sequences as shown in figure 6, revealed two point differences at positions 16111 and 16214.



Comparison of the two sequences as shown in figure 7, revealed no differences between them.



Comparison of the two sequences as shown in figure 8, revealed two point differences at positions 00146 and 00184.

To avoid contamination, tight segregation was kept between pre-PCR and post-PCR areas. DNA extractions and PCR setups were carried out in a dedicated biosafety hood that was located separately from that used for postPCR analysis. The hood was frequently cleaned with 10% hypochlorite and alcohol (Prince 1992) and UV irradiated overnight (Fox 1991). Separate sets of pipetting devices were dedicated for sample preparation and for setting up reactions. All pipettes were used with sterile aerosolresistant tips. All post PCR steps were carried out with a dedicated set of pipettes. Each step of the analysis was conducted in temporal sequence, with specific days being dedicated to extraction, amplification and post-PCR analysis. All solutions used were autoclaved and UV irradiated, while all reagents for PCR were bought in the smallest amount available and were aliquoted. The use of positive controls was avoided. The extractions were always carried out for each sample separately with one negative extraction control (extraction blank) also being

Comparison of the two sequences as shown in figure 9, revealed one point difference at position 16173. Comparison of the two sequences as shown in figure 10, revealed one point difference at position 00073. Discussion Nucleic acids gradually degrade over time through processes such as hydrolysis and oxidation. For these reasons, recovery and amplification of aDNA, when possible, is usually limited to fragments 125 μg/ml.

The ancient DNA extraction and amplification from burials discovered in the South Eastern Villa in Delphi is a primary effort realised with the combination of technical and alternative ways of processing already known methods. Attempts on Greek samples are still limited, similarly perhaps to the knowledge of the potentials offered by these scientific methods to the excavator, for a deeper knowledge of the materials brought to light. This experimental attempt by a Greek laboratory and its prospective, systematic occupation with ancient DNA extraction, apart from its modern and multiple applications for other purposes, are expected to give a significant impetus to this subject and new important data for the research of the past, parallel

DNA precipitates at –700C for 7 minutes, and centrifuged at 14.000 rpm and 40C for 15 minutes. The pellet is dissolved in 20-30 μl of ionised-distilled water. At this stage, it can be stored at –200C. 3. Amplification Typical PCR amplification occurs in 2-7 μl of extract with 1U Taq DNA Polymerase, 160 μg/ml BSA, 200 μΜ out of each dNTP, 20 pmol out of each primer for human bglobin gene, or DR gene of the HLA system, or X and Y–chromosomes identification genes in a PCR bugger 237

Κ. Κoniavitou, Chr. Κroupis, Α. Κapsalis, L. Karali, Pl. Petridis and F. Μavridis to the occupation of other scientists from different institutions.

DNA and the Population Prehistory of Europe, McDonald Institute Monographs, Cambridge: University of Cambridge, 315-320. Bradley, D.G., Machugh, D.E., Cunningham, P., Loftus, R.T., 1996, Mitochondrial Diversity and the Origins of African and European Cattle, Proceedings of the National Academy of Sciences of the USA , 5131-5135. Brown, T.A., and K.A., 1992, Ancient DNA and the Archaeologist, Antiquity , 10-23. Brown, T.A., and Α., Flaherty, C.E., Little, L.M., Prag, A.J.N.W, 1999, DNA Analysis of Bones from Grave Circle B at Mycenae: A First Report, Annual of the British School at Athens , 115-119. Βrown, K.A., Pluciennik, M., 2001, Archaeology and Human Genetics: Lessons for Both, Antiquity , 101-106. Cavalli-Sforza, L., 1996, The Spread of Agriculture and Nomadic Pastoralism: Insights from Genetics, Linguistics, and Archaeology, in D. HARRIS (ed.), The Origins and Spread of Agriculture and Pastoralism in Eurasia, London: UCL Press, 51-69. Evison, M., 1996, Genetics, Ethics and Archaeology, Antiquity , 512-514. Evison, M., Kyparissi-Apostolika, N., Stavropodi, E., Fieller, N., Smile, D., 2000, An Ancient HLA Type from a Palaeolithic Skeleton from Theopetra Cave, Greece, in N. Kυπαρίσση υπαρίσση - Αποστολίκα (ed.), Σπήλαιο Θεόπετρας, Πρακτικά Διεθνούς Συνεδρίου, Αθήνα, 109-118. Halgelberg, algelberg,, E., Sykes, Β., Hedges, R., 1989, Ancient Bone DNA Amplified, Nature, Nov.30, 342 (6249), 485. Halstead, P., 1996, The Development of Agriculture and Pastoralism in Greece: When, How, Who, and What?, in D. Harris (ed.), The Origins and Spread of Agriculture and Pastoralism in Eurasia, London: UCL Press, 296-309. Hanni, C., Brousseau, T., Laudet, V., Steheinn, D., 1995, Isopropanol Precipitation Removes PCR Inhibitors from Ancient Bone Extracts, Nucleic Acid Res., Mar.11, 23, (5) 881-882. Jones, S., 1997, Τhe Archaeology of Ethnicity, London: Routledge. Kalmar, T., Bachrati, C.Z., Marcsik, A., Rasko, I., 2000, A Simple and Efficient Method for PCR Amplifiable DNA Extraction from Ancient Bones, Nucleic Acid Res, June, 15, 28 (12), E67. Kotsakis,, Κ., 2000, H αρχή της Νεολιθικής στην Ελλάδα, in N.. Kυπαρίσση υπαρίσση - Αποστολίκα (ed.), ed.), .), Σπήλαιο Θεόπετρας, Πρακτικά Διεθνούς Συμποσίου, Αθήνα, 173-177. Pääbo, bo,, S., 1999, Ancient DNA, in B. SYKES (ed.), The Human Inheritance, Oxford: Oxford University Press, 119-134. Petridis, PL., 1997, Dèlphes dans l’Antiquité tardive: première approche topographique et céramologique, Bulletin de Correspondance Ηéllenique éllenique, , 681-688. Pluciennik, Μ., 1996, Genetics, Archaeology and the Wider World, Antiquity , 13-14. Renfrew, C., 1998, World Linguistic Diversity and Farming Dispersals, in: R. ΒLENCH, M. SPRIGGS (eds.), Archaeology and Language, Vol. 1. Theoretical and Methodological Orientations, London: Routledge, 82-90. Renfrew, C., 1999, Archaeogenetics: Towards a Population Prehistory of Europe, in C. RENFREW, K. BOYLE (eds.), Archaeogenetics: DNA and the Population Prehistory of Europe, McDonald Institute Monographs, Cambridge: University of Cambridge, 3-12. Shanks, M., Hodder, I., 1995, Processual, Post-processual and Interpretive Archaeologies, in HODDER I. et al. (eds.), Interpreting Archaeology, London: Routledge, 3-29. Sykes, B., 1999, Human Diversity in Europe and Beyond: From Blood Groups to Genes, in: C. Renfrew, K. Boyle (eds.), Archaeogenetics: Towards a Population Prehistory of Europe,

The application of these methods on a larger scale in Aegean Archaeology is expected with particular interest. The progress of research with the systematisation of laboratory procedures and methodology by specialists but also the application of specific questions by archaeologists, or better a series of hypothesis that can be tested, form the next stage of the research process. The continuation of research of prehistoric cemeteries of a specific date in the Aegean would be the next step in an attempt of understanding burial customs, family relationships, and sex diversification, in a long term prospective followed by the approach of matters related with population movements. This paper also aims to connect theory with science that both are very important in archaeological research. Despite the fact that these approaches contradict most of the times with each other in terms of methodology and ideological background, they can be linked systematically for providing better understanding of ancient societies. The Science versus anti-Science (Thomas 1991) concept is no longer useful in Archaeology and both approaches’ results and research potentials should be evaluated as serving a single objective. Taking into consideration the limited purpose of this pilot programme and its primary encouraging results, the successful issue of analogous attempts in the future can be foreseen. The continuous progress in Molecular Biology and the application and adaptation of new and improved techniques will possibly assist for an analysis of “more laborious” samples, even with a small percentage of degraded or contaminated DNA. Notes Under the direction of V. Dèroche and Pl. Petridis, a large group of collaborators consisted of archaeologists and students of Archaeology from Greece, France, Belgium, and Switzerland, architects and conservators from Greece and France, as well as topographers, one numismatologist, and several specialists of clay analysis and environmental studies, specifically shells and bones. References Αllaby, R., 1999, Wheat Domestication, in C. Renfrew and K. Boyle (eds.), Archaeogenetics: DNA and the Population Prehistory of Europe, McDonald Institute Monographs in Prehistory, Cambridge: University of Cambridge, 321-324. Ammerman, A.J., Cavalli-Sforza, L.L., 1984, The Neolithic Transition and the Genetics of Populations in Europe, Princeton: Princeton University Press. Βar-Gal, G.K., Smith, P., Tchernov, E., Greenblatt, C., Ducos, P., Gardeisen, A., Horowitz, L.K, 2002, Genetic Evidence for the Origin of the Agrimi-Goat (Capra aegagrus cretica), Journal of Zoology 256, , 369-377. Bradley, D., 1999, Mitochondrial DNA Diversity and the Origins of Agriculture, in C. Renfrew, K. Boyle (eds.), Archaeogenetics:


Ancient DNA Extraction and Amplification of Human Bone Samples from the Area of Delphi Williams - Sims, P., 1998, Genetics, Linguistics, and Prehistory: Thinking Big and Thinking Straight, Antiquity , 505-527. Zvelebil, M., 1995, At the Interface of Archaeology, Linguistics and Genetics: Indo-European Dispersals and the Agricultural Transition in Europe, Journal of European Archaeology  (1) 13-22.

McDonald Institute Monographs, Cambridge: University of Cambridge, 23-30. Sykes, B., Renfrew, C., 1999: Concepts in Molecular Genetics, in C. Renfrew, K. Boyle (eds.), Archaeogenetics: Τowards a Population Prehistory of Europe, McDonald Institute Monographs, Cambridge, University of Cambridge, 33-70. Thomas, J., 1991 : Science versus Anti- Science ? Archaeological Review from Cambridge, , 1, 27-36.



Department of Oral Pathology and Surgery, Faculty of Dentistry, University of Athens, 2 Thivon Street, 11527, Athens, Greece [email protected] Abstract.: Ancient Greek authors reported the Pygmies, the Sciapodes, the Macrocephali and the Argippeans. The Pygmies were reported as a dwarf-race in the upper Nile; also, little black men living in a sub-desert region and on the west coast of Central Africa. Aristotle verified the existence of pygmies. In the forests of Central Africa still live groups of African pygmies, who fail to respond to growth hormone. The Sciapodes (Shadow-feet) were recorded as an Ethiopian, or a Libyan nation with immense feet. The distribution of the filariasis or non-filariasis endemic elephantiasis of the lower legs in Africa supports the notion that the Sciapodes may well have been sufferers of tropical elephantiasis. For the Macrocephali (Long-headed) Hippocrates considered a custom responsible for the length of their head. A Caucasian tribe, the Siginni, was also practicing the same custom. Ancient Egyptian sculptures and wall paintings depict long–headed individuals. There is enough evidence to support that the Macrocephali had acquired macrocephaly, through culturally prescribed behaviours. The Argippeans, men inhabiting the foothills of high mountains northeast of Scythia, were said to be bald from birth, snubbed-nosed and long-bearded. Opinions stand, for the Argippeans being probably a Mongoloid tribe living north to Tien Shan and south to Altai Mountains. Περίληψη: Αρχαίοι Έλληνες συγγραφείς αναφέρουν τους Πυγμαίους, τους Σκιάποδες, τους Μακροκέφαλους και τους Αργιππαίους. Αναφέρονται οι Πυγμαίοι, ως φυλή νάνων στον Άνω Νείλο, και άνθρωποι «μικροί» κάτοικοι της υπό-Σαχάραν περιοχής και της δυτικής ακτής της Κεντρικής Αφρικής. Ο Αριστοτέλης επιβεβαιώνει την ύπαρξη των Πυγμαίων. Στα δάση της Κεντρικής Αφρικής υπάρχουν ακόμα ομάδες Αφρικανών πυγμαίων που δεν ανταποκρίνονται στην δράση της αυξητικής ορμόνης. Οι Σκιάποδες αναφέρονται ως Αιθίοπες, ή και ως Λίβυοι με πελώρια πόδια. Η κατανομή της φιλαριασικής ή μη-φιλαριασικής ενδημικής ελεφαντίασης των κάτω άκρων στην Αφρική υποστηρίζει την άποψη ότι οι Σκιάποδες μπορεί κάλλιστα να υπέφεραν από τροπική ελεφαντίαση. Για τους Μακροκέφαλους, ο Ιπποκράτης θεωρεί ένα έθιμο ως υπεύθυνο για την επιμήκυνση της κεφαλής τους. Το ίδιο έθιμο εφάρμοζαν οι Σίγιννοι του Καυκάσου. Αρχαία Αιγυπτιακά αγάλματα και τοιχογραφίες απεικονίζουν μακροκέφαλους. Υπάρχουν αρκετές ενδείξεις που υποστηρίζουν ότι οι Μακροκέφαλοι είχαν επίκτητη μακροκεφαλία από εφαρμογή εθιμοτυπικών συνηθειών. Οι Αργιππαίοι, οι οποίοι κατοικούσαν στους πρόποδες υψηλών βουνών βορειοανατολικά της Σκυθίας , αναφέρονται ως φαλακροί εκ γενετής, με πλακουτσωτή (σιμή) μύτη και γένια μεγάλα. Υπάρχουν γνώμες που υποστηρίζουν ότι οι Αργιππαίοι ήταν μια Μογγολική φυλή που ζούσε βόρεια από τα όρη Tien Shan και νότια από τα Altai.


from ancient Greek literary sources. In chronological order, these sources are listed as follows:

Several tribes with physical abnormalities and diseases, distributed in the world known at the time, have been reported in the Archaic and Early Classical Greek literature. Even in antiquity, there were arguments about their existence. In spite of the rational explanations provided in Classical writings, later authors, of the Hellenistic Age, insisted that these tribes were fabrications and myths.

• • • •

Indeed, there have been described tribes, such as the Enotocoetae (men that sleep in their ears), the Monommati (one-eyed men), the Amycteres (people without nostrils) (Strabo, Geography, 15.1.57), the Cynocephali (dogheaded men), the Acephali (headless men) (Herodotus, IV.191), who sound peculiar, but some of them, namely the Pygmies, the Sciapodes, the Macrocephali and the Argippeans, seem to deserve attention. By this study, I intend to propose a presumptive diagnosis of their abnormalities and diseases and the geographic distribution of the reportedly affected tribes, based on relevant evidence

• • • • •


Homer, epic poet, eighth century B.C. (Lesky, 1966, p 14) Hesiod, epic poet, circa 700 B.C. (Lesky, 1966, p 91) Hecataeus of Miletus, historian, sixth century B.C. (Lesky, 1966, pp 2, 220) Herodotus, historian, fifth century B.C. (Lesky, 1966, p 307) Hippocrates, medicus, born in 460 B.C. (Lesky, 1966, p 486) The treatise on Airs Waters and Places (On Environment) was written shortly before 430 B.C. (Lesky, 1966, p 489) Aristophanes, comic writer, born about 445 B.C. (Lesky, 1966, p 425) His comedy, The Birds, was brought out in 414 B.C. (Lesky, 1966, p 438) Archippus, comic writer, fifth-fourth century B.C. (Lesky, 1966, p 426)

(%D]RSRulou-Kyrkanidou • • •

Aristotle, philosopher,384-322 B.C. (Lesky, 1966, pp 548, 552). Apollodorus of Athens, grammarian, second century B.C. (Lesky, 1966, pp 787, 857) Strabo, geographer, circa 64 B.C.- 19 A.D. (Lesky, 1966, p 890)

to land with their ship, would ever leave their towns and flee to the hills;” (Herodotus, IV.43). To interpret this passage we have to keep in mind that, in ancient times, Africa was called Libya. It is well known that the ancient Pillars of Heracles correspond to Gibraltar. The Libyan promontory called Solois is probably Cape Cantin, in the latitude of Madeira (Godley, 1921 p 241). After he rounded the promontory Solois (Cape Cantin), Sataspes sailed for many months southward, as Herodotus reported, presumably, along the west coast of Africa. “When he was farthest distant he sailed by a country of little men who wore palm-leaf raiment.” It becomes evident that, the little men Sataspes met at the west coast were the Pygmies Negroes (Negrillos), who inhabit this area of Central Africa (The New Caxton Encyclopedia, 1967 p 4311).

Tribes with physical abnormalities and diseases The Pygmies The Pygmies (Πυγμαίοι, Πυγμαίοι,, οι)) are mentioned from both Homer and Hesiod. More precisely, in Homer’s Iliad (3. 36), the Pygmies are referred to in connection with their fights with the cranes: “… like the clamor of cranes that arises before the face of heaven when they flee from wintry storms and boundless rain, and with clamor fly toward the streams of Ocean, bringing slaughter and death to Pygmy men, …”

Exploration of young Nasamonians in ancient Libya: “…Nasamonians. These are a Libyan people, inhabiting the country of the Syrtis and the country a little way to the east of the Syrtis. When these Nasamonians on their coming were questioned if they brought any news concerning the Libyan desert, they told Etearchus that there had been among them certain sons of their chief men, proud and violent youths, who, when they came to man’s estate, besides planning other wild adventures, had chosen by lot five of the company to visit the deserts of Libya, and see what they might beyond the outmost range of travellers. It must be known that all the northern seacoast of Libya-from Egypt as far as the promontory of Soloeis, which is the end of Libya- is inhabited through its whole length by Libyans, many tribes of them, except the part held by Greeks and Phoenicians; the region of Libya above the sea and the men of the seacoast is infested by wild beasts; and farther inland than the wild-beast country all is sand, exceeding waterless and wholly desert.” (Herodotus, II. 32).

The fights of Pygmies and cranes had been favoured in many paintings (Lexicon Iconographicum Mythologiae Classicae (LIMC), 1988). Among them there is an early black figure Corinthian altar painting of the later sixth century B.C., with the pygmy as a tiny man like a pituitary dwarf, and an archaic red figure vase painting with realistic, plump pygmies, like achondroplastic dwarfs. The name of the Pygmies is also preserved in two Hesiod fragments: “Nor let anyone mock at Hesiod who mentions...or even the Troglodytes and the Pygmies” (Hesiod, Catalogues of Women and Eoiae, Fr.43). “No one would accuse Hesiod of ignorance though he speaks of the Half-dog people and the Great-Headed people and the Pygmies” (Hesiod, Catalogues of Women and Eoiae, Fr.44).

(The country of the Syrtis corresponds to the Gulf of Surt (Gulf of Sidra, Gran Sirte) coast of Libya, extending from Misratah west, to Benghazi east).

“Little men” are, also, reported by the father of history Herodotus to live in a country at a farthest distance of the Pillars of Heracles, which sounds like the west coast of Central Africa (IV.43) and “Little men, and black” in a sub-desert area of Africa, most probably the Sub-Saharan area (II.32, 33, 34):

“This then was the story told by the young men: - When they left their companions, being well supplied with water and provisions they journeyed first through the inhabited country, and having passed this they came to the region of wild beasts. After this, they travelled over the desert, towards the west, and crossed a wide sandy region, till after many days they saw trees growing in a plain; when they came to these and were plucking the fruit of the trees, they were met by little men of stature smaller than common, who took them and led them away. The Nasamonians did not know these men’s language nor did the escort know the language of the Nasamonians. The men led them across great marshes, which having crossed they came to a city where all the people were of like stature with the escort, and black. A great river ran past this city, from the

“...Sataspes went to Egypt, where he received a ship and a crew from the Egyptians, and sailed past the Pillars of Heracles. Having sailed out beyond them, and rounded the Libyan promontory called Solois, he sailed southward; but when he had been many months sailing far over the sea, and ever there was more before him, he turned back and made sail for Egypt. Thence coming to Xerxes, he told in his story how when he was farthest distant he sailed by a country of little men who wore palm-leaf raiment; these, whenever he and his men put in 242

Pygmies, Sciapodes, Macrocephali, Argippeans:mythical Or Existent Tribes? west towards the rising sun; crocodiles could be seen in it.” (Herodotus, II. 32).

public media (Remy, 2003). Most likely, these Pygmies are descendants of the Pygmies of ancient times. Then, it is a good reason for the name that has stuck to these, the world’s shortest people.

“… the Nasamonians returned-as the men of Cyrene told me- and that the people to whose country they came were all wizards; as to the river that ran past the city, Etearchus guessed it to be the Nile; and that is but reasonable. For flows from Libya, and right through the midst of that country;” (Herodotus, II.33).

PYGMY is listed in the On Line Mendelian Inheritance in Man (OMIM), #265850, and is referred to the African pygmy. In scientific terms, the African pygmy fails to respond to exogenous hormone in the presence of normal serum levels of growth hormone and of somatomedin (Rimoin et al, 1969). Furthermore, pygmies exhibit relative refractoriness to some of the acute metabolic actions of growth hormone. Thus, the ‘defect’ appears to involve end-organ responsiveness to somatomedin. The genetics of the change is unclear. Most, however, interpret the results of crossing with neighbouring non-pygmy tribesmen as indicative of multifactorial (i.e., polygenic) inheritance. More recently, Hattori et al. (1996) concluded that the short stature of African Pygmies is related to insulin-like growth factor 1receptor (IGF1R) insensitivity, and suggested that human stature in general may be genetically regulated via control of the IGF1R gene.

“As it (the Nile) flows through inhabited country, its coarse is known to many; but none can speak of the source of the Nile; for Libya, through which it runs, is uninhabited and desert. Concerning its course I have told all that I could learn by inquiry; and it issues into Egypt” ([Herodotus, II. 34). The existence of the Pygmies in a region about the source of the Nile, and the moving of cranes to the same area, as in Homeric Iliad, is accepted by the philosopher Aristotle: “… for they (the cranes) move from the Scythian plains to the marshes above Egypt from where the Nile flows; this is the region whereabouts the pygmies live (for they are no myth, but there truly exists a kind that is small, as reported - both the people and their horses - and they spend their lives in caves)” (Aristotle, History of Animals, XII. xii 597a 5-9).

The Sciapodes Another tribe is the Sciapodes (Σκιάποδες, Σκιάποδες,, οι), ), Shadefooted, Shadefoots or Shady-feet, a fabulous people in the hottest part of Libya (Africa), with immense feet, which they used as sunshades as they reclined (Liddell et al, 1990). The apparent ability of the Sciapodes to shade themselves with a swollen foot was an attempt to convey the size of the limp (Pliny the elder, 1st century AD, cited by Lawrence, 1990).

In a map on the distribution of races in Africa from the New Caxton Encyclopedia (1967, p 4311), Pygmies Negroes (Negrillos) are shown to live in an area around the north side of Congo river, extending from the west coast of Central Africa to the source of Nile river east. These areas correspond to those referred to by ancient literary sources as being inhabited by short people.

Hecataeus of Miletus (FGrHist I F 327), on a tour of Egypt, recorded the Sciapodes as being an Ethiopian nation; Archippus (Comicorum Atticorum Fragmenta, Fr. 53) recorded them as a Libyan nation.

It seems, however, that the area occupied by the Pygmies has shrunk since 1967, as it is shown in an article published in the National Geographic magazine in 1995 (Chadwick, 1995 p 2-45). In a remote pocket of central Africa runs a river called Ndoki. Its waters flow from a tropical forest, Ituri forest that supports groups of Pygmies and a greater abundance of wildlife than exists perhaps anywhere else on the continent. If land belonged to the people who first staked claim to it, then central Africa’s forest would no doubt belong to the Pygmies. Their name, which derives from the Greek word pygme (πυγμή, as a measure of length: the distance from the elbow to the knuckles= 18 δάκτυλοι,, 0.35 m, Liddell et al, 1990)) meaning “half an arm’s length”, is considered pejorative. Yet the name has stuck to these, the world’s shortest people, although they call themselves Efe. Adults average, the male about four feet, eight inches and the female four feet, 5 inches because they are unable to process the hormones needed for normal growth (Diamond, 1991). The shrinking of the race of Pygmies continues today by brutalities against them, such as ritual cannibalism, from rebels, as it is shown on the

The Sciapodes, were also referred to, without any precise designation, by Aristophanes in his comedy The Birds (1553-1555): CHORUS “Far away by the Shadefoots lies a swamp, where all unwashed Socrates conjures spirits” “The race of the Sciopodes (sic) is said to live in Ethiopia. They have one leg apiece and are of marvellous swiftness, and the Greeks call them Sciopodes (sic) from this, that in summertime they lie on the ground on the backs and are shaded by the greatness of their feet”, From Etymologies of Isodore of Seville, Book XI: 3, 23. (Compiled in Spain A.D. 636, cited by Price, 1974). Illustrations of the Sciapodes come from: Mappa Mundi of Hereford Cathedral; the Osma Beatus map of 1203; Barber 243

(%azopoulou-Kyrkanidou is, a custom was chiefly responsible for the length of their head:

& Riches, A Dictionary of Fabulous Beasts (London: Macmillan, 1971), who observed that the mythological Sciapodes may be one- or two-legged. In the latter case, only one of the legs is enlarged and thus closer to the original model. The story of the leaping about of onelegged Sciapodes evidently grew in the telling (cited by Lawrence, 1990); the Middle Ages, 1493: Hartmann Schedel Illustration aus dem . Diagnosis by Wiedemann: Partial gigantism (Proteus syndrome, Recklinghausen) (skiapode as described by Wittkower) (Kunze and Nippert, 1986, p. 44, fig. 46).

“The races that differ but little from one another I will omit, and describe the condition only of those which differ greatly, whether it be through nature or through custom. I will begin with the Longheads There is no other race at all with heads like theirs. Originally custom was chiefly responsible for the length of the head, but now custom is reinforced by nature. Those that have the longest heads they consider the noblest, and their custom is as follows. As soon as a child is born they remodel its head with their hands, while it is still soft and the body tender, and they force it to increase in length by applying bandages and suitable appliances, which spoil the roundness of the head and increase its length. Custom originally so acted that through force such a nature came into being; but as time went on the process became natural, so that the custom no longer exercised compulsion. For the seed comes from all parts of the body, healthy seed from healthy parts, diseased seed from diseased parts. If, therefore, bald parents have the most part bald children, grey-eyed parents grey-eyed children, squinting parents squinting children, and so on with other physical peculiarities, what prevents a long-headed parent having a long-headed child? At the present time long-headedness is less common that it was, for owing to intercourse with other men the custom is less prevalent”.

The world distribution of filariasis (Rook et al, 1986, p. 1002) includes tropical Africa and the north coast of Africa, areas referred to as being the Sciapodes’ habitat. This supports the notion that the Sciapodes may well have been sufferers of tropical elephantiasis, which involve the lower limbs observed in filariasis, a disease due to infestation, mainly, with the filarial worms Wuchereria bancrofti. Non-filariasis endemic elephantiasis of lower legs also occurs in areas at high altitude, over 1000 metres, and seasonal rainfall of tropical Africa, as in East Africa (Ethiopia, north-western Tanzania, Rwanda). This “highaltitude” elephantiasis of the lower legs is suggested to be a geochemical disease, since it is related to the distribution of red clay soil derived from volcanic rocks rich in colloidal iron oxide, and high proportion of alumino-silicate particles of colloid size in the clays. The number and small size of the particles facilitates entry through the bare foot skin into the lymphatic tissues, where they exert the known irritant and fibrosing effects of silica and alumina and cause irreversible damage to the lymphatic channels (Price, 1974; Price, 1976; Price and Bailey, 1984).

(race: έθνος, number of people living together, nation, people, cast, class of men (Liddell et all, 1990)). Practicing to acquire a long head was also a strange custom of a tribe living in around the Caucasus Mountains, the he Siginni:

Partial gigantism of lower limbs, occasional manifestation of Proteus syndrome or Recklinghausen neurofibromatosis, could only be observed in sporadic cases of these rare disorders. On the contrary, the filariasis or non-filariasis endemic elephantiasis of the lower legs manifests a high prevalence in the population of affected areas.

