The archaeology of geological catastrophes

The Archaeology of Geological Catastrophes Geological Society Special Publications Series Editors A. J. HARTLEY R. E...

Author: Bill McGuire

127 downloads 1950 Views 44MB Size Report

This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!
Report copyright / DMCA form

DOWNLOAD PDF

The Archaeology of Geological Catastrophes

Geological Society Special Publications

Series Editors A. J. HARTLEY R. E. HOLDSWORTH

A. C. MORTON M. S. STOKER

Special Publication reviewing procedures The Society makes every effort to ensure that the scientific and production quality of its books matches that of its journals. Since 1997, all book proposals have been refereed by specialist reviewers as well as by the Society's Publications Committee. If the referees identify weaknesses in the proposal, these must be addressed before the proposal is accepted. Once the book is accepted, the Society has a team of series editors (listed above) who ensure that the volume editors follow strict guidelines on refereeing and quality control. We insist that individual papers can only be accepted after satisfactory review by two independent referees. The questions on the review forms are similar to those for the Journal of the Geological Society. The referees' forms and comments must be available to the Society's series editors on request. Although many of the books result from meetings, the editors are expected to commission papers that were not presented at the meeting to ensure that the book provides a balanced coverage of the subject. Being accepted for presentation at the meeting does not guarantee inclusion in the book. Geological Society Special Publications are included in the ISI Science Citation Index, but they do not have an impact factor, the latter being applicable only to journals. More information about submitting a proposal and producing a Special Publication can be found on the Society's web site: www.geolsoc.org.uk

It is recommended that reference to all or part of this book should be made in one of the following ways: McGuiRE, W. J., GRIFFITHS, D. R., HANCOCK, P. L. & STEWART, I. S. (eds) 2000. The Archaeology of Geological Catastrophes. Geological Society, London, Special Publications, 171. GUIDOBONI, E., MUGGIA, A. & VALENSISE, G. 2000. Aims and methods in Territorial Archaeology: possible clues to a strong IV century AD earthquake in the Straits of Messina (Southern Italy) In: McGuiRE, W. J., GRIFFITHS, D. R., HANCOCK, P. L. & STEWART, I. S. (eds) The Archaeology of Geological Catastrophes. Geological Society, London, Special Publications, 171, 45-70.

GEOLOGICAL SOCIETY SPECIAL PUBLICATION NO. 171

The Archaeology of Geological Catastrophes

EDITED BY

W. J. MCGUIRE

University College London, UK

D. R. GRIFFITHS

University College London/Institute of Archaeology, UK

P. L. HANCOCK University of Bristol, UK

I. S. STEWART Brunei University, UK

2000

Published by The Geological Society London

THE GEOLOGICAL SOCIETY The Geological Society of London was founded in 1807 and is the oldest geological society in the world. It received its Royal Charter in 1825 for the purpose of 'investigating the mineral structure of the Earth' and is now Britain's national society for geology. Both a learned society and a professional body, the Geological Society is recognized by the Department of Trade and Industry (DTI) as the chartering authority for geoscience, able to award Chartered Geologist status upon appropriately qualified Fellows. The Society has a membership of 8600, of whom about 1500 live outside the UK. Fellowship of the Society is open to those holding a recognized honours degree in geology or cognate subject and who have at least two years' relevant postgraduate experience, or who have not less than six years' relevant experience in geology or a cognate subject. A Fellow with a minimum of five years' relevant postgraduate experience in the practice of geology may apply for chartered status. Successful applicants are entitled to use the designatory postnominal CGeol (Chartered Geologist). Fellows of the Society may use the letters FGS. Other grades of membership are available to members not yet qualifying for Fellowship. The Society has its own publishing house based in Bath, UK. It produces the Society's international journals, books and maps, and is the European distributor for publications of the American Association of Petroleum Geologists, (AAPG), the Society for Sedimentary Geology (SEPM) and the Geological Society of America (GSA). Members of the Society can buy books at considerable discounts. The publishing House has an online bookshop (http:Jlbookshop.geolsoc.org.uk). Further information on Society membership may be obtained from the Membership Services Manager, The Geological Society, Burlington House, Piccadilly, London W1V OJU, UK. (Email: [email protected]: tel: +44 (0)171 434 9944). The Society's Web Site can be found at http://www.geolsoc.org.uk/. The Society is a Registered Charity, number 210161. Published by The Geological Society from: The Geological Society Publishing House Unit 7, Brassmill Enterprise Centre Brassmill Lane Bath BA1 3JN, UK (Orders: Tel. +44 (0)1225 445046 Fax +44 (0)1225 442836) Online bookshop: http://bookshop.geolsoc.org.uk First published 2000 The publishers make no representation, express or implied, with regard to the accuracy of the information contained in this book and cannot accept any legal responsibility for any errors or omissions that may be made. © The Geological Society of London 2000. All rights reserved. No reproduction, copy or transmission of this publication may be made without written permission. No paragraph of this publication may be reproduced, copied or transmitted save with the provisions of the Copyright Licensing Agency, 90 Tottenham Court Road, London W1P 9HE. Users registered with the Copyright Clearance Center, 27 Congress Street, Salem, MA 01970, USA: the item-fee code for this publication is 0305-8719/99/S 15.00. British Library Cataloguing in Publication Data A catalogue record for this book is available from the British Library. ISBN 1-86239-062-2

Typeset by Aarontype Ltd, Bristol, UK Printed by The Alden Press, Osney Mead, Oxford, UK

Distributors

USA

AAPG Bookstore PO Box 979 Tulsa OK 74101-0979 USA Orders: Tel. +1 918 584-2555 Fax +1 918 560-2652 e-mail: [email protected] Australia Australian Mineral Foundation Bookshop 63 Conyngham Street Glenside South Australia 5065 Australia Orders: Tel. +61 88 379-0444 Fax +61 88 379-4634 e-mail: [email protected] India Affiliated East-West Press PVT Ltd G-l/16 Ansari Road, Daryaganj, New Delhi 110002 India Orders: Tel. +91 11 327-9113 Fax+91 11 326-0538 e-mail: [email protected] Japan Kanda Book Trading Co. Cityhouse Tama 204 Tsurumaki 1-3-10 Tama-shi Tokyo 206-0034 Japan Orders: Tel. +81 (0)423 57-7650 Fax +81 (0)423 57-7651

TContents

Preface HANCOCK, P. L., CHALMERS, R. M. L., ALTUNEL, E., £AKIR, Z. & BECKER-HANCOCK, A. Creation and destruction of travertine monumental stone by earthquake faulting at Hierapolis, Turkey GRIFFITHS, D. R. Uses of volcanic products in antiquity JONES, R. E & STIROS, S. C. The advent of archaeoseismology in the Mediterranean BUCK, V. & STEWART, I. A critical reappraisal of the classical texts and archaeological evidence for earthquakes in the Atalanti region, central mainland Greece GUIDOBONI, E., MUGGIA, A. & VALENSISE, G. Aims and methods in territorial archaeology: possible clues to a strong fourth-century AD earthquake in the Straits of Messina (southern Italy) FRIEDRICH, W. L., SEIDENKRANTZ, M.-S. & NIELSEN, O. B. Santorini (Greece) before the Minoan eruption: a reconstruction of the ring-island, natural resources and clay deposits from the Akrotiri excavation DRIESSEN, J. & MACDONALD, C. F. The eruption of the Santorini volcano and its effect on Minoan Crete BICKNELL, P. Late Minoan IB marine ware, the marine environment of the Aegean, and the Bronze Age eruption of the Thera volcano RUSSELL, J. K. & STASIUK, M. V. Ground-penetrating radar mapping of Minoan volcanic deposits and the Late Bronze Age palaeotopography, Thera, Greece CIONI, R., GURIOLI, L., SBRANA, A. & VOUGIOUKALAKIS, G. Precursory phenomena and destructive events related to the Late Bronze Age Minoan (Thera, Greece) and AD 79 (Vesuvius, Italy) Plinian eruptions; inferences from the stratigraphy in the archaeological areas PARESCHI, M. T., STEFANI, G., VARONE, A., CAVARRA, L., GIANNINI, F. & MERIGGI, A. A geographical information system for the archaeological area of Pompeii CIONI, R., LEVI, S. & SULPIZIO, R. Apulian Bronze Age pottery as a long distance indicator of the Avellino Pumice eruption (Vesuvius, Italy) CHESTER, D. K., DUNCAN, A. M., GUEST, J. E., JOHNSTON, P. A. & SMOLENAARS, J. J. L. Human response to Etna volcano during the classical period KIRK, W. L., SIDDALL, R. & STEAD, S. The Johnston-Lavis collection: a unique record of Italian volcanism PLUNKET, P. & URUNUELA, G. The archaeology of a Plinian eruption of the Popocatepetl volcano GONZALEZ, S., PASTRANA, A., SIEBE, C. & DULLER, G. Timing of the prehistoric eruption of Xitle Volcano and the abandonment of Cuicuilco Pyramid, Southern Basin of Mexico TORRENCE, R., PAVLIDES, C., JACKSON, P. & WEBB, J. Volcanic disasters and cultural discontinuities in Holocene time, in West New Britain, Papua New Guinea RIEHLE, J. R., DUMOND, D. E., MEYER, C. E. & SCHAAF, J. M. Tephrochronology of the Brooks River Archaeological District, Katmai National Park and Preserve, Alaska: what can and cannot be done with tephra deposits DODGSHON, R. A., GILBERTSON, D. D. & GRATTAN, J. P. Endemic stress, farming communities and the influence of Icelandic volcanic eruptions in the Scottish Highlands

vii 1 15 25 33 45 71 81 95 105 123

143 159 179 189 195 205 225 245 267

vi

CONTENTS

DAY, S. J., CARRACEDO, J. C, GUILLOU, H., PAIS PAIS, F. J., BADIOLA, E. R., FONSECA, J. F. B. D. & HELENO, S. I. N. Comparison and cross-checking of historical, archaeological and geological evidence for the location and type of historical and sub-historical eruptions of multiple-vent oceanic island volcanoes GRATTAN, J. P., GILBERTSON, D. D. & DILL, A. 'A fire spitting volcano in our dear Germany': documentary evidence for a low-intensity volcanic eruption of the Gleichberg in 1783? JAMES, P., CHESTER, D. & DUNCAN, A. Volcanic soils: their nature and significance for archaeology SIDDALL, R. The use of volcaniclastic material in Roman hydraulic concretes: a brief review HUNT, P. Olmec stone sculpture: selection criteria for basalt HUGHES, R. & COLLINGS, A. Seismic and volcano hazards affecting the vulnerability of the Sana'a area of Yemen WAELKENS, M, SINTUBIN, M., MUCHEZ, P. & PAULISSEN, E. Archaeological, geomorphological and geological evidence for a major earthquake at Sagalassos (SW Turkey) around the middle of the seventh century AD STIROS, S. C. Fault pattern of Nisyros Island volcano (Aegean Sea, Greece): structural, coastal and archaeological evidence DE BOER, J. Z. & HALE, J. R. The geological origins of the oracle at Delphi, Greece Index

281

307 317 339 345 355 373 385 399 413

Preface The Archaeology of Geological Catastrophes brings together a diverse collection of papers that address the archaeological identification and cultural significance of large-scale geological events, mainly earthquakes and volcanic eruptions. Major earthquakes and volcanic eruptions typically recur at intervals of anything between a few decades to many tens to hundreds of thousands of years. Yet the instrumentation by which we record and monitor them has only been around for little over a century. To reduce the hazard posed by earthquakes and volcanism, we require a longer record of them than can be provided from modern instrumental snapshots. On the assumption that future earthquake and volcanic activPaul L. Hancock who sadly died during the ity will be like that of the recent past, completion of this volume. we need to understand the history of earthquakes and volcanism over millennial timescales. For the geologist, therefore, archaeology presents a potential tool to illuminate this time window, lying astride the documentary archives of historians and the geological archives of the surficial rock record. For the archaeologist, recognizing the impact of earthquakes or volcanic activity on a site or region more often provides a missing piece of the human history of that area, often explaining the conditions for cultural development or demise. In this context, an individual earthquake or volcanic eruption may often be the solution to an archaeologist's interpretation of inferred local or regional upheavals. By contrast, for the volcanologist or earthquake geologist, the identification of a major prehistorical seismic or volcanic event is generally the starting point from which to go on to derive other parameters (e.g. event magnitude, source etc.), or fit into regional models or datasets. For example, for archaeologists, the 464 BC earthquake at Sparta, Greece, was the trigger for a major change in political conditions in the Peloponnese region; for earthquake geologists it provides key evidence to estimate the earthquake energy released on a major fault that has since been seismically quiescent (Armijo et al. 1991). Furthermore, this difference is more than simply one of perspective, since it is generally underpinned by contrasting philosophical and theoretical frameworks, by varying methodological approaches, and by practitioners using distinct terminologies and presenting results in widely differing forums. In short, in many ways, archaeology and geology are fundamentally distinct disciplines. The distinction leads to uncertainty, or even suspicion, about studies that seek to integrate the two. Many seismologists, for example, will no doubt still

viii

PREFACE

empathize with comments made by Charles Richter 40 years ago, complaining that 'Ancient accounts of earthquakes do not help us much; they are incomplete, and accuracy is usually sacrificed to make the most of a good story' (Richter 1958, cited by Vita-Finzi 1986, p. 8). Guidoboni (1996) is more generous though equally cautious, noting that 'Earthquakes are not kind, and they do not care for researchers. Their traces can travel through strata and upset methods for dating in unexpected ways. This is one reason why so many important pieces of archaeological evidence are lost for seismology'. To some extent, these criticisms may be less resonant for volcanic archaeology investigations, since eruptive activity frequently leaves geochemically or petrographically distinctive 'event horizons' (e.g. Riehle et al. this volume), or even volcanic deposits that preserve the archaeological record more or less intact (e.g. Gonzalez et al. this volume). By comparison, destruction horizons produced by seismic shaking must compete with the often comparable debris traces of warfare and natural collapse of poor constructions. In this regard, the focus in this volume on 'catastrophes' is less to do with the assumption that these events are inherently more important (or interesting) for our understanding of recent geological history or of our cultural heritage. Instead, it reflects the recognition that it is the large-magnitude geophysical events that are most likely to leave the clearest signals in the archaeological record. The diverse ways in which investigators may interpret those signals is arguably the main theme of this volume. The Archaeology of Geological Catastrophes presents a broad spectrum of papers on the geoarchaeology of earthquakes and volcanoes, and here we draw attention only to a few general themes. Although earthquakes and volcanic eruptions are generally viewed as agents of destruction, numerous papers discuss their potential benefits to past cultures - providing materials for tools, building and sculpture, and even the fertile environmental conditions on which societies depended. Perhaps the most intriguing proposal is the suggestion that the power of the Delphic oracle to the ancient Greeks derived from the geological setting of the site, specifically from gaseous emissions from an underlying active fault. The bulk of contributions, however, focus on the destructive power of earthquakes and volcanoes. Several papers deal specifically with 'archaeoseismology' - the study of pre-instrumental earthquakes that, by affecting locations and their environments, have left their mark in the archaeological record. An important debate to emerge from these papers is whether major past earthquakes are more effectively recognized through regional disturbances in occupation or settlement patterns (territorial archaeology) or through the identification of 'diagnostic' structural indicators at individual sites. A suite of papers tackle different facets of arguably the most prominent geological catastrophe in the archaeological record - the Bronze Age eruption of Thera (Santorini, Greece) and its consequent regional impacts on Minoan culture. Human responses to major volcanic eruptions are also discussed, both in terms of local reactions to volcanism in Sicily and Mexico, and far-field effects, such as the impacts of Icelandic eruptive activity on agricultural demise in the Scottish Highlands. In turn, the value (and potential pitfalls) of historical records of past eruptive activity in documenting the capricious character of volcanism in an area are assessed in case studies from Italy, Germany, the Canary Islands and the Cape Verde islands. Other themes covered within the volume include the application of tephrachronology in volcanic archaeology, the value of volcanic soils in archaeological research, the use of geographic information systems in preserving vulnerable archaeological information

PREFACE

ix

at key cultural sites and the assessment of the vulnerability of important cultural centres to seismic and volcanic threats. To those that may dislike the eclectic character of this volume, the editors would argue that this only serves to reflect the rather disparate state-of-play within the burgeoning fields of earthquake and volcanic archaeology. Furthermore, the papers presented here show varying degrees of cross-disciplinary co-operation, but the bulk of the research is still largely being undertaken by archaeologists or by geologists working in relative isolation. It is hoped that by raising some important research questions, volumes like The Archaeology of Geological Catastrophes, will accelerate the move towards the type of interdisciplinary research advocated by Van Andel (1991, p. 324), in which historians, archaeologists and geologists (among others!) participate in a ' ... collaboration which assumes intensive exchange of information, ideas and procedures from the planning stage through to final publication'. Such collaborations are likely to be essential if the past societal impacts of earthquake and volcanic activity are to be effectively unravelled. References ARMIJO, R., LYON-CAEN, H. & PAPANASTASSIOU, D. 1991. A possible fault rupture for the 464 BC Sparta earthquake, Nature, v, 351. GUIDOBONI, E. 1996. Archaeology and historical seismology: the need for collaboration in the Mediterranean Area. In: STIROS, S. & JONES, R. E. (eds.) Archaeoseismology, Fitch Laboratory Occasional Paper 7 British School at Athens, Athens, Greece, 7-13.