“Others are said to practice making their heads appear as long as possible and making their foreheads project beyond their chins.” (Strabo, Geography, 11.11.8). Skull variations can be caused not only from genetic defects, as in the craniosynostoses, but from external causes too. Intentional cranial deformation was used for a number of reasons including beautification, tribal identification, and social stature (Goodrich and Tutino, 2001). The normal infant skull, with its head modelled posterior from delivery, was thought to be the optimum shape in ancient Egypt. The custom of applying external pressure with a series of bandages slowly and deliberately created the desired contours. The skull, extended posterior, was leaving little room for the normal position of the frontal lobes of the brain. Yet, this was done so slowly that the suture lines have been found intact. The brain was allowed to remodel very slowly, and apparently without increasing the intracranial pressure, for there are no markings on the inside of the skull to indicate this (Moncada and Hendel, 1985). Artificial cranial deformation was an accepted practice in Egypt. The best known example remains Queen Nefertiti,

Macrocephali Now we come to the Macrocephali (Μακ Μακρροκέφαλ οκέφαλοι, οι, οι), οι), the Long-headed, the Longheads, the Great-Headed, who were firstly mentioned by Hesiod (Catalogues of Women and Eoiae, Fr.44): “No one would accuse Hesiod of ignorance though he speaks of the Half-dog people and the Great-Headed people and the Pygmies.” Hippocrates also referred to the Macrocephali in the treatise On Airs Waters and Places (XIV. 1-28), written shortly before 430 B.C. In addition, he gave a reasonable explanation for the aetiology of their macrocephaly, that


Pygmies, Sciapodes, Macrocephali, Argippeans:mythical Or Existent Tribes? wife of Akhenaten (14th century B.C.), who had a stone placed at the nape of her neck and then bound to provide the desired elongated head shape. In the Akhenaten period, the desired head shape was an inordinately elongated skull and pointed chin (Goodrich and Tutino, 2001).

the lees of it they make cakes, and eat them. For they have but few of smaller cattle, the pasture in their land not being good. They dwell each man under a tree, covering it in winter with a white felt cloth, but using not felt in summer. These people are wronged by no man, for they are said to be sacred; nor have they any weapon of war. These are they who judge in the quarrels between their neighbours; moreover, whatever banished man has taken refuge with them is wronged by none. They are called Argippeans.”

Ancient Egyptian pieces of art, as a wall bas-relief of the regal Akhenaten and Nefertiti revel in the attention of their daughters (Gore and Garrett, 2001) and a head in granite of one of Akhenaten’s (1353-1336 B.C.) daughters, sculptured in the Amarna style characterized by an intensive search for reality, and wall paintings, as one of a Rumses’ III (c. 1182-1151 B.C.) son (Carpiceci, 1994 pp 86, 121), depict long-headed noble individuals, like those described by Hippocrates.

The location of the habitation of Argippeans is illustrated in a map of The World according to Herodotus, 440 B.C., on the frontispiece of Herodotus Book I (Loeb Classical Library, 1975) in connection with the Black Sea, the Sea of Azov and the country of Scythians, Scythia. The latter was four-sided, located north to the Black Sea (Pontus) and the Sea of Azov, extending between the boundaries of the Danube river west, to the Don river east, a distance of four thousand furlongs (στάδια, i.e. 500 miles or 804 km), ), and vertically towards the inland four thousand furlongs too (Herodotus IV. 101). This area corresponds today to the Ukraine and South Russia.

Nothing is known more about the he long heads of the Siginni than what Strabo described.. Artificially shaped crania occur throughout the prehistoric periods in Cyprus, from the Aceramic Neolithic to Late Bronze Age (Lorentz, 2003). Intentional cranial deformation was widespread throughout the Americas (Goodrich and Tutino, 2001). Lest it be thought that the artists of Central and South America confined their efforts to reproducing copies of human craniofacial deformity in terracotta and stone, it is necessary to describe the ritual of skull binding in Amerindian neonates. Carefully carved cradles or head boards were used to mould the skull of the female infant into shapes which resemble the turricephaly, i.e. a high, pointed condition of the top of the skull, found in craniostosis in man (Poswillo, 1989).

According to Tomaschek (1930), the description of the Argippeans gestalt indicates a tribe of Mongoloid or old-Turkish people, more likely ancestors to Utighuren or Uighuren, who are relatives to Hunnen. The route of the Skythian commercial road, as well as the reference of southern inhabiting Issedones, indicate the location of the Argippeans habitation to be north to Tien Shan and south to the Altai Mountains. Through the so-called Dzungarische people, they should have traveled along the commercial route, to the early eastern-Asiatic peoples (for example Rhabarber) and reached to the Pontus (Black Sea).

Although Hippocrates didn’t mention where the race (people) of the Macrocephali lived, there is enough evidence in his text to support that the Longheads had environmentally induced, acquired macrocephaly. But to the recent knowledge, acquired characteristics are not inherited as Hippocrates thought.

The Argippeans were said to have flattened noses of Mongoloid type, but unlike the Mongols, who have coarse hair and sparse beards, they were all bald from their birth (male and female alike) and with long beards (Herodotus, IV.23). Zenobios (V 25), cited by Tomaschek (1930), attributed the aetiology of baldness to the consumption of salty water from the steppe. However, salty water is not listed among the nutritional or chemical causes of alopecia (Rook et al, 1986, vol. 3, pp 1980-3). The first component of the word ‘Argippeans’ Argippeans’ argim or argin could mean decent and it is referred to the eminent stature of the bald-headed (Tomaschek, 1930).. This appearance, which characterized all the people, originally was attributed to those of the Argippeans “who judged in the quarrels between their neighbours” (Herodotus, IV, 23). They completely shaved their head hair and this was a sign of noble origin and a delicate way of life, in contrast with the nomads who let their hair grow long (Tomaschek, 1930)..

The Argippeans Herodotus (IV.23), described people called Argippeans (Αργιππαίοι, Αργιππαίοι,, οι): ): “As far as the country of these Scythians all the aforesaid land is level and its soil is deep; but thereafter it is stony and rough. After a long passage through this rough country, there are men inhabiting the foothills of high mountains, who are said to be all bald from their birth (male and female alike) and snub-nosed and with long beards; they speak a tongue of their own, and wear Scythian raiment, and their fair comes from trees. The tree wherefrom they live is called “Pontic”; it is about the size of a fig tree, and bears a fruit as big as a bean, with a stone in it. When this fruit is ripe, they strain it through cloth, and a thick black liquid flows from it, which they call “aschu”; they lick this up or mix it with milk for drinking, and of the thickest of

Opinions stand (Tomaschek, Tomaschek, 1930) for the Argippeans being a tribe of Mongoloid or old-Turkish people, the location of their habitation to be north to Tien Shan and south to the Altai Mountains. 245

E %D]opoulou-Kyrkanidou Arguments on the existence of these tribes

+RPHULFVWRU\RIWhe battle between the cranes and the “pygmies”, who they said, were three spans tall.” (Strabo, Geography, 2.1. 9).

As early as in the sixth century B.C., the historian Hecataeus of Miletus (Fr. 328) was ironic with Homer saying that, the story about the fleeing of cranes from wintry storms and boundless rain, and with clamour flying toward the streams of Ocean, bringing slaughter and death to Pygmy men, amuses those who are unfamiliar with this story and increases the noise.

(span span:: σπιθαμή, 0.18 m) “But Megasthenes, going beyond all bounds to the realm of myth, speaks of people five spans long and three spans long, some without nostrils, having instead merely two breathing orifices above their mouths; and he says that it is the people three spans long that carry on war with the cranes...” (Strabo, Geography, 15.1.57). (“...carry on war”: the war to which Homer refers)

The opposite opinion and criticism of the Pygmies, the Macrocephali and other tribes, were expressed later by authors of the Hellenenistic Age, in spite of the rational explanations already provided by Aristotle about the Pygmies (History of Animals, VII, xii 597a 5-9) and by Hippocrates about the Macrocephali (Airs, Waters, Places, XIV. 1-28).

Conclusion From the analysis of the names, the description of physical characteristics, the geographic distribution and the presumptive diagnosis of the Pygmies, the Sciapodes, the Macrocephali and the Argippeans, we come to the conclusion that these have been existent tribes. More precisely, the Pygmies have lived, under the same name and in the same area, as referred to by Homer and Hesiod, for almost three thousand years. The Argippeans, apparently a Mongoloid tribe, have been living in about the same area as mentioned by Herodotus in the fifth century B.C. The Sciapodes, as sufferers of filariasis or non-filariasis endemic elephantiasis of the lower legs, are still found at a high prevalence in populations of affected areas of tropical Africa, as referred to by Hecataeus of Miletus, sixth century B.C., and Archippus, fifth-fourth th-fourth century B.C. The Macrocephali, hopefully, belong to the past, since the custom that was inducing their long head seems not to be practised anymore.

Apollodoros of Athens, the great grammarian of the second century B.C., in his Catalogue of Ships (FGrHist, 244 F 157), argues that it is evident, that Hesiod and Homer and other ancient authors weave in myths about the Macrocephali and the Pygmies and other monstrosities willingly, not from ignorance of the existing, but for moulding the impossible for the sake of talking marvels and of enjoyment. The geographer Strabo travelled widely. The Inhabited World according to Strabo is illustrated on the frontispiece of volume I of The Geography of Strabo (Loeb Classical Library, 1960). He comments on what had been written about the Pygmies by Homer and Hesiod, and about people of three and five spans long and other monstrosities, spoken by Megasthenes, a historian histor istorian ian fof ofthe the the4th-3rd 4th-3rd century B.C. (FGrHist 715 F 27).

Acknowledgments “… though the story is surely not told in ignorance of its local setting but rather in the guise of myth; and the same is true of the stories that Apollodorus cites from Hesiod and the other poets without even realising in what way he is comparing them with the stories in Homer. For he compares what Homer says about the Pontus and Egypt and charges him with ignorance, on the ground that, though he wanted to tell the truth, he did not do so, but in his ignorance stated as true what was not true. Yet no one could charge Hesiod with ignorance when he speaks of “men who are half-dog,” of “long-headed men” and of “Pygmies”; nRPRUHVKRXOGRQHFKDUJH+RPHUZLWK ignorance when he tells these mythical stories of his, one of which is that of these very Pygmies;” (Strabo, Geography, 1. 2. 35).

The numerous short passages in English from ancient Greek authors were reprinted by permission of the publishers and the Trustees of the Loeb Classical Library, Cambridge, Mass.: Harvard University Press. The Loeb Classical Library® is a registered trademark of the President and Fellows of Harvard College. I express my gratitude. References Archippus, Fr.53. T. Kock editor (1976) Comicorum Atticorum Fragmenta, H&S Hes Publishers, Utrecht, Netherlands. Reprint of the edition Leipzig,Teubner (1880-1888). Apollodoros Of Athens, Catalogue of Ships, FGrHist. 244 F 157 (D. Zeittafeln). Aristophanes, The Birds, 1553-1555. Edited and Translated by J Henderson BB. Loeb Classical Library Vol. III, Cambridge, Mass: Harvard University Press/London, England, second edition 2000. © 2000 by the President and Fellows of Harvard College. Aristotle, History of Animals, VII. xii 597a 5-9. Loeb Classical Library Books VII-X. Translated by DM Balme 1991. London: Heinemann/ Cambridge, Mass.: Harvard University Press, 1991.

“But especially do Deimachus and Megasthenes deserve to be distrusted. For they are the persons who tell us about the “men that sleep in their ears”, and the “men without mouths”, and “men without noses”; and about “men with one eye”, “men with long legs” “men with fingers turned backward”; and they revived, also, the 246

Pygmies, Sciapodes, Macrocephali, Argippeans:mythical Or Existent Tribes? (1843, 9th ed. of 1940 reprinted in 1990). New York: Oxford University Press. Lorentz, KO. 2003 Cultures of physical modifications: Child bodies in ancient Cyprus. Stanford J Archaeol, vol 2. Megasthenes, FGrHist 715 F 27. Moncada, GA. & Hendel, PM. 1985 Craniofacial deformities: An historical and etiological perspective. Plast Surg Nurs 5(1), 4-13 Omim (Online Mendelian Inheritance in Man), # 265850 PYGMY Poswillo, DE. 1989 Myths, masks and mechanism of facial deformity. Eur J Orthodont 1(1), 1-9. Price, EW. 1974 Endemic elephantiasis of the lower legs- natural history and clinical study. Trans R Soc Trop Med Hyg 68(1), 44-52. Pice, EW. 1976 The association of endemic elephantiasis of the lower legs in East Africa with soil derived from volcanic rocks. Trans R Soc Trop Med Hyg 70(4), 288-95. Price, Ew. & Bailey, D. 1984 Environmental factors in the etiology of endemic elephantiasis of the lower legs in tropical Africa. Trop Geogr Med 36(1), 1-5. Remy,, J-P. -P. P.. 2003 Κανίβαλοι εναντίον Πυγμαίων στο Κονγκό. Η ανείπωτη φρίκη του εμφυλίου πολέμου. ΤΟ ΒΗΜΑ, Κυριακή 9 Μαρτίου, σσ Α34-Α35. Rimoin,, Dl., ., Merimee,, Tj., ., Rabinowitz,, D.,., Cavali-sforza, -sforza, sforza,, Ll.. & Mckusick, VA, 1969 Peripheral subresponsiveness to human growth hormone in African pygmies. New Eng J Med 281, 1383-1388. Rook, A., Wilkinson, Ds., Ebling, Fjg., Champion, Rh., Burton, Jl. 1986 Textbook of Dermatology, fourth edition. Oxford: Blackwell Scientific Publications, vol. II pp. 1001-1003, vol. III, pp 1980- 1983. Strabo, Geography, 1. 2. 35; 2. 1. 9; 15. 1. 57. Loeb Classical Library Vol. I. Translated by HL Jones 1917, London: Heinemann/Cambridge, Mass.: Harvard University Press, 1969. Strabo, Geography, 7. 3. 6. Loeb Classical Library Vol. III. Translated by HL Jones 1924, London: Heinemann/ Cambridge, Mass.: Harvard University Press, 1983. Strabo, Geography, 15. 1. 57. Loeb Classical Library Vol VII, Translated by HL Jones 1930, London: Heinemann/ Cambridge, Mass.: Harvard University Press, 1966. Tomaschek, S. Argippaioi. In: Paulys RE (1883- 193028): Realencyclopadie der Classischen Altertumswissenschaft. Stuttgart: Druckenmuller, vol. II1, pp 719-721. The New Caxton Encyclopedia 1967. London: The Caxton Publishing Company Limited, vol.13, p 4311. Wayne, D. Melrose 2002 Lymphatic philariasis: new insights into an old disease. Int J Parasitol 32(8), 947-60.

Carpiceci, AC. 1994 Art and History of Egypt. Casa Editrice Bonechi: Florence, pp 86, 121. Chadwick, DH. 1995 Ndoki-Last place on earth. National Geographic 188 (1), 2-45. Diamond, JM. 1991 Why are pygmies small? Nature Nov 14; 354(6349): 111-2. Godley, AD.1921 Translator: Herodotus, Book IV. 23, 43, Loeb Classical Library II 1971, London: Heinemann / Cambridge, Mass.: Harvard University Press, pp 223, 241. Goodrich, JT. & Tutino, M. 2001 An annotated history of craniofacial surgery and intentional cranial deformation. Neurosurg Clin N Am 12(1), 45-68. Gore, R. & Garret, K. 2001 Pharaohs of the sun. National Geographic 199(4), 34-57. Hattori, Y, Vera, Jc, Rivas, Ci, Bersch, N, Bailey, Rc, Geffner, Me, Golde, Dw. 1996 Decreased insulin-like growth factor I receptor expression and function in immortalized African Pygmy T cells. J Clin Endocr Metab 81, 2257-2263. Hekataios Of Miletus FGrHist I F 327 Herodotus Book II. 12, 32, 33, 34. Loeb Classical Library Vol. I. Translated by AD Godley 1920. London: Heinemann / Cambridge, Mass.: Harvard University Press, 1975. Herodotus Book IV. 22, 23, 42, 43, 99, 101, 191. Loeb Classical Library Vol. II. Translated by AD Godley 1921. London: Heinemann / Cambridge, Mass.: Harvard University Press, 1971. Hesiod Catalogues of Women and Eoiae, Fr. 43, 44. In: Hesiod The Homeric Hymns and Homerica. Loeb Classical Library. Translated by HG Evelyn-White 1914. Appendix added 1936. Cambridge, Mass.: Harvard University Press/ London: Heinemann, 1982. Hippocrates Airs Waters Places, XIV. 1-28. Loeb Classical Library. Translated by WHS Jones 1923. London: Heinemann / Cambridge, Mass.: Harvard University Press, 1957. Homer The Iliad, 3. 3-6. Loeb Classical Library Vol. I, Books112. Translated by AT Murray 1924. Revised by WF Wyatt, Cambridge, Mass.: Harvard University Press/ London, England, second edition 1999. © 1999 by the President and Fellows of Harvard College. Kunze, J. & Nippert, I. 1986 Genetics and Malformations in Art. Berlin: Grosse Verlag, p. 44 fig.46, p. 59 fig. 71. Lawrence, BR. 1990 Sciapodes in tropical medicine. Trans R Soc Trop Med Hyg, 84(1), 174-175. Lesky, A. 1966 A History of Greek Literature. Translated by Willis J. and de Heer C. London: Methuen, pp. 2, 14, 91, 220, 307, 425, 426, 438, 486, 489, 548, 552, 787, 857, 890. Lexicon Iconographicum Mythologiae Classicae (LIMC) 1988 VII/2, pp. 466-486. Zurich and Munchen: Artemis Verlag. Liddell, Hg, Scott, R & Jones, Hs. 1990 A Greek-English Lexicon


THE SIGNIFICANCE OF FABRIC DIVERSITY IN NEOLITHIC KNOSSOS CERAMICS S. Dimitriadis Dept. of Mineralogy and Petrology, Aristotle Univ. 54124, Thessaloniki, Greece. sarantis @ Abstract: A deep trench was excavated during restorations carried out by the 23rd Ephorate of Prehistoric and Classical antiquities of Heraklion, Crete in the winter of 1997. It was cut, 8.5 m deep, in the northeastern sector of the central court of the Minoan palace of Knossos, close to the Room of the Medallions and uncovering the neolithic deposits of the site spanning from the Aceramic to the Late Neolithic. A study of all the neolithic material retrieved is underway. This paper presents the results of the petrographic analysis of the pottery found during the 1997 salvage excavation. The report focuses on the diversity of fabrics that characterize the EN I and EN II ceramics in particular, and discusses it in terms of technology and provenance. Since the stylistic study of the corresponding ceramics is not yet completed this report includes some preliminary observations only regarding pottery types and how they may or may not be related to the fabric types distinguished. Based on the nature of the a-plastic tempering materials, the examined ceramics were classified into three groups: One was tempered with various kinds of carbonate grains; a second with a variety other than carbonate rock and mineral fragments, and a third with approximately equal amounts of carbonate and non-carbonate grains. The matrix in the first group is either calcareous or noncalcareous. In the second and third group the matrix is always non-calcareous. According to mineralogical and microtextural characteristics of the tempering grains, varieties and sub-varieties of ceramic fabrics were distinguished within each of the above three groups. The spectrum of coexisting fabrics changes in the various occupation phases. Form, surface treatment and firing indicate that the potters were skilled; ceramics were fired at temperatures less than ~750 ˚C in a not fully oxidized atmosphere. The materials used are compatible with a Cretan origin for nearly all of these ceramics. However, no obvious potential sources for quite a number of these materials were found within the vicinity of Knossos (within a range of 15 km from it). Such sources exist however at more distant places; two which are most likely to host the actual sources used, are at map distances of no more than 30 km from Knossos. One is the area between Psiloritis and Talea ori, the other is mount Dikti. Based on this evidence we can argue that a considerable part of the neolithic ceramic assemblage of Knossos has been “imported” from areas (settlements) not far from it, with which Knossos could have established continuous relationships and a rather systematic exchange of artefacts. Περιληψη: Μία οχτώμιση μέτρων βάθους εκσκαφή που έγινε για τεχνικούς λόγους το χειμώνα του 1997 από την 23η Εφορία Προϊστορικών και Κλασσικών Αρχαιοτήτων στο βορειοανατολικό τμήμα της κεντρικής αυλής του Μινωικού ανακτόρου της Κνωσσού, διέτρησε το σύνολο των υποκείμενων νεολιθικών στρωμάτων που περιλαμβάνουν φάσεις από την Ακεραμική μέχρι τη Νεότερη Νεολιθική. Αν και η πλήρης μελέτη της κεραμικής που απέδωσε η εκσκαφή δεν έχει ακόμη περατωθεί, κάποια σημαντικά συμπεράσματα μπορούν να συναχθούν από την πετρογραφική μελέτη που έγινε σε 268 λεπτές τομές κεραμικών οστράκων, κυρίως από αποθέσεις που αποδίδονται σε φάσεις της Πρώιμης Νεολιθικής (ΕΝ Ι και ΕΝ ΙΙ). Διαπιστώθηκε μια ευρεία ποικιλότητα ως προς την κεραμική ύλη των οστράκων που εξετάστηκαν. Η ποικιλότητα αφορά κυρίως τα απλαστικά κοκκώδη ένθετα αλλά και τη λεπτόμαζα. Η τελευταία είναι είτε ασβεστιούχος, προερχόμενη από αρχικές μάργες, είτε σιδηροαργιλλούχος, προερχόμενη από αρχικές σιδηρομιγείς αργίλλους, είτε ενδιάμεσης σύστασης μεταξύ των δύο προηγουμένων. Τα απλαστικά κοκκώδη ένθετα είναι όμως εκείνα που η φύση τους χρησιμοποιήθηκε για μια επί υλικής βάσεως ταξινόμηση της κεραμικής και για το διαχωρισμό ομάδων και τύπων κεραμικής ύλης. Διακρίθηκαν τρεις ομάδες κεραμικής ύλης, μια στην οποία τα απλαστικά κοκκώδη ένθετα είναι σχεδόν αποκλειστικά θραύσματα διαφόρων ανθρακικών ορυκτών και πετρωμάτων, μια άλλη στην οποία αυτά είναι θραύσματα μη ανθρακικών πετρωμάτων, και μια τρίτη όπου ανθρακικά και μη ανθρακικά κοκκώδη ένθετα συμμετέχουν με περίπου ίσες αναλογίες. Στην πρώτη ομάδα η λεπτόμαζα μπορεί να είναι ασβεστιούχος, μη ασβεστιούχος ή ενδιάμεση. Στη δεύτερη και τρίτη ομάδα η λεπτόμαζα είναι πάντοτε μη ασβεστιούχος. Με βάση τα ορυκτολογικά και μικροϊστολογικά χαρακτηριστικά των απλαστικών ένθετων διακρίθηκαν τέσσερις κύριοι τύποι κεραμικής ύλης στην πρώτη ομάδα και ένας αριθμός παραπέρα ποικιλιών σε κάθε τύπο. Στη δεύτερη και τρίτη ομάδα αντιστοιχούν από ένας κύριος τύπος με αρκετές και στις δύο περιπτώσεις επί μέρους ποικιλίες. Διαπιστώθηκε μια αξιοπρόσεκτη χρονική μεταβολή του φάσματος των τύπων κεραμικής ύλης που συνυπάρχουν στις διαδοχικές φάσεις. Κυριότερες τεχνολογικές παρατηρήσεις είναι η διαπίστωση ότι οι κεραμείς της εποχής κατείχαν καλά την τέχνη τους, έψηναν σε όχι πολύ υψηλές θερμοκρασίες (≤ 750˚ C) και, κατά κανόνα, σε όχι πλήρως οξειδωτικές συνθήκες. Σε ό,τι αφορά την προέλευση των υλικών κατασκευής της κεραμικής που εξετάστηκε, με βάση γεωλογικά κριτήρια αυτή μπορεί κάλλιστα να είναι στο σύνολό της από το εσωτερικό της Κρήτης. Διαπιστώνεται εν τούτοις μια δυσκολία εντοπισμού πηγών για αρκετές και σημαντικές από αυτές τις πρώτες ύλες στην περιοχή της Κνωσσού, σε απόσταση μέχρι ~ 15 km από αυτήν. Πιθανές πηγές για σχεδόν όλες τις πρώτες ύλες ανιχνεύονται όμως σε πιο απομακρυσμένες από την Κνωσσό περιοχές της κεντρικής Κρήτης. Δύο τέτοιες, που θα μπορούσαν να εμπεριέχουν τις πραγματικές πηγές των υλικών που χρησιμοποιήθηκαν, βρίσκονται σε απόσταση όχι μεγαλύτερη από ~30 km από την Κνωσσό. Η πρώτη είναι η περιοχή μεταξύ του Ψηλορείτη και των Ταλέων ορέων και η δεύτερη στο όρος Δίκτη. Θεωρείται


S. Dimitriadis λοιπόν με βάση τις διαπιστώσεις αυτές ότι ένα σημαντικό μέρος της νεολιθικής κεραμικής της Κνωσσού μπορεί να είναι επείσακτο, προερχόμενο όμως από θέσεις παραγωγής (οικισμούς) όχι πολύ απομακρυσμένους από αυτήν και με τους οποίους είναι πιθανό να είχε η Κνωσσός συνεχείς και σταθερές σχέσεις και μια μάλλον συστηματική ανταλλαγή χρηστικών αντικειμένων και υλικών αγαθών.


layer 12. This new set of C14 dates from Knossos provides evidence for a sequence of human occupation of the site starting sometime round 7040 BC. There is however a significant chronological gap of 1400 years between the Aceramic layer no. 39 and the first stratum where pottery is present (no. 37) which is difficult to bridge with the intervening relatively thin layer 38. The reasons for such a chronological gap are not clear and it could be attributed to a number of events ranging from settlement abandonment to intentional removal of occupational debris between layers 39 and 37.

Restoration work was carried out in the winter of 1997 by the 23rd Ephorate of Prehistoric and Classical antiquities of Heraklion in the northeastern sector of the central court of the Minoan palace of Knossos close to the room of Medallions (fig. 1). It involved the excavation of a small trench, 8.5 meters deep, cutting through the Neolithic deposits and reaching the Aceramic strata. This excavation provided a unique opportunity to revisit the Neolithic occupation of Knossos. This paper presents the first results of the petrographic analysis of 268 thin sections from ceramic sherds retrieved during the 1997 salvage excavation. A short discussion follows on what are perceived as significant implications of these results in terms of technology and raw material provenance and on how they may contribute to the emerging picture of settlement existence and pottery production and exchange during the Early Neolithic in Crete.

The stylistic analysis of the ceramic material to date confirms the general stylistic characteristics identified by Furness (1953) and Evans (1971) for the Neolithic pottery of Knossos, as well as for Neolithic pottery excavated at earlier sites nearby (e.g. Katsambas). The ceramic material unearthed at Knossos in 1997 may be attributed to the phases identified by Furness and Evans and consequently the occupation layers identified correspond roughly to the strata distinguished by the previous excavators of the site. Thus, strata 37 to 31 (C14: 5365 – 4990 BC) correspond to Early Neolithic I as identified by Furness and Evans; strata 29 to 12 (C14: 4961 – 4789 BC) roughly coincide with the Early Neolithic II, and despite the absence of recent C14 dates the occupation layers above stratum 12 may be attributed to the Middle and Late Neolithic based on the general similarities of pottery styles (Appendix I).

The fragmentary condition of the ceramic material retrieved, and the fact that the stylistic analysis is not yet complete, do not allow detailed references to vessel shapes or surface decoration of pottery corresponding to the several fabric types identified to be made. This paper therefore focuses on the diversity of fabrics characterizing these ceramics and discusses it in terms of technology and provenance. The stratigraphic evidence: an overview

Methodology The 1997 excavation revealed a sequence of 39 distinct stratigraphic layers corresponding to different occupation phases covering the entire span of the Neolithic from the Aceramic to the Late Neolithic (fig. 2). These layers were numbered from 1 to 39 with 1 being the top layer (last neolithic occupation) and 39 the bottom (earliest occupation). Layer 37 at a depth of 7.5 meters is where ceramic material appears for the first time.

Samples for petrographic analyses were taken from sherds that had been selected on the basis of their stratigraphic relevance, macroscopically perceptible fabric differences, wall thickness, surface treatment and decoration. Vessel shape was not a primary selection criterion, the reason being that due to the fragmentary nature of the ceramic material vessel shapes were very difficult, in most cases impossible, to reconstruct. In a few cases samples were extracted from vessel fragments with use-ware marks (such as: sedimentation and sooting). The primary objective however was to collect samples representative on a material basis of all the different types of pottery identified for each distinct occupation layer in numbers proportional to their relative frequencies. Early Neolithic I and Early Neolithic II are better represented because of the extent of the occupation layers from which the samples could be taken. Middle Neolithic and Late Neolithic strata were thinner and yielded fewer diagnostic sherds. For this reason this paper focuses on Early Neolithic I and Early Neolithic II.