VAN ANDEL

> T- H- 1991.Geo-archaeology and archaeological science. In: NICK-KARDULIAS, P. (ed.) Beyond the Site: regional Studies in the Aegean Area, University Press of America Inc. Maryland, 25-44. ViTA-FiNZi, C. 1986. Recent Earth Movements: an introduction to Neotectonics, Academic Press, London. j

in Stewart

London, December 1999

This page intentionally left blank

Creation and destruction of travertine monumental stone by earthquake faulting at Hierapolis, Turkey P. L. HANCOCK*'1, R. M. L. CHALMERS1, E. ALTUNEL2, Z. £AKIR3 & A. BECKER-HANCOCK4 1

Department of Geology, University of Bristol, Wills Memorial Building, Queen's Road, Bristol BS8 1RJ, UK Jeoloji Muhendisligi Bolumu, Muhendislik Mimarlik Fakiiltesi, Osmangazi University, Eskisehir, Turkey 3 Engineering Faculty, Mersin University, Ciftlik Koyu, Mersin, Turkey 4 Department of English, University of Bristol, 3-5 Woodland Road, Bristol BS8 1TB, UK Abstract: The presence of travertines adjacent to the city and their value for construction was well known to the Greek, Roman and Byzantine residents of Hierapolis (modern Pamukkale). The travertines were mainly extracted from quarries on the outer slopes of a low plateau below the city. The distinctive attribute of most of the quarries is that they are narrow but deep vertical-sided trenches. Each trench is the site of a nearly vertical fissure that was filled by banded fissure travertine, one type of so-called Phrygian marble. Trench walls, formerly the contacts between vertical banded travertines and outward dipping bedded travertines, bear a well-defined herringbone pattern of tool marks identical to those on many of the stone blocks that were used for building Hierapolis. Deposition of the travertines in 21 major fissure-ridges was a consequence of precipitation following ' degassing of carbonate-rich hot waters emerging from springs aligned along active faults and associated fissures. Whereas the dense and attractively banded travertine in fissures was principally used as an ornamental stone, the bedded travertines of ridge sides were mainly employed as a dimension stone and for making columns. After many of the monuments at Hierapolis had been constructed from travertine, itself a faulting-related material, some of them were subsequently destroyed or damaged by earthquake fault reactivation, which caused them to be either shaken or displaced. The zone of greatest seismic damage coincides with the trace of the Hierapolis fault zone, whose location was detected from an alignment of offsets of walls and petrified irrigation channels. The kinematic class of this fault zone could be deduced because offsets of the linear archaeological features permitted opening directions to be determined, thus allowing the fault zone to be reinterpreted as a normal fault zone achieving a small downthrow to the southwest. The knowledge that the Hierapolis fault zone is a structure across which there is active stretching and increased hydrothermal flow helps to explain why the present-day area of hot pools and travertine deposition is situated immediately downslope of the fault trace. If this relationship between displaced features and recent travertine deposits occurs elsewhere it might be employed for finding the locations of earthquake faults.

The purpose of this paper is to explain how earthquake faulting in the city of Hierapolis is associated with the deposition of a large body of travertine that was quarried for stone in Greek and Roman times, and how subsequent reactivation of faults in the same area during these periods and later was responsible for damaging the city. The site of Hierapolis, one of several * Deceased, reprint requests should be addressed to A. Becher-Hancock.

Greek and Roman cities in the Maeander River valley (the present Menderes River) (Fig. 1), is roughly coincident with the present tourist village of Pamukkale, a settlement that in the last 50 years has served the needs of visitors not only to Hierapolis but also to the famous white travertine deposits that give Pamukkale its name, 'cotton castle'. Excluding the extensive northern necropolis and smaller southern necropolis, much of Hierapolis lies within its late Roman city wall (Peres 1987) (Fig. 2).

From: McGuiRE, W. G., GRIFFITHS, D. R., HANCOCK, P. L. & STEWART, I. S. (eds) The Archaeology of Geological Catastrophes. Geological Society, London, Special Publications, 171, 1-14. 1-86239-062-2/OO/S 15.00 © The Geological Society of London 2000.

Fig. 1. (a) Turkish sector of the Aegean extensional province, (b) Geological, topographic and historical setting of Hierapolis within the Denizli basin (modified after Altunel & Hancock 19930).

CREATION AND DESTRUCTION OF TRAVERTINE MONUMENTAL STONE

3

Fig. 2. Sketch map of selected principal monuments within Hierapolis.

Neotectonic, topographic and historical setting Many landforms in regions of active extensional tectonics directly reflect recent earth movements and hence they can have a great influence on routeways and settlements. This is especially true of western Anatolia, which is situated in the east of the Aegean extensional province (Fig. la), a region currently experiencing normal faulting and the formation of rift valleys (grabens) and intervening horst-block mountains as a consequence of roughly NNE-SSW stretching (Jackson 1994). Hierapolis is sited on the northeastern edge of the Denizli basin, a structure within the Menderes graben but close to its confluence with the Gediz graben. The Denizli basin has been subsiding since Miocene time (Westaway 1993). It is framed to the south by a major E-W trending normal fault, which is part of the Menderes system, but to the northeast the principal faults

trend NW-SE, that is, they follow the Gediz graben trend. The Denizli basin and the Gediz graben are separated by a zone, to the northwest of Buldan (Fig. Ib), that does not contain large normal faults and hence is not expressed by a graben or basin. The floor of the Denizli basin is mainly underlain by Neogene and Quaternary clastic sediments. The Quaternary travertine masses of the basin, of which the Pamukkale mass is only one, rest on these clastic sediments. External to the Denizli basin there are outcrops of metamorphic and igneous basement rocks unconformably overlain by Neogene clastic deposits. The Pamukkale range-front fault separates the sediments of the Denizli basin from the basement rocks to the northeast, with a downthrow southwest of at least 450m (Altunel & Hancock 1993^). The large-scale topography of the region is a direct expression of Neogene and Quaternary tectonic activity. For example, the Denizli basin

4

P. L. HANCOCK ET AL.

is an area of low relief, the nearly level Quaternary flood plains of the Maeander and Lykus (modern £uruksu) rivers being at about 280m above sea level and about 60m lower than the surrounding incised plateau underlain by Neogene sediments. The mountain massif of Kuciik £6kelezdag to the northeast of the basin rises to a maximum of 1739m, whereas to the south of Denizli the higher mountain of Babadag reaches

2300 m and the even higher peak of Honazdag, south of Colossae, achieves 2571 m. Between the Denizli basin and the Alasehir valley of the Gediz graben is an area of high but not rugged ground, rising to about 1324m. The Pamukkale travertines, which were used for building Hierapolis, are situated within what we call the Pamukkale plateau but some parts of northeastern Hierapolis are sited on the

Fig. 3. Map of the distribution of morphological varieties of travertine and active faults in the Hierapolis area (based on Altunel & Hancock 19930). It should be noted that the large outcrop of actively depositing terraced-mound travertine near Pamukkale is sited in the hanging wall of the Hierapolis fault zone.


Fig. 4. Vertically banded fissure travertine cutting horizontal bedded travertine in the Yarikkaya fissure-ridge, about 1400m north of Develi.

lower slopes of the Pamukkale range front. The plateau is bounded to the northeast by the 300m-high Pamukkale range front that defines the southwest edge of the Kiiciik £6kelezdag massif. There are, except in the south, two levels within the Pamukkale plateau. Hierapolis is situated on the upper level, adjacent to the

5

Pamukkale range front, and about 30m above the lower level, which contains most of the travertine deposits (Fig. 3). The southern segment of the slope between the two terraces of this divided plateau is the site of the most spectacularly white of the actively depositing travertines. The major valleys, which coincide with the Menderes and Gediz grabens, were important routeways for peoples and armies travelling either west or east between the Aegean coastlands and the central Anatolian plateau, a region that gave access to the upper Euphrates valley and from there to Persia and further east (see Ramsay's (1890) map of routes in ancient Asia Minor). Although the Menderes graben provided the easiest pathway from the Aegean coast via Tralles (modern Aydin) and Nyssa to the interior, the Gediz graben was also a vital route connecting Magnesia (modern Manisa), King Croesus' city of Sardis and Philadelphia (modern Alasehir) in the Gediz graben to the cities of Tripolis, Hierapolis, Laodicea and Colossae in the Denizli basin. Because the Denizli basin lies at the confluence of two major routes from the coast to the interior and because it is a large area of relatively level and well-watered ground close to the Anatolian plateau it is no surprise that cities developed within it in ancient times. In addition to Hierapolis (of Greek foundation, although mainly Roman and Byzantine monuments remain), there were Tripolis, Laodicea and Colossae (Fig. Ib). Hierapolis was probably occupied before Classical times, its hot springs, both then and later, being a great attraction. Furthermore, the presence of a holy spring (the Plutonium next to the Temple of Apollo; see later discussion) in Hierapolis was critical to

Fig. 5. A vertical crestal fissure of approximately 5 m width within an inactive NW-trending fissure-ridge about 1500m north of Develi. (Note the gentle dip of the bedded ridge travertines away from the fissure, which is the site of a Roman quarry from which Phrygian marble was obtained.)

6


Fig. 6. Terraced-mound travertine with metre-scale pools, 800 m NE of Pamukkale village. Water supplying these pools issues from springs sited on the Hierapolis fault zone. (Note the palisades of stalactites fringing each pool.)

Fig. 7. A perched self-built and petrified water channel of approximately 10m height that is now ruptured, possibly as a result of earthquake ground shaking; about 1600m east of Develi.

maintaining the city's continuing importance throughout Classical times. It is also noteworthy that Herodotus (484-420 BC), who in Book 7 writes about this area, refers to a city in the present neighbourhood of Hierapolis as Cydrara, and later mention is also made by him to Hydrela, both places possibly being the city that we now think of as Hierapolis. Laodicea was a Greek city, mainly built of travertine, but with important Roman modifications in the form of aqueduct pipes bringing water from a spring in Denizli, 8km away, across a valley and into a water tower within the city centre. The pipes of this damaged water tower are now furred-up with calcareous deposits, testifying to the widespread presence of dissolved calcium carbonate in the ground waters of the entire Denizli basin. Colossae, which is much less well preserved than Hierapolis or Laodicea, was, again, both a Greek

and Roman city. Tripolis, a mainly Greek city in the extreme northwest of the Denizli basin, is also built of travertine quarried from a small mass within the footwall block of the rangefront fault. Until excavated, the city was largely covered by slope deposits derived from the range front. Herodotus (1954), in Book 7 states that this is the route taken by Xerxes, who is reputed to have discovered, near what was probably Tripolis, a plane tree so beautiful that he decorated it with golden ornaments. Travertine deposits In this paper we use the term 'travertine' to embrace all 'freshwater' limestone products of deposition from hot carbonate-rich spring waters irrespective of whether they are compact (i.e. travertine as often defined) or whether they are


7

Fig. 8. Plan and profiles of the £ukurbag ridge (900 m east of Develi), the profiles emphasizing that Roman (and possibly later) quarrying has given rise to trench-like excavations where vertically banded fissure-ridge travertine has been selectively extracted.

porous, and might be called 'tufa' (Ford & Pedley 1996). Neogene clastic sediments are the most abundant materials underlying the Pamukkale plateau but the most distinctive rocks are travertines of Quaternary age, mainly less than 400 ka (Altunel & Hancock 19930). Although the older travertines are of middle Pleistocene age (>400ka), travertines are still being deposited, testifying to the continued action of hydrothermal flow in this area of active faulting. All the travertines are products of the degassing and

consequential precipitation from hot carbonaterich waters that emerge from springs aligned along fissures and faults that opened during the late Quaternary stretching of the Denizli basin. Stretching was also responsible for the increments of slip on the normal faults that frame the basin. The Pamukkale range-front fault is the closest of these faults to the travertines of the Pamukkale plateau. Of the five types of landform constructed of travertine (Altunel & Hancock I993a,b, 1996),

Fig. 9. A well-defined trench-like quarry corresponding to the central fissure of a ridge 900m NNW of Pamukkale (visible in the background). The bedded travertines of the ridge are higher on the east side of the fissure because it expresses the location of an underlying normal fault downthrowing west.

8


Fig. 10. Part of an abandoned Roman column cut from bedded travertine in the £ukurbag ridge, about 900 m east of Develi.