C14 dating was conducted at the NCSR Democritos, Athens, on samples of carbonized material and charred seeds. The series of 13 new radiocarbon C14 dates obtained (Appendix I) allowed a re-evaluation of the chronological sequence of the site which was presented in more detail by N. Efstratiou during the November 1999 Heraklion conference on the 100 years of excavation at Knossos. A brief outline of this sequence will be given here. The Aceramic occupation layer 39 can be dated to between 7040 and 6770 BC (calibrated). The succeeding layers 37 to 12 cover the period from 5365 to 4789 BC (calibrated). There are no C14 dates from layer 38 and from strata above 250

The Significance of Fabric Diversity in Neolithic Knossos Ceramics Two hundred and sixty eight sherds were collected, thin sectioned and examined under the polarizing microscope. In addition, a provisional geological survey of the Knossos area in conjunction with systematic collection of sands from streams within the vicinity was conducted so that any potential sources of the raw materials used in ceramic manufacture could be identified. The sand samples were resin impregnated, thin sectioned and then examined under the polarizing microscope. All thin section preparations and microscopic work were done at the Department of Mineralogy and Petrology of the University of Thessaloniki. Ceramic fabrics: general remarks Tempering appears to have been extensively practiced as is evident from the sherds of all the layers excavated in 1997. Three groups of fabrics can be distinguished on the basis of the material(s) used as temper: the first group has almost exclusively a-plastic tempering grains of various carbonate minerals and rocks, the second mainly grains of a variety of other than carbonate rocks and minerals, whereas the third is characterized by approximately equal amounts of carbonate and non-carbonate tempering grains. In none of these groups was evidence of thermal decomposition of the included carbonate constituents found, therefore in all probability the ceramics examined had been fired at temperatures of less than about 750º C. Firing, in most cases not fully oxidized, was nevertheless successful. Moreover, both the forming and the surface treatment of even the earliest ceramics unearthed denote significant technological knowledge and mastering of the art by these early potters, as has also be noted by Furness (1953) and Evans (1994).

Figure 1. Plan view of the area of the Palace of Knossos showing the position of the 1997 excavation (full circle) and the approximate outskirts of the Neolithic settlement by the end of: Aceramic (Acer.), Early Neolithic I (EN I) and Early Neolithic II (EN II). Compiled from figs 2 and 3 of Evans (1994).

Under the polarizing microscope careful examination at magnifications of at least 250× allows the distinction of three kinds of matrixes in the examined samples: calcareous (from original marls), non-calcareous (from original ferruginous mudstones) and semi-calcareous (from original ferruginous marlly clays). The less-calcareous matrixes are richer in admixtures of silt sized quartz and chert, whereas typically calcareous matrixes are nearly free of such admixtures. A-plastic carbonate tempering grains were combined with all three kinds of matrixes; non-carbonate tempering grains on the other hand were always combined with noncalcareous matrixes. It is possible to further distinguish specific fabric types by considering the detailed mineralogical and microtextural characteristics of the a-plastic tempering grains and the nature of the including matrix. Besides their technological significance, these fabric types in conjunction with geological data can provide clues as to the likely provenance

Figure 2. Section of the west face of the 1997 trench.


S. Dimitriadis

Photo 1: Carbonate grains of various kinds and sizes in calcareous matrix; layer 36; EN I; f.w. 15 mm, XP.

Photo 2: Partly recrystallized dolomitic grain in calcareous matrix; layer 34; EN I; f.w. 0.80 mm, PPL.

Photo 3: Fabric 1B; general view of a whole thin section; calcareous matrix; layer 34; EN I; f.w. 20 mm, XP.

Photo 4: Fabric 1B1; calcareous matrix; layer 34; EN I; f.w. 4 mm, XP.

Photo 5: a dolomite rhombohedron in fabric 1B1; layer 35; EN I; f.w. 0.4 mm, XP.

Photo 6: Fabric 1B3; calcareous matrix; layer 34; EN I; f.w. 2.3 mm, XP.

PLATE I. (f.w.: field width; PPL: plain polarized light; XP: crossed polars)


The Significance of Fabric Diversity in Neolithic Knossos Ceramics

Photo 7: Fabric 1B3; dolomitic monocrystalls and succrosic dolostone fragment; calcareous matrix; layer 34; EN I; f.w. 0.4 mm, XP.

Photo 8: Fabric 1C1; calcareous matrix; layer 30; EN I; f.w. 0.8 mm, PPL.

Photo 9: Fabric 1C2; non-calcareous matrix; layer 18; EN II; f.w. 2.3 mm, PPL.

Photo 10: Fabric 1C1; fragment of an aggregate of zoned dolomitic crystals; calcareous matrix; layer 32; EN I; f.w. 0.8 mm, PPL.

Photo 11: Fabric 1C1; conversion of dolomite to calcite via recrystallization; layer 32; EN I; f.w. 0.8 mm, PPL.

Photo 12: Fabric 1D; dolomitic monocrystals and cleaved sparry calcite; layer 29; EN II; f.w. 2.3 mm, XP.

PLATE II. (f.w.: field width; PPL: plain polarized light; XP: crossed polars)


S. Dimitriadis of the raw materials utilized. It must be stressed however that despite the compositional differences of the several fabric types identified in the Neolithic ceramic material of Knossos, many of these fabrics are to some degree related. Certain types are apparently kindred and, though uncommon, there are transitional fabrics bridging most of the compositional differences among the main fabric types.

common than 1B1 and was found only in Early Neolithic I layers. 1B3 was also found only in Early Neolithic I layers. Fabric 1C is characterized by combining a calcareous or non -calcareous matrix with monocrystalline, cleaved calcite fragments. 1C appears to be the most diachronic and the most commont among all Neolithic ceramic fabrics in Knossos. It is further subdivided into three varieties:

The fact that the area excavated in 1997 was very small hinders generalizations and strictly quantitative inferences. This study probed in time but not much in space. However, within these limitations we considered the mere presence, the consistently abundant presence, or the complete absence of each fabric type within and among stratigraphic levels as parameters indicating significant differences. Also relevant is the estimated quantitative participation of any one fabric in the spectrum of coexisting fabrics within each layer and the changes of this spectrum through time.

Variety 1C1 has calcareous matrix (Plate II, photo 8). It is missing from layers 37 and 36 but is very common in all the subsequent layers, that is continuously from sometime within the Early Neolithic I until the Late Neolithic. Varieties 1C2 and 1C3 have semi-calcareous and noncalcareous matrixes respectively (Plate II, photo 9). These varieties are missing from layers deeper (earlier) than 18, but become common from 18 upwards, that is after sometime within the Early Neolithic II.

Fabric types

The presence of quartz and chert, in addition to the fabrics sparry-calcite grains typical for 1C, is noticeably higher in 1C2, and more so in 1C3, than it is in 1C1.

Group 1. Fabrics tempered with carbonate grains This group includes four fabric types:

In many samples of nearly all the varieties of 1C, in addition to sparry-calcite there are also fragments of iron-stained, zoned or laminated aggregates of curve-faced dolomite crystals (Plate II, photo 10). Within these aggregates clear calcitic neoblasts have been developed in places (Plate II, photo 11). This is taken to be an indication that the crushed calcite crystals used as the main tempering material in 1C fabrics may have been taken from coarsely crystalline (recrystallized and in places de-dolomitized) dolostones.

Fabric 1A is tempered with grains of various kinds of carbonates (dark micritic, fine to medium sparitic, partly or completely dolomitized and/or recrystallized limestones, Plate I, photos 1, 2). This fabric can be further subdivided into two varieties: 1A1 with calcareous and 1A2 with non- calcareous matrix. Ceramics of the 1A1 variety are common in Early Neolithic I but scarcely present in later strata. Ceramics of the 1A2 fabric variety are present in Early Neolithic I and Early Neolithic II, and sporadically also in subsequent strata.

Fabric 1D is intermediate between 1B and 1C and bridges them by containing both kinds of tempering grains: fine rhombohedral monocrystalline dolomite plus larger cleaved sparry-calcite (Plate II, photo 12). This fabric first appears in late Early Neolithic 1 and, contrary to 1B and 1C which appear to be the two most common fabrics in the samples examined, its presence is sporadic.

Fabric 1B is easily recognized macroscopically, having a fine spotted appearance (white specks on a dark matrix, Plate I, photo 3). It is tempered with tiny, turbid, angular, subangular or subrounded grains, easily recognized as abraded rhombohedra of dolomite monocrystals (Plate I, photos 4, 5). Fabric 1B can be further subdivided into three varieties:

Group 2. Fabrics tempered with non-carbonate grains This “group” includes one main fabric (2E), which can however be further subdivided in a number of varieties and sub-varieties:

Variety 1B1 has calcareous matrix. Variety 1B2 has non-calcareous matrix and in addition contains also a lot of subrounded quartz grains apart from the dolomite monocrystals. Variety 1B3 has in most cases calcareous matrix and in addition to dolomite monocrystals contains also fragments of a polycrystalline, equigranular, sucrosic dolosparite (Plate I, photo 6, Plate II, photo 7).

Variety 2E1 is characterized by the presence of fragments of slate and calcareous or pelitic phyllites, with subordinate sandstone and quartzite, in addition to always present chert and quartz grains (Plate III, photos 13, 14). A few sparrycalcite carbonate grains are also present in some cases; they are most likely the products of mechanically disintegrated veined calcareous phyllites. The non-calcareous matrix in this fabric is of at least two different kinds or origins. One of them (fabric sub-variety 2E1a) is characterized by the conspicuous presence of fragments of siliceous sponge

A single representative of 1B1was found in layer 37; its presence however really starts from layer 35 and is continuously abundant in all Early Neolithic I and Early Neolithic II layers, gradually becoming less common in Middle and Late Neolithic layers. 1B2 is much less 254

The Significance of Fabric Diversity in Neolithic Knossos Ceramics

Photo 13: Fabric 2E1; fragments of schists and slates; layer 36; EN I; f.w. 2.3 mm, PPL.

Photo 14: Fabric 2E1; fragments of various schists; layer 35; EN I; f.w. 4.0 mm, PPL.

Photo 15: Sponge spicules (longitudinal section middle up, two cross sections right down) and Foraminifera (left up) in the matrix of fabric variety 2E1a; layer 35; EN I; f.w. 0.8 mm, PPL.

Photo 16: Oblique section through a sponge spicule; fabric 2E1a; layer 37; EN I; f.w. 0.4 mm, PPL.

Photo 17: Fabric 2E2; two serpentinite grains; layer 36; EN I; f.w. 0.8 mm, XP.

Photo 18: Fabric 2E2; a large greenschist fragment; layer 35; EN I; f.w. 0.8 mm, XP.

PLATE III (f.w.: field width; PPL: Plain polarized light; XP: crossed polars)


S. Dimitriadis

Photo 19: Fabric 2E2; two grains of altered basic rocks; layer 37; EN I; f.w. 0.8 mm, PPL.

Photo 20: Fabric 2E2; two grains of altered basic rocks and a sponge spicule (middle right) in the matrix; layer 35; EN I; f.w. 0.8 mm, PPL.

Photo 21: Fabric 2E3; layer 33; EN I; f.w. 20 mm, XP.

Photo 22: General view of the whole thin section of a sherd with fabric 3F; layer 35; EN I; f.w. 20 mm, XP.

Photo 23: Fabric 3F2; layer 35; EN I; f.w. 2.0 mm, XP.

Photo 24: Fabric 3F3; a granitic fragment (right down); layer 16; EN II; f.w. 0.8 mm, XP.

PLATE IV (f.w.: field width; PPL: plain polarized light; XP: crossed polars)


The Significance of Fabric Diversity in Neolithic Knossos Ceramics spicules and Foraminifera (Plate III, photos 15, 16), completely missing from the other kind of matrix (fabric sub-variety 2E1b). Both these are fairly common in the earlier part of Early Neolithic I but sporadically present in later phases.

photo 24). Few representatives of this uncommon variety were found in Early Neolithic I and in Early Neolithic II. Discussion Stratigraphic and technological aspects of fabric diversity in EN Knossos ceramics

Variety 2E2 has greenschist, serpentinite and altered basic volcanic fragments in addition to slaty, phyllitic and psammitic ones (Plate III, photos 17, 18 and Plate IV, photos 19, 20). Two varieties (2E2a and 2E2b) can be similarly distinguished on the basis of the presence or absence of sponge spicules and Foraminifera in the matrix. Fabric 2E2 is also common only in Early Neolithic I with a single representative found also in a Middle Neolithic layer.

The degree of fabric diversity of the Early Neolithic Knossos ceramics, as revealed in this study and summarized in fig. 3, is significantly wider than anything suggested in earlier studies of these ceramics (Furness 1953, Evans 1971, Noll 1982). Recently, Tomkins and Day (2001) re-examined Neolithic ceramics from earlier excavations in Knossos and they refer to a variety of fabrics featuring a compositional range comparable to the one found in this study. The material they examined, apparently covering the whole Neolithic site, must for this reason be more representative than the material of the 1997 spatially restricted excavation; yet, it seems that no important fabric or group of fabrics were missed in our sampling, an exception being a fabric tempered with blueschist fragments, noticed by Tomkins and Day (2001) but not encountered in the 1997 material.

Variety 2E3 is characterized by abundant quartz, quartzite, chert and various types of carbonate fragments, in addition to the tempering constituents of 2E2 (Plate IV, photo 21). This fabric is in fact transitional to the next group and could have been attributed to that as well. It was found primarily in Early Neolithic I layers and only scarcely in Early Neolithic II ones. Group 3. Fabrics tempered with approximately equal amounts of carbonate and non-carbonate

In general, Early Neolithic I ceramics appear to have a wider tempering material variation than Early Neolithic II ones. There also seems to exist a difference concerning the nature of matrixes (calcareous or non-calcareous) among similar fabrics from the earlier phases and the later ones; in the former calcareous matrixes dominate whereas in the latter non-calcareous matrixes appear to be more common.

This “group” includes just one main fabric (3F), which can be further subdivided into three varieties. All three have non-calcareous matrixes and the tempering grains are fragments of various rock types, similar or identical to the ones present in the other two fabric groups, primarily those of group 2, with strained quartzite being the most common. The commonly sub-rounded to rounded shape of the grains in this group implies tempering with stream or beach sand. Shell fragments, extremely rare in the other two groups, are occasionally seen in ceramics of this group, which are also relatively more porous and friable than those of the other two (Plate IV, photo 22).

In most Early Neolithic I fabrics sizable (a few mm) subangular to subrounded tempering grains are evident macroscopically. They apparently correspond to “fragments of gravel or grit” observed by Furness (1953: 103) in many Early Neolithic I sherds. These are in fact fragments of various kinds of carbonate rocks, schists, serpentinites, quartzites and metabasites.

Fabric variety 3F1 is tempered with fragments of quartzite plus various schists, plus metabasites and serpentinites, plus a variety of calcitic or dolomitic carbonate types (micritic, or sparitic, incipiently or intensely recrystallized). Only one representative of this subclass was recovered in Early Neolithic I; it is however more common in Early Neolithic II. It was also found in a single specimen from a Middle Neolithic layer.

Although fabric 1B1 can be found sporadically in later phases, it is nevertheless typical of both Early Neolithic I and Early Neolithic II. This fabric, with the characteristic fine spotted appearance, most likely corresponds to ceramics which Furness (1953: 103) described as showing “small white flecks” on their broken surfaces. She suggested that these were tempered with powdered gypsum from the “hill of Gypsades” near Knossos. However, the microscopic study has not identified any gypsum, powdered or not, intact or thermally transformed in any of the ceramics examined. The fine white “flecks” must correspond to the abraded dolomite monocrystals. These latter were either produced by grinding an equigranular dolosparitic rock or, alternatively, were detrital. In the latter case they may have been the sole constituents of dolomitic sands used for tempering or the constituents of raw dolomite bearing clays which were used untempered.

Fabric variety 3F2 is tempered with quartz and quartzite plus various (as above) carbonates (Plate IV, photo 23) and is apparently more common in Early Neolithic I than in Early Neolithic II. Fabric variety 3F3 is tempered with quartzite, feldspars and various carbonates, plus biotite or amphibole. Some polymineralic fragments are clearly granitic (Plate IV,


S. Dimitriadis In the case of fabric 1B3, typical of Early Neolithic I, the dolomite crystals constituting the polycrystalline sucrosic dolosparitic fragments are identical to the individual dolomite monocrystals dispersed in the same matrix (Plate I, photo 6, plate II, photo 7). One possibility is that this fabric was tempered with what produced the incomplete crushing and grinding of a sparitic dolostone; another is that its raw material was an immature sediment containing polycrystalline dolosparitic fragments plus the single-crystal detrital dolomite grains produced by the natural disintegration of the parental dolosparite. In any case, the raw material used in fabric 1B1 is a naturally or technologically more refined cognate of the raw material used in fabric 1B3; it may be relevant that fabric 1B3 appears to be more common than 1B1 in the earlier phases, disappearing later, whereas 1B1 persists much longer.

part of Early Neolithic II. However, starting sometime within the Early Neolithic II, (especially from layer 18 upwards) non-calcareous and semi-calcareous matrixes are becoming noticeably more common than calcareous ones. This trend seems to continue in the Middle Neolithic and Late Neolithic. It is not clear what brought about this change. Was it due to natural raw material restrictions or an intended technological shift, perhaps after realization that the non-calcareous clays and/or ceramic matrixes are more suitable for certain purposes? Group 2 fabrics are common in Early Neolithic I only. They dominate the bottom of the stratigraphic sequence (layers 37 and 36); they become less common in layer 35 after which they decline. This fading out of group 2 fabrics may be somehow related to the nearly simultaneous appearance and quick domination of group 1 fabrics (especially fabric 1C). Was a technological reorientation, a spatial reorganization within the settlement, or something more radical responsible for these changes?

Fabric 1C is macroscopically easily identifiable by the conspicuous presence of its yellowish to light brown angular (cleaved) crystalline calcite tempering grains, which were most likely produced by crushing a coarse sparitic, partly or completely dedolomitized carbonate rock or crystalline aggregate. 1C with its varieties is the most common fabric of all and the one with the more extended and continuous stratigraphic presence. However, it is missing from the bottom of the sequence (layers 37 and 36) in the material of the 1997 excavation. It makes a quantitatively impressive first appearance in layer 35 and remains the commonest fabric continuously after that. This may signify a spatial reorganization within the settlement, the beginning of massive introduction of a new ceramic product from elsewhere, or a major technological turning point by the “enthusiastic” approval of a new tempering material. The sudden appearance and subsequent “popularity” of this fabric may also be related to what Neolithic potters perceived as being advantageous qualities of a “novel” ceramic mixture, in conjunction perhaps with new uses of pottery vessels and new demands on pottery production as well as with the fact that sparry-calcite rocks can be easily ground to produce grit proper for tempering.

Group 3 fabrics (3F and its varieties) are not prevalent anywhere in the stratigraphic sequence and their rather discontinuous presence is mainly restricted to Early Neolithic I and Early Neolithic II. As we have already pointed out, tempering components found in fabrics of the other two groups are also participating in group 3 fabrics. The presence of group 3 fabrics therefore may indicate either proximity of the sources of the various raw materials seen in all the identified fabric types or, simply, the fact that there are many repetitions of various rock types across Crete. It may be worth adding here that a preliminary effort to study the correlation of certain variables such as fabrics to vessel shape, surface treatment and decoration (some results of which have already been presented by N. Efstratiou during the 1999 meeting at Heraklion) indicate that the correlations detected are not at all significant. It is therefore safe to say here that the choice of clays and tempers does not appear to be affected by the type of vessel shape or surface treatment or decoration at any one period. However, more detailed analysis of the material ascribed to each distinctive Neolithic occupation phase is necessary.

Fabric 1D, bridging fabrics 1B and 1C indicates that both tempering materials, dolomite monocrystals and sparrycalcite, were, at least under some circumstances, available to the same potters. We have already suggested that the sparry-calcite tempering grains were produced by crushing coarse-grained, recrystallized and partly dedolomitized carbonates; the coexistence of calcite and dolomite grains is therefore less of a problem. 1D “mixed tempering” is uncommon and appears later in relation to “single tempering” with either of the components; however, it is likely that this difference is due to compositional characteristics of the temper source(s) rather than due to changes in the technological choices of the potters.

Raw material provenance - implications The wide diversity of ceramic fabrics and the sequential changes of the spectrum of coexisting fabrics in Neolithic Knossos (described above and summarized in fig. 3) may be attributed to environmental, social and technological factors, acting independently or in combination. The most important among these, apart from changes in the spatial organization of the settlement, are pottery exchanges between coeval settlements and changes in technological practices. Despite all these however, fabric diversity could

In group 1 fabrics calcareous and non-calcareous matrixes are both present but the calcareous ones appear to be significantly more common during Early Neolithic I and 258

The Significance of Fabric Diversity in Neolithic Knossos Ceramics discussion however is to find out whether potential sources of the raw materials identified in the various ceramic fabrics exist in the vicinity of Knossos or not, in order to test the likelihood of local (strictly Knossian) origin of these ceramics. A thorough field survey, not undertaken until now, with this particular objective in mind is certainly necessary for a conclusive answer to the above question. In the course of this study however a provisional field survey was conducted and sand samples from streams running through the area of Knossos and discharging at the north coast, east and west of Heraklion, were collected and thin sectioned. Field observations within an area up to 15 km around Knossos in conjunction with the analyses of the collected sand samples and the study of the geological maps, failed to locate apparent potential sources for a significant number of the components seen in the examined ceramics. Especially the components of group 2, and to some extent group 3 fabrics, which geologically are most likely coming from the phyllite-quartzite and the ophiolitic formations of the island, are practically missing from Knossos area. Only some serpentinite and metabasic fragments were found in sands collected from Amoudara, ~10 km westnorthwest of Knossos. A small exposure of phyllitic rocks opposite Ano Arhanes was examined but the results were not convincing that this might have been the source of raw materials of group 2 and 3 fabrics. As regards group 1 fabrics (those with carbonate tempering components) the situation is not clear. Limestones and dolomites are plentifully exposed all over Crete and near Knossos. Therefore, utilization of local carbonate rocks by potters working in the vicinity of Knossos was not a problem. However, some characteristic types, such as equigranular sucrosic dolosparites or dolomitic sands, potential sources of the materials used in 1B fabric varieties, or coarsegrained and partly dedolomitized sparry-carbonates, potential sources of the tempering grains in 1C fabrics, were neither located in the surveyed area nor found in the sand samples collected from it. It is important to point out once again that 1B and especially 1C are the most common fabrics among the Neolithic ceramics of Knossos.

Figure 3. Approximate relative frequencies of fabric types coexisting in the Neolithic of Knossos (especially EN I and EN II).

not have existed if a wide variety of raw materials were not accessible to the potters. It may be interesting to compare the results of the present petrographic study with the results of similar studies made on ceramics of Bronze Age Knossos. However, since the origin of many of the latter is still disputed such a comparison will probably recycle uncertainties and will not lead to firm conclusions; for this reason such a comparison has not been attempted here. In the first place and in relation to provenance of the raw materials used in the ceramics studied it is important to point out that all could have been procured from within Crete. The various geological formations and the petrographic types existing in the island are fully capable providing all the mineral and rock components seen in these ceramics (for reviews on the geology of Crete see: Bonneau et al 1977, Fytrolakis 1980, Bonneau 1984, Hall et al 1984, Jacobshagen 1986, Fasoulas 2000. See also the 1:50,000 sheets of the geological map of Crete, IGME publications).

Calcareous clays (marls), the original material of the calcareous matrixes, are widespread in the Neogene formations in Crete and are plentifully exposed near Knossos. Non-calcareous clays might have been extracted from argillaceous intercalations within marls; however, the natural admixtures seen in many non-calcareous matrixes of the studied ceramics imply a connection of the original clays with the phyllite-quartzite and the ophiolithic formations of the island that are not exposed near Knossos. Non-calcareous clays with Foraminifera and sponge spicules, the original material of the matrix of many samples of group 2 fabrics, are also probably connected with sediments accompanying ophiolitic formations.

The above point is not to be taken as evidence that all the early Neolithic ceramics of Knossos are definitely Cretan in origin. Geological formations and rock types similar to the ones present in Crete are also present in other islands of the Southern Aegean as well as in mainland Greece and Anatolia. A Cretan origin is simply the least complicated interpretation, supported, albeit not proven by the mineralogical evidence. What matters more for this

On the basis of the evidence available so far it is therefore 259

S. Dimitriadis possible to argue that a significant part of the Neolithic ceramic assemblage of Knossos may not be of strictly local origin.

examples of “transitional fabrics” bridging nearly all the compositional gaps of the main fabric types. This has been interpreted to indicate proximity of the various raw material sources.

Fabric diversity, like the one seen in the Neolithic ceramics of Knossos, could be interpreted in two ways:

As we have already pointed out sources of materials used in certain important fabrics have not been located within an area extending up to ~ 15 km from Knossos. Only fabric 1A appears to be fully compatible with a local origin. If we consider a wider area however (extending up to ~ 30 km map distance from Knossos) we can find appropriate parental rock types from which potential sources for nearly all the components used in the ceramics studied could stem from. Within this range two areas stand out as having such rock types in close proximity to each other (fig. 4).

It may reflect the fact that the sources of the different raw materials used in these ceramics are geographically clearly separated and spread all over Crete (and perhaps beyond it), in which case fabric diversity is the result of an extended pottery exchange network among coexisting settlements. Alternatively, it may reflect pottery production within a single (or a few) geographically restricted area(s) where several geological formations and rock types coexist and form sources of a variety of raw materials available to the potters working there. In this latter case fabric diversity is not necessarily related to the existence of a widespread pottery exchange network.

The first is the region between Psiloritis and Talea ori in north central Crete, west of Knossos. Almost all the geological formations and rock types existing elsewhere in the island are also exposed there by mere geological coincidence and feed the intermountain valleys with their weathering products. All kinds of carbonates (limestones and dolomites of various ages and states of recrystallization, especially coarse grained calcitic and dolomitic marbles of the “plattenkalk series”, slates, phyllites and quartzites of the “phyllite-quartzite series”, and ophiolites with greenschists, serpentinites and basic volcanics are present in this region. Granitic rocks (fragments of which characterize fabric 3F3) have not been noticed in this region however or anywhere else in north-central Crete. They are known to exist in Mirabello district, eastern Crete, and in south central Crete where granitic rocks are components of the “Asterousia nappe”. Any (or both) of these two areas could therefore be the place(s) of origin of ceramics with 3F3 fabric. However, since a small isolated part of the Asterousia nappe is also exposed between Psiloritis and Talea ori, the possibility exists that granitic fragments may be present in the intermountain sediments of the above area as well; this has to be checked in a future survey.

Fabric diversity can however also be the result of the above two alternatives holding together. In such a case it is almost impossible to estimate the relative importance of each of them on the evidence of fabric characteristics only. Tomkins and Day (2001) considered the first of the above alternatives and interpreted the variability of ceramic fabrics in the material they studied in terms of settlement existence and pottery production and exchange in Early Neolithic Crete. They concluded that production and exchange of pottery during that period on the island must have been more complicated than anything previously suggested. They ascribed a local origin to a large proportion of the fabrics identified by them and a non-local (from elsewhere in north-central Crete, from further a-field or from outside Crete) to the rest. They did not consider the second of the above alternatives with its implications however and for this reason the discussion in the following will be focused on this. The results of the present study indicate that the amount of the non- strictly local ceramics that were probably “imported” to Knossos during the Early Neolithic may be large. Moreover, no positive evidence was found during our provisional survey that two very common fabrics (1C and 1B) are strictly local (manufactured within a range of less than ~ 15 km from Knossos).

The fabric with blueschist fragments noticed by Tomkins and Day (2001) indicates a more distant origin. It may come from Spilli-Preveli area in southwestern central Crete (the only place in the island where blueschists are known to exist). Alternatively, this fabric, as well as fabric 3F3, may both be of no Cretan origin at all. Ceramics with these two fabrics may have come from the Cyclades where granitic rocks and blueschists are plentifully exposed. Early contact between Knossos and the Aegean islands are anyway attested by the unretouched obsidian flakes (of possibly Melos and/or Giali origin) found in Early Neolithic I strata at Knossos (Evans 1994).

Based on published geological information in conjunction with the study of geological maps, a search was attempted to find those geological formations and parental rock types from which appropriate sources for the raw materials seen in the Neolithic ceramics of Knossos could have formed. It was assumed that areas where potential sources of as many as possible of the components seen in the described fabrics may coexist are more likely to be the places from where these components came. This is not an arbitrary assumption, since among the studied ceramics there are

The second area where rock types exist that may have been parental to most of the materials participating in the fabrics described in this work is that of Dikti mountain in east-central Crete, east south-east of Knossos. Coarse grained dolomitic carbonates of the “plattenkalk series” 260

The Significance of Fabric Diversity in Neolithic Knossos Ceramics b.

The apparent change of preference (?) of the fine clays used by potters from calcareous to non-calcareous ones sometime within Early Neolithic II.

Acknowledgments I wish to thank first of all A. Karetsou who as director of the 23rd Ephorate of Prehistoric and Classical Antiquities of Heraklion allowed access to the ceramic finds of the 1997 excavation. She and N. Efstratiou of the University of Thessaloniki generously passed the material for petrographic analysis to me. I also thank an anonymous reviewer for the constructive comments and D. Margomenou for her help during the sampling and for her comments on an earlier version of the manuscript.

Figure 4. Sketch map of the island of Crete. The lined areas are places where rock types that could be parental to the raw materials seen in the Neolithic ceramics of Knossos coexist in close proximity to each other.

and rocks of the “phyllite-quartzite series” are exposed there, whereas ophiolitic lithologies are exposed more to the south near Vianos.


The first of the above two areas, the one between Psiloritis and Talea ori, is perhaps somewhat nearer to Knossos; however, considering the geomorphology, access to it from Knossos is not as easy as to the second area.

Bonneau, M. 1984. Correlation of the Hellenide nappes in the south-east Aegean and their tectonic reconstruction. In: J. E. Dixon and A. H. F. Robertson (eds). The Geological Evolution of the Eastern Mediterranean. Geol. Soc. Spec. Publication No 17: 517-527. Blackwell Scientific Publications. Bonneau, M., Angelier, J., Epting, M. 1977. Reunion extraordinaire de la societe geologique de France en Crete. Bull. Soc. Geol. Fr. 19: 87-102. Evans, J. D. 1971. Neolithic Knossos: the growth of a settlement. PPS 37 (2): 95-117. Evans, J. D. 1994. The Early Millennia: Continuity and Change in a Farming Settlement. In: D. Evely, H. Hughes-Brock and N. MamiGgliano (eds). KNOSSOS A Labyrinth of History. The British School at Athens, 1-20. Fasoulas, Ch. 2000. Geological field guide of Crete. Natural History Museum of Crete Publication. pp104. Furness, A. 1953. The Neolithic pottery of Knossos. BSA 48: 94-134. Fytrolakis, N. 1980. The Geological Structure of Crete. Unpubl. Thesis (in Greek). Athens. pp147. Hall, R., Audley-Charles, M. G., Carter, D. J. 1984. The significance of Crete for the evolution of the Eastern Mediterranean. In: J. E. Dixon and A. H. F. Robertson (eds). The Geological Evolution of the Eastern Mediterannean. Geol. Soc. Special Publication No 17: 499-516 Blackwell Scientific Publications. I.G.M.E., Geological mapping of Greece, 1:50,000. Institute of Geology and Mineral Exploration, Athens. Jacobshagen, V. 1986. Geologie von Griechenland. BerlinStuttgart: Gebruder Borntraeger. Noll, W. 1982. Mineralogie und Technik der Keramiken Altkretas. N. Jb. Miner. Abh. 143 (2): 150-199. Tomkins, P. and Day, P. M. 2001. Production and exchange of the earliest ceramic vessels in the Aegean: a view from Early Neolithic Knossos, Crete. Antiquity 75: 259-260.

A third area where sources may be found of the materials seen especially in fabrics 1B and 1C is that of Lefka ori (“Plattenkalk series” and “Trypali unit”) in western Crete; this area is however significantly more distant from Knossos (map distance ~ 100 km) and access to it from Knossos in early Neolithic times should have been quite problematic. If valid, the origin of a large proportion of the early Neolithic ceramics found in Knossos from a single (or a few) production centre(s) existing at the time somewhere else in central Crete may signify well organized and systematic exchanges of goods and artifacts among the communities of the area rather than casual exchanges of pottery among distant communities. An aspect of the fabric diversity of the ceramics of Neolithic Knossos that certainly deserves attention in future work is the sequential changes in the spectrum of coexisting fabrics as revealed in this study. More pronounced among these are: a.

The change from the dominance of group 2 fabrics at the bottom of the sequence, where fabric 1C is practically absent, to the rather sudden appearance of 1C fabric, simultaneous with the decline of group 2 fabrics, whose dominance after that lasted the whole of the rest of the Neolithic. This change could reflect an important event. However, it may simply be due to a spatial reorganization within the settlement.


S. Dimitriadis Appendix I KNOSSOS 1957-60 vs. 1997 Comparison of Calibrated C14 Dates J.Evans strat./phase BC. vs. 1997 strat. X Aceramic 7292-6693 39 IX 7033-6649 6755-6426 IX Early EN I 6565-6235 VIII VII VI EN I 5611-5305 37 5411-4845 35 5312-4961 33 32 32 31 V Late EN I 5258-4858 IV EN II 4905-4787 29 4902-4617 28 4957-4502 28 24 14 III MN 4782-4502 12 4685-4363 4547-4357 II LN 4600-4253 I 4456-4265 4456-4255 LN/FN 4041-3529

BC. 7040-6770

5210-5060 5365-5305 5210-4990 5210-5000 5207-5037 5260-5070 4930-4800 4961-4852 4940-4800 5061-4945 4907-4799 4917-4789


EARLY HELLADIC II POTTERY FROM THEBES: AN INTEGRATED TYPOLOGICAL, TECHNOLOGICAL AND PROVENANCE STUDY J. Hilditch, E. Kiriatzi Fitch Laboratory, British School at Athens, Souidias 52, Athens 106 76, Greece [email protected]

K. Psaraki, V. Aravantinos 9th Ephorate of Prehistoric and Classical Antiquities, 32200 Thebes, Greece Abstract: This paper concerns the integrated archaeological and scientific study of Early Helladic II ceramics from Thebes, Boeotia, which aims to shed light on aspects of pottery production and consumption. Detailed typological study shows two co-existing ceramic traditions; a ‘local’ with vessel shapes derived from earlier local Helladic contexts and an ‘Anatolianising’, relating to the so-called ‘Lefkandi I’pottery and linking the Theban assemblage to other sites in the Aegean. The analytical programme, incorporating petrographic analysis and planned ICP-AES analysis, aims to achieve compositional, technological and, potentially, provenance characterisation of the products of the two ceramic traditions. Petrographic analysis of pottery and the procurement of, and experimentation with, geological samples from the broader area has laid the foundations for the consideration of ‘local production’ and the ‘Anatolianising’ vessel shapes; the latter as possible imports to the site, or as locally derived adaptations to a wider regional trend. In this way, the present study has significant social and cultural implications for understanding EHII society in Thebes but, furthermore, provides a new basis for assessing a more generalised phenomenon; the appearance of a small number of ‘Anatolianising’ drinking and pouring shapes in many sites across the Aegean during the later part of EHII. Περιληψη: Το άρθρο αυτό αφορά στη μελέτη της κεραμικής της Πρωτοελλαδικής ΙΙ από τη Θήβα, στη Βοιωτία, που ενσωματώνει την εφαρμογή τεχνικών από τις θετικές επιστήμες, με στόχο να φωτιστούν όψεις του φαινομένου της παραγωγής και κατανάλωσης της κεραμικής. Η λεπτομερής μελέτη της τυπολογίας υποδεικνύει δύο κεραμικές παραδόσεις που συνυπάρχουν: μια «τοπική», με σχήματα που προέρχονται από πρωιμότερα τοπικά Ελλαδικά σύνολα και μια «Ανατολίζουσα», που σχετίζεται με την επονομαζόμενη κεραμική «Λευκαντί Ι» και συνδέει το θηβαϊκό σύνολο με άλλες θέσεις στο Αιγαίο. Το αναλυτικό πρόγραμμα, το οποίο ενσωματώνει αρχικά πετρογραφική ανάλυση και στη συνέχεια ανάλυση φασματομετρίας πλάσματος επαγωγικής σύζευξης, έχει ως στόχο τον χαρακτηρισμό της σύστασης, της τεχνολογίας και, εν δυνάμει, της προέλευσης της κεραμικής των δύο διαφορετικών παραδόσεων. Η πετρογραφική ανάλυση της κεραμικής, μαζί με τη συλλογή και την πειραματική επεξεργασία γεωλογικών δειγμάτων από την ευρύτερη περιοχή, θέτουν τις βάσεις για τη μελέτη του φαινομένου της «τοπικής παραγωγής» αλλά και της κεραμικής με «ανατολίζοντα» σχήματα, με την τελευταία να θεωρείται πιθανώς επείσακτη ή τοπική απομίμηση μιας ευρύτερης τάσης. Έτσι, η παρούσα μελέτη έχει σημαντικές κοινωνικές και πολιτισμικές προεκτάσεις για την κατανόηση της κοινωνίας της Πρωτοελλαδικής ΙΙ στη Θήβα, αλλά, επιπλέον, παρέχει μια νέα βάση για την εκτίμηση ενός γενικότερου φαινομένου, δηλαδή την εμφάνιση μικρού αριθμού κεραμικών σχημάτων, που συνδέονται με την πόση ή το σερβίρισμα υγρών, σε αρκετές θέσεις στο Αιγαίο κατά τα τέλη της Πρωτοελλαδικής ΙΙ περιόδου.


In addition, this study also provides a new basis for assessing a more generalised phenomenon; the appearance of a small number of ‘Anatolianising’ drinking and pouring shapes in many sites across the Aegean during the later part of EHII (Renfrew 1972, Rutter 1979). Detailed typological study of the Early Helladic II ceramics by Psaraki (2004) has shown a remarkably standardised assemblage reflecting the introduction of ‘Anatolianising’ elements to the wider ‘local’ ceramic tradition found at Thebes. These ‘Anatolianising’ elements relate to the socalled ‘Lefkandi I’ pottery and therefore link the Theban assemblage to other sites in the Aegean.

This paper presents the preliminary petrographic analyses of Early Helladic II ceramics from recent excavations in Thebes, Boeotia1. The preliminary analyses presented here form part of an integrated scientific study based at the Fitch Laboratory, British School at Athens, aimed at shedding light on aspects of Theban pottery production and consumption. The present study has been designed to incorporate petrographic and chemical analyses, using inductively coupled plasma atomic emission spectroscopy (ICP-AES)2, in order to assess social and cultural implications for understanding EHII society in Thebes.

The ceramic assemblage The ceramics under study were recovered during recent excavations for the extension of the Thebes Archaeological Museum, which revealed a unique set of EHII architectural remains. Substantial deposits of late Early Helladic II pottery were found within a three-roomed apsidal house and its surroundings (Aravantinos 1997). The material assemblage from the site also included tools and a seal made from flint and obsidian, as well as a ‘hoard’ of metal

The excavations were carried out under the direction of Dr Vassilis Aravantinos, Ephor of the Θ’’ Ephoreia of Prehistoric and Classical Antiquities, whilst Ms Kiriaki Psaraki has undertaken the detailed study of the pottery for the final publication (in press). This study would not have been possible without the generous assistance and invaluable experience of the members of the Ephoreia in Thebes. The authors would also like to thank the Institute for Aegean Prehistory (INSTAP) for funding the current project. 2 The ICP-AES results will not be presented here due to the preliminary nature of the data at the time this paper was presented. 1


J. Hilditch, E. Kiriatzi, K. Psaraki and V. Aravantinos weapons. The site offers a unique insight into the late EHII horizon within Boeotia and the tightly constrained chronology provides an excellent opportunity to evaluate technological variability in ceramic production without having to consider differences through time. It is important to note that all of the vessels within the Theban assemblage are handmade. In general the late EH II pottery assemblage displays a high degree of standardisation in terms of vessel shape, with clear boundaries between each vessel type. However, earlier typological and macroscopic investigation revealed a number of different surface treatments for vessels of the same shape and variation in the inclusions of the vessel fabrics (Psaraki 2004). The main shapes of the assemblage include pithoi, pans, cooking pots, basins with t-rims, ellipsoid bowls with straight and in-curved rims, onehandled or Trojan cups (Fig. 1), two-handled cups, twohandled tankards (Fig. 2), as well as askoi, hydrias and pithamphorae. The assemblage contains predominantly shapes from the ‘local’ tradition, such as the two-handled tankard, the cooking pots, the basins with t-rims and the ellipsoid bowls with in-curved rim. In addition to these local shapes are vessels with varying degrees of, what has been described in other Lefkandi I group literature as ‘Anatolianising’ elements (Psaraki 1997).

Figure 1 – One handled ‘Trojan’ cup

Figure 2 – Two handled tankards with wavy rims (height approx. 17cm).

The one-handled cup (Fig. 1) is a very common shape during this late Early Bronze II horizon, found from the Cyclades to Thessaly, and has become known as the Trojan cup due to its similarity with vessels from the Troy III ceramic assemblage in Anatolia (Rutter 1979, Mellink 1986). Also present at Thebes are the shallow bowls with simple rims that Rutter has included within the Lefkandi I group (1979:3). An interesting shape within the late EHII Theban assemblage is the two-handled cup. This shape has strong parallels with the local two-handled tankard shape, yet the concept of a small two-handled drinking vessel finds parallels in the ‘Anatolianising’ bell-shaped cups. The smaller two-handled cup found at Thebes has a wavy rim, a characteristically Boeotian stylistic trait, similar to the rims found on the larger local tankard (Psaraki 2004). However, macroscopic investigation has recognised important similarities of fabric, surface treatment and body thickness between ‘local’ vessels and those containing ‘Anatolianising’ elements.

Figure 3 – Simplified geological map of the area around Thebes, with sampling locations outside of the city highlighted by white dots.

Research questions and methodology So, the initial investigations have raised three important questions that must be asked of the Theban ceramic assemblage: •

(height approx. 11cm).

relate directly to technological decisions during the production process? Do the ‘Anatolianising’ elements represent a foreign ceramic presence within the Thebes assemblage or a local adaptation of a wider regional influence?

A total of 145 samples were carefully selected to represent the existing variation within the EH II pottery assemblage. These samples were then made into thin sections for petrographic analyses. In addition to the petrographic analyses, refiring tests were carried out at 1050ºC in an oxidising atmosphere to qualitatively assess the variation of the clays within the assemblage. The results obtained have begun to help us answer the main questions surrounding the Theban ceramic assemblage.

What is the relationship between the mineralogy of the ceramic fabrics and the local geological environment - can we confidently propose a ‘local’ Theban centre of ceramic production? Does the macroscopic variation found within fabrics 264

Early Helladic II Pottery From Thebes: an Integrated Typological, Technological and Provenance Study The results

on the elemental compositions of these fabrics and the sub-groups just mentioned from ICP-AES analyses, currently underway, will give us a more secure means of characterising these vessels and determining their relationship to the local production system.

Local production Integral to the consideration of ‘local’ ceramic production is an understanding of the relationship between the fabric of a ceramic and the local geological environment. In the immediate area surrounding Thebes, several geological facies dominate the landscape (Fig. 3). Thebes itself sits upon a large unit of Pleistocene conglomerate, consisting of sandstones and red-loams, all derived from nearby underlying deposits of carbonates and the Shale-SandstoneChert Formation. Within a radius of 5km from Thebes also lie Pliocene sediments, a series of sandstones, marls and clays (all containing microfauna associated with lacustrine to brackish conditions) and a unit of limestone containing ophiolitic and Shale-Sandstone-Chert Formation lenses (I.G.M.E. 1970).

Technological decisions We have already established the compatibility of the majority of the fabrics analysed to the local geology, and therefore possible raw materials; however the technological decisions behind Theban ceramic production are a little more difficult to define. The problem again stems from the variability within the abundant source material, the Pleistocene conglomerate. The alluvial run-off from the conglomerate constitutes a substantial area between the unit itself and the Theban plain and would have contained the ideal association of coarse sand deposits, finer sediments and clays required for ceramic production. It is highly likely that a number of different clay compositions could have developed from the weathering of the conglomerate. It is even possible that these clays may have varied in composition within a single source due to the range of parent material present within such a small area. This is reinforced by the results of our geological sampling program, which was unable to link specific areas to specific associations of minerals.

As a means to identify the relationship between the ceramic fabrics and the local geology, an extensive geological sampling program of local clays, sediments and outcrops has been undertaken within the vicinity of Thebes, including locations such as the Mycenaean chamber tombs of Kastellia, and various streambeds throughout the city. The highly variable nature of the underlying Pleistocene conglomerate, evidenced by repeated sampling throughout the city itself, is reflected within the fabric variation of the largest defined petrographic group from our analyses. This group composes the majority of the samples studied and, although highly heterogeneous, is entirely compatible with the local geological environment, containing a mixture of serpentinites, sandstones, cherts and various forms of calcareous inclusions. A comparison of the geological samples collected against the archaeological fabrics studied can be seen in Figure 7. Vessels of coarse, medium and fine fabric fall within the ‘local’ group and, importantly, all of the vessel shapes sampled can also be found within this group. This suggests a significant level of local ceramic production within Thebes during the later part of the EH II period, in a range of fairly standardised shapes, fabrics and surface treatments.

Therefore it is not possible at this stage to distinguish between the use of a single or multiple source areas for the raw materials used in ceramic production within the vicinity of Thebes. Refiring tests carried out on all 145 samples within the study have revealed a range of clays with highly variable iron and calcium contents. Interestingly, no relationship could be defined between members of the same set. If we highlight the buff firing fabrics (14% of the total sample): this set contains a range of coarse through to very fine fabrics, but even the coarser buff fabrics cannot be distinguished by mineralogy or texture from other coarse vessels that display red or pink fabrics after refiring. The results of the fine ware analyses by ICP-AES will give us hopefully a clearer look at the relationship between these clays and the local geological environment.

There does not appear to be a significant amount of ‘foreign’ or imported pottery within the Theban ceramic assemblage at this time. In the coarse wares of the assemblage there are only two confirmed non-local vessels, unrelated to the local geological environment of Thebes, a pithamphora and a fragment of an unidentified brown burnished cup, both with a mica-schist derived fabric. The natural variation within the local sediments however has produced a number of petrological sub-groups based on the main mineralogical constituents of the Pleistocene conglomerate i.e. with their respective quantities and associations within the vessel fabrics. Almost 17% of the sample assemblage were fine wares and, as such, too fine to allow detailed characterisation by petrography. Additional information

The inability of the geological sampling to distinguish between specific compositional areas within the conglomerate or the alluvium may not be as disappointing as first thought. A tankard from the assemblage, showing what appears to be a coil-join, is highlighted in Figure 5. From the photomicrograph it is possible to see that the fabric types on either side of the join are different. Both fabrics are still geologically compatible with the local geological environment and also have parallels with other petrographic sub-groups defined within the study. This exploitation of multiple raw material sources within a single vessel and therefore, by extension, a single production unit, is difficult to interpret archaeologically and we must be cautious when considering how technologically meaningful this feature 265

J. Hilditch, E. Kiriatzi, K. Psaraki and V. Aravantinos

Figure 4 – Example of the fabric variation seen within the ellipsoid bowls with in-curved rim XPL, height = 24mm.

may be. For instance, there may have been a deliberate technological choice to use two distinct clays due to slightly different mechanical properties of the raw clays with regards to shaping. Perhaps the potter was unaware that the raw clay used to continue the vessel had a slightly different composition or, if they did, then perhaps this difference was immaterial to the vessel making process and another factor, such as shape or surface treatment, was the defining characteristic of the vessel for the potter. Technological decisions made with regard to coarseness of fabric are clearly illustrated by the initial petrographic analyses, which have highlighted an obvious distinction between vessel shapes of coarse and fine fabrics. At the coarse end of the spectrum the pithoi, pans and cooking pots form a coherent group, whilst the finest fabrics encountered belong predominantly to the one- or twohandled tankards/cups. However the analyses also reveal a continuum between the coarse and fine wares, highlighted in particular by the pronounced fabric variation within the ellipsoid bowls with in-curved rims (Fig. 4). The same also applies to askoi, jugs, hydria and pithamphora shapes of the assemblage. The fabric variation seen within the bowls reveals a very flexible approach towards paste preparation and may indicate more than one unit of production within Thebes during the late EHII period. If we consider the ‘local’ fine and medium fabrics alongside the coarser examples, then it is possible to recognise the addition of coarse alluvial sediments derived from the Pleistocene conglomerate as temper.

Figure 5 – Sample 81, a tankard, showing two different types of clays. XPL, width = 15mm.

the Anatolianising phenomenon, represent only 1% of the ceramic assemblage. They have predominantly fine fabrics but petrographic analyses have been able to identify 9 of the 12 cups in our sample as mineralogically compatible with the local geological environment of Thebes. The remaining 3 cups, of noticeably finer fabric, have insufficient inclusions to enable petrographic characterisation, so we await the results of the ICP-AES analyses to determine the relationship between these vessels and the local suite of fabrics. It is possible at this stage though to argue that the majority of these vessels represent locally derived Boeotian adaptations of the ‘Anatolianising’ drinking/table ware phenomena.

‘Anatolianising’ wares The consideration of our third and final question, the status of the vessels with ‘Anatolianising’ elements within the ceramic production system at Thebes, has only been permitted by combining typological analyses with the petrographic results and the geological sampling program. Two main possibilities were initially recognised through the typological analysis. Firstly, that these vessels represent a foreign presence at Thebes, either by import or through the introduction of a new ceramic technology, or alternatively, that these vessels represent locally derived adaptations to a wider regional trend at this time.

The shallow ellipsoid bowls compose 4% of the ceramic assemblage at Thebes and they form an interesting picture in terms of fabric. The three coarsest samples within our study are similar in terms of their inclusions i.e. the coarseness of the grains and their relative proportions within the matrix. However, as Figure 6 demonstrates, they have different vessel wall widths, exhibit varying fabric densities and appear to consist of different clay matrices. There is a mixture of buff and red firing vessels within this shallow bowl category. The coarse inclusions are all compatible with the local geological environment of Thebes, and the variation in clay composition can

The Trojan cups, the vessels most strongly associated with 266

Early Helladic II Pottery From Thebes: an Integrated Typological, Technological and Provenance Study

Figure 6 – Comparison of shallow ellipsoid bowls with simple rim (the central sample is a buff fabric). XPL, height = 24mm.

Figure 7.a

Figure 7.b

Figure 7.d

Figure 7.c

Figure 7.e

Figure 7 – Comparison of geological vs. archaeological fabrics. Geological sampling helped to assess the compatibility of local raw materials to the proposed local fabric: a) One handled cup (S02/64), b) Silt sample (G02/02), c) Alluvial sand (G03/02), d) Coarse alluvial deposit (G07/02), e) Pithos (S02/02) XPL, width = 3.7mm.

also be demonstrated in other vessel categories from the site, such as the pithoi. These samples fit the local fabric criteria and, as such, can be confidently ascribed to a local production unit (or units). Whether the difference in clay composition reflects the existence of different production units within Thebes at this time is uncertain. The remaining two samples are of very fine fabric and are difficult to characterise petrographically. They also display higher levels of optical activity within the matrix, indicative of lower firing temperatures than their coarser counterparts. These samples are included within the current ICP-AES analysis program so we can gain more insight to the relationship between these vessels and the local fabric suite.

tradition, particularly with regards to the two handled tankards and the wavy rim feature, but the concept of a small two handled drinking vessel within this period is derived from the ‘Anatolianising’ phenomenon at the end of the Early Bronze II period. The six examples within our study are all of a medium to fine fabric with red firing clays; however the inclusions of these vessels appear to vary from the coarse local fabrics. There are much lower amounts of serpentinites, in association with higher levels of quartz and feldspars. This is a general feature of the medium to fine fabrics present at Thebes and as such there is nothing within these fabrics to discount provenance from the immediate vicinity of Thebes. This observation strengthens the possibility of more limited technological processing of raw materials for some, if not all, of the medium and fine vessels within the assemblage. The two handled cup however appears to stem from entirely local technological traditions.

The final category to discuss under this question is that of the two handled cups. As mentioned earlier, these shapes are more stylistically grounded in the wider ‘local’ 267

J. Hilditch, E. Kiriatzi, K. Psaraki and V. Aravantinos Conclusions

with many sites at which Lefkandi I type material has been found, there is no complete package of ‘Anatolianising’ vessels at Thebes. Instead, as illustrated by our study, local communities appear to have chosen a variety of these core shapes and adapted them to their own requirements. This case study has shown the importance of concentrating on local technological choices in order to gain a more detailed understanding of the regional ‘Anatolianising’ phenomenon. We believe that the application of this locally focused approach would be a valuable tool in the study of other ceramic assemblages at Lefkandi I/Kastri group sites, in order to better understand the nature of the ‘Anatolianising’ phenomenon across the Aegean during the end of the Early Bronze II period.

To conclude, this paper has drawn on the results of the petrographic analysis, combined with refiring tests and geological sampling, in order to characterise, and thereby assess the relationship between the variations of style and fabric within the Theban ceramic assemblage. The geological samples taken not only have clarified the results of this project but also will form an important background for future analytical work in this region. The bulk of the fabrics analysed, representing the majority of the pottery of the assemblage, are compatible with the local geology. We can, therefore, assume a broadly local production (within the wider vicinity of Thebes) for the majority of the pottery. There is significant variation, both in the ceramic fabrics and the geological samples/units, which will hopefully be better characterised and understood, with respect to raw material variability, production technologies and choices, through the intended future combination of petrographic and chemical analyses.

References Aravantinos, V., 1997, “Θήβα”, ArchDeltio, 52, 354-359. Broodbank, C., 2000, An Island Archaeology of the Early Cyclades, Cambridge: Cambridge University Press. Mellink, M., 1986, The Early Bronze Age in West Anatolia: Aegean and Asiatic Correlations, In The End of the Early Bronze Age in the Aegean (G. Cadogan ed.), Leiden, p.139152. Popham, M.R. and Sackett, L.H. (eds), 1968, Excavations at Lefkandi, Euboea, 1964-1966, London. Psaraki, K., 2004. A New EHII Pottery Assemblage from Thebes, In Die Frühbronzezeit in Griechenland. Mit Ausnahme von Kreta (Eva Alram-Stern ed.), Die Ägäische Frühzeit, Forschungsbericht 1975-2002, II, 2, Band Teil 1 und 2. Veröffentlichungen der Mykenischen Kommission, Wien, 1259-1265. Psaraki, K., 1997, Μεταβολές στην Παραγωγή και Χρήση της Κεραμικής της ΠΕΧ στη Βοιωτία και την Εύβοια.. Το πρόβλημα της κεραμικής “Λευκαντί Ι”, Μεταπτυχιακή Εργασία. Αριστοτέλειο Πανεπιστήμιο Θεσσαλονίκης, Θεσσαλονίκη. Renfrew, C.A., 1972, The Emergence of Civilisation: The Cyclades and the Aegean in the Third Millenium BC, London. Rutter, J.B., 1979, Ceramic Change in the Aegean Early Bronze Age: the Kastri Group, Lefkandi I and Lerna IV, Institute of Archaeology, Occ. Paper 5, University of California, LA. Sampson, A., 1993, Manika and Mainland Greece in EHIII: Ceramic Evidence for Relations with the Aegean and Anatolia, In Wace and Blegen: Pottery as Evidence for Trade in the Aegean Bronze Age 1939-1989 (Zerner, C with P. Zerner and J. Winder eds.), Amsterdam, 159-164. I.G.M.E., 1970, Geological Map of Thebes, Scale = 1:50,000.