Fig. 11. A toppled wall constructed of travertine blocks quarried from bedded travertine. Toppling of this wall to the northeastern side of the Colonnaded Street probably occurred during an earthquake. (Note that a small petrified water channel (arrows) formed on the upstanding edges of one row of fallen blocks.) three are especially important from the perspective of the construction and destruction of Hierapolis (Fig. 3): (1) The 21 fissure-ridges, mainly younger than 80 ka, each comprise a crestal fissure filled by vertical colour-banded travertine cutting white- to yellow-bedded travertine dipping away from ridge crests (Figs 4 and 5). Ridges range in length from about 100 to 1500m in width from about 5 to 500m, and in height they rise up to 25m above the surrounding nearly flat land of the lower level of the Pamukkale plateau. (2) Terraced-mound travertines, many of which are still accumulating, mantle the hillslopes between the upper and lower levels of the Pamukkale plateau down which the carbonaterich waters have flowed and cascaded. Where they are being actively deposited these travertines are snow-white but where they are dry they

have weathered to an unattractive dark brown to black colour. Numerous hot pools encircled by palisades of travertine stalactites characterize the southern slopes of the main mass of terraced-mound deposits near Pamukkale village (Figs 3 and 6). (3) Self-built channel travertines, the sites of most of which were artificially determined, are wall-like features of l-2m width developed where carbonate-rich, hot spring waters flow in a confined channel used for irrigation or other purposes. Degassing during turbulent flow leads to the precipitation of travertine on the floors and walls of these channels, which grow in height until the channel becomes perched far above its original level. Some of these petrified water channels date from Roman times because: (a) Vitruvius described them at the time of Augustus (27BC-AD14) (D'Adria, in Peres


9

subsurface water flow, as Muir Wood (1993) has suggested happens elsewhere in the world, is not known. Thus whether travertine deposition at Hierapolis was greater or less at such times is a major question requiring an answer. Travertine as an ancient building material

Fig. 12. Fissures cutting the walls of part of the Northern Baths, a monument built of bedded travertine blocks. Fissuring is a characteristic form of damage in dry masonry walls shaken during an earthquake. 1987), and (b) Hierapolis is the centre of the network they define. Some channels, such as those passing through the Northern City Gate or the Byzantine Basilica, are clearly younger; indeed, some of them are still in use. On the outer and steeper slopes of the Pamukkale plateau some petrified irrigation channels have grown as high as 10m (Fig. 7). The genetic connection between travertine deposition and earthquake faulting is that the stretching responsible for normal faulting also opened associated vertical fissures striking parallel to the faults (Altunel & Hancock 19930). It is mainly these fissures that have allowed the carbonate-rich hydrothermal waters to rise to the surface. In addition, some waters ascend via the steeply inclined fault zones, some of which curve to become nearly vertical fissures close to the ground surface. The Pamukkale area and surrounding region are characterized by abnormally high heat-flow values. This is vividly reflected in the geothermal field near Cubukdagi, about 15 km to the west of Pamukkale (Simsek & Okandan 1990). Whether individual slip increments on faults are accompanied by changes in

Much of Roman and Byzantine Hierapolis is built of travertine extracted from quarries in the fissure-ridge deposits situated below the city. The distinctive attribute of these quarries is that many of them are narrow (2-10m) but deep (5-20 m) vertical-sided trenches. Each trench is the site of a nearly vertical fissure that was filled by banded fissure travertine, one type of the socalled Phrygian marble (Figs 5, 8 and 9). Trench walls, formerly the contacts between vertical banded travertines and outward dipping bedded travertines, display scaffolding holes and many are decorated by a herringbone pattern of tool marks identical to those on sarcophagi and many of the blocks that were used for constructing buildings at Hierapolis. The dense and attractively banded travertine from fissures was principally used as an ornamental stone, whereas the bedded travertine from ridge sides was mainly employed as a dimension stone and for making columns (Fig. 10). Bean's (1971) remark that local marble was not widely used in Hierapolis does not accord with our experience, unless he was referring only to the banded travertine from fissures. Notable buildings constructed of bedded travertine blocks include, from north to south (Fig. 2): (1) the Northern Roman Baths dating from the second-third centuries and including a Byzantine (fifth-century) Basilica built within it; (2) the Hellenistic Theatre, which was largely demolished in the first century AD when the Roman Theatre was built; (3) the Monumental (Frontinus) Gateway dedicated to Domitian, dating from the end of the first century AD; a structure that was rebuilt after a devastating earthquake; (4) the Nymphaeum adjacent to the Monumental Gateway; (5) the early Christian Martyrion in honour of St Philip from the end of the fourth or beginning of the fifth century; (6) the Northern Gate through the city wall; (7) the Temple of Apollo; (8) the Roman Theatre; (9) the Southern Baths (now a museum); (10) a Byzantine Basilica of the sixth century; (10) a 12th-13th century Byzantine fort; (11) the sixth-century Southern City Gate. In addition, most tombs in both the Northern and Southern Necropolis, beyond the city walls, are built of travertine blocks, as is the city wall itself. According to D'Adria (in Peres 1987)

Fig. 13. Plan of the Hierapolis fault zone. The shapes of buildings are schematic and not all modern buildings are shown (after Hancock & Altunel 1997). It should be noted that in section A-B the petrified water channel is downfaulted in a mini-graben where it has been stretched over two normal faults that are oriented roughly at right angles to the line of the channel; the appearance of this mini-graben is illustrated in Fig. 15.


11

Earthquake history of Hierapolis

0

50

Fig. 14. Sketch plan of a petrified irrigation channel that has been offset at two places by the Hierapolis fault about 90 m SSE of the northern city wall. It should be noted that where the channel locally trends E-W the offset as determined from piercing points combines horizontal opening with sinistral motion, and where it locally trends N-S opening is combined with a dextral sense of motion. These apparently contradictory senses of horizontal offset have arisen because the observed 'strike-slip' sense is controlled by the angle between a channel long axis and the opening direction across the trace of the fault. There is also an overall downthrow of about 7 cm to the west across the fault (after Hancock & Altunel 1997).

blocks from many buildings were cannibalized to construct the city wall, which was built as a result of a law introduced in AD 396, that is, at a time late in the Roman Empire. Furthermore, many of the Doric columns lining the sides of the Colonnaded Street (Via Domizianea) are of travertine and directly comparable with the stone of the abandoned column at the £ukurbag quarry (Fig. 10).

Although faulted archaeological features are readily analysed indicators of deformation patterns associated with earthquakes they may be more difficult to employ as guides to their timing. Historical earthquake catalogues are needed for this aspect of their analysis. The scholarly catalogue of pre-tenth century AD earthquakes by Guidoboni et al. (1994) indicates that the following earthquakes after the birth of Christ but before the 10th century were destructive at Hierapolis: 47, 60, an unknown date in the third century, an unknown date in the fourth century, 494, and early in the seventh century. It is also possible that an earthquake recorded by Guidoboni et al. (1994) as having occurred in about 27 BC, which was responsible for rebuilding work at Laodicea, might have also damaged Hierapolis, less than 10km from Laodicea. Soysal et al. (1981) also reported the AD 60 earthquake in their catalogue, and added to it events of MSK intensities VIII and VII in 65 and 20 BC, respectively. Most archaeological and historical writers focus on the AD 60 earthquake; for example, Peres (1987) recorded that Hierapolis was rebuilt after it, as did Bean (1971), and McDonagh (1989) also reported that Laodicea was rebuilt after the event. Earthquakes of MSK intensity VII or more that affected the Hierapolis area between the 10th and 20th centuries include those of 1354, 1651, 1703, 1887 and 1899, according to Soysal et al. (1981), Ates & Bayiilke (1982) and Ambraseys (1988). In the 20th century, the only event of Ms greater than 6.0 to have affected the Hierapolis area is that of 1900 reported by Ergin et al. (1967) and Gencoglu et al. (1990).

Earthquake damage at Hierapolis Many of the monuments that had been constructed of travertine related to older Quaternary episodes of faulting were then destroyed or damaged by reactivation of the faults, which caused continued earthquake shaking or ground displacement. For example, the toppled wall (Fig. 11) part of the Colonnaded Street, collapsed without losing its essential form and then became the foundation of a later petrified water channel, which follows part of its length. Vertical fissures, which elsewhere are regarded as a characteristic form of earthquake shaking damage (Stiros 1996), rupture the walls of the thirdcentury Northern Baths (Fig. 12). Examples of buildings reported by D'Adria (in Peres 1987) as having been reconstructed after the

12


Fig. 15. A slightly raised petrified irrigation channel cut by two fissure-faults (arrows) reflecting the locations of underlying normal faults that dip towards each other. The nearer of the faults, which achieves the greater displacement, downthrows west; the further but subordinate fault downthrows east. A section along the channel close to the gendarme post is shown in section A-B in Fig. 13.

Fig. 16. The so-called 'sacred pool' containing a submerged section of the Roman Colonnaded Street and fallen columns. The pool is sited on the trace of the Hierapolis fault zone; the flooding of the hollow possibly is a result of the subsidence of a small graben within the fault zone. A thin veneer of travertine is being unconformably deposited on the columns, which, being aligned, might have toppled during an earthquake.

AD 60 earthquake in Nero's reign include the Temple of Apollo and the Roman Theatre. In the southeastern Necropolis a sarcophagus made of travertine was overturned by earthquake shaking (presumed here to be the AD 60 event because Ronchetta (in Peres 1987) described it as 'the' earthquake). The zone of greatest earthquake damage coincides with the trace of the Hierapolis fault zone, whose location was detected from an alignment of offset walls and petrified irrigation channels (Fig. 13) (Hancock & Altunel 1997). The kinematic class of this fault zone could be determined precisely from the many offsets of the linear archaeological features, which are cut by vertical fissures expressing faults that are steeply

inclined a few metres below the surface (Hancock & Altunel 1997) (Fig. 13). The offset archaeological features permitted piercing points, and hence opening directions, to be determined, thus allowing the fault zone, previously thought by Altunel & Hancock (19930) to be a sinistral strike-slip, to be reinterpreted as a normal fault zone achieving a small downthrow to the southwest. In plan, the sense of the angle between a channel and the trace of a fault determines whether a channel is offset horizontally in a dextral or sinistral sense (Fig. 14). Where the line of a channel is subparallel to the opening direction there will be horizontal opening and a normal component of displacement of the channel. This gives rise to a mini-graben, where a

CREATION AND DESTRUCTION OF TRAVERTINE MONUMENTAL STONE subsidiary normal fault that is antithetic to the main one also cuts the channel (Figs 13 (section line A-B) and 15). The knowledge that the Hierapolis fault zone is a structure across which there is active stretching and increased hydrothermal flow, as reflected by the concentration of hot springs in the zone, helps to explain why the area of greatest present-day travertine deposition is situated in its immediate hanging wall and just downslope of the fault trace (Fig. 3). In the once-beautiful 'sacred pool' (now within the Pamukkale Motel and not be confused with the holy cavern of the Plutomium) a thin veneer of travertine is being unconformably deposited on columns that have fallen alongside a submerged paved area, possibly a continuation of the Colonnaded Street. The pool, possibly sited on a small graben within the Hierapolis fault zone, is being fed from a hot spring within the fault zone. The nearly uniform direction of toppling of the fallen columns might be a reflection of earthquake shaking (Fig. 16) (Nur & Ron 1996), and, if this is so, the deposition of travertine on the columns testifies to the intimate relationship between faulting, ground rupture, earthquake shaking and travertine deposition. Hierapolis, like Delphi (Greece), possesses a holy cavern about which legends have grown up. At Delphi the Oracle's mantic sessions might have been a result of the Oracle inhaling light hydrocarbon gases arising from buried limestones rich in bitumens (De Boer 1999). At Hierapolis, the legend that animals and men, other than certain priests, who entered the holy cavern, known as the Plutomium (adjacent to the Temple of Apollo) died in the so-called strong-smelling 'steams' we think is likely to be a consequence of the concentration of carbon dioxide in addition to other gases in such a small subsurface chamber into which hot waters flowed. The Plutomium's proximity to the eastern branch of the Hierapolis fault zone means that beneath it there is likely to be a higher than normal concentration of vertical fissures that are the conduits for such waters (Fig. 13). Summary (1) The travertine deposits at Hierapolis are secondary products of the area having been stretched during earthquake normal faulting for at least 400 000 years. (2) The bedded travertines of fissure-ridges have been quarried since before Roman times for dimension stone and were the principal building materials used by Greeks, Romans and Byzantines in the construction of monu-

13

ments. Finely banded travertines quarried from the vertical fissures cutting ridges were mainly used for ornamental purposes. (3) Many of the monuments and other features built of travertine have been damaged by earthquake shaking or faulting during the period since at least AD 60. Damage is concentrated along a narrow corridor coincident with the Hierapolis fault zone. (4) Opening directions determined from piercing points defined by displaced features allow the Hierapolis fault zone to be reinterpreted as a normal fault zone. The area of contemporary greatest travertine deposition is just downslope and in the immediate hanging wall of the fault zone. If this relationship occurs elsewhere it might be employed for finding the locations of earthquake faults. On site, P. Arthur of the University of Lecce explained to us the significance of numerous archaeological monuments. We thank the Universities of Bristol, Osmangazi and Mersin, and the Natural Environment Research Council of the UK for grants supporting our research.

References ALTUNEL, E. & HANCOCK, P. L. 19930. Active fissuring and faulting in Quaternary travertines at Pamukkale, western Turkey. Zeitschrift fur Geomorphologie Supplementary Volume, 94, 285-302. 19936. Morphological features and tectonic setting of Quaternary travertines at Pamukkale, western Turkey. Geological Journal, 28, 335-346. 1996. Structural attributes of travertine-filled extensional fissures in the Pamukkale plateau, western Turkey. International Geology Review, 38, 768-777. AMBRASEYS, N. N. 1988. Engineering seismology. Earthquake Engineering and Structural Dynamics, 17, 1-105. ATES, R. C. & BAYULKE, N. 1982. The 19 August 1976 Denizli, Turkey, earthquake: evaluation of the strong accelograph record. Bulletin of the Seismological Society of America, 72, 1635-1649. BEAN, G. E. 1971. Turkey Beyond the Maeander. Ernest Benn, London. DE BOER, J. Z. 1999. Could emission of light hydrocarbon gases have played a role in the mantic sessions at Delphi (Greece)? This volume. ERGIN, K., GUCLU, U. & Uz, Z. 1967. A Catalog of Earthquakes for Turkey and Surrounding Area (HAD to 1964AD). ITU Faculty of Mining Engineering, Istanbul. FORD, T. D. & PEDLEY, H. M. 1996. A review of tufa and travertine deposits of the world. EarthScience Reviews, 41, 117-175. GEMCOGLU, S., INAN, E. & GULER, H. 1990. Tiirkiye'nin Deprem Tehlikesi (Earthquake Hazard of Turkey). Chamber of Geophysical Engineers of Turkey, Ankara.

14


GUIDOBONI, E., COMASTRI, A. & TRAINA, G. 1994. Catalogue of Ancient Earthquakes in the Mediterranean Area up to the 10th Century. Institute Nazionale di Geofisica, Rome. HANCOCK, P. L. & ALTUNEL, E. 1997. Faulted archaeological relics at Hierapolis (Pamukkale), Turkey. Journal of Geodynamics, 24, 21-38. HERODOTUS 1954. The Histories, Book 7 trans Selvincourt, Pengiun Books. JACKSON, J. 1994. Active tectonics of the Aegean region. Annual Review of Earth and Planetary Sciences, 22,239-271. McDoNAGH, B. 1989. Blue Guide: Turkey: the Aegean and Mediterranean Coasts. A. & C. Black, London. MUIR WOOD, R. 1993. Neohydrotectonics. Zeitschrift fur Geomorphologie, Supplementary Volume, 94, 275-284. NUR, A. & RON, H. 1996. And the walls came tumbling down: earthquake history in the Holyland. In: STIROS, S. & JONES, R. E. (eds) Archaeoseismology. Fitch Laboratory Occasional Paper, 7, 75-85.

PERES, A. (ed.) 1987. Hierapolis di Frigia 1957-1987. Settore Cataloghi D'Arte del Gruppo Editoriale Fabbri, Torino. RAMSAY, W. M. 1890. The Historical Geography of Asia Minor. John Murray, London. SIMSEK, S. & OKANDAN, E. 1990. Geothermal energy development in Turkey. Transactions of the Geothermal Research Council, 14, 257-266. SOYSAL, H., SlPAHIOGLU, S., KOLCAK, D. & ALTI-

NOK, Y. 1981. Turkiye ve £evresinin Tarihsel Deprem Katalogu (M.6.2100-M.S. 1900). (Historical Earthquake Catalog of Turkey and its Environment, 2100BC to 1900AD). TUBITAK Publications, Ankara. STIROS, S. C. 1996. Identification of earthquakes from archaeological data: methodology, criteria and limitations. In: STIROS, S. & JONES, R. E. (eds) Archaeoseismology. Fitch Laboratory Occasional Paper, 7, 129-152. WESTAWAY, R. 1993. Neogene evolution of the Denizli region of western Turkey. Journal of Structural Geology, 15, 37-53.