This study has added to general discussion of the Lefkandi I phenomenon by approaching the problem from a technological perspective. Through studying the technology of the Theban ceramic assemblage we can clearly see in the late EHII levels a continuation of ‘local’ ceramic tradition. This very much echoes the words of Broodbank (2000:310) who talks about the “Kastri group shapes form[ing] a statistically minor component that slots into existing repertoires”. This can be seen with respect to the Lefkandi I horizons on the mainland and surrounding regions also (Manika – Sampson 1993; Lefkandi – Popham & Sackett 1968). There was no radical change in the technology of production, merely the introduction of certain stylistic developments into the established local tradition. Our approach can be used to evaluate theories on migration or invasion that have surrounded the appearance of Lefkandi I/ Kastri Group horizons over the years. In the case of Thebes, our study does not support the pervasive appearance of a brand new culture or people into Boeotia at this time; it merely suggests the partial incorporation of a regional trend into a strong well-defined local culture. As


REGIONAL PRODUCTION CENTERS OF IRON AGE POTTERY WARES IN PHILISTIA: A PROVENANCE STUDY EMPLOYING INDUCTIVELY-COUPLED PLASMA (ICP-AES AND ICP-MS) D. Ben-Shlomo The Institute of Archaeology, The Hebrew University of Jerusalem [email protected] Abstract: The paper will discuss a provenance research, still in progress, applying chemical fingerprinting (ICP) and involving two case studies of different pottery groups sharing geographical and cultural characteristics. The study is an examination the so called “Ashdod Ware” pottery of the Iron IIA (10th-9th centuries BC, also termed LPDW), principally appearing in the Philistine cities of Tel Ashdod and Tel es-Safi-Gath. The Iron IIA “Ashdod Ware” was hardly studied and the question is whether this pottery is local as well and if its production centers can be identified. According to the initial results of the pilot study at least two production centers of LPDW can be identified within Philistia, representing Ashdod and Tell es-Safi/Gath. While some inner trade between these sites is evident there is no evidence for importation of this ware from other regions. The competence of ICP for the purpose of intra-regional provenance can be demonstrated and it is suggested that ICP can be widely used for provenance determination of pottery while the existing NAA data bank can still be used to some extent for comparison with ICP chemical profiles. Περιληψη: Η παρούσα εργασία συζητά τη διερεύνηση προέλευσης δύο διαφορετικών ομάδων κεραμικής με κοινά γεωγραφικά και πολιτιστικά χαρακτηριστικά, εφαρμόζοντας Επαγόμενη Φασματοσκοπία Σύζευξης Πλάσματος (ICP). Η μελέτη αφορά στην επονομαζόμενη κεραμική τύπου “Ashdod” της εποχής του Σιδήρου IIA (10ος-9ος αιώνας π.Χ.), η οποία εμφανίζεται κυρίως στις πόλεις των Φιλισταίων Tel Ashdod και Tel es-Safi-Gath. Ο τύπος αυτός κεραμικής δεν έχει μελετηθεί εκτεταμένα και το ερώτημα είναι αν πρόκειται για ντόπια κεραμική και αν μπορούν να εντοπιστούν τα κέντρα παραγωγής της. Σύμφωνα με τα πρώτα αποτελέσματα της πιλοτικής μελέτης, μπορούν να εντοπιστούν δύο τουλάχιστον κέντρα παραγωγής στη Φιλισταία, στο Ashdod και το Tell es-Safi/Gath. Παρ΄όλο που υπάρχει ένδειξη για εσωτερικό εμπόριο μεταξύ αυτών των δύο θέσεων, δεν υπάρχει καμία ένδειξη για εισαγωγή της εν λόγω κεραμικής από άλλες περιοχές. Επιδεικνύονται οι δυνατότητες της τεχνικής IPC για τη διερεύνηση προέλευσης της συγκεκριμένης κεραμικής και προτείνεται η ευρύτερη χρήση της τεχνικής στη διερεύνηση προέλευσης κεραμικής ενώ η υπάρχουσα τράπεζα δεδομένων NAA μπορεί επίσης να χρησιμοποιείται σε κάποια έκταση για σύγκριση με τα προφίλς ICP.


Bichrome pottery. This pottery is the so-called ‘Ashdod Ware’ (Dothan and Freedman 1967, pp.130-1), and what is now suggested to be defined as ‘Late Philistine Decorated Ware’ (LPDW) (Ben-Shlomo et al. 2004). In several sites as at the cemetery of Azor and Tell Qasile (see fig. 1) this ware is preceded by Philistine vessel types decorated in red slip. The red-slipped Philistine vessels can be viewed as a short-lived intermediate style of the Philistine decorated pottery.

The study of decorated pottery wares has long been an important component in the evaluation of the material culture of Philistia and its surroundings during the Iron Age. The appearance of the Philistine pottery in the beginning of the 12th century BCE is seen to be one of the main components of a new material culture brought to the southern coastal region of Israel by immigrants from the Aegean region and Cyprus. Decorated Philistine wares of the Iron Age I (12th-11th centuries BCE) are divided into an earlier Monochrome style and a later Bichrome style; both styles show high resemblance in form, decoration and technology to contemporary Mycenaean wares from the Aegean and Cyprus (Dothan 1982, Killebrew 2000). Nevertheless, this pottery has been shown to be locally made in Philistia according to INAA and petrography (Asaro et al. 1971, Gunneweg et al. 1986, Killebrew 1998). The pottery wares appear principally in the Philistine cities of Tel Ashdod, Tel Miqne-Ekron, Tell es-Safi/Gath and Ashkelon (Gaza has hardly been excavated) (see fig. 1).

The LPDW is red-slipped, hand-burnished and decorated in black and white, and appears mostly in closed forms (fig. 2). The ware was initially identified at Tel Ashdod and therefore was termed ‘Ashdod Ware’. Over the years, vessels decorated in the LPDW style have been uncovered at several sites of southern Israel in strata dating, in most cases, to the Iron Age IIA. Recently, a large group of these vessels was found in Tell es-Safi/Gath (Maeir 2001, Maeir and Ehrlich 2001). In this study the provenance of the late Philistine decorated ware is examined, attempting to identify its production center/centers. As some of the shapes of this ware are similar to contemporary Phoenician and Cypriote vessels, sources outside Philistia may be suggested for this pottery as well. Moreover, as the intra-regional provenance of

During the subsequent period, the Iron Age IIA (10th9th centuries BCE), a distinct type of decorated pottery is also confined mainly to the Philistine cities and their surroundings (fig. 1); it seems to replace the Philistine 269

D. Ben-Shlomo

Figure 2: Major forms of LPDW pottery.

meticulous vertical hand burnish on all or part of the vessel and, in some cases, wheel burnishing on the lower part of the vessel (see figs. 3-4). The painted decoration is applied over the red slip, and includes, in most cases, horizontal black bands on various parts of the vessels and/or white bands in between them. The visual appearance of the clay is relatively uniform, having a reddish-brown or orange color. However, several of the vessels seem to be made of finer, better-levigated clay; these vessels often also have a more lustrous burnish. An important additional characteristic of the LPDW pottery is the specific vessel forms on which the above decoration and surface treatment appears. These forms could be classified as a coastal type (Gitin 1998, pp.1657). The coastal forms most typical to the LPDW pottery are closed kraters, amphorae and jugs (fig. 2). Kraters include large specimens with globular-hemispherical body, a wide vertical neck and two loop handles connecting the rim to the shoulder (fig. 2:2); in many cases they have ‘pinched ring bases’. An additional krater type has a globular body a short vertical neck and two slanting horizontal handles. This krater has parallels from Phoenician sites, e.g. from Akhziv (Dayagi-Mendels 2002, p.118, fig. 5.4, K1), the Lebanese coast (Sa‘ideh 1966 fig. 17, Chapman 1972 figs. 20 & 32) and Cyprus (Karageorghis 1970 pls. CCIV:5, 1824). Another krater type decorated in the LPDW style is a hole-mouth krater with a globular body, rounded base and triangular or ridged rim (fig. 2:1). The sack shaped jars (fig. 2:3), which are very typical of the coastal area in the Iron Age IIA-B, appear in several cases with vertical burnished red slip and, at times, with black and white bands on the neck and/or shoulder. One of the most prominent LPDW forms is the globular amphora (figs. 2:5, 3). These vessels have a globular body, wide long vertical neck, ridged or triangular rim, pinched ring base and two loop handles. The vertical burnish occurs on the neck. LPDW jug types

Figure 1: Map of Israel showing sites with LPDW (Ashdod Ware) pottery.

both the early and late Philistine wares has not yet been sufficiently established, material from the different Philistine cities should be compared. The provenance study of the pottery includes petrography and chemical fingerprinting. This paper presents chiefly the preliminary chemical results. The investigation of production centers of the various Philistine wares can shed light on cultural-political and economical aspects of this region in this period, and possibly aid in solving some chronological problems relating to the relevant strata as well. Such a provenance study will further clarify the evolution of pottery production technology and trade patterns during the Iron Age. Late Philistine decorated ware (‘ashdod ware’) The predominant feature characterizing the LPDW pottery is the readily identifiable surface treatment and decoration technique. The typical surface treatment includes red slip, 270

Regional Production Centers of Iron Age Pottery Wares in Philistia As noted above, the LPDW appears primarily in Philistia and particularly at Tel Ashdod and Tell es-Safi/Gath. Fewer examples come from other sites in Philistia, the inner Shephelah and Judea, while isolated examples come from more northern sites (fig. 1). This distribution pattern appears to indicate that the LPDW pottery was produced in Philistia and was traded primarily within the southern land of Israel. Accordingly, the most probable production centers are Tel Ashdod and Tell es-Safi/Gath. On the other hand, it is possible that the Philistine cities were the major consumers of this ware, which was imported from somewhere outside Philistia. The second possibility cannot be ruled out, since from a stylistic point of view, the LPDW pottery displays both Cypriote and Phoenician characteristics in addition to its Philistine ones. Recently LPDW from Ashkelon was suggested to be imported from the Phoenician coast (Master 2001, p.35). However, this was based on a petrographic analysis of a single, very small sherd. The most noticeable archaeological evidence of LPDW pottery production in Philistia comes from Tel Ashdod. At Ashdod, several of these vessels were found within pottery kilns: in Area M, Strata Xb-Xa (Dothan and Porath 1982, pl. II:3-4; Loci 7027-7029, 7202, 7271, 7277, 7072, 7083) and in Area D Strata VIIIb-a (Dothan 1971 pp.89-92, Loci 1167, 1169). The kilns in Area M are relatively small and irregularly shaped (Dothan and Porath 1982 pp.7,16, plans 3-5). Kiln 7083 from Stratum Xa has a rectangular shape, is 1 m long, having several subdivisions (Dothan and Porath 1982 p.16, Pl. II:3-4). This latter kiln in particular contained several complete LPDW vessels. In Area D, the kilns are more regularly shaped. They are built within pits as rectangular one-chambered brick structures measuring 3 x 1.5 m (Dothan 1971 pp.89-92). Usually the roofs of the kilns were not preserved, but these were probably vaulted ones. The group of kilns in Area D can be understood as a pottery workshop that was constructed on the southwestern edges of the city. Although not many Iron Age II kilns have been published, the Ashdod kilns seem to belong to a rather common type (see Killebrew 1996 p.153, Table 1, Wood 1991 pp.26-33). Recently, in a salvage excavation conducted by Y. Israel of the Israel Antiquities Authority, an Iron Age II kiln site near Tell es-Safi/Gath was unearthed (Kfar Menahem). The site includes five or six well-built rectangular kilns having a uniform size and orientation (fig. 5). According to the preliminary review of the pottery, the site may be dated to the 8th century BCE (I wish to thank Y. Israel for the information on this site).

Figure 3: A LPDW amphora from Tel Ashdod (courtesy of IAA).

Previous research on the pottery from Tel Ashdod included a provenance study of pottery from Area M employing INAA (Perlman and Asaro 1982). Of the 85 vessels that were sampled, 17 were defined ‘Black on Red’, the apparent equivalent of our LPDW group (Perlman and Asaro 1982 p.73) (rather than the Cypro-Phoenician Black on Red Ware). Most of these vessels were attributed to chemical Groups Ia, Ib and Ic; of them, eight belong to

Figure 4: A LPDW jug from Tel Ashdod (courtesy of IAA).

include a jug with large globular body, short narrow neck and pinched ring base (figs. 2:4, 4). Other rarer forms of LPDW are rounded and carinated bowls, amphoriskoi, globular jugs, juglets, and zoomorphic vessels and kernoi. 271

D. Ben-Shlomo used for pottery making (idem, p.66). However, loess clay from the nearby alluvial bed of the Lachish River could also have been used. If large amounts of sand temper were not added to the loess matrix then it would be difficult to distinguish between clay originating in the vicinity of Tell es-Safi or of Tel Ashdod. Method Sampling In the first stage of the research fifty samples were selected for chemical analysis (table 2). The samples came from three sites: Tell es-Safi/Gath, Tel Ashdod and Kfar Menahem. About twenty-five of the samples belong to the late LPDW, while most of the other samples are used as reference material. A group of vessels found in pottery kilns at Tel Ashdod were used as reference for Ashdod while a group of common pottery forms from Tell es-Safi/ Gath, excluding cooking pots and storage jars, were used as reference for Gath. Cooking pots are excluded as they are usually made of a different clay recipe, while storage jars are more likely to be transported. The samples from the kiln site near Tell es-Safi (Kfar Menahem), though dated some 50-100 years later, were also anticipated to produce a reference group for Gath. In the second stage of the research an additional 150 samples will be subjected to chemical analysis. Most of these samples will come from the four Philistine cities excavated (Ashdod, Ekron, Ashkelon and Gath) and from other regional sites, and belong to the various Philistine decorated wares of the Iron Age. All of the samples are also analyzed by thin section petrography.

Figure 5: An Iron Age II pottery kiln near Tell es-Safi/Gath (Kfar Menahem) (courtesy of IAA and Y. Israel).

Group Ia. All three groups were chemically identified as being local to Ashdod, and can be normalized to the major Group Ia. Perlman and Asaro also noted that within the Group Ia samples, the eight ‘Black on Red’ examples are more compactly grouped. They interpret this as a result of a production line of a specific potter. The archaeometric results indicating that the LPDW pottery was made at Ashdod is not surprising, since, as noted above, some of these vessels were found in pottery kilns at Ashdod. However, interestingly, all of these vessels are local to Ashdod, including those with stronger Cypro/ Phoenician stylistic characteristics. Given the results from Ashdod and the recently excavated assemblage from other sites, especially at Tell es-Safi/Gath, further analysis was undertaken. The goal was to examine the provenance of this ware at other sites and to try to identify patterns of trade.

Analytical method

The geological setting

The chemical method used was Inductively-Coupled Plasma (ICP). Relatively few provenance studies of ancient pottery utilizing the ICP method have been published so far, although compositional results obtained by this method are comparable in their quality to INAA results, without the requirement of a nuclear facility at hand. In this method the powdered samples are dissolved in an acid solution and injected into high-temperature Argon plasma. The atomic emission (ICP-AES) and mass (ICP-MS) spectrums are measured in two separate instruments.

The major sites sampled in this study, Tell es-Safi/Gath and Tel Ashdod, are about 20 km apart and illustrate several differences in their geological settings. The geological setting of Tell es-Safi/Gath and its vicinity (Buchbinder 1969, Kfar Menahem sheet, Sneh, Bartov and Rosensaft 1998 Sheet 2) includes the Pleshet formation (a kurkar conglomerate) which is present on the upper parts of the tell, with other elements of conglomerate formations containing limestone, chalk and chert, appearing as nodules (the Adulam, Zor’a-Maresha and Ahuzam formations). Nevertheless, alluvial brown soils are abundant in the Ha’elah riverbed, which is located just to the north of the tell. It would appear that most of the pottery was made from this type of soil.

Sample preparation and analysis procedure A sample of c. 200-300 mg of powder was extracted from each vessel by a diamond drill. The powder was dried for 12 hours in an oven temperature of 110° C and then weighed precisely to a 200 mg (or 100 mg if too small) sample. The samples were dissolved in an acid cocktail of 5 ml of HF 40%, 5ml of HNO3 2.5 %, and 1 ml of HClO4. This mixture was left in covered beakers for at least 12 hrs and then heated at 100° C for 2 hours; thence the solution was dried at 230° C. Finally the sample was leached in a

In comparison, the geological setting of Tel Ashdod is more uniform. Bakler noted four types of sediments in the region of Tel Ashdod: kurkar, hamra, more recent alluvium and sand dunes (1982 p.65). He notes that Hamra was the main outcrop on the tel and probably the main raw material 272

Regional Production Centers of Iron Age Pottery Wares in Philistia HNO3 1% solution. The solution was mixed in a glass flask with HNO3 1% and an internal standard (10 ppb in solution of Bi, In, Re and Ru—which are very rare elements in soils) to a volume of 100ml: a final dilution of 1:500. The internal standard is used chiefly to monitor the drift of the instrument. Somewhat similar procedures were described in several other cases in which ICP was used for chemical fingerprinting of ceramics and other materials (Hart et al. 1987; Beith et al. 1988; Porat et al. 1991; Ponting and Segal 1998).

organic (P, Ca, Sr), or by post-depositional processes (P, Na, K, Rb). Several elements are susceptible to contamination from drilling or other apparatus (Ni, Cu, Zn), while other elements may behave erratically for unknown reasons (Ba). It is important to select elements having high-precision compositional values, as the elemental composition within a limited geographical zone does not vary considerably (Adan-Bayewitz 1993 pp.43-45, Hein et al. 1999). Some of the rare earth elements are considered to be redundant in obtaining chemical fingerprinting, as they behave in a similar compositional pattern (Glascock 1992 p.16). Nevertheless, as the total group of elements used for chemical grouping in this case is large and the errors are not too big, almost all of the elements were eventually used. Ca and Ba were excluded on account of high variability and Cu, Ni and Zn were excluded on account of contamination.

The analytic equipment used for the analyses was located in the Bristol University Geochemical Laboratory and included an ICP MS—VG PQ2 turbo (Plasma Quad) and an ICPAES—YJ Ultima II (sequential) (Plasma temperature— 8000° K; max vacuum—1.7 x 10-7 mbar). Each run included twenty-five samples, five synthetic standards (standards 1-5, with graded 0%-100% concentrations), a blank (only 1% HNO3), a wash (5% HNO3 in ICP-MS only), several international rock standards (as BE-N, BHVO3, JA2, JB1, JB3, GA), and an in-house standard. If the values of internal standards change within the duration of the run, the instrument compensates accordingly automatically. While the calibration in ICP-MS was made according to the five synthetic standards—the cocktail of elements prepared according to the elements obtained and their required concentration range—in ICP-AES it is made according to both synthetic and international rock standards. The rock standards are furthermore used in both cases as quality controls.

The groups were formed by Principal Component Factor Analysis (PCA) performed on all thirty-three ‘good’ elements. The results are illustrated according to two major components (Factors 1 and 2) reflecting 68.3% of the variance (fig. 6). Hierarchical Cluster Analysis (squared Euclidean distance; “distance between groups” clustering method) was also used (SPSS version 11 and Excel programs). The grouping according to PCA seemed to be more sensitive than the grouping based on Cluster Analysis; therefore, tentatively, the chemical groups were defined on the basis of PCA. It should be noted, though, that the general results of both statistical methods were similar. The groups’ mean concentrations and standard deviations of a selection of twenty-six elements are presented in table 1.

Major and minor elements (Na, Mg, Al, P, K, Ca, Ti, Mn and Fe) were obtained by ICP Atomic Emission Spectroscopy (AES), while the heavier trace elements and rare earth elements (Sc, V, Cr, Co, Ni, Cu, Zn, Rb, Sr, Y, Cs, Ba, La, Ce, Pr, Nd, Sm, Eu, Gd, Tb, Dy, Ho, Er, Tm, Yb, Lu, Hf, Ta, Th and U) were obtained by ICP Mass Spectrometry (MS). In order to get better precision and accuracy of the results, the heavier elements (La-Lu) and U and Th were run separately from the lighter ones. Several of the elements may be considered as “problematic” for chemical fingerprinting; this may be due to various reasons (see discussion below). Therefore, not all the elements obtained were used for the chemical grouping; thirtythree elements were chosen excluding Ca, Ba, Cu, Ni, and Zn. The compositional results (log10 transformed) were subjected to Principal Component Analysis (PCA) and Cluster Analysis (squared Euclidean distance and Ward clustering method). Results and discussion

It is reasonable that pottery produced in a limited geographical region may have a relatively similar chemical composition, although each production center will still maintain its own ‘chemical fingerprint’. This can clearly be seen by the dispersion of the chemical composition of the fifty samples (fig. 6). The majority of the samples are clustered together in one group including 23 of 25 LPDW samples. This group could be possibly divided into three initial subgroups, Groups 1-3. In addition 5-6 outliers are of a different composition (KM1, KM2, KM10, SF39, SF40 and possibly SF5). As the main cluster includes reference material from Ashdod it can be concluded that the LPDW were produced in Philistia. For comparison sample SF39, a Cypriote White Slip milk bowl, and SF40, a typical example of a Cypro-Phoenician juglet, both have significantly different chemical profiles (see fig. 6 and table 1).

Several elements should be excluded or related to with caution in the statistical procedure defining the chemical profiles of the pottery. A few of the elements are obtained with a lower degree of precision or accuracy (Ta, Ti, Cr, Gd); other elements were influenced by the potter’s intervention—the use of temper, mostly calcareous or

The initial sub groups defined by PCA, Groups 1-3 (figs. 7-8; table 1, columns A, B, D) are relatively close in their composition. Group 1 contains five Tell es-Safi LPDW vessels and non-decorated “common” types (Table 1 Column A); Group 2 contains all LPDW vessels from Ashdod and three LPDW vessels from Tell es-safi (Column 273

D. Ben-Shlomo B) and Group 3 (Column D) is located, chemically speaking, in between Groups 1 and 2. Outliers (samples not falling in a group) are one LPDW sample from Tell es-Safi (SF5), three Kfar Menahem samples (KM1, KM2, KM10), and a Cypro-Phoenician Black on Red vessel from Tell es-Safi (SF40) (Columns G-J). The compositions of elements such as Co, Ti, Mn, Sc and U is overlapping between the groups, while other elements as Cr, Y, La, Ce, Sm and Pr show a certain difference. This difference is not great (about 15%-20% of the mean values) but is nevertheless distinct. It should be noted that although all the elements in Group 2 are of a lower concentration than in Group 1, the factor varies considerably between the elements, and therefore they cannot be considered as a single group with different dilution factors. However, it can be noticed that nearly all concentrations in Group 3 are about 10% lower than those in Group 1. Such a dilution may result in a higher silica concentration in these vessels (possibly reflecting an addition of more quartz temper). Thus, when a 0.91 dilution factor is applied (Column E) Groups 1 and 3 may be considered identical (see figs. 78). Consequently, all members of the initial Group 3 were reassigned to Group 1. Two samples coming from kilns in Area D of Tel Ashdod (AS5-6) are somewhat similar in elemental composition to Group 1. They may represent a slightly different clay recipe at Tel Ashdod, and were thus denoted as Group 2A (Table 1: Column C, fig. 7). The high variability of calcium contents in all the groups is probably due to the addition of calcite-rich temper to the clay.

Figure 6: Scatter plot showing samples from Ashdod (AS= Δ), Tell es-Safi/Gath (SF= ○) and Kfar Menahem (KM= □), according to the general chemical composition reflected by the two major components (PCA). Filled symbols indicate LPDW vessels.

Several of the samples from Ashdod were also previousely analyzed by INAA (Perlman and Asaro 1982, table 1, Group Ia). The elemental results relevant are presented (table 1 Column F). Although the grouping made according to both ICP and INAA analyses is compatible, there are variations in some of the specific elemental concentrations. While differences between concentrations of Ca, Fe, Al, Ti, Co and Th are within the measurement errors, other elements illustrate a significant difference between ICP and NAA measurements. These are Cr, U, Sc, Mn and especially Hf and Lu. Several recently published studies (Hein et al. 2002, Tsolakidou and Kilikoglou 2002) discuss compatibility between different chemical fingerprinting methods and will hopefully achieve some manner of calibration between these two methods.

Figure 7: ‘Zoomed in’ PCA scatter plot according to initial chemical groups, showing samples after dilution factor applied on Group 3.

According to Cluster Analysis two main clusters are formed (Clusters 1 and 2); moreover, eight outliers are identified. In addition to the six outliers identified by PCA (KM1, KM2, KM10, SF5, SF39, SF40), the clustering produced another two outliers—KM8 and SF4. The additional outliers, though, are relatively close to Cluster 2. Cluster 1 and Cluster 2 were formed in an additional run omitting the outliers. They included sixteen and six LPDW vessels respectively. Cluster 1 includes all samples from Tell esSafi, while Cluster 2 includes most samples of Ashdod and PCA Group 2. There are, however four samples of PCA

Figure 8: ‘Zoomed in‘ PCA scatter plot according to pottery wares and site of samples, after dilution factor applied on Group 3.



1.02 0.21 0.26 0.04 0.20 0.07 0.79 76.48 5.16 2.58 0.66 0.20 0.49 1.08 1.72 0.20 0.81 0.03 0.29 0.02 0.02 0.06 0.05 0.01 0.42 0.22

23.87 5.22 4.72 6.82 17.28 11.55 6.79 10.83 6.44 14.06 3.81 19.51 13.86 4.07 3.00 3.07 3.06 2.50 5.19 2.24 2.53 3.04 2.89 2.99 5.97 15.42

5.09 4.64 5.96 0.67 1.04 0.66 13.35 780.37 93.62 19.51 19.24 1.39 3.61 29.46 61.84 7.19 28.90 1.41 5.95 0.77 0.82 2.25 1.98 0.28 7.42 1.65

%CV Mean (2) 2A 4.81 4.30 5.73 0.62 1.17 0.62 11.74 737.51 84.77 17.75 19.24 1.10 4.27 29.43 62.19 7.03 28.38 1.38 6.06 0.74 0.80 2.20 1.96 0.29 7.94 1.32

0.87 0.30 0.33 0.04 0.22 0.08 1.20 61.30 6.39 1.03 0.76 0.24 0.72 1.46 2.10 0.25 0.92 0.04 0.27 0.02 0.02 0.05 0.04 0.01 0.55 0.23

Mean SD (15) “Group 3”

Col. C Col. D

18.17 7.07 5.69 5.98 19.06 12.48 10.26 8.31 7.54 5.78 3.97 21.48 16.95 4.97 3.38 3.59 3.23 3.00 4.43 3.00 2.61 2.40 2.13 3.91 6.88 17.30

5.29 4.73 6.29 0.68 1.29 0.68 12.91 810.45 93.15 19.51 21.15 1.21 4.69 32.35 68.35 7.73 31.18 1.52 6.66 0.81 0.88 2.42 2.15 0.32 8.72 1.45

%CV Mean 3/0.91 5.0 +-0.7 4.05 +-0.15 5.52 +-0.2 0.667 +-.033 --0.67+-0.058 13.38 +-0.6 809 +-32 116 +-6 18.70 +-0.91 --1.65+-0.26 11.48+-1.02 30.4 +-1.1 ------------------0.433 +-0.03 7.78 +-0.31 2.12 +-0.1

5.34 5.51 11.09 0.54 2.48 0.27 21.63 455.37 117.96 20.55 15.75 3.91 2.58 26.10 58.01 6.82 27.78 1.40 5.70 0.75 0.77 2.15 1.81 0.26 7.76 2.73

0.01 0.06 0.24 0.002 0.04 0.002 0.39 1.68 3.02 0.53 0.16 0.05 0.12 0.33 1.00 0.09 0.60 0.03 0.15 0.02 0.03 0.03 0.06 0.01 0.28 0.04

Col. G Group Ia* KM2 +/-

Col. E Col. F

24.18 3.10 4.20 0.27 0.18 0.30 8.95 200.92 135.89 9.09 29.48 0.37 1.97 25.69 35.52 5.57 23.35 1.15 4.87 0.71 0.91 2.66 2.29 0.36 4.51 6.06

0.15 0.01 0.03 0.002 0.01 0.002 0.41 0.56 3.96 0.21 0.58 0.02 0.11 0.51 0.50 0.08 0.53 0.002 0.08 0.02 0.01 0.08 0.04 0.01 0.11 0.03

KM10 +/-

Col. H

3.37 3.60 4.86 0.48 0.95 0.49 10.65 598.00 65.21 14.90 16.20 0.69 3.62 24.08 49.15 5.58 22.41 1.14 4.98 0.59 0.65 1.77 1.62 0.26 5.78 1.17


Col. I

Table 1: Elemental compositions of the chemical groups formed by PCA.

Concentrations in PPM unless noted otherwise; SD= standard deviation; %CV= (SDx100)/Mean; *= Perlman and Asaro 1982, Table 2 (27 samples).

4.26 4.10 5.48 0.61 1.18 0.62 11.66 706.41 80.15 18.33 17.44 1.02 3.50 26.47 57.31 6.61 26.57 1.31 5.67 0.71 0.76 2.12 1.87 0.28 7.10 1.43

40.93 6.32 4.35 6.89 16.62 20.64 10.38 8.84 6.45 9.59 6.29 19.70 12.86 3.89 3.47 3.22 2.85 2.86 3.42 4.81 6.34 7.61 6.85 7.47 4.16 10.96

2.17 0.30 0.27 0.05 0.20 0.15 1.33 67.37 6.04 1.82 1.35 0.25 0.60 1.23 2.29 0.25 0.88 0.04 0.22 0.04 0.06 0.19 0.15 0.02 0.35 0.17

Ca% Fe% Al% Ti% K% Na% Sc Mn Cr Co Y Cs Hf La Ce Pr Nd Eu Sm Tb Ho Er Yb Lu Th U

5.31 4.77 6.29 0.67 1.21 0.73 12.87 762.50 93.62 18.97 21.47 1.29 4.69 31.73 65.91 7.66 30.97 1.52 6.56 0.83 0.90 2.48 2.23 0.33 8.53 1.55

%CV Mean SD (16) Group 2

Col. B

Element Mean SD (ppm) (11) Group 1

Col. A

0.01 0.03 0.04 0.002 0.03 0.01 0.38 5.07 3.80 0.41 0.60 0.04 0.15 0.43 0.78 0.08 0.38 0.01 0.09 0.01 0.002 0.02 0.01 0.01 0.20 0.06


2.50 4.58 6.98 0.43 1.09 0.31 11.65 1082.52 87.33 23.88 16.86 2.27 3.78 28.34 68.42 6.87 27.61 1.35 6.23 0.71 0.74 1.92 1.73 0.25 9.83 1.03

0.01 0.02 0.11 0.01 0.01 0.002 0.63 13.13 2.33 0.88 0.71 0.10 0.13 0.26 0.96 0.03 0.31 0.03 0.17 0.002 0.02 0.04 0.13 0.01 0.17 0.01

SF40 +/-

Col. J

Regional Production Centers of Iron Age Pottery Wares in Philistia

D. Ben-Shlomo

Sample Site Number

Basket No.


Stratum Type

AS1 AS2 AS3 AS4 AS5 AS6 AS7 AS8 AS9 AS10 AS11 AS12 AS13 AS14 KM1 KM2 KM3 KM4 KM8 KM9 KM10 KM13 SF1 SF2 SF3 SF4 SF5 SF6 SF7 SF8 SF9 SF10 SF11 SF12 SF13 SF14 SF15 SF16 SF17 SF18 SF19 SF24 SF25 SF26 SF28 SF33 SF34 SF37 SF39 SF40

K1479/1 H3341/1 H853/1 H3055/1 D4950/1 D4891/1 M785/3 M788/1 M987/1 M827/10 M822/1 M827/9 M827/2 M827/3 181 225 173 142 158 162 302 187 450225 450306 440175 330111 220331 220244 320146 220387 420141 450342 420291 420142 230150 230285 440170 290397 220356 220346 450743 450163 510250/5 540022/53 510370/5 320198 220389 510370/3 490211 210119/40

Ashdod VI: Pl. 53:6 Ashdod VI: Pl. 53:7 Ashdod VI: Pl. 60:4 Ashdod VI: Pl. 63:7 Ashdod II-III: Fig. 41:7 Ashdod II-III: Fig. 41:22 Ashdod IV: Fig. 3:1 Ashdod IV: Fig. 3:17 Ashdod IV: Fig. 4:3 Ashdod IV: Fig. 7: 3 Ashdod IV: Fig. 7:12 Ashdod IV: Fig. 7:13 Ashdod IV: Fig. 7:14 Ashdod IV: Fig. 8:2


Ashdod Ashdod Ashdod Ashdod Ashdod Ashdod Ashdod Ashdod Ashdod Ashdod Ashdod Ashdod Ashdod Ashdod KM KM KM KM KM KM KM KM Safi Safi Safi Safi Safi Safi Safi Safi Safi Safi Safi Safi Safi Safi Safi Safi Safi Safi Safi Safi Safi Safi Safi Safi Safi Safi Safi Safi

as TS: Fig. 7:2 as TS: Fig. 10:9 as TS: Fig. 7:1 as TS: Fig. 6:12 as TS: Fig. 8:5 as TS: Fig. 7:6 as TS: Fig. 6:6 as TS: Fig. 6:6

4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4

as TS: Fig. 9:9


TS: Fig. 6:13 TS: Fig. 6:14 TS: Fig. 8:3 TS: Fig. 8:2 TS: Fig. 9:8

TS: Fig. 8:8

TS: Fig. 8:1 TS: Fig. 8:4


Chemical Group (ICP) Bottle LPDW 2 AM1 LPDW 2 JG2 LPDW 2 KR3 LPDW 2 JGT? 2A JG RS 2A KR1C LPDW 2 JR 2 JR RS 2 BL RS 2 BL RS 2 KR1B LPDW 2 KR1B LPDW 2 JG 2 JG Outlier JG4 LPDW Outlier JGT 1 BL 1 JG1 1 JGT 2 JGT Outlier BL 1 KR1B LPDW 2 JR-KR 1 KR1B LPDW 1 KR4 LPDW 1 KR1B LPDW Outlier AM1 LPDW 1 AM1 LPDW 1 JGT LPDW 1 JG LPDW? 1 JG4 LPDW 1 JG4 LPDW 1 AM1 LPDW 1 Flask LPDW 1 AM1 LPDW? 2 AMPS1 LPDW 1 AM1A LPDW 2 JG-CUP LPDW 1 JG4C LPDW 1 Pithos 1 JR 1 KR RS 1 BL RS 1 JG 1 Lamp 1 BL1 RS 1 BL1 RS 1 WS BL Cyp WS Outlier JGT BOR Outlier


INAA Group (P&A)

From kiln From kiln

From kiln From kiln From kiln From kiln From kiln From kiln 1, with slag Kiln entrance From kiln 1 From kiln 1 From kiln 1entrance From kiln 4 entrance From kiln 3 entrance From kiln 1

Ib (#694) Ib (#700) Ib (#718) Ib (#709) Ib (#713) Ia (#711) Ib (#714) Ib (#708)

no decoration no decoration

Warped Warped

Cypriote origin

KM= Kfar Menahem; Ashdod II-III= Dothan 1971; Ashdod IV= Dothan and Porath 1982; Ashdod VI= Dothan and BenShlomo 2005; TS= Maeir 2001; BL= bowl; KR= krater; JG= jug; JGT= juglet; AM= amphora; RS= red-slipped; WS= White Slip; BOR= Black on Red; P&A= Perlman and Asaro 1982; Table 2. List of samples analyzed by ICP


Regional Production Centers of Iron Age Pottery Wares in Philistia Group 2 in Cluster 1 and two samples of PCA Group 1 in Cluster 2.

installations in the site as pottery kilns, or at least their final usage as such, should be reexamined.

Groups 1, combining initial sub group 1 and 3 (twenty six samples altogether), should be viewed as representing a Tell es-Safi/Gath production center. This is not due to reference kiln material from the site’s vicinity, but due to the fact it includes all other pottery vessel forms found in the site. This group of undecorated pottery included various common pottery forms of these warped vessels, and was used as a tentative reference for the clay composition of the site. Group 2 is clearly of an Ashdod provenance, as it includes all the vessels found in the kilns of Ashdod, and is similar to Group Ia of Perlman and Asaro 1982 (Column F). Thus, it represents an Ashdod production center.

It should be stressed that this research is continuing and some changes in the chemical groupings may be made in the future, especially concerning the reference material of Tell es-Safi/Gath. Other statistical methods, as the ‘Modified Mahalanobis Distance’ (Mommsen et al. 1988, Beier and Mommsen 1994) may bring somewhat better distinction between the groups. Moreover, further sampling of reference material and possibly of raw material from the vicinity of the sites, together with detailed petrographic analysis, may sharpen or modify the distinction between the chemical groups. The vessels were also thin sectioned for petrographical analysis. Preliminary results show that most are of a loess or alluvial soil matrix with 20%-30% quartz inclusions of sand and silt sizes. The inclusions vary somewhat in sorting and roundness. A relatively small quantity of limestone and chalk inclusions also occurs, and hardly any other inclusions are visible within the samples. A few of the LPDW vessels are made of clay mixed with hamra soil. The dark and isotropic appearance of many of the samples and the lack of calcite testify to a firing temperature above 800 degrees.

According to the chemical groups formed by the preliminary analysis, most of the LPDW vessels found in Tell es-Safi were made at the site or its vicinity (Group 1), while three of the vessels seem to have been made at or near Tel Ashdod. All of the LPDW vessels sampled from Tell Ashdod were made at the site (Group 2), yet the composition of the clay varies between the kilns of Area D and those of Area M. The three vessels from Tell esSafi matching the Tel Ashdod composition (Group 2) are an intact amphora (SF16; Maeir 2001, Fig. 8:10), another amphora fragment (SF14) and a vessel of the KR1B krater type (SF1). All other LPDW vessels fit the common pottery reference group or a diluted composition of this group. One of the LPDW vessels from Tell es-Safi (SF5), a KR1B krater type, is possibly a compositional outlier, although it may be a diluted sample of the chemical profile of Group 1. Thus, none of the LPDW vessels analyzed seems to have been imported from Phoenicia, Cyprus or other distant regions (compare the composition of a Cypro-Phoenician juglet seen in table 1 Column I, SF40).

Conclusions The presently available results indicate quite clearly that the LPDW is a local ware of Philistia, produced primarily at Tell es-Safi/Gath and Tel Ashdod. The provenance study has shown that some intra-regional trade in the LPDW did occur, for example between Ashdod and Tell es-Safi/Gath. During the Iron Age IIA, both the archaeological data and the historical sources appear to indicate that Ashdod and Tell es-Safi/Gath were the primary cities in Philistia. Thus, it is not surprising that one of the most typical elements of the Iron Age IIA Philistine material culture, the LPDW pottery, was produced in, and most likely distributed from these two sites.

Interestingly the eight samples from the kilns of Kfar Menahem are divided between the three main groups. They also include two or three outliers, one of them an LPDW jug (KM2) and another a juglet (KM10), which according to both petrography and chemical composition may be imported from a distant site. Another jug (KM1) is also a chemical outlier, while a juglet (KM9) has a composition that fits Group 2 from Ashdod. These results seem to be contrary to the accepted view that kiln material could be easily used as a chemically homogenous reference material. This discrepancy possibly occurs because several different clay recipes are used in the same workshop (using various soil types as Loess, Hamra etc.); moreover, wasters could represent faulty and/or different clay recipes. Thus, paradoxically, the chemical variability observed in a pottery production site may in some cases be higher than observed in a consumer site. A similar phenomenon was reported in pottery kiln from Late Minoan Kommos, Crete (Buxeda I Garrigós et al. 2001 pp.366-9) and in modern workshops in northwest Spain (Buxeda I Garrigós et al. 2003 pp.1516). Another possibility is that the interpretation of the

Further research and analysis of vessels from sites inside and outside Philistia (as Tel Batash, Gezer, Beth Shemesh, Tel Miqne-Ekron, Ashkelon, Tell Hamid, Ruqeish, Beer Sheba, Tel Massos and Khirbet el Qom) will determine whether this pottery was also exported to other sites and regions or was locally produced at these sites (as an imitation). The regional production centers of the early Philistine pottery should be identified as well. Thence, it will be possible to compare trade patterns in decorated pottery between the early and late Iron Age in Philistia. The use of ICP for large-scale provenance studies of pottery is not yet widely spread. To date, the INAA method is still the major method used. However, it is clear the ICP can give a large array of elements (even more than INAA) with highly reliable precision suitable for the provenancing of 277

D. Ben-Shlomo pottery. The ICP method is also relatively speedy and inexpensive. The major drawbacks are the destruction of the sample and dissolution difficulties (see Tsolakidou and Kilikoglou 2002). Another impediment is the lack of substantial data banks for chemical profiles obtained by ICP. For this reason it is relatively difficult to assign isolated chemical profiles obtained by ICP to known reference groups, thus, identifying the exact geographical provenance of the pottery. Therefore, combining ICP results with INAA data bank is very desirable objective and a feasible one too (see Hein et al. 2002). For this a calibration system between different laboratories is needed; comparable blocks of elements should be measured and related calibration standards (preferably with certified values) should be used in both methods.

Procedure: Geochemical Orientation Survey. Geological Survey of Israel Report GSI/15/88. Ben-Shlomo, D., Shai, I. and Maier, A.M., 2004, Late Philistine Decorated Ware (‘Ashdod Ware’): Typology, Chronology and Production Centers, Bulletin of the American Schools of Oriental Research, , 1-34. Buchbinder, B., 1969, Geological Map of Hashephela Region, Israel, Jerusalem: The Geological Survey of Israel. Buxeda i Garrigos, J., Kilikoglou, V., and Day. P.M., 2001, Chemical and Mineralogical Alteration of Ceramics from a Late Bronze Age Kiln at Kommos, Crete: The Effect on the Formation of a Reference Group, Archaeometry, (3), 349371. Buxeda i Garrigos, J., Cau Ontiveros, M.A. and Kilikoglou, V., 2003, Chemical Variability in Clays from a Traditional Cooking Pot Production Village: Testing Assumptions in Pereruela, Archaeometry, 45(1), 1-17. Chapman, J., 1972, A Catalogue of Iron Age Pottery from the Cemeteries of Khirbet Silm, Joya, Qraye and Qasmieh of South Lebanon, Berytus, 21, 51-194. Dayagi-Mendels, M., 2002, The Akhziv Cemeteries. The BenDor Excavations, 1941-1944. IAA reports No. 15, Jerusalem: Israel Antiquities Authority. Dothan, M., 1971, Ashdod II-III (‘Atiqot 9-10), Jerusalem: The Department of Antiquities and Museums. Dothan, M. and Ben-Shlomo, D., 2005, Ashdod VI. Excavations of Areas H and K: The Fourth and Fifth Seasons of Excavation, Jerusalem: IAA Reports Series, No 24. Dothan, M. and Freedman, D.N., 1967, Ashdod I (‘Atiqot 7), Jerusalem: The Department of Antiquities and Museums. Dothan, M. and Porath, Y., 1982, Ashdod IV (‘Atiqot 15), Jerusalem: The Department of Antiquities and Museums. Dothan, T., 1982, The Philistines and Their Material Culture, Jerusalem: Israel Exploration Society. Gitin, S., 1998, Philistia in Transition: The Tenth Century B.C.E. and Beyond, In Mediterranean Peoples in Transition (S. Gitin, A. Mazar and E. Stern eds.), Jerusalem: Israel Exploration Society: 162-183. Glascock, M.D., 1992, Characterization of Archaeological Ceramics at MURR by Neutron Activation Analysis and Multivariate Statistics. In Chemical Characterization of Ceramic Pastes in Archaeology (H. Neff ed.), World Archaeology Monographs No. 7, Madison WI: Prehistory Press: 11-26. Gunneweg, J., Perlman, I., Dothan, T., and Gitin, S., 1986, On the Origin of Pottery from Tel Miqne-Ekron, Bulletin of the American Schools of Oriental Research, 264, 3-16. Hart, F.A., Storey, J.M.V., Aadams, S.J., Symonds, R.P. and Walsh, J.N., 1987, An Analytical Study, Using Inductively Coupled Plasma (ICP), of Samian and Colour-Coated Wares from the Roman Town at Colchester together with Related Continental Samian Wares, Journal of Archaeological Science, 14, 577-598. Hein, A., Mommsen, H., and Maran, J., 1999, Element Concentration Distributions and Most Discriminating Elements for Provenancing by Neutron Activation Analyses of Ceramics from Bronze Age Greece, Journal of Archaeological Science, 26, 1053-1058. Hein, A., Tsolakidou, A., Iliopoulos, I., Mommsen, H., Buxeda i Garrigos, J., Montana, G. and Kilikoglou, V., 2002, Standardization of Elemental Analytical Techniques Applied to Provenance Studies of Archaeological Ceramics: An InterLaboratory Calibration Study, Analyst, , 542-553. Karageorghis, V., 1970. Salamis II. Nicosia: The Department of Antiquities of Cyprus. Killebrew, A.K., 1996, Pottery Kilns from Deir el-Balah and Tel Miqne-Ekron, In Retrieving the Past. Essays on Archaeological Research and Methodology in Honor of Gus W. Van Beek (J.D. Seger ed.), 135-162, Winona Lake, IN: Eisenbrauns.

Identifying pottery imported from distant sites— interregional trade—is relatively straightforward according to its chemical fingerprint. Yet, when dealing with relatively close-by production centers—intra regional trade—distinguishing distinct chemical profiles is much more difficult. This type of provenancing calls for sampling larger groups, obtaining more elements with better precision and accuracy, and probably combination of both mineralogical and chemical methods together with more sophisticated archaeological reasoning. Acknowledgements This provenance study is part of the author’s Ph.D. dissertation conducted under the supervision of Dr. A.M. Maeir and Dr. I. Sharon (the Hebrew University). The chemical analysis of the pottery was conducted in the ICP laboratory of the Earth Sciences Department of Bristol University, England. The analyses was made possible by a grant from the European Commission program for Access to Research Infrastructures, contract HPRI-1999CT-00008, given to A.M. Maeir and D. Ben-Shlomo. I wish to thank the staff of the Bristol laboratory: Dr. Tony Kemp, Dr. Chung Choi and Dr. John Dalton. I thank A.M. Maeir and Y. Israel for enabling me to analyze the pottery from their sites. The study of the Late Philistine Decorated Pottery is a joint project by I. Shai, A.M. Maeir and the author. References Adan-Bayewitz, D., 1993, Common Pottery in Roman Galilee: A Study in Local Trade, Ramat Gan: Bar-Ilan University Press. Amiran, R., 1969, Ancient Pottery of the Holy Land, Jerusalem: Massada Press. Asaro, F., Perlman, I. and Dothan, M., 1971, An Introductory Study of Mycenaean IIIC:1 Ware from Tel Ashdod, Archaeometry, 13, 169-175. Bakler, N., 1982, The Geology of Tel Ashdod, DOTHAN, M. and PORATH, Y., Ashdod IV, 65-69. Beier, Th. and Mommsen, H., 1994, Modified Mahalanobis Filters for Grouping Pottery by Chemical Composition, Archaeometry, 36, 287-306. Beith, M., Shirav, M., Halicz, L. and Bogash, R., 1988, The


Regional Production Centers of Iron Age Pottery Wares in Philistia Perlman, I. and Asaro, F., 1982, Provenience Studies on Pottery of Strata 11 and 10, Dothan, M. and Porath, Y., Ashdod IV: 70-90. Ponting, M. and Segal, I., 1998, Inductively Coupled PlasmaAtomic Emission Spectroscopy Analyses of Roman Military Copper-Alloy Artifacts from the Excavations at Masada, Israel, Archaeometry, (1), 109-122. Porat, N., Yellin, J., Heller-Kallai, L., and Halicz, L., 1991, Correlation Between Petrography, NAA, and ICP Analyses: Application to Early Bronze Egyptian Pottery from Canaan, Geoarchaeology, 6(2), 133-149. Saidah, R., 1966, Fouilles de Khalde, Bulletin de Musée de Beyrouth, 19, 51-90. Sneh, A., Bartov, Y. and Rosensaft, M., 1998, Geological Map of Israel, Jerusalem: Geological Survey of Israel. Tsolakidou, A. and Kilikoglou, V., 2002, Comparative Analysis of Ancient Ceramics by Neutron Activation Analysis, Inductively Coupled Plasma-Optical Emission Spectrometry, Inductively Coupled Plasma-Mass Spectrometry and X-Ray Fluorescence, Analytical and Bioanalytical Chemistry, , 566-572. Wood, B.J., 1990, The Sociology of Pottery in Ancient Palestine, Sheffield: Sheffield Academic Press.

Killebrew, A.E., 1998, Ceramic Typology and Technology of Late Bronze II and Iron I Assemblages from Tel Miqne Ekron: The Transition from Canaanite to Philistine Culture, In Mediterranean Peoples in Transition (S. Gitin, A. Mazar and E. Stern eds.), 379-405, Jerusalem: Israel Exploration Society. Killebrew, A.E., 2000, Aegean-Style Early Philistine Pottery in Canaan During the Iron I Age: A Stylistic Analysis of Mycenaean IIIC:1b Pottery and Its Associated Wares, In The Sea Peoples and Their World: A Reassessment (E.D. Oren ed.), 233-253, Philadelphia: University Museum. Maeir, A.M., 2001, The Philistine Culture in Transition: A Current Perspective Based on the First Seasons of Excavation at Tell es-Safi/Gath, In Settlement, Civilization and Culture. Proceedings of the Conference in Memory of David Alon (A.M. Maeir and E. Baruch eds.), Bar Ilan University, 111129 (in Hebrew). Maeir, A.M. and Ehrlich, C.S., 2001, Excavating Philistine Gath, Biblical Archaeology Review, 27, 22-31. Master, D.M., 2001, The Seaport of Ashkelon in the Seventh Century BCE: A Petrographic Study, Ph.D. thesis, Harvard University, Cambridge. Mommsen, H., Kreuser, A., and Weber, J., 1988, A Method of Grouping Pottery by Chemical Composition, Archaeometry, , 47-57.


NON-DESTRUCTIVE QUANTITATIVE XRF ANALYSIS ON GREEK ARCHAIC POTTERY OF THE ALAIMO SANCTUARY IN LENTINI (SICILY) L. Grasso CNR, Istituto per i Beni Archeologici e Monumentali, Sezione di Catania, Via A. di Sangiuliano 262, 95100, Catania, Italy

L. Pappalardo, F.P. Romano CNR, Istituto per i Beni Archeologici e Monumentali, Sezione di Catania, Via A. di Sangiuliano 262, 95100, Catania, Italy and INFN, Laboratori Nazionali del Sud, LANDIS, Via S. Sofia 44, 95100 Catania, Italy, [email protected] Abstract: The archaeological investigation of the finds within the votive deposit of Alaimo in Lentini (Sicily, Italy) showed the presence of three reference classes of pottery. Nevertheless, some artefacts were difficult to classify both for their stylistic typology and for the colour and consistence of the clay pastes Among these samples, a group of Rhodian-like aryballoi has been evidenced. In order to group the finds and to identify the doubtful artefacts, the content of some trace elements (Rb, Sr, Y, Zr and Nb) has been determined in 28 well preserved samples of different pottery shapes. Measurements have been performed by using a portable non destructive XRF spectrometer at the Archaeological Museum of Lentini. Compositional data of trace elements have been determined by using a new method based on a multilinear regression approach. A best relative fit and a multivariate cluster analysis have been performed; analytical results allowed for the verification of the presence of groups of membership among the analysed objects and the classification of some artefacts of doubtful archaeological categorization. Περιληψη: Η αρχαιολογική έρευνα των ευρημάτων στην αναθηματική θέση Alaimo στο Lentini (Σικελία, Ιταλία) αποδεικνύει την παρουσία τριών ομάδων αναφοράς στην κεραμική. Παρ΄ όλα αυτά, μερικά αντικείμενα ήταν δύσκολο να ενταχθούν σε ομάδες, τόσο λόγω της καλλιτεχνικής τους τεχνοτροπίας όσο και εξ΄ αιτίας του χρώματος και της σύστασης των πηλών που χρησιμοποιήθηκαν. Ανάμεσα στα δείγματα αυτά βρίσκεται και μια ομάδα αρύβαλλων Ροδιακού τύπου. Με σκοπό την ομαδοποίηση των ευρημάτων και την αναγνώριση των αμφίβολων αντικειμένων, προσδιορίστηκε η περιεκτικότητα σε ιχνοστοιχεία (Rb, Sr, Y, Zr και Nb) σε 28 δείγματα από καλά διατηρημένα αγγεία διαφορετικού σχήματος. Οι μετρήσεις πραγματοποιήθηκαν με μη καταστρεπτική, φορητή συσκευή XRF στο Αρχαιολογικό Μουσείο του Lentini. Τα δεδομένα για τα ιχνοστοιχεία έχουν προσδιοριστεί με τη χρήση μιας νέας μεθόδου που βασίζεται σε μια προσέγγιση πολλαπλής γραμμικής παλινδρόμησης. Βρέθηκε η βέλτιστη σχετική προσαρμογή και πραγματοποιήθηκε ανάλυση ομαδοποίησης πολλών μεταβλητών. Τα αναλυτικά αποτελέσματα επιβεβαίωσαν την παρουσία ομάδων στα αντικείμενα που αναλύθηκαν και επέτρεψαν την ένταξη σε ομάδες κάποιων αντικειμένων αμφίβολης αρχαιολογικής προέλευσης.


trace elements generally used in provenance studies of archaeological pottery (Mommsen 2001, Hall and Minyaev 2002). When the analysis is limited to these five elements, it is possible to discriminate among samples only when they present a different pattern of concentrations; in principle nothing could be stated about the samples provenance, and the different content of Rb, Sr, Y, Zr and Nb only suggests the use of different clay paste recipes for manufacturing artefacts.

The use of the non destructive and portable XRF technique in Cultural Heritage field allows for the obtaining of information about the analysed sample without destroying any part of it and without moving it from its place of custody. It is well known that measurements have to be performed carefully, and only in those cases in which the analysed material is homogeneous enough to be representative of the whole sample (Cuomo Di Caprio 1995).

However, it has been shown at least in one case (Pappalardo et al. 1998) that combining analytical results on these five trace elements with archaeological information made it possible to highlight differences between various typological classes of pottery.

In this paper, the quantitative determination of some trace elements (Rb, Sr, Y, Zr and Nb) obtained by a portable XRF spectrometer (Pappalardo et al. 1994) is used to identify and classify the categories of fine pottery most commonly represented within the votive deposit of Alaimo in Lentini (Sicily, Italy) (Rizza G. 2003, Grasso 2004).

Archaeological studies on the Alaimo votive deposit evidenced the presence of three different typologies of pottery (consisting both of imported wares and of local production) as well as artefacts of doubtful classification;

It is known that Rb, Sr, Y, Zr and Nb are among the 281

L. Grasso, L. Pappalardo and F.P. Romano in particular a class of Rhodian-like aryballoi has been evidenced (Grasso 2004).

like aryballoi; they were, in fact, similar in typology but could be distinguished, on the bases of clays colour and consistence, into two completely different groups. The first one was classified as product of the East-Greek area, though the second one was of doubtful collocation, but with technical characteristics that suggested it should be placed in the Corinthian class of pottery. This last hypothesis (Grasso, Pappalardo & Romano 2002) was not confirmed by the published literature, even if the presence of Phoenicians in the town of Corinth and their involvement in the Corinthian pottery production has been previously reported in (Morris & Papadopulos 1998).

In order to verify these hypotheses, a set of well preserved artefacts has been chosen on the basis of archaeological considerations (based both on the stylistic typologies of objects and on the colour and consistence of the clay pastes) as reference samples. Additionally, some samples of uncertain collocation have been also included among the artefacts to be analysed. In order to avoid the problem of moving artefacts to the laboratories of analysis, measurements have been performed in situ at the Archaeological Museum of Lentini by using the portable XRF spectrometer, designed and realised at the LNS/INFN laboratories of Catania (Italy) (Pappalardo et al. 1994).

In order to group the finds of the votive deposit, 28 samples of different pottery shapes have been analysed by using a non-destructive portable XRF technique. Samples have been accurately chosen in order to create the three references groups for each of the three classes mostly represented (the local class, the Corinthian class and the East-Greek class). Moreover, some samples classified as local imitation production of imported artefacts and some samples belonging to the Rhodian-like aryballoi were selected for the non destructive analysis. Table 1 list the analysed artefacts and their archaeological grouping.

The compositional data of the trace elements has been obtained by using a new method based on a multilinear regression approach (Pappalardo et al. 1998, Gigli et al. 2002). Raw data has first been filtered by a best relative fit (Mommsen, Kreuser & Weber 1988) and then used to perform a multivariate cluster analysis. The non destructive analytical approach allowed to cluster reference samples in three groups, confirming what suggested by archaeological studies. Additionally, samples of uncertain categorization have also been classified and, in particular, the hypothesis previously suggested by the authors on the Rhodian-like aryballoi of the Alaimo sanctuary (Grasso, Pappalardo and Romano 2002) has been verified.

The portable XRF spectrometer The portable XRF spectrometer used for the analysis of the Alaimo artefacts has been designed and realised at the LNS/INFN laboratories (Pappalardo et al. 1994, Gigli et al. 2002). The system is shown in Figure 1; it consists of a Si(Li) detector coupled to a measurement device (Figure 109 2) containing three Cd radioactive sources arranged in an annular configuration.

The alaimo votive deposit in Lentini (Sicily) The spectrometer is particularly suited to efficiently excite Rb, Sr, Y, Zr and Nb used, in the present work, to classify and group the fine pottery samples; also major elements like iron and calcium are well excited.

The deposit was found in an area called Alaimo just outside the modern town of Lentini, at the site of an extraurban sanctuary dedicated to the Dioscuri, twin divinities of the Greek Pantheon. The votive deposit was inside a walled enclosure measuring 3x3.5m and consisted mainly of pottery. A few clay-figures and objects made of bronze, iron, amber and bone were also found. The chronology of the deposit materials is from between the second half of the VII century and the beginning of the VI century B.C. (Grasso 2004). The pottery consisted of vases imported from different districts of Greece, mainly from the town of Corinth and from the East-Greek area (the west coast of Asia Minor and opposite islands). However, the majority of the finds were made in local or colonial workshops. Nevertheless, some vases were difficult to classify; if they were placed in one of three classes according to their typology, then this would have been done according to technical characteristics which could indicate imitation products.

Figure 1 The portable XRF spectrometer realised at the LNS/ INFN laboratories of Catania. The radioactive sources are visible on the top of the Si(Li) detector.