Uses of volcanic products in antiquity D. R. GRIFFITHS Institute of Archaeology, University College London, 31-34 Gordon Square, London WC1H OPY, UK Abstract: Since the advent of mankind many human societies have lived in volcanically active zones. The geological, archaeological and historical records provide a rich and diverse source of evidence for both archaeology and volcanology concerning the nature of volcanic processes and the effects of volcanism on the environment and on human society. To achieve a balanced understanding of the effects of volcanism on past cultures, it is important to consider the attractions as well as the hazards of life in an actively volcanic zone. This paper gives an overview of some of the ways in which a wide range of volcanic products were used by mankind in antiquity. These include the use of volcanic rocks as stone tools, as substrates for rock carvings, as materials for building and sculpture, as millstones, as additives to make cements that set under water, and, more indirectly, as precursors of fertile earths for agriculture and as sources of metals and semi-precious stones. The paper also considers some of the properties of volcanic products that may have made them attractive. Assessment of the interaction of past cultures with volcanism is highly relevant to the present: it can provide the temporal perspective needed to deal appropriately with the human aspects of contemporary and future volcanic hazards.

Much consideration has been given to the effects of volcanic events on past societies. This ranges from investigating legends of Atlantis and lost civilizations to the detailed debates in recent decades on the effects and dates of eruptions of the volcano Thera (or Santorini). In these and other considerations of the interplay between mankind and volcanoes, the dramatic aspects of volcanic activity understandably receive the most attention and volcanoes are seen primarily as agents of destruction. Dramatic and destructive volcanic events tend to leave their mark clearly in the geological, archaeological and, on occasion, the historical record, whereas the effects of day-to-day normal levels of volcanic activity are more subtle and harder to discern. Destruction is, however, far from being the only aspect of volcanic activity worthy of attention in studying the past relationships of mankind with volcanoes. Dramatic and destructive volcanic events may leave their mark but they are comparatively rare in day-to-day human experience, even in highly volcanically active areas. Generations might live their entire lives reaping the benefits of their volcanic location while being aware of the potential for destruction only through history or folklore. This paper attempts to provide a brief overview of the ways in which mankind has in the past used the natural resources made available by volcanic activity. It will also consider some of

the properties of these volcanic products that have made them seem attractive and useful. Awareness of the benefits mankind has derived from the use of volcanic products may encourage a more positive awareness of the beneficial aspects of living in a volcanic environment when the complex interactions of past societies with volcanic activity are being considered. Achieving a proper balance between our awareness of the beneficial aspects of volcanism and of its dangers is obviously important in interpreting the interaction of past societies and volcanism. This balance is important not only for our understanding of the past: it is also highly relevant to the present. A detailed and balanced understanding of past events can provide the temporal perspective needed to deal appropriately with the human aspects of contemporary and future volcanic hazards. Although dramatic volcanic events will cer-, tainly disrupt and sometimes end the lives of those who live within reach of their effects, it is worth considering that in the greater scheme of things these disasters may be seen primarily as triggers of change. The destructive effects of a volcanic event may partly or wholly destroy a particular culture but at the same time it may allow the remnants of that culture to re-evolve in a new direction. The more profound the destruction, the less certain the outcome, at least in terms of the survival or re-evolution of the

From: McGuiRE, W. G., GRIFFITHS, D. R., HANCOCK, P. L. & STEWART, I. S. (eds) The Archaeology of Geological Catastrophes. Geological Society, London, Special Publications, 171, 15-23. l-86239-062-2/00/$ 15.00 © The Geological Society of London 2000.

16

D. GRIFFITHS

culture immediately affected by the volcanic event. It is also the case that the destruction or dispersal of one human group by a volcanic event may make way for another to rise to new heights in the new circumstances. A loss for an individual person, a family, a community or a culture may present an opportunity for gain to another such entity. Volcanic events may thus be seen as triggers of change, natural events (natural disasters to those who suffer) that cause a sudden discontinuity in the rise and fall of communities or civilizations. In the, longer term, such disasters and discontinuities may have provided the possibility of more rapid advancement of human civilization than might have been possible under the more restrictive circumstances of an established order. It may not be absurd to ask whether, in the development of mankind as a whole, volcanic events might not on occasion have accelerated the process. Taken together with some of the observations presented here on the usefulness of volcanic products, it is not inconceivable that living in a volcanic zone might be conducive to the development of human civilization. Although it is worth while to indicate the greater questions that might be in part elucidated by the subject of the paper, they are not the central concern here. Where volcanic and human activity are interrelated, the volcanic materials and events can provide a potentially rich source of archaeological information on the human procurement of raw materials (quarrying practice, routes and means of transport, etc.) and the dates of archaeological events. This in turn may serve to assist our understanding of past human behaviour and society (trade and exchange, organization of society, rates and geographical direction of change, etc.). Similarly, there may be occasions when archaeologically datable assemblages of artefacts in association with volcanic deposits can assist in the dating and improved understanding of volcanic processes, such as the effects of ashfalls on the development of soils and agriculture (Olson 1983). Although the immediate aim is to draw attention to the wide variety of use that mankind has made of natural resources that derive from volcanic activity, one of the greater goals of this paper (and many others in this volume) is to improve our understanding of the complex interactions of past societies with the effects and products of volcanism. In this area of endeavour at least, understanding of the past should provide far more than purely academic rewards: understanding of the effects of volcanic events on past societies should also improve enormously our ability to plan for and deal with the

human aspects of volcanic hazards now and in the future. The way of life and the material possessions of some modern people may seem very different from those of the past but, in the event of a major volcanic disaster, the basic human needs may seem remarkably unchanged. Knowledge of the past may well provide the vital information to enable us to make an informed decision on how best to adapt to volcanically driven societal and environmental change.

Criteria for the selection of volcanic materials in the past In general, the criteria used for selecting a particular raw material will depend on the role the material it is intended to fulfil in a particular time and place. These criteria may include the accessibility or availability of the material, its workability, durability, aesthetic qualities and the like. Physical properties and aesthetic qualities are often intricately intertwined in the perception of the artist-craftsman. Before considering individually the volcanic resources that have been used and the more obvious aspects of their properties that may have made them attractive, it is worth noting the possibility that in the past the criteria directing the selection of materials for a particular purpose may not have been the same as they might be now. Methods available for extracting, working or transporting a material may have strongly influenced the material chosen. It should also be noted that there might in the past have been constraints on access to given geographical regions imposed by the extents of political, economic or military control. It is also true that particular sources or uses of materials might not yet have been discovered. Slightly less obviously, social or religious taboo might effectively restrict accessibility. Furthermore, it is possible that the desirability of given materials might be influenced by fashion or by symbolic significance, be that symbolism political, cultural, mystical, magical, religious or some combination of these. Archaeological evidence for the influence of such criteria in determining the selection of materials is usually circumstantial. It may thus be appropriate to consider at the outset the possible symbolic or metaphysical significance of volcanic resources so that this possibility may be borne in mind when considering more physically based criteria for selecting volcanic materials for particular functions. Volcanic activity has probably always been an object of awe in the minds of all those

USES OF VOLCANIC PRODUCTS IN ANTIQUITY who behold it, awe which may have manifested itself as fear and reverence. It would not seem unreasonable that past cultures might have perceived volcanoes as all-providing, all-engulfing instruments of mighty power and, to some, vengeance. The act of witnessing that some materials were the products of volcanic activity may have caused those materials to assume mystical or symbolic significance in the perceptions of eyewitnesses and their society and descendants. Even if volcanic events are rare on the scale of a human lifetime, they are often dramatic enough to ensure the preservation of their memory in history, be that in a formal written sense, as oral tradition or as what might be termed folklore. In some cases visual depictions may also survive. The choice of volcanic materials for use in a given context may thus have been influenced by the very fact that they were known to be of volcanic origin. It may be worth distinguishing between symbolic significance that is based on and associated with the knowledge that a given material is of volcanic origin, and symbolic significance that is based on some other aspect such as the appearance of the material, the circumstances of its discovery or its geographical origin. A material that is difficult to acquire might be intended to symbolize wealth or power rather than anything religious or mystical. In these cases, the user may be wholly ignorant of the fact that the material is more or less directly the product of volcanic activity. Despite our attempts at rationalizing motives for choosing particular volcanic materials, the reality may often be far more complex than our interpretation, and combinations of practical and symbolic criteria may influence selection of a particular material. It is often difficult to determine whether or not symbolic or metaphysical factors may have had an influence in the selection of raw materials. Part of the difficulty in inferring what criteria may have influenced the selection of particular materials arises from the nature of archaeological evidence. Archaeology gathers its information from the material remains of past human activity. This evidence relates most directly to past material culture, and it is very difficult to proceed from this with any degree of certainty to draw conclusions about the thoughts associated with past actions or the motivation driving those actions. Archaeology may (generally with some difficulty) be able to discover some aspects of what people did, where they did it, when they did it and how they did it, but to discover why they did it is the hardest question of all, a question that cannot generally be answered with any certainty.

17

Although the pitfalls of too ready a recourse to ritual explanation are fairly obvious, there are a number of examples of possible ritual influence in the choice of raw material that may be worthy of note. One of the most remarkable may be the vast heads carved out of basalt boulders by the Olmecs. It is possible that the carvers or their predecessors had seen such boulders flung from erupting volcanoes and made their carvings mindful of the volcanic origin of the boulders. Another example of the choice of a volcanic material possibly being influenced by ritual or symbolic considerations may be the choice of andesite, a dark grey volcanic rock whose phenocrysts glisten in the sun, for the Inca Koricancha or Temple of the Sun in Cuzco, Peru. Field work undertaken with Hunt and Protzen in 1988 and subsequent petrographic analysis strongly suggests that this material was transported from a quarry at Rumiqolqa 35km away when ample alternative sources of building stone were near at hand and nearer sources of andesite existed (Hunt 1990). In this case, even though there might be a symbolic element in the selection of the material, it is possible that the builders were unaware of the volcanic origin of the material. Numerous other examples of a possible symbolic or ritual element in the choice of volcanic stone might be cited but the great majority would suffer from the common problem that although archaeology may be able to establish many facts about life in past times, one can never be certain of people's motivations in acting as they did.

Stone tools made of volcanic rock Many of the earliest known stone tools, believed to be about 2.3 Ma old, were made in east Africa by flaking volcanic lava (Musty 1999). The fine-grained, isotropic texture of many lava samples means that the distribution of stress within the material upon the application of a force is fairly predictable and the mechanical strength is similar at all points in all directions. The way in which the material will flake when struck is therefore fairly predictable and a good degree of control over the flaking process can be achieved with practice. The more homogeneous the mechanical properties of the material, the sharper the edges of the flakes are likely to be, as the fracture surfaces will tend towards perfect conchoidal fracture and not be deflected by intergranular weaknesses or differences in tensile

18

D. GRIFFITHS

strength (Griffiths et al 1987; Cotterell & Kamminga 1990). Perfect conchoidal fracture can be obtained in samples of volcanic glass that have solidified as a single phase. Many samples of volcanic glass do approach this ideal although some contain crystalline phases. Wholly or almost wholly vitreous acidic rock is termed obsidian. This has been a highly sought-after volcanic product for millennia, being transported over considerable distances from the sources. The excellent flaking qualities of obsidian mean that it can be used to produce extremely sharp cutting edges, although these are rather brittle. The predictability of fracture means that long blades and very fine, precisely shaped flaked stone objects can be produced. Given the high degree of skill necessary to produce some of the finest objects and their fragility, some such objects may have had more symbolic than practical use, but this possibility should not overshadow consideration of the practical aspects of the appeal of obsidian. The main use of obsidian has been for tools and weapons, but it was also used for carving figurines and for mirrors in Mexico (Kunz 1971, pp. 204-205) and for Neolithic period mirrors at £atal Huyiik in Anatolia (Shackley 1977, p. 54).

One example where knowledge of volcanic origin, mechanical properties and good visual contrast might each have had an influence may be found in the rock carvings on pahoehoe lava at Puuloa in the district of Puua on Hawaii. Topographic location may also have had an influence (Eleftheriou 1990). These carvings may show up particularly well because of textural contrast between the glassy rapidly cooled surface and the more granular more slowly cooled interior exposed by the carving. Similar contrast is seen with carvings made on the walls of lava tubes. Other rock carvings on Hawaii and neighbouring islands are made on large boulders, and here their initial visibility is attributable to the colour contrast between the weathered surface and the unweathered interior of the rock. Other rock carvings on the Hawaiian islands are, however, made on sedimentary rock. This highlights the difficulty of determining the motives behind the selection of a volcanic material for a function: was the selection fortuitous, was there an element related to the fact that the material was known to be of volcanic origin, was the selection based on the physical properties of the material or were other factors dominant in the minds of the people who chose to carve where they did?

Rock carvings on volcanic rock substrates

The use of volcanic rock as building stone

Thousands of rock carvings on more or less flat rock surfaces (as opposed to three-dimensional sculptures) survive from antiquity in many parts of the world. It is sometimes interesting to question whether particular preference may have been shown for using volcanic as opposed to other rock as a substrate for the depiction of images. Certainly, there are many examples from around the world of carvings on volcanic substrates. In some cases, the choice of substrate may have been influenced by knowledge of the volcanic origin of the rock. There are, however, many other possible criteria that might have influenced the selection of particular substrates including geographical location or orientation. Two further factors that might cause a particular rock type to be favoured are the mechanical properties of the rock and the visual contrast between the worked line or area and the pre-existing uncarved rock surface. A number of examples of the occurrence of rock carvings on volcanic substrates might be cited where one or both of these factors may have caused the rock to be favoured.

The use of andesite to build the Koricancha in Cuzco was mentioned above but there are many other andesite buildings in Cuzco and elsewhere where the possibility of a symbolic aspect in the choice of raw material may be less likely. Studies of the provenance of the rock using thin-section petrology to compare fragments from buildings with samples from known ancient quarry sources and geological exposures, showed convincingly that the stone used by the Incas to build the Koricancha and various other buildings in Cuzco was being imported from some distance over very difficult terrain (Hunt 1990). It is clear that the Incas were not using the ample stone material that was available near at hand and that had seemed acceptable to their predecessors as a material for construction of Sacsaywayman which overlooks Cuzco. What is less clear is the balance of criteria that drove them to select andesite from quarries 35km away from the town. Volcanic origin might have played a role but other qualities include the aesthetic appeal of phenocrysts glittering in the sunlight against the matt blackness of the fine-grained groundmass of the rock.