A particular problem arose from the class of the Rhodian282

Non-destructive Quantitative XRF Analysis on Greek Archaic Pottery of the Alaimo Sanctuary in Lentini

Sample name

Inventory Number

Stylistic typology

Archaeological classification of samples

Len1 Len5 Len6 Len8 Len9 Len35 Len36 Len37

LE4121 LE4128 LE4291 LE4278 LE4398 LE4992 LE4269 LE4420

Cup (Kylix) Oil Mug (Lekythos) Cup (Kotyle) Cup (Kotyle) Oil Mug (Aryballos) Pixis’s Lid Jewellery box (Pixis) Mug (Lekythos)

Corinthian production “ “ “ “ “ “ “

Len3 Len12 Len13 Len16 Len17 Len18 Len19 Len20 Len24

LE4148 LE5067 LE4984 LE4228 LE4938 LE4198 LE4368 LE4153 LE4135

Bowl Bowl (Lekane) Bowl (Kalathos) Small Krater Bowl Spouted Mug (Guttus) Mug Miniaturistic Mug (Olpe) Miniaturistic Mug (Idrya)

Local production “ “ “ “ “ “ “ “

Len29 Len31 Len32 Len33 Len38 Len39 Len40

LE4907 LE4116 LE4864 LE4328 LE4875 LE4273 LE4432

Oil Mug (Aryballos) Pointed Oil Flask Oil Flask (Flat Alabastron) Bowl (Bird Type) Pointed Oil Flask Oil Mug (Aryballos) Oil Mug (Aryballos)

Rhodian production “ “ “ “ “ “

Len28 Len30 Len10 Len42

LE4936 LE4335 LE4134 LE5132

Oil Mug (Rhodian-like Aryballos) Oil Mug (Rhodian-lyke Aryballos) Oil Mug (Corinthian-like Aryballos) Oil Mug (Corinthian-like Aryballos)

Doubtful “ “ “

Table 1 – The artefacts analysed at the Archaeological Museums of Lentini by using the portable XRF spectrometer. The stylistic typologies as well as the archaeological classification of artefacts are given in the last two column of the table.

The system present fixed experimental conditions and reproducibility of data, estimated by using geological standards, is about 1.5-2%. Minimum Detection Limits for the above trace elements range from 5 to10 ppm. Additionally, the x-ray energies emitted by Rb, Sr, Y, Zr and Nb are large enough to reduce surface effects of the analysed sample; the estimated analytical depth is about 200 microns. Quantitative analysis of some trace elements The content of Rb, Sr, Y, Zr and Nb has been determined by using a new method based, through a multilinear regression approach, on the existing relationship between the net fluorescence counts and the concentration (Pappalardo el al. 1998). It could be written as a first approximation as:

Figure 2 The annular device containing the 109Cd radioactive sources used for the non destructive measurements of the Alaimo artefacts.


Ki = αi + βiN(Ca+K) + γi N(Fe+Ti)

The function:


where Ci and Ni are respectively the concentration and the counts of a specific element.


(where N(Ca+K) and N(Fe+Ti) represent net counts of Ca, K, Fe and Ti) has been obtained by using a multilinear regression 283

L. Grasso, L. Pappalardo and F.P. Romano

Sample name

Rb (ppm)

Sr (ppm)

Y (ppm)

Zr (ppm)

Nb (ppm)

dilution factor

































































Average σ(%)

31 35.5

355 3.4

19 5.3

120 6.7

9 12.0

















Len32 Len33







































Average σ(%)

76 32.8

272 15.4

23 17.4

228 9.6

17 5.8









































































Average σ(%)

46 26.1

422 5.9

19 5.3

191 5.2

16 12.5


Table 2 Quantitative data of Rb, Sr, Y, Zr and Nb for samples constituting the three reference classes obtained by the best relative fit; mean values and the standard deviation for each class are also reported. Raw quantitative data can be obtained by dividing concentrations for the dilution factors reported in the table. Sample name

Rb (ppm)

Sr (ppm)

Y (ppm)

Zr (ppm)

Nb (ppm)

Dilution factor


Class of reference



































Local ?

Table 3 Quantitative data of the five trace elements in samples of doubtful origin. Values obtained for dilution factors and χ2 suggest the possible classification.

method. The αi, βi and γi coefficients were determined through a best fit procedure using geological standards of known composition.

Expressions (1) and (2) were used to determine concentrations of the above mentioned trace elements in the 28 measured artefacts; mean errors in compositional data (due to the counting statistics in x-ray spectra and to the accuracy of the quantitative procedure) were 7.8 % for Rb, 7 % for Sr, 10.7 % for Y and Nb, 10.4 % for Zr.

Accuracy of the quantitative procedure is 4% for Rb, 7% for Sr, 8% for Y and Nb and 10% for Zr. 284

Non-destructive Quantitative XRF Analysis on Greek Archaic Pottery of the Alaimo Sanctuary in Lentini The non destructive quantitative method above discussed should be applied only to analyse fine pottery samples – i.e. homogeneous material. In order to take into account possible dilution of clay pastes and to evidence the presence of matching samples directly by using scatter concentrations (Mommsen 2001), a weighted best relative fit procedures has been performed (Mommsen, Kreuser & Weber 1988). Dilution factors were estimated for each sample towards the mean values of Rb, Sr, Y, Zr and Nb content in the three reference groups; a χ2 parameter was also estimated during calculation (Mommsen, Kreuser and Weber 1988). Table 2 reports quantitative data obtained for the artefacts composing the Corinthian class (first series of data), the Rhodian class (second series of data) and the Local Production class (third series of data); best relative fit factors and the χ2 results are also reported. Additionally, average concentrations and standard deviations (in percentage) are indicated for each of the three reference classes.

Figure 3 The dendogram obtained from the cluster analysis. Samples resulted grouped in three well defined classes.

Finally, the results of the cluster analysis for sample LEN10 are not in agreement with that obtained by performing the best relative fit. In this case, more accurate analyses are necessary in order to classify the sample.

The same approach has been applied also for samples of doubtful archaeological classification; results were accepted when dilution factor approximate unity and for the minimum of values obtained for the χ2 parameter. Results and class of reference that better represent concentrations pattern of the five doubtful artefacts are summarized in Table 2.

Conclusions Results of a non destructive XRF analysis performed on 28 artefacts of the Alaimo sanctuary in Lentini allowed to confirm the archaeological hypothesis on the fine pottery typologies mostly represented in the votive deposit.

Analytical results suggested to classify as Corinthian the samples labelled LEN28 and LEN30 and as local manufactured the sample labelled LEN42.

A set of well preserved artefacts of different pottery shapes were used as reference samples for the local manufactured and imported artefacts (mainly from Corinth and from the East-Greek area) .

The large χ2 value obtained for sample labelled LEN10 do not allow to classify with certitude the artefact.

Non destructive measurements were performed in situ and they were focused towards the analysis of some trace elements (Rb, Sr, Y, Zr and Nb) typically present in archaeological pottery. Concentrations were determined by using a new method based on a multilinear regression approach developed at LNS/INFN laboratories.

Mutlivariate cluster analysis Quantitative data obtained by the best relative fit of compositional data have been used to perform a multivariate cluster analysis. Results are reported in Figure 3; calculation has been performed by using the Ward algorithm and the square Euclidean distance (Dillon & Goldstein 1984).

A best relative fit performed on quantitative data of the above trace elements allowed for the verification the existence of three reference classes. Additionally, some typologies of artefacts of doubtful classification were assigned to their classes of membership; in particular, Corinthian imitations were grouped to the Local production class and some Rhodian-like aryballoi clustered on the Corinthian class of pottery.

Reference samples were clustered in three defined and well separated groups, confirming the archaeological hypothesis on the existence of three classes of pottery (the Local class, the Corinthian class and the East-Greek area class) mainly represented in the Alaimo votive deposit. Sample LEN42 grouped in the Local production class; the aryballoi labelled LEN28 and LEN30 result grouped in the Corinthian class, confirming the archaeological hypotheses of the existence of a Rhodian-like aryballoi production in Corinth (Grasso, Pappalardo & Romano 2002).

Finally, results were confirmed by a multivariate cluster analysis performed on compositional data. References Cuomo Di Caprio, N., 1995, In La ceramica in archeologia.


L. Grasso, L. Pappalardo and F.P. Romano Mommsen H., 2001, Provenance determination of pottery by trace element analysis: problems, solutions and applications, Journal of Radioanalytical and Nuclear Chemistry, 3, 657662 Morris, S., Papadopoulos, J.K., 1998, Phoenicians and Corinthian Pottery Industries, In Archaeologische Studien in Kontaktzonen der Antiken Welt (R. Rolley & K. Schmidt eds.), 251, Gottingen. Pappalardo, G., Di Pietro, A., Musumarra, A., Pappalardo, L., 1994, Un sistema di spettrometria XRF portatile: Applicazioni nel settore dei beni monumentali, In Bollettino Accademia di Gioenia Scienze Naturali, 27(346), 329-340. Pappalardo, G., Loreto, G., Musumarra, A., Pettinato, S., Rizza, G., Fichera, V., Marino, R.M., Cirrincione, R., Pappalardo, L., 1998, Non destructive determination of trace-elements in Greek pottery by use of a XRF portable system, In Proceedings of the 1st International Congress on “Science and Technology for the Safeguard of Cultural Heritage in the Mediterranean Basin, 803-806, December 1995, Acireale (Italy), Eds. Tipolitografia Luxograph, Palermo (Italy). Rizza, G., 2003, Scoperta di un santuario dei Dioscuri a Lentini, In Rendiconti dell’Acccademia Nazionale dei Lincei, IX, 14(4), 563-564.

Antiche tecniche di lavorazione e moderni metodi d’indagine, “L’Erma” di Bretschneider, Roma. Dillon, W.R., and Goldstein, M., 1984, In Multivariate Analysis: methods and applications (J. Wiley & Sons eds.), 172-178, New York. Gigli, R., Pappalardo L., Pautasso, A., Romano, F.P., Pappalardo, G., Carastro, A., 2002, Identification of a class of pottery within the votive deposit of Demetra Sanctuary in Catania by using a non destructive XRF method, In Proceedings of 33th International Symposium on Archaeometry, 22-26 April 2002, Amsterdam: in press. Grasso, L., Pappalardo, L., Romano, F.P., 2002, In merito agli aryballoi rodio-cretesi, In Proceedings of “Το Αιγαίο στην πρώιμη εποχή του Σιδήρου”, 1-4 November 2002, Rodi: in press. Grasso L., 2004, Il Santuario di Alaimo: primi risultati dello studio della Stipe, In Proceedings of “Leontini: il mare, il fiume, la città” (M. Frasca, Siracusa ed.), 117–122, 2 May 2002, Lentini (Italy). Hall, M., Minyaev, S., 2002, Chemical analyses of Xiong-nu pottery: A preliminary study of exchange and trade on the inner Asian steppes, Journal of Archaeological Science, 29, 135-144. Mommsen, H., Kreuser, A., Weber, J., 1988, A method for grouping pottery by chemical composition, Archaeometry , 30(1), 47-57.



Faculty of Chemistry, University of Sofia, Bulgaria, [email protected] 2

Institute for Archaeometallurgy, TU Bergakademie Freiberg, Germany,

T. Stoyanov3 3

Faculty of History, Chair of Archaeology, University of Sofia, Bulgaria,

Abstract: The content of 25 elements in the investigated 74 stamped (and some unstamped) amphorae finds from 9 Thracian cities (Bulgaria) dated 5th – 3rd century B.C. have been determined using INAA. The investigated samples were grouped by cluster analysis on the basis of the similarity in their chemical composition. The chemical profiles of the 16 clusters formed were determined, and common production centers for some of the stamped amphorae are identified. Περιληψη: Με τη χρήση Ανάλυσης με Νετρονική Ενεργοποίηση προσδιορίστηκε η περιεκτικότητα σε 25 στοιχεία 74 αμφορέων με σφραγίδα, από εννιά Θρακικές πόλεις της Βουλγαρίας που χρονολογούνται στον 5ο-3ο αιώνα π.Χ. Σύμφωνα με τα αποτελέσματα της ανάλυσης, τα δείγματα ομαδοποιήθηκαν με βάση τις ομοιότητες στη χημική τους σύσταση. Έτσι προσδιορίστηκε το χημικό ‘αποτύπωμα’ των 16 ομάδων που σχηματίστηκαν και αναγνωρίστηκαν τα κέντρα παραγωγής κάποιων ομάδων αμφορέων.

amphorae with the stamp ΜΑΤΡΟ/ΒΙΟΥ (Stoyanov 2000, 2003, 2005). This pottery production is probably part of the Parmeniskos group, to which some of the stamped amphorae found in different Thracian towns also belong – Kabile, Sevtopolis, Pistiros, etc. (Balkanska 1984:125, Domaradzki 1993:47, Getov 1995:92-94, 2000, Kiachkina 1994).

Introduction Amphorae are an especially valuable class of pottery which was produced for the shipping of liquid goods – wine and oil. Transport amphorae are characterized by their shape, vertical handles, size and pointed toe. Each state in the ancient world engaged in amphorae manufacture, maintained uniformity in the shapes and sizes of its vessels. This stylistic difference forms the basis for the typological classification of the amphorae. Amphorae production was important to the states (they regulated the packaging and shipping of goods) and potters used different signs for identification of its production. Therefore the amphorae production of different sites is marked with its own specific identifying stamps. The stamps are directly related to the time and place of manufacture.

The present investigation was undertaken in order to test the hypothesis regarding the local production of some of the stamped amphorae found in the Thracian territories, as well as to present some objective data obtained by chemical methods of investigation for archaeological pottery, for the proposed production center of the Parmeniskos group of amphorae. Archaeological background

The reason for this investigation was the identification of the few stamps found in the large fortified Getic town in Sboryanovo (near the present town of Isperih, see Fig. 1), to the Parmeniskos group determined by Grace (1956). Amphorae of this type are spread over the northern Aegean, Corinth, Troy and the Black Sea basin during the Hellenistic period, but the production centers of these amphorae are unknown and in dispute. There are several different proposals: • • • •

The petrological investigation of the stamped amphorae from the Parmeniskos group shows that clay with similar mineralogical composition could be found among the amphorae produced in Mende, Akanthos and so called Soloha I (Whitbread 1995). However, in the literature there are also some proposals about the similarity of pottery from the Parmeniskos group and early pottery production from Thasos and Akant.

North Aegean region (Brashinskiy 1966a, 1966b, Grace & Savvatianou-Pétropoulakou 1970); Pella (Akamatis 1992:80-83, 2000); Torone (Whitbread 1995:210-219); Meliboia (Garlan 1997:48).

The early importation from Thasos to the interior of Thrace is not well documented. The only exceptions to this are Kabile, Pistiros and the village near Simeonovgrad (see Bozhkova 1999, Domaradzki 1995, Getov 1995, 1999, Titz 2002). Less is known about importation to Thrace from other big Aegean centers like Mende, Chios or the centers connected with amphorae of the type Soloha I etc. (Getov 1995:113-116).

The investigation of the stamps from Sboryanovo, as well as some numismatic and epigraphic data together with the analogous stamps found in Mesambria (today Nessebar) support the hypothesis of a local production of the

For archaeologists it is difficult to determine the provenance 287

I. Kuleff, E. Pernicka and T. Stoyanov

Figure 1. Map of Bulgaria with excavation locations 1 = Sboryanovo; 2 = Orizare; 3 = Burgas; 4 = Prilep; 6 = Polski gradets; 7 = Kabile; 8 = Dana bunar; 9 = Pistiros

of amphorae when they are from earlier periods or are in fragments, especially when they are not stamped. In this respect the possibility of investigating finds from the settlements along the trade route to inner Thrace – along the river Maritsa and its feeders, as well as the finds from Kabile, from the sanctuary of Dana Bunar (near Lyubumets), the finds from Emporion Pistiros and Polski Gradets with later material for which the provenance is better determined (mainly from the Thracian town at Sboryanovo) could be a step forward in the identification of the production center for the Parmeniskos group of amphorae.

Kuleff et al. 1986, Kuleff & Djingova 1990, Kuleff & Pernicka 2002) so that only a short description will be given here. 3.

Experimental 1.

Materials. Seventy four pottery finds from: Sborjanovo (37); Pistiros (7); Varna (7); Kabile (3); Polski gradets (5); Burgas (2); Prilep (2); Danabunar (7); Orisare (6) were subjected to analysis. The finds originate from the excavated archaeological sites indicated on the map in Fig.1. A short description of the investigated finds is presented in Table 1, as well as photos (drawings) of some of them in Fig. 2. The finds are dated back to 5th – 3rd century B.C.


Methods of analysis. The sampling, the preparation of the samples for analysis, and the analysis itself are described in detail in several previous papers (see e.g. 288

Analysis. The investigated finds are fragments of different vessels and the samples for analysis were obtained by breaking off small pieces, which were pulverized in an agate mill. About 100-150 mg of the fine powdered samples were packed in polyethylene vials and irradiated for 12 h in the nuclear research reactor of Institute of Nuclear Chemistry at University of Mainz (Germany) with pile neutrons (thermal neutron flux 1.1012 cm-2.s-1). The activity of the generated radionuclides was measured twice: The first measurement was after cooling for 6-7 days and the second - after 30 days with a HP Ge detector Ortec (energy resolution of 1.8 keV and 45 % efficiency at 1332.5 keV) coupled with a 4096 channel analyzer (Nuclear Data). As standard the home made clay standard (HD-Tony) was used. The concentration of 25 elements are determined: arsenic (As), barium (Ba), cerium (Ce), cobalt (Co), chromium (Cr), caesium (Cs), europium (Eu), iron (Fe), hafnium (Hf), potassium (K), lanthanum (La), luthetium (Lu), sodium (Na), neodimum (Nd), nickel (Ni), rubidium (Rb), antimony (Sb), scandium (Sc), samarium (Sm), tantalum (Ta), terbium (Tb), thorium (Th), uranium (U), ytterbium (Yb), zirconium (Zr).

On the Origin of Stamped Amphorae from Thracian Cities in Bulgaria





K-818.PIS K-817.PIS














K-795.SBO Figure 2. Drawing and photos of some of the investigated archaeological finds.


I. Kuleff, E. Pernicka and T. Stoyanov

Lab.code K-780.SBO K-781.BUR K-782.KAB K-783.KAB K-784.KAB K-785.VAR K-786.VAR K-787.VAR K-788.VAR K-789.VAR K-790.VAR K-791.VAR K-792.SBO K-793.SBO K-794.SBO K-795.SBO K-796.SBO K-797.SBO K-798.SBO K-799.SBO K-800.SBO K-801.SBO K-802.SBO K-803.SBO K-804.SBO K-805.SBO K-806.SBO K-807.SBO K-808.SBO

Description Fragment of an amphora (Paremeniskos group?) handle, Sboryanovo. Isperih A 1078, stamp ΜΑΤΡΟ / ΒΙΟΥ Fragment of an amphora handle, Manastir tepe (anc. Aquae Calidae). Bourgas Arch.No 343, stamp ΑΝΤΙ / ΦΙΛΟΥ, Fragment of an amphora handle, Amphora type II, Kabyle. Yambol 585.; stamp [ΑΝ]ΤΙ / [Φ]ΙΛΟΥ Fragment of an amphora handle, Amphora type II, Kabyle. Yambol 3038, stamp ΑΝΤΙ / ΦΙΛΟΥ Fragment of an amphora handle, Amphora, Kabyle. Yambol 2729, stamp ΜΕΛ[ΣΕ] / ΩΝ[ΟΣ] Fragment of an amphora handle with a part of wall and the mouth, Odessos. Varna Arch.No 5117, stamp ΗΓΗΣΙΝΟ[Υ] Fragment of an amphora handle, Amphora type II, Odessos. Varna Arch.No. II 1846, stamp ΑΝΤΙ / ΦΙΛΟΥ Fragment of an amphora handle with part of wall, Amphora type II, Odessos. Varna Arch.No II 1688, stamp ΑΝΤΙ / ΦΙΛΟΥ Fragment of an amphora handle, Amphora type II, Odessos. Varna Arch.No II 1815, stamp ΜΙΚΙ / ΩΝΟΣ Fragment of an amphora handle, Amphora type II, from Kavarna (anc. Bizone). Varna Arch.No II 1630, stamp ΑΝΤΙ / ΦΙΛΟΥ Fragment of an amphora handle with a part of wall and mouth, Amphora type II, Odessos. Varna Arch.No II 5120, stamp ΜΙΚΙ / ΩΝΟΣ Fragment of an amphora handle with a part of wall and mouth, Amphora type II, Odessos. Varna Arch.No II 5111, stamp ΜΙΚΙ / ΩΝΟΣ Fragment of a handle of amphora of unidentified centre, Sboryanovo. Isperih A 383, circular stamp E and Υ Fragment of a Heraclian amphora mouth, Sbotyanovo. Isperih A 445, stamp HPA. Fragment of an amphora with two handles, Sboryanovo. Isperih A 482. Circular stamp divided into 4 fields – legible Д and Π. On the other handle triangle stamp – legible Α and К Fragment of an amphora handle of unknown centre, Sboryanovo. Isperih A 523. Quadrangle stamp – amphora with 4 letters around Λ Α and Μ Ε Fragment of a Thasian amphora handle, Sboryanovo. Isperih A 830, stamp of ΑΠΟΛΛΟΔΩΡΟΣ, crane Fragment of a Thasian amphora handle, Sboryanovo. Isperih A 1072, stamp of ΑΠΟΛΛΟΔΩΡΟΣ, dagger Fragment of a Thasian amphora handle, Sboryanovo. Isperih A 981. Stamp of КЛЕПЦЩН, dolphin Fragment of a Thasian amphora handle, Sboryanovo. Isperih A 1113. stamp of ΑΠΟΛΛΟΔΩΡΟΣ, indistinct emblem Fragment of a Thasian amphora handle, Sboryanovo. Isperih A 1294. stamp of ΘΑΣΩΝ, dog Fragment of a Thasian amphora handle, Sboryanovo. Isperih A 1098 stamp of ΘΑΣΩΝ, dog Fragment of a handle from Akanthos’ (?) amphora, Sboryanovo. Isperih A 780, circular stamp – legible - Σ and Μ (?) Fragment of a Thasian amphora handle, Sboryanovo. Isperih A 1050, stamp of ΑΠΟΛΛΟΔΩΡΟΣ, cuirass Fragment of a Thasian amphora handle, Sboryanovo. Isperih A 830, stamp of ΑΠΟΛΛΟΔΩΡΟΣ, thymiaterion Fragment of a Thasian amphora handle, Sboryanovo. Isperih A 1008 stamp of ΘΑΣΩΝ, kantharos Fragment of a handle of an amphora from Ainos (?), Sborjanovo. Isperih A 1015, quadrangle stamp - kerykeion (?) and Α/Ι/Ν/[Ι]/Ο/Ν around Fragment of a Heraklean (?) amphora mouth, Sboryanovo. Isperih A 980 part of stamp, legible - Ω and Τ Fragment of a Thasian amphora handle, Sborjanovo. Ispeirh A 1573 stamp of ΘΑΣΩΝ, kantharos


Remark/Citation Stoyanov 2005, No 153 Kiachkina 1994, p.189 Getov 1995, p.95 No 227 Getov 1995, p.95 No 226 Getov 1995, p.96 No 230 Toncheva 1974, No 69 Mirchev 1958 No 273 Mirchev 1958, No 274 Mirchev 1958, No 80 Mirchev 1958, No 275 Toncheva 1974, No 66 Toncheva 1974, No 68 Bozhkova 2005, No 46 Bozhkova 2005, No 43 Bozhkova 2005, No 45 Stoyanov 2005, No 155 Stoyanov 2005, No 82 Stoyanov 2005, No 83 Stoyanov 2005, No 34 Stoyanov 2005, No 85 Stoyanov 2005, No 14 Stoyanov 2005, No 13 Stoyanov 2005, No 152 Stoyanov 2005, No 81 Bozhkova 2005, No 26 Stoyanov 2005, No 10 Stoyanov 2005, No 154 Stoyanov 2005, No 146 Stoyanov 2005, No 11

On the Origin of Stamped Amphorae from Thracian Cities in Bulgaria

Lab.code K-809.SBO K-810.SBO K-811.PGR K-812.SBO K-813.SBO K-814.SBO K-815.PIS

K-816.PIS K-817.PIS

Description Fragment of a Thasian amphora handle, Sboryanovo FIN 17-99. stamp of ΘΑΣΩΝ, kantharos Fragment of an amphora handle from an unidentified centre, Sborjanovo FIN 23-99. quadrangle stamp legible ΩΣ Or image of insect Fragment of a Heraclean (?) amphora, from a pit in Thracian site near the village of Polski Gradets. Nova Zagora NSF 1509. Englyphic stamp on the neck Fragment of a Thasian amphora handle, Sboryanovo. Isperih A 1111 stamp of ΑΠΟΛΛΟΔΩΡΟΣ, flower Fragment of a Thasian amphora handle, Sboryanovo. Isperih A 1174 stamp of ΘΑΣΩΝ, alabastron Fragment of a Thasian amphora handle, Sboryanovo. Isperih A 425 stamp of ΑΠΟΛΛΟΔΩΡΟΣ, double axe Fragment of an amphora handle, Pistiros. Septemvri Arch. No 1.506 stamp ΑΝΤΙ / ΦΙΛΟΥ Fragment of an amphora handle, Pistiros. Septemvri Arch. No 1. 1034, stamp ΜΑΤΡ / ΟΔΩΡΩ, grape cluster Fragment of a Thasian amphora neck and handle, Pistiros. Septemvri Arch. No 25, stamp ΑΡΙΣΤΟ[ΜΕΝΙΣ]/ … / ΘΑΣΙΩ[Ν] emblem of kneeling Heracles with a bow


Fragment of a Thasian amphora handle, Pistiros. Septemvri Arch. No 1. 1018, Stamp ΔΑ]ΜΑΣ (ΤΙΣ) / Θ]ΑΣΙΟΝ / Μ]ΥΕΣΚΟΣ retrograde


Fragment of a Thasian amphora handle, Pistiros. Septemvri Arch. No 1.1162, stamp ΔΕΙΝ[...] Fragment of a Thasian amphora neck and handle, Pistiros, Septemvri; Arch. No 1. 1020, stamp head of bearded man (satyr?) [ΘΡΑΣΩΝ]ΙΔΗΣ ?



K-848.DNB K-849.SBO

Fragment of a Thasian amphora neck and handle, Pistiros. Septemvri Arch. No AB I 416, stamp head of man [?]ΙΡΙΑ or Δ[?] Thasian (?) amphora, Thracian sanctuary at Danabunar site II, near the town of Lyubimets, pit 9, depth 0.4-0.45 m. Thasian (?) amphora, Thracian pit sanctuary at Dana bunar site II, pit 9, depth 0.4-0.45 m. At the neck dipinto Amphora, Thracian pit sanctuary at Dana bunar site II,, pit 9, depth 0.4-0.45 m. Thasian amphora, Sborianovo. Stamp of ΙΣΟΔΙΚΟΣ, pomegranate (?)


Fragment of an amphora handle, unidentified centre. Isperih A 494. Stamp ΒΟΗΘ


Fragment of a Coan amphora handle, Sboryanovo. Isperih A 1195. stamp МПC[---]


Fragment of the wall of Heraclean (?) amphora, Sboryanovo. Isperih A 1489. Circular stamp - ΣΑΤ]ΥΡΟ ΣΚΥΘΑ Fragment of an “Solocha I” amphora, FIN 80-93, 1993, Sborianovo.