USES OF VOLCANIC PRODUCTS IN ANTIQUITY More pragmatic considerations of the workability of the rock may also have influenced the selection of andesite as the stone of choice. Initial extraction of andesite from at least some of the quarries is facilitated by the existence of large-scale fractures in the rock, perhaps arising from thermal stresses as the extruded rock originally cooled. Although andesite is hard it is considerably easier to work than one might at first expect. It is brittle and being largely fine grained exhibits fairly poor but still useful conchoidal fracture. Protzen has demonstrated that an approximately rectangular block, of andesite can be fashioned fairly quickly and without great labour from an irregular chunk extracted from a quarry. Protzen showed the author how a large hammerstone can be bounced on the surface of the andesite block removing some material by crushing on impact. If the hammerstone is given a slight flick towards one as it is about to hit the block, the transverse motion often removes flakes from the surface of the relatively brittle fine-grained andesite. This greatly enhances the rate of removal of material and hence the speed of shaping the blocks (Protzen 1986). Extraction and rough forming are thus fairly easy. Many hammer stones and partially finished blocks of andesite were found at quarry sites. The final and most impressive stage was the fashioning of the blocks so that large structures could be built without any mortar, each block sitting in an irregular but smoothly contoured depression on top of the course below, perfectly formed to receive the shape of the next block without leaving room for even a razor blade to be inserted. Thousands of other examples of ancient building and carving in andesite or basalt may be found elsewhere in the world, the vast Buddhist monument of Borobodur in Java being but one. Kempe (19836, p. 96) cited many instances of ancient buildings constructed from andesite or basalt but said of the latter: 'its use, however, has been restricted to areas where other more suitable stones are not available'. This contrasts with Hunt's suggestion (1991, and a specific instance of his wider arguments developed in a paper in this volume) that in many parts of the world, perhaps partly for metaphysical reasons, basalt and andesite may have been rocks of choice for various building and sculptural purposes. This illustrates how widely people may differ when it comes to inferring the motivation of past cultures, although Kempe himself elsewhere admitted that: 'Rocks and minerals have long appealed to man's mythical or superstitious sense, as Pliny and Agricola amply testify' (Kempe 1983a, p. 78.)

19

The use of pyroclastic deposits for making hydraulic cements Pyroclastic deposits of various forms (ash, tuff, pumice, etc.) or other crushed volcanic materials were used in antiquity for making cement that would set under water by reaction of lime with the water and with silicates and aluminates in the volcanic material (Torraca 1988). Pure lime cement hardens by drying of the slaked lime paste followed by reaction with atmospheric carbon dioxide to form a carbonate cement: drying is required for hardening. Hydraulic cements, on the other hand, set by reaction with water and no carbon dioxide is needed. This means that large blocks of hydraulic cement or concrete may be preformed or cast in situ without the concern that air needs to reach the centre of the block to permit hardening. Hydraulic cement is water resistant once set. The incorporation of volcanic material in a cement makes it hydraulic because the volcanic material provides a source of silicates and aluminates in a reactive form. Their reactivity may derive from a combination of the presence of relatively unstable and internally stressed glass phases resulting from rapid cooling and a high specific surface area resulting from the fine particulate state of some components together with the highly vesicular nature of others such as pumice and frothy lava. (The vesicles result from the exsolution of gases dissolved in the liquid rock.) Hydraulic cements incorporating volcanic earths in a slaked lime mortar were used in the Hellenistic period around the fourth century BC. The Romans used hydraulic cements widely in marine and other architecture, incorporating pumice in the hydraulic concrete used to build the dome of the Pantheon in Rome and thereby reducing its bulk density (Torraca 1988). Materials (natural and artificial) that can produce a hydraulic reaction with slaked lime are called pozzolanas after the town of Pozzuoli, near Naples, that was a famous source of pozzolanic earths used by the Romans to make hydraulic cements (Torraca 1988). Further discussion of the use of volcanic materials in Roman hydraulic cements is provided by Siddall (this volume). Other uses of pumice In addition to its use as a weight-reducing component of the Pantheon dome, pumice was also used in Roman times to remove bodily hair, as an abrasive and by scribes for smoothing the ends of rolls of parchment (Shackley 1977, pp. 30-31).

20

D. GRIFFITHS

The use of volcanic rocks as millstones and quernstones Criteria influencing the choice of material for making millstones can be seen to be practically based. The use of vesicular volcanic rocks as quernstones and millstones is widespread in many regions throughout the world. This choice of vesicular volcanic material, although far from universal, seems far too widespread in time and place to be a reflection of anything other than the selection of a particular type of raw material for a particular task. In addition to the advantages of vesicularity discussed below, factors such as ease of extraction and forming (alluded to above in describing the advantages of andesite and basalt as building stones) may have contributed to the popularity of these rocks as quernstones and millstones. Fine-grained volcanic rock may also be resistant to the plucking of mineral grains that might contaminate flour from, for example, a sandstone millstone (Hunt & Griffiths 1992). The author hypothesizes that as a vesicle (which arises from the exsolution of gases from a cooling but still liquid volcanic rock) is first worn into by one stone rubbing against the other, the vesicle will at first act like the teeth of a cheese grater, cutting slivers from the passing cereal grains. As a given vesicle becomes progressively more worn down it tends towards becoming an approximately hemispherical depression in the millstone surface, and it ceases to present an acute-angled peripheral cutting edge to passing cereal grains. The hollow may act as a holder that holds a cereal grain static with respect to one millstone while dragging it across the edges of the open vesicles on the opposing stone. As a vesicle wears down further still, it becomes a shallow depression with an oblique-angled periphery. Too shallow to hold a grain in place, the worn down vesicle may still act as a hollow into which a passing grain may partially sink to be bruised or broken as it collides with the blunt periphery of the vesicle. At any one time, each face of a pair of millstones will present many vesicles in each of these stages of wear. It is emphasized that the proposed mode of action is at present purely speculative. Irrespective of the mechanisms of milling, however, the transport of vesicular millstones far from their sources is a testament to their effectiveness (see, e.g. Williams-Thorpe & Thorpe 1988). Fertile agricultural soils derived from volcanic deposits Weathering of volcanic deposits is widely recognized as having the potential to result in highly

fertile agricultural soils. Reasons behind this may include the fact that some of the constituents of the deposits are particularly susceptible to leaching of mineral ions into solution in the soil water. Provided these dissolved ions are retained within the soil water rather than being washed away, they provide a ready source of plant mineral nutrients. The presence of relatively unstable and perhaps internally stressed glass phases together with a high ratio of solid surface area to soil water solution volume facilitates relatively rapid rates of movement of ions into solution. Many other factors are, however, important in the formation of fertile soils (Nahon 1991). It also needs to be recognized that, even where conditions are favourable, the formation of rich soils will generally take many hundreds of years of weathering and soil development. In the shorter term of a few hundreds of years, lava flows and heavy ash falls generally yield relatively sterile ground (Olson 1983). Hydrothermal deposits associated with volcanism The process of subduction of oceanic crust generates volcanoes in the overlying crust, earthquakes from below the subduction zone and the concentration of great mineral wealth through deposition from hydrothermal waters rising from the subducted crust. This mineral wealth is thus associated with the same processes that create volcanoes and earthquakes along subduction zones. Whether or not people in antiquity associated the volcanoes, earthquakes and the mineral deposits is uncertain and perhaps unlikely. Volcanoes can be long extinct whereas the mineral deposits remain, and other processes than subduction can give rise to precious metal concentration. The vast amounts of silver extracted in the Potosi region of Bolivia during and before the Spanish conquest are, however, the result of hydrothermal processes in a subduction zone, and similar rich metal deposits exist in many regions around the Pacific Ocean. Where hydrothermal waters are driven to near the surface by subduction zone processes or by other processes, extensive deposits of minerals such as noble metals and metal sulphides leached from the rocks below may form as the water cools and/or boils because of the pressure drop near the surface. Escape of hydrogen sulphide can result in precipitation of gold, cinnabar (mercury sulphide) and realgar (arsenic sulphide). Other processes can result in precipitation of gold or copper (Hibbard 1995, pp. 395-423).

USES OF VOLCANIC PRODUCTS IN ANTIQUITY Where hydrothermal waters reach the surface precipitates may form and the water may have been used for health spas or baths. Where gases and vapours reach the surface, sulphur and other minerals may be deposited and reactions may form haematite and other minerals (Hibbard 1995, pp. 424-427). Materials such as bright red cinnabar might be used medicinally, for personal adornment or as a pigment. Similar possibilities exist for a number of substances that may result from hydrothermal activity, and such activity is often associated with volcanic activity. Much use was made in antiquity of the red chalcedony semi-precious stone cornelian or carnelian. It is formed by precipitation in amygdaloidal (literally almond-shaped) cavities in vesicular lavas from silica-rich hydrothermal or meteoric waters percolating through the rock (Hibbard 1995, p. 413). Bloodstone and agates were also prized and were similarly formed. Much secondary material eroded out of the host lava may be found in rivers and along beaches, however, so users were not necessarily aware that these materials were derived from volcanic lavas.

Fire derived from volcanoes We have dealt hitherto in this paper primarily with the solid raw material resources made available to people in antiquity by the agency of volcanic eruption. In some instances, however, the gift of fire may have been one of the more apparent resources received by mankind from volcanoes in antiquity. The high temperature of volcanic emanations often gives rise to fire in trees that stand in the path of lava. Lava may have been an important early source of fire in some parts of the world, the fire once being fed with fuel to keep it burning. Fire itself came to have manifold uses for humanity, from providing warmth and light, providing power in many forms (first over animals and enemies, later to drive machines), to permitting the modification of materials (e.g. changing the colour of pigments and jewels) and the manufacture of entirely new materials with entirely new properties such as ceramics and metals. This process of development continues to this very day. The gift of fire as such may not have been the only contribution of volcanic activity to the creation and growth of pyrotechnology. Volcanoes may also have played a role as mother of invention, a source of inspiration by example. It is not at all unlikely that the effects of heat in creating ceramics may have been noted where

21

lava flows baked adjacent earth. Nor is it impossible that the creation of metals by natural smelting of ores by volcanic activity may on occasion have been observed by people in antiquity. The archaeological importance of volcanic materials The brief overview presented in this paper has sought to show that for a variety of reasons volcanic products have been very widely used by mankind throughout antiquity. It follows from this that the study of artefacts made from volcanic material provides a potentially very valuable source of archaeological information. These studies can answer basic questions concerning provenance and technology of manufacture, and, in some cases, provide dates for archaeological events. This basic information provides clues for the drawing of many higher-level inferences about such topics as cultural interaction, trade systems and criteria influencing the choice of raw materials for particular purposes, and rates of change and development. Identification of the geographical source of raw materials gives information about the movement of materials in antiquity, which in turn may indicate cultural contacts. More detailed studies of distribution may give insight into transport mechanisms (e.g. marine, riverine or by land) and mechanisms of trade (e.g. single-step procurement or multiple stages of exchange). For example, obsidian analysis undertaken with the aim of determining provenance has been the source of much archaeological attention in recent decades. Perhaps predictably, some of the earlier clear-cut attributions of provenance have been clouded by increasing awareness of variation of composition within a source and by increasing awareness of the occurrence of obsidian in secondary contexts. Improved sampling and analytical strategies are being developed to address these problems. Volcanic materials also provide many archaeologically important opportunities for dating. Some of these approaches provide a date through the stratigraphic association of archaeological artefacts or human remains with datable volcanic deposits. Volcanic activity results in the formation of new materials as molten material crystallizes or vitrifies. This new material is often free of radiogenic gases and free of radiation damage but from the moment it solidifies radiogenic argon and radiation damage start to accumulate. In such circumstances, potassium-argon dating can be used to date the solidification by

22

D. GRIFFITHS

radiogenic gas accumulation, and thermoluminescence, electron spin resonance or fission-track dating can date the solidification by accumulation of radiation damage. Radiation damage dating can also be applied to material heated by volcanic emanations sufficiently to anneal away previous radiation damage. Living organisms killed by volcanic events during the last 40 ka years or so may provide material suitable for radiocarbon dating. All of these approaches date a volcanic event and may by association provide archaeologically useful dates. Although archaeology has much to gain from studying aspects of volcanic materials and events, it is not a one-way process. Archaeological and historical evidence may provide dates that may be useful in dating associated volcanic events or material. Thermoluminescence, electron spin resonance or radiocarbon dating may again be used to date archaeological samples if circumstances permit. Although less accurate because of the influence of environmental factors such as temperature, other dating techniques involve measuring the extents of chemical changes that occur after a volcanic product such as an andesite block or an obsidian flake has a fresh surface exposed to the environment by human shaping and it starts to weather or hydrate (Hunt 1991; Shackley 1998). Further work is needed but progress is being made. Archaeological or historical information may also serve as a valuable source for improving the understanding of the effects on habitation, society, environment and agriculture of geologically recent volcanic events of various types. Understanding the past effects of volcanic events may help us to mitigate the ill effects of comparable events in the future. Conclusions It is clear that in antiquity mankind made considerable use of a wide range of volcanically derived resources. Physical, practical, economic, political, symbolic and metaphysical factors may all have played a role in influencing the selection of particular materials under particular circumstances. The extent of use of volcanic materials in the past makes them potentially a very rich area for future archaeological research. It is hoped that this brief attempt to increase our appreciation of the breadth of mankind's use of volcanic resources will stimulate further interdisciplinary studies in the future. It is clearly important that the positive and attractive aspects of volcanic activity, some of which have been noted in this paper, should be

considered alongside the destructive aspects in developing a fuller understanding of the complex influence of volcanic phenomena on human societies in antiquity. A fuller understanding of the detailed nature and effects of a variety of past volcanic events should significantly improve our ability to deal effectively with present and future volcanic disasters.

References COTTERELL, B. & KAMMINGA, J. 1990. Mechanics of Pre-industrial Technology. Cambridge University Press, Cambridge. ELEFTHERIOU, T. 1990. An endangered legacy: ancient rock art of Hawaii. BA dissertation, Institute of Archaeology, University College London. GRIFFITHS, D. R., BERGMAN, C. J., CLAYTON, C. J., OHNUMA, K., ROBINS, G. V. & SEELEY, N. J. 1987. Experimental investigation of the heat treatment of flint. In: SIEVEKING, G. DE G. & NEWCOMER, M. H. (eds) The Human Uses of Flint and Chert. Cambridge University Press, Cambridge, 43-52. HIBBARD, M. J. 1995. Petrography to Petrogenesis. Prentice-Hall, Eaglewood Cliffs, NJ. HUNT, P. N. 1990. Inca volcanic stone provenance in the Cuzco province, Peru. Papers from the Institute of Archaeology, 1, 24-36. 1991. Provenance, weathering and technology of archaeological basalt and andesite. PhD dissertation, Institute of Archaeology, University College London. 1999. Olmec stone sculpture: selection criteria for basalt. This volume. & GRIFFITHS, D. R. 1992. The suitability of basalt, andesite and other volcanic stone for querns, millstones and grinding purposes. Quern Study Group Newsletter, 2, 4-6. KEMPE, D. R. C. 19830. Raw materials and miscellaneous uses of stone. In: KEMPE, D. R. C. & HARVEY, A. P. (eds) The Petrology of Archaeological Artefacts. Clarendon Press, Oxford, 53-79. 19836. The petrology of building and sculptural stones. In: KEMPE, D. R. C. & HARVEY, A. P. (eds) The Petrology of Archaeological Artefacts. Clarendon Press, Oxford, 80-153. KUNZ, G. F. 1971. The Curious Lore of Precious Stones. Dover, New York. MUSTY, J. 1999. The earliest tool makers. Current Archaeology, 164, 312. NAHON, D. B. 1991. Introduction to the Petrology of Soils and Chemical Weathering. John Wiley, New York. OLSON, G. W. 1983. An evaluation of soil properties and potentials in different volcanic deposits. In: Sheets P. D. (ed.) Archaeology and Volcanism in Central America. University of Texas Press, Austin, 52-56.

USES OF VOLCANIC PRODUCTS IN ANTIQUITY PROTZEN, J.-P. 1986. Inca stonemasonry. Scientific American, 254, 80-88. SHACKLEY, M. 1977. Rocks and Man. George, Allen & Unwin, London. SHACKLEY, M. S. 1998. Current issues and future directions in archaeological volcanic glass studies. In: SHACKLEY, M. S. (ed.) Archaeological Obsidian Studies: Method and Theory. Advances in Archaeological and Museum Science, 3, 1-14.

23

SIDDALL, R. 1999. The use of volcaniclastic material in Roman hydraulic concretes: a brief review. This volume. TORRACA, G. 1988. Porous Building Materials, 3rd edn. ICCROM (International Centre for the study of the Preservation and Restoration of Cultural Property), Rome. WILLIAMS-THORPE, O. & THORPE, R. S. 1988. The provenance of donkey mills from Roman Britain. Archaeometry, 30, 275-290.


The advent of archaeoseismology in the Mediterranean R. E. JONES1 & S. C. STIROS2 1

Department of Archaeology, Gregory Building, University of Glasgow, Lilly bank Gardens, Glasgow G12 8QQ, UK 2 Department of Civil Engineering, University of Pair as, Pair as, Greece Abstract: This paper presents a brief historical overview of the development of archaeoseismology from the observations of Lanciani at ancient sites in Rome, of Kritikos in Athens, Evans at Knossos and Blegen at Troy, to the emergence in the last years of the twentieth century of archaeoseismology as a distinct sub-discipline of palaeoseismology. Some current issues are explored, beginning with major seismic events such as that in AD 365 in the Eastern Mediterranean whose effects were geographically widespread but uneven in their destructive severity; generalizations are hard to come by, and each case has to be examined on its own merits. The need to examine the suitability of building methods and materials in areas of seismic risk is emphasized. Finally, the contribution of seismic events to destruction horizons in two contrasting cases in the prehistoric Aegean is considered: at Mycenaean centres in the Argolid in the 13th-12th centuries BC, and in the Peloponnese at the end of Early Bronze II.

The seismicity of the Mediterranean needs little introduction. It is a phenomenon transcending time and affecting many parts of this region, a fact of life even, awesome and terrifying in its potential destructive power. Despite the notable progress that has been made in recent decades in calculating seismic risk, this sense of the unknown when a tremor is about to strike applies perhaps almost as much to the region today as it surely did in antiquity. In this brief non-technical paper, the impact of seismic events on the archaeological record in parts of the Mediterranean, mainly the Aegean, is examined, together with how archaeologists and seismologists have interpreted the archaeological record there for evidence of seismic damage or destruction. This well-trodden path can, we believe, bear further exploration because the implications that may follow upon an interpretation of ancient destruction by seismic or other natural, rather than anthropogenic, causes can be considerable. From sometimes uncertain but by no means always naive beginnings earlier this century has emerged the sub-discipline of palaeoseismology, archaeoseismology. This has sharpened and refined the whole approach of identifying past earthquakes during the last few millennia, and of placing tectonic activity into a broader context. For seismologists, the new subdiscipline has provided the means for expanding the historical and instrumental seismological record. As discussed below, what surely hastened

its emergence in Mediterranean archaeology was the controversy generated by two major events (amongst many others): the collapse of the Mycenaean world, and the destruction in AD 365. Historical overview Our recent research suggests that one of the earliest instances in which seismic damage was explicitly invoked to explain the archaeological record at an excavation was in central Greece in a notoriously seismically active area near Atalanti (Fig. 1), scene of the great series of shocks in April 1894, the first measuring 7.0 in magnitude in Athens. Blegen (not dated, 1926), the US excavator renowned for his work at Troy in the 1930s and later at the Mycenaean palace at Pylos in SW Greece, explored in 1911 the area around Atalanti for indications of small Greek states known from classical texts, in particular Opous and Ion. Below the hill of Oion near Kyparissi he excavated city walls that were, unbeknown to him, remarkably close to the 1894 seismic fault. What he observed were some undulations and tilting in the foundations in the walls, which he suggested, were due to earthquakes, and there was also evidence of repairs (Fig. 2). This early work was to become just a foretaste of what has emerged in the last 15 years of field work: the fifth century BC stoa at Kyparissi with its foundations deformed by the activity of the

From: McGuiRE, W. G., GRIFFITHS, D. R., HANCOCK, P. L. & STEWART, I. S. (eds) The Archaeology of Geological Catastrophes. Geological Society, London, Special Publications, 171, 25-32. l-86239-062-2/00/$15.00 © The Geological Society of London 2000.

26

R. E. JONES & S. C. STIROS

Fig. 1. Map showing locations of the sites mentioned in Greece.

Locris fault, deformed tombs at a Hellenistic cemetery at Lamia, and at Kynos a late prehistoric site (Stiros & Dakoronia 1989; Dakoronia 1996; Buck & Stewart this volume). The first person, however, to make a systematic study of seismic effects 'fossilized' in archaeological remains and, furthermore, to make a typology of the indications of earthquake activity was Lanciani. As one of the foremost Italian scholars concerned with meticulous study of ancient Rome, he wrote about earthquake activity in that city, such as the parallel arrangement of the collapsed columns and their capitals from the main forum of the port of Rome at Ostia (Fig. 3) (Lanciani 1918). This observation, taken together with his recording in an earlier publication of the inundations of the River Tiber in recent historical times and the consequent

damage caused, point to the significance he attached to natural destructive phenomena in antiquity. In Athens, meanwhile, the earliest identification of archaeoseismic events is probably that by N. Kritikos (Galanopoulos 1956), who explained the sinusoidal offset of columns in the fifth century BC temple of Hephaistion (the Thesion) in terms of seismic deformation. Decades later, this type of seismic deformation was modelled through analytical calculations (e.g. Sinopoli 1991; Augusti & Sinopoli 1992). Earthquake damage at the Minoan Palace at Knossos on Crete is well known. In the course of the excavations at the Palace from the turn of the century, Evans was struck by the notable feature of a sequence of destruction levels associated structurally with deformed or more usually collapsed walls. But it was over two decades

ARCHAEOSEISMOLOGY IN THE MEDITERRANEAN

27

Fig. 2. Blegen's excavations at Kyparissi in 1911. Left, wall on the west side of trench IV (taken from the south). Deformation of the wall is clearly visible along the course of the wall, 5-10 m from the corner. Right, trench I, SE corner from the south. Atalanti islet in the background. Both illustrations are courtesy of the American School of Classical Studies, Athens (photographic archive: N67, 5 June 1911, and N76, 9 June 1911, respectively; Kyparissi, C. Blegen).

later that he cautiously associated them with earthquakes. While working one evening on his Palace of Minos in June 1926 at his home, the Villa Ariadne close to the Palace, north-central Crete was hit by a major earthquake, causing much structural and some fire damage to houses throughout the region. Yet the Villa Ariadne, built with the best seismic protection technology

Fig. 3. The Forum at Ostia (from Lanciani 1918).

of the day, survived unscathed. The breaks in continuity that Evans had observed at the Palace now crystallized in his mind; they were equated with seismic shocks of differing intensity, and, moreover they were to form one of the foundations of the chronology of Minoan Crete during the second millennium BC to 1400 BC, and indeed the rest of the Aegean (2190BC, Middle Minoan (MM) I A, to 1500BC, Late Minoan (LM) I A, and to 1400BC, LM II end of Palace) (Evans 1928). Some 50 years later it was to be the severe earthquake of c. 1700BC signalling the end of the First Palace Period in Crete that was invoked in more dramatic manner by Sakellarakis & Sapouna-Sakellarakis (1979) at the Minoan shrine or small temple of Anemo Spilia just to the south of Knossos. In their imaginative reconstruction based on the discovery of four human skeletons, vessels and other artefacts, in the west chamber, a priest, attended by a woman, apparently cut the carotid artery of a young man and collected blood. As an earthquake struck, an attendant dropped the libation jar before reaching the deity whose altar included the rock of the sacred hillside. The falling roof crushed and entombed the four people, and the place was set on fire. At Troy, Blegen confidently ascribed the end of the sixth city in the 13th century BC to seismic destruction (Blegen 1963), a claim since reexamined by Rapp (1982). The same event was to be linked by Schaeffer (1948) to a wave

28


of destruction (dated by him to 1365BC) that enveloped much of the Levant, extending to the site that he was excavating, Ras Shamra (Ugarit). Schaeffer's publication was to be a turning point, proposing as it did that the IndoEuropean migrations of the third and second millennia BC in the East Mediterranean and western Asia were triggered in part by the destructive effects of successive earthquakes. Destruction horizons in well-separated areas were to be associated with single seismic events. These are just some of the landmarks of the first half of this century. But despite the lack of enthusiasm with which Schaeffer's theory was held in the 1950s, as the pace of excavation throughout the Mediterranean quickened after World War II, so the seismic explanation of damaged buildings and structures (especially when the site in question and its stratigraphy were complex) became a convenience, to be used sometimes uncritically; judgements were subjective at best. It was not until the 1970s that signs of change were afoot. There was, first of all, a need for modelling, from the seismological and construction engineering viewpoints, of earthquake damage of known seismic intensity to the full range of ancient structures from temples to mud brick houses. But, equally urgently, a lead had to come from seismologists to define diagnostic criteria that could be used objectively and rigorously to identify earthquakes in the archaeological record. One of the first of these was published by Karcz & Kafri (1978), who dealt with the rich data set from Israel, particularly around Jericho, work that has since been extended throughout Israel by Nur & Ron (1996). To document an ancient seismic destruction, according to Karcz & Kafri, it is necessary to show that the observed effects can unhesitatingly be assigned to an earthquake, and that any other cause, in particular anthropogenic causes, can be excluded. Key issues were now claming more attention, such as the nature of the structure and its constructive materials, the type and extent of damage, the existence of similar damage at neighbouring sites, and the location of the site. Other researchers, such as Stiros (1996, p. 152), have since put forward their own criteria. This writer emphasized structural considerations in elucidating the response of buildings to earthquakes (main shocks and after-shocks from a single episode, as well as the cumulative effect of several earthquakes), separating the effects on buildings' foundations and superstructure, and he identified some of the natural phenomena, such as storms and rock falls, that may induce dynamic failures in a building similar to those resulting from an earthquake.

Archaeoseismology In principle, then, no longer was the identification of earthquake damage entirely in the hands of the excavator or those concerned with the conservation and restoration of ancient buildings; there had to be some dialogue with seismologists and other earth scientists, drawing where possible on comparable work carried out elsewhere in the world, notably in California and Japan. International meetings during the 1980s and especially the 1990s, and their subsequent publication, have aided the process of dialogue between seismologists, earth scientists, engineers, historians and field archaeologists, recognizing the congruence of interests in archaeology and seismology. In the Mediterranean these events have ranged from a treatment of how writers in the Greek and Roman worlds depicted seismic events and their social and economic consequences on society (what, for example, were the social consequences of the event that destroyed Sparta in 464 BC? (Ducat 1984)); to broad coverages of methods and case studies (e.g. Guidoboni 1989; Frohlich & Jacobelli 1995 (see section on Pompeii, below); Stiros & Jones (1996), and catalogues of seismic events (e.g. Guidoboni et al. 1994). In turn, this has led naturally and legitimately to the creation of a recognizable sub-discipline, archaeoseismology, by a process that has occurred often enough in other branches of archaeology, whereby the parent science has adapted to meet the needs of, and the questions arising from, archaeology. We can now briefly review a few issues that have emerged from research in this field during the last decade. Many of the case studies illustrating these issues are drawn from Greece but they are likely to have wider application to the Mediterranean as a whole.

Major seismic events That there have been major earthquakes that caused widespread damage and loss of life over a large area, and, furthermore, that they had in some cases marked socio-political consequences, whereas others decidedly did not, is undoubtedly acknowledged. The point is that each case has to be examined on its own merits and, crucially, within its own historical framework. That in AD 365 is one good but complex example that seems to have had a tectonic expression, with an epicentre to the south of Cyprus, preceded by tremors up to 5 years previously and with subsequent tsunamis that hit Egypt, Turkey and Greece: this event, which was very well

ARCHAEOSEISMOLOGY IN THE MEDITERRANEAN documented by historians of the time and has been dated accurately numismatically, can be placed within the tectonic history of the region. Some of the relevant sites that have been explored, such as Kourion and Corinth, have left signs that meet the criteria very well, whereas this may not consistently be the case at other sites, for example, of the Roman period in North Africa, for which Di Vita (1990) has described the archaeological evidence for their destruction. At Kourion in southern Cyprus, which was completely devastated and abandoned after AD 365, much attention has been paid to the destroyed mid-first century BC temple of Apollo (Soren 1981; Soren & Davis 1985). In their remarkable work in reconstructing the events at Kourion, Soren & Lane (1981) argued for a seismic shock wave from the SW, which threw down blocks north and east. However, Rapp (1987, p. 375) has argued that the shock wave may have been from the opposite direction, as it cannot be assumed that motion causes the destruction (duration of vibration is nearly always important) nor that wall collapse always results predominantly from horizontal vibration. Secondary shear waves (i.e. vertical motion) are known to cause much damage, in part by effects on soil and of the other unconsolidated material underlying walls. At Corinth, the rich commercial Late Roman and post-Roman city was hit not only by the major event of AD 365 but also by another in AD 400, destroying the large temples and other formal buildings. However, the interest here rests not with the mechanism of destruction but with the political and social implications of the calamity, for there was no attempt to rebuild: these buildings were stripped of their stone and reused (Rothaus 1996). Corinth retained its wealth but the wealthy class evidently decided not to fund the rebuilding of major civic buildings. With the consequent demise of the temples, the pagan cult centres of the city hastened the end of the late antique city, this being one factor that led to its transformation into a Byzantine city. Some other, contrasting cases are worth mentioning. The AD 1296 earthquake at Pergamon had considerable impact on the city, but, according to Rheidt (1996), was no more than a transient episode with little influence upon the overall development of the city. In comparison, the event of 464 BC that destroyed Sparta triggered a revolt of slaves and a war that led to the end of the city (Ducat 1984). And Gortyn in southern Crete, which was destroyed by the AD 620 earthquake, was fully rebuilt and flourished, yet a later earthquake, in AD 670, ruined the city for ever (Di Vita 1996).

29

Other seismic events Much more common in the East Mediterranean is the apparently less serious earthquake. Here the historical record of seismicity in a given region may tell a cautionary tale, as it does in the case of the Middle East, as reported by Ambraseys et al. (1994). Notwithstanding the incomplete historical record, the picture emerges that earthquakes in parts of this region are indeed numerous, of medium size, preceded and followed by damaging shocks causing localized destruction and relatively small loss of life. The same also applies to central Greece, about which Ambraseys (1996, p. 32) said, 'Occasionally earthquakes precipitated more than that but their lasting effects would not seem to be very significant'. The historical dimension to archaeoseismology in the work of Ambraseys and others in the Eastern Mediterranean and the Middle East has had at least two beneficial effects: first, of emphasizing the need to look more carefully at building materials and methods in the archaeological record and their suitability in areas of seismic risk (although anti-seismic building methods in Minoan Crete and Hittite Anatolia have long been recognized, if imperfectly); and, second, of resisting the temptation to claim contemporaneity of earthquakes at well-separated locations. A reminder that the absence of historical accounts of an earthquake does not weaken or preclude the case that this was the mechanism of destruction, especially if the site in question lies in a supposedly aseismic zone, is well demonstrated at the site of Pella, a capital of Macedonia. Its destruction dated to c. 90 BC, followed by wide-scale abandonment, is not reported in ancient texts, nor could it be attributed by the excavators to anthropogenic factors or to natural causes, such as foundation instability. Instead, the nature of the destruction fulfilled the criteria to be classified as seismically based; the site lies within 100 km of the epicentre of the earthquake that occurred at Grevena in 1995, this area having been regarded by seismologists as essentially aseismic (Stiros 1995, 1996, p. 143).