K-846.DNB K-847.DNB

K-853.SBO K-854.SBO K-855.SBO K-856.SBO K-857.SBO K-858.SBO K-859.SBO K-860.DNB

Fragment of a Coan amphora handle, Sborianovo. Isperih A 1297 stamp МПC[---] Fragment of an amphora from an unidentified centre. Isperih A 1096; Stamp – monogram in a circular field Fragment of Coan amphora handle, Sboryanovo FIN 30-99. Stamp ДЗМЗ[ФСЙПХ] , club below Fragment of an amphora from an unindentified centre, Sboryanovo. Isperih A 536. Stamp - H and P in monogram Fragment of an “Solocha I” amphora, Sboryanovo FIN 190-93 Fragment of an amphora from an unidentified centre, Sboryanovo. Isperih A 1407. Stamp – monogram in an oval field. Mendean (?) amphora, Thracian sanctuary. Danabunar site, Pit 7


Remark/Citation Stoyanov 2005, No 12 Stoyanov 2005, No 156 Not published Stoyanov 2005, No 84 Stoyanov 2005, No 15 Bozhkova 2005, No 25 Domaradzki 1993, 37, fig. 13. 4; Domaradzki 1995, 59, No 7 Domaradzki 1995, 59, No 6 Domaradzki 1995, 59, No 4 Domaradzki 1993, 37, fig. 13. 3 Domaradzki 1995, p. 59, No 3; Cf. Garlan 1999, No 97 Not published Domaradzki 1995, p. 59, No 1; Cf. Garlan 1999, No 65 Domaradzki 1995, p. 59, No 2 Not published Not published Not published Stoyanov 2005, No 4 Bozhkova 2005, No 49 Stoyanov 2005, No 150 Stoyanov 2005, No 145 Stoyanov 2005, §6 Stoyanov 2005, No 149 Stoyanov 2005, No 157 Stoyanov 2005, No 147 Bozhkova 2005, No 47 Stoyanov 2005, § 6 Stoyanov 2005, No 158 Not published

I. Kuleff, E. Pernicka and T. Stoyanov

Lab.code K-861.DNB K-862.DNB K-863.DNB K-864.DNB K-865.DNB K-866.DNB K-867.PGR K-868.PGR K-869.PGR K-870.PRI K-871.PRI K-872.ORI K-873.ORI K-874.ORI K-875.ORI K-876.ORI K-877.ORI

Description Mendean (?) amphora,Thracian sanctuary. Danabunar site, Pit 7 Mendean (?) amphora, Thracian sanctuary. Danabunar site, Pit 42 Mendean (?) amphora, Thracian sanctuary. Danabunar site. Pit 17 Mendean (?) amphora, Thracian sanctuary. Danabunar site, Pit 17 Mendean (?) amphora, Thracian sanctuary at Danabunar site, Pit 17, depth 0.45-0.55m. Mendean (?) amphora, Thracian sanctuary. Danabunar site, Pit 44 Fragment of a Thasian (?)amphora from a pit 43 in a Thracian site near the village of Polski Gradets Fragment of a Thasian (?)amphora from a pit 43 in a Thracian site near the village of Polski Gradets Fragment of a Thasian (?)amphora from a pit 43 in a Thracian site near the village of Polski Gradets Fragment of the mouth of an amphora of unknown origin, Tumulus, near the village of Prilep, 1998 (11) Fragment of an amphora Tumulus, Prilep, 1998 (12) Fragment of an amphora handle. Surface find. Preroman, Roman & late antique settlement. Village of Orizare. Fragment of an amphora wall. Surface find. Preroman, Roman & late antique settlement. Village of Orizare. Fragment of an amphora wall. Surface find. Preroman (?), roman & late antiquity settlement. Village of Orizare. Fragment of the wall of a vessel made on a potter’s wheel. Surface find. Preroman, Roman(?)& late antique settlement. Village Orizare. Fragment of the wall of a vessel made on a potter’s wheel. Surface find. Preroman, Roman & late antique (?) settlement. Village Orizare Fragment of the wall of a vessel made on a potter’s wheel. Surface find. Preroman, Roman & late antique (?) settlement. Village Orizare

Remark/Citation Not published Not published Not published Not published Not published Not published Not published Not published Not published Not published Not published Not published Not published Not published Not published Not published Not published

Table 1. Description of the investigated finds


Statistical interpretation. By analyzing a large number of standard reference materials we are able to check the accuracy and precision of the analysis. The obtained analytical data are published in (Kuleff & Pernicka 2002) and can be used for evaluation and comparison with data for similar archaeological pottery obtained in other laboratories.

The obtained analytical data for the 74 pottery samples analyzed were subjected to cluster analysis. For clustering we used all determined elements except As, Ni, Nd, Sb and in some cases K and Na. On the basis of the similarity in the chemical composition the investigated samples are grouped into 16 clusters while 4 samples remain as outliers. The result of clustering procedure is presented in Table 3 and was also checked using the package of statistical programs SPSS 11. The chemical profiles of the clusters with more than 2 members, formed on the basis of the similarity in the chemical composition are presented in Table 4. The relatively large number of clusters formed is not surprising, if the diversity of the investigated finds and places where they were presumably made/found is taken into account.

The cluster analysis of the analytical data for investigated archaeological finds was performed using Brookhaven National Laboratory Program Package for multivariate statistical analysis (Sayer 1988). The program is written in Fortran and provides a variety of different types of hierarchical clustering. The results of clustering were checked using the statistical program package SPSS 11 (see Bühl & Zöfel 2002).

The calculated RSD-values in the chemical profiles of the clusters (Table 4) could be used for the evaluation of the tightness of the clusters. Thus the clusters no. 1, 3, 4, 7 to 10 and 12 can be considered as statistically significant. Only clusters no. 2, 6 and 14 together with some isolated finds show a wider spread in the chemical composition. There are few RSD-values (excluding As, Nd and Ni) over 35 %, which indicates that the grouping of the finds in these clusters is more or less accidental.

Results and discussion The precision of the analysis carried out in the present work is demonstrated with the data in Table 2, where the results of three independent determinations of the content of 25 determined elements in the sample 877.ORI are presented (the sample was selected by chance). The calculated uncertainty (relative standard deviation) is below 10 % (with one exception for Ni) for all determined elements.

The broad concentration range of some elements, like As 292

On the Origin of Stamped Amphorae from Thracian Cities in Bulgaria




877-3.ORI 7.4±0.5 1 300±25

877.ORI mean 7.4±0.6 2 287±10

RSD 3 (%) 8.3 3.6

As Ba

6.6±0.4 1 275±45

8.1±0.5 1 285±35

Ce Co Cr Cs Eu Fe(%) Hf K (%) La Lu Na (%) Nd Ni Rb Sb Sc Sm Ta Tb Th U Yb Zn Zr

64.2±0.6 19.9±0.3 98.1±1.1 7.05±0.09 1.23±0.03 5.05±0.05 5.40±0.08 1.96±0.22 25.4±1.0 0.356±0.092 1.12±0.03 35±7 120.9±2.7 0.57±0.05 20.4±0.2 5.2±0.3 1.07±0.07 0.82±0.05 12.6±0.1 2.5±0.2 2.3±0.2 48±4 202±17

68.3±0.7 19.9±0.3 101.2±1.1 7.32±0.09 1.31±0.03 5.20±0.06 5.48±0.09 2.09±0.23 28.2±0.5 0.339±0.021 1.19±0.04 27±6 125.0±2.5 0.68±0.06 21.0±0.2 5.4±0.3 1.07±0.05 0.85±0.06 13.2±0.1 2.6±0.2 2.3±0.2 49±4 188±18

66.6±0.7 19.9±0.3 102.8±1.1 7.39±0.10 1.30±0.03 5.30±0.07 5.74±0.09 26.3±1.0 0.315±0.023 32±7 15±6 120.7±2.9 0.55±0.05 21.4±0.2 4.5±0.3 1.13±0.06 0.79±0.05 12.9±0.1 2.4±0.2 2.5±0.2 50±7 220±20

66.4±1.7 19.9±0.3 100.7±2.0 7.25±0.15 1.28±0.04 5.18±0.10 5.54±0.15 2.03±0.07 26.6±1.2 0.336±0.015 1.16±0.04 31±3 15±6 122.2±2.0 0.60±0.06 20.9±0.4 5.0±0.4 1.09±0.03 0.82±0.03 12.9±0.2 2.5±0.1 2.4±0.1 49±1 197±7

2.5 6.5 2.0 2.0 3.1 2.0 2.6 3.2 4.4 4.3 3.0 10.5 40 1.6 9.5 2.0 7.7 2.6 3.0 1.9 4.0 4.2 2.0 3.6

1 = standard deviation from 3 parallel analyzed samples 2 = total uncertainty of the analysis 3 = RSD – relative standard deviation Table 2. Evaluation of the precision of the analysis (concentration is given in [mg/kg]).

and Sb, can be explained by the partial volatilization of these elements during the firing of the pottery (see e.g. Kuleff & Djingova 1990). For elements like Nd and Ni the larger RSD-values are the result of insufficient activity of the radionuclides generated during the neutron irradiation of the samples and consequently bad counting statistics of the measurement. This is the reason why these elements (As, Ni, Nd and Sb) are not included in the multivariate statistical evaluation. The large spread of Ba as well as of U is very difficult to understand, since the reproducibility of measurements of these two elements is usually very good (see Table 2, as well as Kuleff & Pernicka, 2002).

for the production of amphorae decorated with the stamps ΜΑΤΡΟ/ΒΙΟΥ and ΜΕΛΣΕ/ΩΝΟΣ. 2. The finds with the stamp ΜΙΚΙ/ΩΝΟΣ (3 amphorae), found in Varna, are grouped together and may also have been produced in one pottery workshop. These finds together with the sample with a stamp of ΗΓΗΣΙΝΟ[Υ] belong to the Parmeniskos group. Further samples from different sites (Varna, Sborianovo, Prilep) included in cluster 1 show that the relations between this unknown production center and Thrace were intensive. At this stage of the investigation we are not in the position to provide information on the origin of this group;

From the grouping of the investigated samples on the basis of the similarity in their chemical composition the following inferences can be made:

3. Amphorae with the stamp ΑΠΟΛΛΟΔΩΡΟΣ (6 samples) from different sites (Varna, Sboryanovo, and Pistiros) belong to three different clusters and, accordingly, are made from different raw materials, probably at different sites or workshops. It is also possible that this stamp was used to identify the production of workshops from the same chain of producers, but situated in geochemically different regions.

1. All investigated samples with the stamp ΑΝΤΙ/ΦΙΛΟΥ (7 amphorae) were produced from the same raw clay material, although they derive from different find sites (see Table 1 and Fig. 1). We believe that they were produced in the same workshop. The same raw material was also used


I. Kuleff, E. Pernicka and T. Stoyanov

Cluster (number) 1 (7) 2 (4) 3 (6) 4 (2) 5 (9)

6 (3) 7 (6) 8 (3) 9 (9)

10 (4) 11 (2) 12 (5) 13 (2) 14 (3) 15 (2) 16 (2) Outlier (4)

Member of the cluster


785.VAR 788.VAR, 790.VAR, 791.VAR 855.SBO, 857.SBO, 870.PRI 792.SBO 859.SBO, 861.DNB, 866.DNB 864.DNB, 871.PRI, 874.ORI, 875.ORI, 876.ORI, 877.ORI 810.SBO 856.SBO 780.SBO 781.BUR, 782.KAB, 783.KAB, 786.VAR, 787.VAR, 789.VAR, 815.PIS 784.KAB 793.SBO 807.SBO 852.SBO 799.SBO, 803.SBO 808.SBO 819.PIS 847.SBO, 867.PGR 805.SBO, 813.PIS 806.SBO 796.VAR, 797.SBO, 804.SBO 798.SBO 800.SBO, 801.SBO 817.PIS 818.PIS 860.DNB 853.SBO, 863.DNB, 865.DNB, 868.PGR 820.PIS, 821.PIS 794.SBO 802.SBO 858.SBO, 872.ORI, 873.ORI 862.DNB, 869.PGR 812.SBO 851.SBO, 854.SBO, 846.SBO, 848.SBO 849.SBO 850.SBO 795.SBO 809.SBO 811.PGR 816.PIS

ΗΓΗΣΙΝΟ[Υ] ΜΙΚΙ/ΩΝΟΣ ? Legible Ε and Υ ? ? ? Legible Ω and Σ ΔΗΜΗ[ΤΡΙΟΥ] ΜΑΤΡΟ/ΒΙΟΥ

Origin Pella (?) Chalkidiki ? ?






local Thasos ? ? local ? ? Thasos ? ? Thasos ? ?

Table.3. Membership of the investigated samples to clusters, formed on the basis of similarity in chemical composition.


On the Origin of Stamped Amphorae from Thracian Cities in Bulgaria 4. A similar conclusion is valid for the amphorae with stamp ΘΑΣΩΝ - 6 samples, all of which are found in Sboryanovo, where there is no doubt about the trade connections with Thasos. The samples with this stamp again belong to three different clusters and one amphora is among the outliers. It is interesting to note that the half of the amphorae with the stamp ΘΑΣΩΝ mainly cluster together with amphorae stamped with ΑΠΟΛΛΟΔΩΡΟΣ (clusters 7 and 9). (Amphorae with stamp ΘΑΣΩΝ can also be found in cluster 8 and one amphora stamped with ΑΠΟΛΛΟΔΩΡΟΣ is in cluster 14). This is an indication that the production of amphorae in Thasos, decorated with different stamps, was prepared using identical materials - possibly different workshops at the same site or at least within a small region drawing on the same clay sources.

1998) with the data for the chemical composition of the members of cluster 2 in Table 4. According to ceramic epigraphy the stamps ΑΝΤΙ/ΦΙΛΟΥ and ΜΕΛΣΕ/ΩΝΟΣ may belong to the Parmeniskos type (see e.g. Stoyanov 2003), and the membership of stamp ΜΑΤΡΟ/ΒΙΟΥ to the same cluster explicitly supports the affiliations of this stamp to the Parmeniskos group. From the formation of clusters 1 and 5, in which the members have stamps belonging to the Parmeniskos group, it is evident that this kind of stamp was used in at least two production centers using different raw materials. At this stage of the investigation we are not able to determine the location of these workshops. Among the amphorae analyzed by Slusallek et al. (1983) is one sample (№ 1285) also associated with the Parmeniskos group. The comparison between the chemical profiles of the amphorae associated with the Parmeniskos group investigated in the present work (cluster 1 and 5) and the chemical composition of the sample analyzed by Slusallek et al. (1983), does not show any similarity – the contents of Co, Hf, and Th were outside of the range of the two standard deviation interval. The same result was obtained by comparison of the profiles of clusters 1 and 5 to the whole group to which the sample 1285 belongs (see Slusallek et al. 1983).

5. The density of the clusters containing amphorae with different stamps show that they were produced from raw material with the same chemical composition. This means that these clay beds are situated in one and the same geochemical region. Therefore, although these amphorae were probably produced in different workshops, they are produced in the same production center. Based on the above considerations we propose that the production of amphorae with stamps ΗΓΗΣΙΝΟ[Υ] and ΜΙΚΙ/ΩΝΟΣ had been carried out in the same place. The same is also valid for amphorae with the stamps ΜΑΤΡΟ/ΒΙΟΥ, ΑΝΤΙ/ΦΙΛΟΥ and ΜΕΛΣΕ/ΩΝΟΣ, as well as ΚΛΕΟΦΩΝ, ΑΡΙΣΤΟ[ΜΕΝΙΣ]/…/ΘΑΣΙΩ[Ν] and ΔΑ]ΜΑΣ(ΤΙΣ)/ Θ]ΑΣΙΟΝ/Μ]ΥΕΣΚΟΣ together with some of the samples (800.SBO, 801.SBO) with the stamp ΘΑΣΩΝ. Taking into account that stamp ΘΑΣΩΝ indicates Thasos as a production center, the provenance of the members of cluster 7, 8, and 9 seems to be clearly identified with Thasos.

7. The archaeological information shows that amphorae with stamp ΘΑΣΩΝ are produced in pottery workshops on the island of Thasos. The investigation of the raw materials used to make Thasian amphorae support the existence of at least of two chemically and petrologically distinct clay sources on the island (see e.g. Jones 1986, Whitbread 1995). The samples with the stamp ΘΑΣΩΝ investigated in this study, are members of three different clusters. This gives us reason to propose that the amphorae with the stamp ΘΑΣΩΝ were produced not only on the island but probably also in workshops located on the continent in the region of southern Thracia or Macedonia (North Greece), the so called perea of the island of Thasos. Such a claim is also expressed by Bökert (1983). It is one of the possible explanations for the heterogeneity of the chemical composition of the amphorae from the Parmeniskos group mentioned by some of the investigators (e.g. Whitbread 1995).

6. At this stage of the investigation, the discussed results and the limited number of the investigated finds, do not give us enough evidence to support the hypothesis expressed by Stoyanov (2000, 2003, 2005), that the samples with the stamp ΜΑΤΡΟ/ΒΙΟΥ were produced in the region of the Bulgarian Black Sea coast, not far from the former Greek colonies of Mesambria and Appolonia (today Nessebar and Sozopol). The reason for this uncertainty is: a.

the forming of statistically significant cluster (cluster 5) including stamps ΜΑΤΡΟ/ΒΙΟΥ, ΑΝΤΙ/ΦΙΛΟΥ and ΜΕΛΣΕ/ΩΝΟΣ;


the difference in the chemical composition of the pottery produced from the clay near Nessebar and Sozopol (data in Kuleff et al. 1998) and the amphorae with the stamp ΜΑΤΡΟ/ΒΙΟΥ. This was established by comparison of the chemical profile of the pottery samples from the region of Nessebar (Kuleff et al.

8. Analysis of pottery samples from the Greek settlements belonging to the Thasos perea and determination of the chemical profile of the production may eventually lead to the identification of the production sites of amphorae from the Parmeniskos group. Conclusion The results of this study could be summarized as follows:


I. Kuleff, E. Pernicka and T. Stoyanov

Element As Ba Ce Co Cr Cs Eu Fe (%) Hf K (%) La Lu Na (%) Nd Ni Rb Sb Sc Sm Ta Tb Th U Yb Zr Element As Ba Ce Co Cr Cs Eu Fe (%) Hf K (%) La Lu Na (%) Nd Ni Rb Sb Sc Sm Ta Tb Th U Yb Zr

Cluster 1 (7)1 M±SD2 17.0±4.3 482±83 69.4±4.8 20.8±3.6 252±40 6.6±0.9 1.25±0.14 4.60±0.71 5.2±0.5 2.18±0.41 32.3±2.8 0.419±0.037 1.05±0.25 38±18 30±12 106±13 1.03±0.37 19.0±2.1 6.3±0.5 1.18±0.09 1.01±0.07 13.7±0.9 2.8±0.5 3.2±0.4 219±43 Cluster 6 (3) M±SD 12.7±2.9 741±276 54.9±13.9 15.9±1.9 136±42 4.0±0.4 1.14±0.14 3.97±0.54 4.3±0.6 1.66±1.03 28.7±4.8 0.330±0.025 0.49±0.17 24±14 11±3 58.6±4.7 0.81±0.20 16.5±2.6 5.2±0.7 0.86±0.08 0.81±0.19 7.9±0.6 1.7±0.1 2.6±0.2 161±52

RSD3 25.7 17.3 6.9 17.3 16.0 13.6 11.2 15.5 9.6 18.8 8.7 8.9 23.8 47.4 40.0 12.3 35.9 11.1 7.9 7.6 6.9 6.6 17.6 12.5 19.6

Cluster 2 (4) M±SD 9.6±7.2 290±108 58.5±8.2 28.4±6.6 311±106 7.5±4.2 1.08±0.12 4.54±0.28 3.9±0.8 2.36±0.55 31.8±6.4 0.349±0.057 0.75±0.25 30±9 30±9 107±18 0.69±0.3 19.6±1.9 5.1±0.8 1.05±0.31 0.84±0.13 11.3±1.8 2.4±0.5 2.3±0.3 156±56

RSD 75.5 37.2 14.0 23.2 34.1 56.0 11.1 6.2 20.5 23.3 20.1 16.3 33.3 30.0 30.0 16.8 4.3 9.7 15.7 29.5 15.5 15.9 20.8 13.0 35.9

Cluster 3 (6) M±SD 9.0±2.4 283±60 69.6±7.2 15.1±2.7 105±23 6.9±2.1 1.31±0.20 4.06±0.73 5.7±1.4 1.93±0.13 30.5±4.1 0.350±0.042 1.02±0.01 30±6 12±5 108±16 0.77±0.25 16.7±2.6 5.4±0.8 1.18±0.21 0.84±0.20 12.5±1.3 2.5±0.6 2.5±0.3 222±58

RSD 26.4 21.2 10.3 17.9 21.9 30.4 15.3 18.0 24.6 6.7 13.4 12.0 9.8 20.0 41.7 14.8 32.5 15.6 15.7 17.8 23.8 10.4 24.0 12.0 26.1

Cluster 5 (9) M±SD 10.4±3.6 466±45 55.1±2.9 12.6±1.0 96.6±7.2 4.6±0.3 1.14±0.06 3.67±0.28 4.7±0.4 2.52±0.34 26.1±1.7 0.287±0.038 1.45±0.16 32±9 8±2 104±16 0.46±0.07 15.1±1.1 5.2±0.3 0.85±0.07 0.69±0.10 10.0±0.6 2.9±0.4 2.2±0.4 181±19

RSD 34.6 9.7 5.3 7.9 7.5 6.5 5.3 7.6 8.5 13.5 6.5 13.2 11.0 28.1 25.0 15.4 15.2 7.3 5.8 8.2 14.4 6.0 13.8 18.2 10.5

RSD 22.8 37.2 25.3 11.9 30.9 10.0 12.3 13.6 14.0 62.0 16.7 7.5 34.7 58.3 27.3 8.0 24.7 15.8 13.5 9.3 23.5 7.6 5.9 7.7 32.3

Cluster 7 (6)1 M±SD2 10.2±3.5 484±137 90.7±13.6 8.4±1.9 64.7±19.3 7.8±2.6 1.36±0.29 2.93±0.53 5.6±0.8 2.50±0.53 47.8±7.2 0.418±0.014 0.89±0.22 43±14 11±5 101±30 1.74±1.0 12.1±2.5 7.3±1.4 1.46±0.09 0.97±0.19 18.7±2.1 2.7±0.8 3.1±0.4 223±60

RSD3 35.1 28.3 15.0 22.6 29.8 33.3 21.3 18.2 14.3 21.2 15.1 3.3 24.7 32.6 45.5 29.1 58.0 20.7 19.4 6.2 19.6 11.2 29.6 12.9 26.9

Cluster 8 (3) M±SD 27±24 456±37 81.4±1.7 9.0±1.9 66.4±12.4 6.6±0.6 1.38±0.25 3.01±0.20 5.3±0.4 0.91±0.31 38.3±6.3 0.322±0.43 0.60±0.07 39±4 10±3 98±28 1.78±0.71 13.0±0.9 4.6±1.1 1.44±0.21 1.06±0.23 18.2±2.4 5.8±0.7 2.1±0.3 225±23

RSD 88.9 8.1 2.1 21.1 18.7 9.1 18.1 6.6 7.5 34.1 16.4 13.4 11.7 10.3 30.0 28.6 39.9 6.9 23.9 14.6 21.7 13.2 12.0 14.3 10.2

Cluster 9 (9) M±SD 17.0±4.8 703±114 63.1±5.0 17.0±0.9 160.0±17.6 14.5±2.6 1.21±0.07 4.33±0.31 4.0±0.3 2.59±0.60 29.7±2.3 0.316±0.027 1.01±0.13 27±7 16±2 117±19 1.4±0.7 18.6±1.4 5.5±0.4 1.00±0.08 0.78±0.13 13.2±2.9 3.0±0.5 2.2±0.2 205±83

RSD 28.2 16.2 7.9 5.3 11.0 17.9 5.8 7.2 7.5 23.2 7.7 8.5 12.9 25.9 12.5 16.2 50.0 7.5 7.3 8.0 16.7 22.0 16.7 9.0 40.5


On the Origin of Stamped Amphorae from Thracian Cities in Bulgaria

Element As Ba Ce Co Cr Cs Eu Fe (%) Hf K (%) La Lu Na (%) Nd Ni Rb Sb Sc Sm Ta Tb Th U Yb Zr

Cluster 10 (4) M±SD 25±33 514±104 75.9±2.4 16.6±0.6 136.0±8.7 16.3±3.4 1.36±0.12 4.28±0.33 5.0±0.4 2.81±0.41 35.0±1.7 0.320±0.035 0.92±0.21 47±11 18±6 133±19 2.4±1.7 18.7±1.2 5.7±0.4 1.18±0.10 0.85±0.13 14.5±0.9 2.5±0.2 2.5±0.3 234±7

RSD 132.0 20.2 3.2 3.6 6.4 20.9 8.8 7.7 8.0 14.6 4.8 10.9 22.8 23.4 33.3 14.3 70.8 6.4 7.0 8.5 15.3 6.2 8.0 12.0 3.0

Cluster 12 (5) M±SD 9.5±12.9 431±60 81.2±7.1 21.4±4.0 229±49 11.1±2.2 1.43±0.09 4.99±0.59 5.1±0.7 2.25±0.57 41.3±6.6 0.425±0.036 0.86±0.18 43±14 31±7 131±23 1.14±1.06 21.4±2.3 7.0±1.0 1.51±0.16 0.97±0.10 16.9±4.1 3.2±0.5 3.2±0.2 215±25

RSD 135.8 13.9 8.7 18.7 21.4 20.0 6.3 11.8 8.6 25.3 16.0 8.5 20.9 32.6 22.6 17.6 93.0 10.7 14.3 10.6 10.3 24.3 15.6 6.3 11.6

Cluster 14 (3) M±SD 24.1±15.7 485±185 89.0±28.6 19.1±0.5 220±72 14.4±3.6 1.54±0.17 4.54±0.24 5.8±2.1 2.08±0.39 53.5±18.9 0.490±0.043 0.91±0.18 32±13 20±7 146±11 1.58±0.42 17.7±2.9 6.9±1.0 1.53±0.62 1.06± 19.9±9.5 6.2±2.7 2.7±0.3 258±123

RSD 65.1 38.1 32.1 2.6 32.7 25.0 11.0 5.2 36.2 18.8 35.3 9.0 19.3 40.6 35.0 7.5 26.6 16.4 14.5 40.5 23.1 47.7 43.5 11.1 47.7

1 = Number of samples in the cluster; 2 = Mean±standard deviation; 3 = Relative standard deviation in percent Table 4. Chemical profile of the clusters, formed on the basis of the similarity in chemical composition




Thasos on the production of other pottery centers on the northern Aegean coast.

Based on their chemical composition it is possible to differentiate the group of amphorae with stamps ΜΑΤΡΟ/ΒΙΟΥ, ΑΝΤΙ/ΦΙΛΟΥ and ΜΕΛΣΕ/ ΩΝΟΣ (cluster 5) from the amphorae with stamps ΗΓΗΣΙΝΟΥ and ΜΙΚΙ/ΩΝΟΣ (cluster 1), which had been attributed to the Parmeniskos group by Grace and Savvatianou-Petropouilakou (1970).

Acknowledgement The authors thank their colleagues M. Dontsheva (Archaeological Museum Varna), Prof. L. Getov (University of Sofia), L. Domaradzka and Ch. Preshlenov (Archaeological Museum Sofia), P. Kiashkina (Archaeological Museum Burgas) and D. Kozhukharov (Archeological Museum Nessebar) for their support and for samples used in this investigation.

Of interest, from the point of view of the production center or centers (?) of the Parmeniskos group of amphorae, is the membership of samples from amphorae identified as of Mendean production (samples 861.DNB; 864.DNB; 866.DNB). This observation could indicate that the production site of the Parmeniskos group should be sought between Chalkidiki and Pella (for which the largest archaeological data bank exists).

One of the authors (I.K.) is indebted to the VolkswagenStiftung for financial support, which gave him the opportunity to carry out this investigation. References

The stamp (sample 806.SBO) which indicates the production of Enos (not far from the mouth of river Maritsa), shows a typological similarity with the production of the pottery workshops from Thasos, which is confirmed by the similarity in the chemical composition of the raw materials (clay). The position of the sample 794.SBO illustrates the influence of

Akamatis, I., 1992, Η αγορά της Πέλλας κατά το 1989, ΑΕΜΘ, 3, 75-90 Akamatis, I., 2000, Ενσφράγιστες λαβές αμφορέων από την αγορά της Πέλλας. Balkanska, A. 1984, Amfori i amforni petshati, (Amphorae and stamp of amphora) (Bulg.), in Sevtopolis, I. Bit i kultura (Way of life and culture), (Sofia), 115-158.


I. Kuleff, E. Pernicka and T. Stoyanov Scientific Studies, Fitch Laboratory Occ. Paper 1, British School at Athens, (Athens). Kiachkina, P., 1994, Les contacts commerciaux des Thraces de la region de Burgas avec le monde Egéen (IVe – IIe s. av. J.C.) (D’aprés les trouvailles archéologiques dans le sanctuaire Thrace aux environs d’Aquae Calidae). Thracia Pontica V, 175-190. Kuleff, I., Djingova, R., 1990, Activation analysis in archaeology, in Activation Analysis (ed. Z. Alfassi), CRC-Press Inc., Boca Raton, 2, 427-489. Kuleff, I., Djingova, R. & Balabanov, P., 1998, Archaeometric investigation of pottery from the Thracian town Deultum (VIIV c. BC), Berliner Beiträge z. Archäometrie, 15, 199-236. Kuleff, I., Djingova, R. & Penev, I., 1986, INAA of pottery, Journal of Radioanalytical and Nuclear Chemistry, 99, 345358. Kuleff, I., & Pernicka, E., 2002, INAA of geological standard reference materials, Journal of Radioanalytical and Nuclear Chemistry, 251, 139-143. Mirchev, M., 1958, Amfornite peshtati ot museya vav Varna, (The amphorae stamps in museum in Varna), (in Bulg.), (Sofia). Sayer, E., 19