Seismic destructions in Late Bronze HI and Early Bronze II Greece The case in support of earthquake destruction increases proportionately as the evidence at one site is mirrored independently at neighbouring sites. The collective picture should be the goal, because of the way it strengthens the validity of identifying the traditional criteria of, typically,

30


deformed walls and skeletons close to collapsed doorways. An excellent case in point is the Argolid in the Peloponnese at the end of the Late Bronze Age. Did a series of earthquakes catalyse the demise of the Mycenaean palace culture? The answer must be no, and yet the scepticism on the part of some Aegean prehistorians that seismic effects had any place in the late Mycenaean world must surely now be regarded as misplaced. There is an earthquake horizon at Mycenae itself: damage in the cult centre on the Citadel, and skeletons in doorways covered with fallen stones from houses outside the Citadel, all dated to 1250BC (mid Late Helladic (LH) IIIB), and yet no destruction to large, contemporary buildings such as the Lion Gate and the Treasury of Atreus (French 1996; Nur 1998). On the Unterburg at Tiryns, Kilian found evidence of probably several earthquakes, somewhat earlier as well as later than at Mycenae; curving, deformed walls of early LH IIIC (1190-1150 BC) were notable (Kilian 1996). At Midea a skeleton was found recently in a room by the East Gate dating to the end of LH IIIB2, together with collapsed, distorted curved and tilting walls (Astrom & Demakopoulou 1996). Individually, the evidence at each site is far from watertight, and it is only when the three sites are treated collectively that the picture becomes more convincing. Midea and Tiryns suffered a probably severe earthquake around 1200BC, and Mycenae a less severe one around 1250BC; at Tiryns earthquakes seem to have been frequent, and all the indications are that they were followed by rapid and repeated repair work. Limited corroboration of the multiple seismic events comes from Papadopoulos' (1996) calculations of the probability of such events at the Mycenaean centres in the Argolid (and Corinthia) based on the seismic intensity v. frequency distributions from Greek data of this century; there was a high probability of destructive shock at roughly 30 year intervals. This case study greatly helps put earthquake damage into context: there were indeed earthquakes but they were neither devastating nor 'fatal'. An interesting, contrasting situation is emerging regarding the marked destruction horizon at the end of the Early Bronze (EB) II period in southern Greece around 2500-2400 BC, which neatly illustrates the dangers of invoking single events to explain destruction horizons in a given region. The traditional explanation of this phenomenon in EB II, which has been under attack for some time, is attributed to invasion by Indo-Europeans. But it is now becoming clear that events at that crucial period of time, the end of Early Helladic (EH) II, were not uniform throughout southern and central Greece.

We may contrast the picture at the type site of the Early Helladic period, Lerna, where there was major change (the House of the Tiles burnt down) with, on the one hand, the lack of any destruction in the comparable horizon at Kolonna on Aegina, and, on the other hand, the strong evidence for earthquake destruction in the EH II levels not only at Ayios Dimitrios and Voidokilia situated on the seismically active western flank of the Peloponnese (Zachos 1996) but also probably at Thebes (Sampson 1996, p. 115).

Pompeii A feature of recent work has been the emergence of highly detailed site-specific work. At Pompeii, for example, Frohlich & Jacobelli (1995) have documented the large amount of evidence that exists for the structural damage, repairs and changes of use that occurred during the lifetime of this town. Central to their enquiry work was to ascertain whether a second earthquake occurred after that of AD 62 and before the volcanic eruption of AD 79; sadly, this was not fully resolved, Conclusions These issues and case studies have illustrated some of the progress that seismology has made in its interaction with Mediterranean archaeology concerning the last 4000-5000 years. However, it is gratifying to note that the more distant past has not been neglected: broadening the scope of archaeoseismology to consider the role of tectonic processes on a wide time scale in relation to human adaptation to the environment, work stimulated by the British excavations of the Palaeolithic rock shelter at Klithi in Epirus is surely relevant and a likely corrective to most preconceptions. In brief, Bailey et al. (1993) have drawn attention to the active tectonics of northwest Greece giving rise to landscapes that could be beneficial to human survival, and the history of these processes shows that 'Palaeolithic sites were located to take advantage of tectonically created features at both local and regional scales'. Much remains to be done, but in looking to the future, we identify archaeoseismology's need to take a more pro-active role in formally assessing the seismic risk to ancient monuments, a process already under way in some Mediterranean countries, as the subjects' greatest challenge in the coming years.

ARCHAEOSEISMOLOGY IN THE MEDITERRANEAN

References AMBRASEYS, N. N. 1996. Material for the investigation of the seismicity of Central Greece. In: STIROS, S. & JONES, R. E. (eds) Archaeoseismology. Fitch Laboratory Occasional Paper, 7, 23-36. , MELVILLE, C. & ADAMS, R. 1994. The Seismicity of Egypt, Arabia and the Red Sea. CUP, Cambridge. ASTROM, P. & DEMAKOPOULOU, K. 1996. Signs of an earthquake at Midea? In: STIROS, S. & JONES, R. E. (eds) Archaeoseismology. Fitch Laboratory Occasional Paper, 7, 37-40. AUGUSTI, G. & SINOPOLI, A. 1992. Modelling the dynamics of large block structures. Meccanica, 27, 195-211. BAILEY, G., KING, G. & STURDY, D. 1993. Active tectonics and land use strategies: a Palaeolithic example from Northwest Greece. Antiquity, 67, 292-313. BLEGEN, C. not dated. Excavations to the S.W. of Kyparissi, 1911. American School of Classical Studies (Athens), School Papers 1899-1913 (unpublished). 1926. American Journal of Archaeology, 30, 403. 1963. Troy and the Trojans. Thame & Hudson, London. DAKORONIA, PH. 1996. Earthquakes of the Late Helladic III period (12th century BC) at Kynos (Livanates, Central Greece). In: STIROS, S. & JONES, R. E. (eds) Archaeoseismology. Fitch Laboratory Occasional Paper, 7, 41-44. Di VITA, A. 1990. Sismi, urbanistica e cronologia assolute - terremoti e urbanistica nelle citta di Tripolitania fra il 1 secolo, A.C. ed il IV D.C. In: L'Afrique dans I 'Occident Romain. Collection de 1'Ecole Francaise de Rome, 134, 425-494. 1996. Earthquakes and civil life at Gortyn (Crete) in the period between Justinian and Constant II (6th-7th century AD). In: STIROS, S. & JONES, R. E. (eds) Archaeoseismology. Fitch Laboratory Occasional Paper, 7, 45-50. DUCAT, J. 1984. Le treblement de terre de 1864 et 1'histoire de Sparte. In: HELLY, B. & POLLING, A. (eds) Tremblements de terre: histoire et archeologie. Proceedings of International Meeting on Archaeology and History, Antibes. APDCA, Valbonne, 73-86. EVANS, A. 1928. The Palace of Minos II. Macmillan, London. FRENCH, A. B. 1996. Evidence for an earthquake at Mycenae. In: STIROS, S. & JONES, R. E. (eds) Archaeoseismology. Fitch Laboratory Occasional Paper, 7, 51-54. FROLICH, T. & JACOBELLI, L. (eds) 1995. Archaologische und seismologie: la regione Vesuviana dal 62 al 79 n.C. Problemi archeologici e sismologici. Biering & Brinkmann, Munich. GALANOPOULOS, A. 1956. The seismic risk at Athens. Praktika Akadimias Athinon, 31, 461-472 [in Greek]. GUIDOBONI, E. (ed.) 1989. / Terremoti Prima del Mille in Italia e nell'Area Mediterranea: Storia, Arche-

31

ologia, Sismologia. Storia Geofisica Ambiente, Bologna. , COMASTRI, A. & TRAINA, G. 1994. Catalogue of Ancient Earthquakes in the Mediterranean Area up to the 10th Century. Institute Nazionale di Geofisica, Rome KARCZ, I. & KAFRI, U. 1978. Evaluation of supposed archaeoseismic damage in Israel. Journal of Archaeological Science, 5, 237—253. KILIAN, K. 1996. Earthquakes and archaeological context at 13th century BC Tiryns. In: STIROS, S. & JONES, R. E. (eds) Archaeoseismology. Fitch Laboratory Occasional Paper, 7, 63-68. LANCIANI, R. 1918. Segni di terremoti negli edifizi di Roma antica. Bulletino della Archeologica Communale Roma, 1-30. NUR, A. & RON, H. 1996. And the walls came tumbling down: earthquake history in the Holyland. In: STIROS, S. & JONES, R. E. (eds) Archaeoseismology. Fitch Laboratory Occasional Paper, 7, 75-85. PAPADOPOULOS, G. A. 1996. An earthquake engineering approach to the collapse of the Mycenaean Palace civilisation of the Greek mainland. In: STIROS, S. & JONES, R. E. (eds) Archaeoseismology. Fitch Laboratory Occasional Paper, 7, 205-209. RAPP, G. 1982. Earthquakes in the Troad. In: RAPP, G. & GIFFORD, J. A. (eds) Troy: the Archaeological Geology. Princeton, NJ, 43-58. 1987. Assessing archaeological evidence for seismic catastrophes. Geoarchaeology, 1, 365-379. RHEIDT, K. 1996. The 1296 earthquake and its consequences for Pergamon and Chliara. In: STIROS, S. & JONES, R. E. (eds) Archaeoseismology. Fitch Laboratory Occasional Paper, 7, 93-103. ROTHAUS, R. M. 1996. Earthquakes and temples in Late Antique Corinth. In: STIROS, S. & JONES, R. E. (eds) Archaeoseismology. Fitch Laboratory Occasional Paper, 7, 104-112. SAKELLARAKIS, Y. & SAPOUNA-SAKELLARAKIS, E. 1981. Drama of death in a Minoan temple. National Geographic, 159, 204-222. SAMPSON, A. 1996. Earthquakes at Mycenaean and pre-Mycenaean Thebes. In: STIROS, S. & JONES, R. E. (eds) Archaeoseismology. Fitch Laboratory Occasional Paper, 7, 113-117. SCHAEFFER, C. F. A. 1948. Stratigraphie comparee et chronologic de VAsie Occidentale. OUP, Oxford. SINOPOLI, A. 1991. Dynamic analysis of a stone column excited by a sine wave motion. Applied Mechanics Review, 44, S246-S255. SOREN, D. 1981. Earthquake: the last days of Kourion. In: BIERS, J. C. & Soren, D. (eds) Studies in Cypriot Archaeology. Institute of Archaeology UCLA Monograph, XVIII, 117-123. & DAVIS, T. 1985. Seismic archaeology at Kourion: the 1984 campaign. Report of the Department of Antiquities of Cyprus, 193-301. & LANE, E. 1981. New Ideas About the Destruction of Paphos. Report of the Department of Antiquities of Cyprus, 178-183. STIROS, S. C. 1995. Unexpected shock rocks an 'aseismic' area. Eos Transactions, American Geophysical Union, 76, 513-514.

32

R. E. JONES & S. C. STIROS -1996. Identification of earthquakes from archaeological data: methodology, criteria and limitations. In: STIROS, S. & JONES, R. E. (eds) Archaeoseismology. Fitch Laboratory Occasional Paper, 7, 129-152. - & DAKORONIA, PH. 1989. Ruolo storico e identificazione di antichi terremoti nei siti della Grecia. In: GUIDOBONI, E. (ed.) / Terremoti Prima del Mille in Italia e neil"Area Mediterra-

nea: Storia, Archeologia, Sismologia. Bologna, 422-439. & JONES, R. E. (eds) 1996. Archaeoseismology. Fitch Laboratory Occasional Paper, 7. ZACHOS, K. 1996. Tracing a destructive earthquake in the south-western Peloponnese (Greece) during the Early Bronze Age. In: STIROS, S. & JONES, R. E. (eds) Archaeoseismology. Fitch Laboratory Occasional Paper, 7, 169-185.

A critical reappraisal of the classical texts and archaeological evidence for earthquakes in the Atalanti region, central mainland Greece VICTORIA BUCK1'2 & IAIN STEWART2 1

British School at Athens, Odos Souedias 52, GR106 76, Athens, Greece (e-mail: [email protected]). Neotectonics Research Centre, Department of Geography and Earth Sciences, Brunei University Uxbridge, UBS 3PH, UK Abstract: Despite numerous damaging earthquakes in central Greece only the Atalanti Fault is considered to have ruptured successively, most recently in 1894 and in a historical event in 426 BC. Although the pre-Christian earthquake is now firmly entrenched in the tectonic literature, classical literary accounts are inconsistent and do not unequivocally tie the event to a particular time and place. Archaeological evidence from sites close to the Atalanti Fault similarly remains ambiguous, and fails to convincingly corroborate the rupture of the Atalanti Fault in 426 BC. In this paper, the main thrust of the argument is not to try to define the seismogenic source of the 426 BC event, but to illustrate the level of uncertainty that accompanies literary and archaeological information, and to highlight the need for caution when using interdisciplinary methods or datasets in earthquake seismology.

Archaeological records of seismic disturbance allow historical earthquake catalogues to be extended into pre-historical times or serve as an independent dataset against which to assess documentary reports of past earthquakes. Where archaeological records are long and well constrained, they are potentially a valuable tool in reconstructing the recent earthquake history of faults, particularly in terms of shedding light on the likely 'recurrence intervals' of faults, that is, the time elapsed between successive earthquakes on an individual fault. Away from plate boundaries such intervals range from many centuries to a few millennia, and therefore fall into the realm of archaeological study. Despite numerous damaging earthquakes in Greece, both in modern times and those documented from antiquity, no individual fault is unequivocally known to have ruptured twice. However, perhaps the best candidate for a fault widely considered to have ruptured successively is the Atalanti Fault, known to have ruptured in AD 1894, and also thought to have hosted a widely reported damaging event in 426 BC. This paper will examine the ancient literary accounts of the 426 BC event and the archaeological records of two excavated sites in the vicinity of the Atalanti Fault to assess independently contemporary evidence for this histori-

cal event. The paper will show that although accounts of the 426 BC event locate the earthquake to the region of the Gulf of Evia, they fail to tie the event convincingly to the same fault segment that ruptured in AD 1894, rather than a number of other potential seismic sources in this region. In addition, the paper will illustrate the problems involved in correlating seismic events with specific fault segments in this region. Background The Bay of Atalanti lies within the extensional rift system of the Gulf of Evia, central mainland Greece. The rift system is dominated by a segmented belt of large, 35-40 km long, NW-SE striking normal faults (Roberts & Jackson 1991; Ganas et al. 1996) predominantly dipping north (Fig. 1). The rift is considered active with numerous small (M < 5.5) earthquakes instrumentally recorded since the 1960s, and at least two large events (M > 6) accounted for in the ancient and recent historical literature: the AD 1894 and 426 BC earthquakes. The 1894 Atalanti earthquake, which consisted of two main shocks of M6.4 and M6.9 (Ambraseys & Jackson 1990) and hundreds of aftershocks, caused extensive damage from Agios

From: McGuiRE, W. G., GRIFFITHS, D. R., HANCOCK, P. L. & STEWART, I. S. (eds) The Archaeology of Geological Catastrophes. Geological Society, London, Special Publications, 171, 33-44. 1-86239-062-2/00/S15.00 © The Geological Society of London 2000.

Fig. 1. Map of Lokris region in relation to Greece (inset) showing the location of modern towns and villages (indicated in bold type), archaeological sites mentioned in the text (indicated by stars and italic type), and the neotectonic faults (after Ganas et al. 1996) with downthrow indicated by barbs.

Fig. 2. Panoramic view of the Atalanti plain looking north from the footwall above Kyparissi. Donkey Island, which was formed as a result of tectonic subsidence accompanying the 1894 earthquake on the Atalanti Fault, is in the middle of the picture and indicated by an arrow. Atalanti Island, on the left, is widely considered to have formed as a result of the 426 BC earthquake.

36

V. BUCK & I. STEWART

Konstantinos to Larymna with surface rupture along the Atalanti Fault (Anon. 1894a, b; Davidson 1894#, b). This was accompanied by numerous secondary surficial effects such as ground liquefaction, fissuring, marine inundation (between Arkitsa and Almyra; see Fig. 1); rockfalls (in the Kyparissi area; see Fig. 1), landslides, and a seismic sea wave (Davidson 1894&; Skouphos 1894; Mitsopoulos 1895; Ambraseys & Jackson 1990). The general tectonic subsidence of the Atalanti plain also resulted in the formation of Gaidaronisi (Donkey Island) through the immersion of the low-lying marsh area (Fig. 2). The Atalanti/Lokris Fault segment is considered to be the seismogenic source of the 1894 event; however, numerous other potential seismogenic faults lie in the vicinity (Fig. 1). In addition to the 1894 event, the Atalanti region is noted in a number of publications as having been affected by several historical earthquakes, including those of 426 BC, AD 105, AD 551 and AD 1544 (Stiros & Rondogianni 1985; Guidoboni et al 1989; Papazachos & Papazachou 1989). This paper, however, focuses only on the preChristian event of 426 BC. The contemporary account by Thucydides and the complementary descriptions by Diodorus Siculus and Strabo (both of whom wrote several centuries after the event) of the effects of the earthquake on named towns, forts and harbours suggest that the seismogenic source lay within the Gulf of Evia rift system. Although the size and location of the 426 BC event are imprecisely known, the apparent similarity of the observed surficial effects (e.g. the formation of an island and sea inundations) with those of the 1894 earthquake has led to the suggestion that both earthquakes, probably of comparable magnitude, occurred along the Atalanti Fault (Stiros & Pirazzoli 1995; Ganas et al. 1998). However, the fragmentary nature of the information, and the inconsistencies between the information provided by the three classical authors (Appendix), suggest that an independent data source is needed before the proposed 426 BC event can be unequivocally assigned to the Atalanti Fault. Classical accounts of the 426 BC event The 426 BC earthquake is dated from its inclusion in the contemporary account of the Peloponnesian War by Thucydides (Thucyd., iii. 89.1-4). It is unfortunate that Thucydides, who wrote the only contemporary account of the earthquake, provides the least geographical detail regarding the extent of destruction, mentioning damage in three separate geographical areas, the Isthmus near Corinth leading into Attica, Orobiae on the

island of Euboea, and Atalante in Lokris (see Appendix) Although Thucydides gives few details of the overall destruction, his account focuses on the effects of marine inundations on coastal sites, which reflects the subject matter of his writing: a chronological account of a historical event, i.e. the Peloponnesian War. A significant difference between this and the later derived sources of Diodorus Siculus (Diod. Sic., xii. lix) and Strabo (Strabo, I. iii. 20) is his specific reference to the island of Atalante as a pre-existing landform. Similarly, the historian Diodorus Siculus, writing in the first century, provides little geographical detail, perhaps reflecting the fact that his writing is derived from other earlier sources. The earthquake effects he briefly outlines are assumed to be those caused by the 426 BC event on the basis of similarities between this and the text of Thucydides, i.e. the invasion of Attica by the Peloponnesians and the location of their camp at the Isthmus leading to Attica. However, it is notable that in this account the formation of Atalante Island has now become the focus of the account of this earthquake, in direct contrast to Thucydides, who states that the island was already in existence before the 426 BC event. In comparison with the first two chronologically based texts, the Geographies, written by Strabo in the first century BC, is a geographically based compilation of knowledge. Quoting at least two sources, Strabo gives an undated account of a large destructive earthquake that affected the Gulf of Evia region. The first section of the relevant passage (Strabo, I. iii. 20) (Appendix) details precisely the localities and the extent of damage, including death tolls, citing the origin of the information as a written account (now lost) by Demetrius of Callatis. However, the final section (Strabo, I. iii. 20) (Appendix), which focuses on the genesis of Atalante Island, switches from a precise to a vague, almost anecdotal, account of the effects farther east in the Bay of Atalanti, and should therefore be viewed with caution as a description of the effects of the same earthquake discussed earlier in the chapter. Again, it is the association with Atalanti Island that leads subsequent workers to assume that this is an account of the 426 BC event. Geological interpretations of the literary sources Although the texts are fragmentary, some important inconsistencies emerge from a comparison of the previous accounts. First, the dating of the

CLASSICAL AND ARCHAEOLOGICAL EVIDENCE FOR EARTHQUAKES event is derived from the chronological accounts of the Peloponnesian Wars by Thucydides and Diodorus Siculus. Strabo's account is assumed to be the same event because of the inclusion of Atalanti in the description of areas experiencing damage. Second, whereas Thucydides and Diodorus Siculus discuss the 426 BC earthquake in relation to the Atalanti area, Strabo focuses most attention on the Malakios Gulf (see Fig. 1), with only minor reference to Atalanti. The question arises, therefore, whether these accounts are selective reports of a regionally extensive earthquake or testify to separate events that have in some texts been combined. Third, whereas Atalanti Island is already in existence before the 426 BC earthquake in the contemporary account of Thucydides, the two later reports of Diodorus Siculus and Strabo ascribe the formation of the island to the earthquake itself. This is of importance because the formation of the island in the earthquake strongly suggests subsidence of the Atalanti plain, probably as a result of rupture of the Atalanti Fault, as was the case with the detachment of Donkey Island in the 1894 event (Fig. 3). By contrast, Thucydides' reports of widespread marine inundation around the Atalanti coast may equally be attributed to tsunami attack by rupture on one of a number of seismogenic faults in the area (Fig. 1). Most geologists employing these ancient sources have assumed that all three texts refer to a single major earthquake in 426 BC, but their subsequent seismotectonic interpretations vary markedly. Some, using the combined destruction and geographical information, have constructed isoseismal maps with the long axis of the derived damage ellipse trending east-west, with the maximum intensity located in the Malakios Gulf (Sieberg 1932; Bousquet & Pechoux 1977) (Fig. 4). Other workers (Stiros 19880, c\ Stiros & Dakoronia 1989; Dakoronia 1993), on the basis of the occurrence of possible archaeoseismic damage in the Atalanti area, and the apparent similarity of geomorphological effects observed here during the 1894 event to those described in the ancient texts (particularly the coastal inundation and formation of islands), ascribe the event to a 70km long NW-trending fault linking Atalanti with Lamia. It should be noted that this latter view implies that the Atalanti Fault is up to 70km long (Fig. 5), something which is disputed by the recent work on fault segmentation in the Atalanti region (Roberts & Jackson 1991; Ganas et aL 1996), which suggests a maximum segment length of 3 5-40 km (in the Atalanti area) as shown in Fig. 1. In a detailed review of the ancient texts, Mouyiaris (1988) proposed that the narrative of

37

Strabo is in fact an agglomeration of several events that cannot be securely dated, and he accepted only the contemporary account by Thucydides as that of the 426 BC event. However, on the basis of the texts alone, he was unable to define securely the seismogenic origin of the earthquake, preferring to conclude that the Atalanti plain area had clearly been the subject of previous earthquake events. Clearly, the seismogenic source of the 426 EC event cannot be unequivocally identified from the ancient literature alone, because of inconsistencies between the texts, which may simply be the result of differing lexical or writing styles. However, validation of literary evidence from an independent data source, such as archaeoseismological evidence, may be able to define which scenario is the more probable.

Archaeological data A number of coastal and inland archaeological sites are located around Atalanti Bay (Fig. 1). Their proximity to the active Atalanti Fault makes it reasonable to expect that if an earthquake had occurred then diagnostic indicators (sensu Nikonov 1988; Stiros 1996) would be present in the archaeological stratigraphies of the sites. In the following section, the archaeological record of two sites (Alai and Kyparissi) is considered in relation to the validation of the 426 BC event.

Alai The acropolis site of Alai is located on the southeastern shore of Atalanti Bay in the vicinity of the modern village of Theologos, c. 6 km NE of the Atalanti Fault (Fig. 1). The site contains evidence of occupation from the Neolithic to the Byzantine period (Coleman 19926), with an occupational hiatus of around 4000 years between the Neolithic period and the foundation of Alai in the Archaic period (Wren 1996). The site was excavated annually during 1911-1914 by the American School of Classical Studies at Athens under the directorship of Hetty Goldman and Alice Walker, with four additional campaigns during 1921-1935 (Coleman 19926; Goldman 1940). 'Seismic disturbances' in the area are mentioned in a footnote to the initial excavation report (Walker & Goldman 1915, p. 419) particularly with reference to an earthquake in 1893 (possibly a misquote for 1894) producing submergence in the Almyra area of

38


Fig. 3. Comparative maps showing the formation of Donkey Island, (a) Pre-1894 coastline, showing Donkey Island (indicated by a bold arrow) attached to the relatively smooth shoreline of Atalanti Bay (British Admiralty 1980). (b) Post-1894 coastline, showing Donkey Island as a post-1894 earthquake landform (British Admiralty 1984).

the Atalanti shoreline. However, there is no description of any earthquake-induced damage at the site, probably because of the general nature of the report. Goldman (1940, p. 454) discussed in slightly more detail the evidence for destruction in 426 BC: The earliest building, which never had a stone entablature or cornice, was razed and buried some time after 510BC, together with its altar, under a thick covering of earth.' A tenuous link is made between the observed architectural damage and earthquakes known from the classical literature: 'In the great earthquakes of 426 and 425 BC, the latter accompanied by cata-

strophic inundations, which were recorded by ancient authors as of such exceptional violence that they tore the island of Atalante across the bay from Halae in two, the superstructure of the building collapsed, and the temple area was strewn with blocks' (Goldman 1940, p. 454). The link, which appears to provide an explanation of the architectural remains, is further supported by dating the second building through correlation of architectural styles: The capital [head of pillar or column] No. 3 and geison [cornice or ornamental moulding just below the ceiling] are very close in style to those of the Argive Heraion, and if we accept for the second temple of Hera

Fig. 4. An isoseismal map derived from the combined use of the classical texts by Thucydides, Strabo and Diodorus Siculus. It should be noted that the maximum damage isoseismal (11) is located in the Maliakos Gulf, particularly around the site of Phalara (redrawn from Sieberg (1932)).

CLASSICAL AND ARCHAEOLOGICAL EVIDENCE FOR EARTHQUAKES

39

Fig. 5. Map showing a proposed 70 km fault system linking the Atalanti Fault with Lamia (redrawn from Stiros & Dakoronia (1989)).

the date proposed by Professor Dinsmoor 423 BC - the Halae building, which is slightly later in style, was rebuilt not long after the disastrous earthquakes' (Goldman 1940, p. 454). More recently, the remains at Alai have been the focus of the Cornell Halae and East Lokris Project (CHELP), a US interdisciplinary archaeological and environmental project under the directorship of John Coleman. The ArchaicClassical remains and stratigraphy have been carefully reconsidered and a date of 480 BC has been assigned for the collapse of the first temple (Wren 1996), thus throwing into question the correlation between the earthquake described in the ancient literature and the observed archaeological field evidence. On the basis of this new date, Wren puts forward the hypothesis that the destruction level identified in Alai could equally be the result of the Persian advance towards Athens under the command of Xerxes, which is known to have taken place at this time. Further excavation in 1996 of the Archaic-Classical levels at the site, however, has revealed a toppled wall, indicative of a natural rather than a military destruction, and cultural finds from deposits associated with it tend to confirm a 480 date, although to date there is no unambiguous evi-

dence for the cause of the destruction (Coleman, pers. comm.). Shoreline constructions at Alai, identified as harbour installations such as 'ship sheds' (Fossey 1990), also have been reinterpreted as fortification walls in the light of an extensive site survey. This, coupled with the apparent lack of hinterland associated with the remains, suggests a relative rise in sea level (Coleman I992a). Although, the mechanism (e.g. eustatic rise or tectonic land subsidence) and the magnitude of change cannot be precisely determined from the data currently available, the site is known to have subsided in 1894 and so some tectonic contribution is likely (Ganas & Buck 1998).

Kypans si (Opus) At the base of the Kokkinovrachos hill in the SE end of the Atalanti plain, 2 km southeast of the modern village of Kyparissi, lies an archaeological settlement site that is now considered as the 'most likely candidate for the city of Opus' (Dakoronia 1993, p. 117). This site consists of two localities, both lying 200m from the trace of the Atalanti Fault. A fortified acropolis is well

40


preserved and visible at the top of Kokkinovrachos hill, and an associated settlement site is located somewhere on the north and northwestern parts of the same hill (although there are now no remains visible above ground level, there is a high density of ceramic material visible on the cultivated surface of the olive groves). The acropolis site has been surveyed and described several times since the 1880s (see Fossey (1990) for references), with no specific mention of earthquakes or any resulting seismic damage. The original excavations of the settlement site were carried out in 1911 by the American School under the directorship of Carl Blegen. Their report, too, gives no account of earthquakes or earthquake damage, particularly noteworthy given that the site was excavated less than 20 years after the 1894 event (Blegen 1926). More recently, however, two seasons of excavation, in 1978 and 1979, were carried out by the Greek Archaeological Service under the directorship of Dr Ph. Dakoronia, which exposed an Archaic (sixth-century BC) stoa [a porch or portico not attached to a larger building] considered to have been 'destroyed by an earthquake; the site is frequently afflicted by earthquakes, lying as it does near the well known Atalante seismic fault' (Dakoronia 1993, p. 117). The configuration of the damage, i.e. 'the conspicuous upward bending' (Stiros 1988c, p. 1634) of the stoa wall, has led to the site being considered as direct evidence of vertical offset by movement of the Atalanti Fault (Stiros 1988a-c, 1996; Stiros & Dakoronia 1989; Stiros & Pirazzoli 1995). 'This seems to be a plausible explanation, since the stoa lies very close to, or just on the surface trace of the Locris 1894 earthquake' (Stiros 1988c, p. 1634). A succession of studies has dated the destruction of the stoa to between 540 and 425 BC (Dakoronia 1988; Stiros 1988a-c) on the basis of typological comparison of terracottas, specifically acroteria. It would appear that the stoa was in use until c. 480 BC, when it sustained a roof repair. However, the lack of excavation material dating to the beginning of the fifth century BC, and the redating of the Alai destruction, throws into question the later possible dates for the destruction of the stoa. Reassessment of the Kyparissi sherds in the light of new excavation data from Alai may result in a more concisely constrained destruction date in the future. It should be noted that the dates of destruction quoted by researchers relate to the date of abandonment, i.e. the latest possible occupation based upon the dating of cultural material such as pottery including terracottas, or absolute dating. This may not necessarily equate to the

date of deformation of the stoa, however, because, lying close to the rupture trace of the 1894 break as it does, it is conceivable that the stoa has been affected by multiple events. Other local sites There are a number of less published sites of archaeological remains, including the Islet of Mitrou, and the sites of Palaeomagaza near Skala, and Kynos near Livanates, all located along the shore of Atalanti Bay (Fig. 1). However, these sites will not be considered further in this paper, because of the paucity of published data on either the sites as a whole or the Archaic-Classical period within the site. Discussion To summarize the archaeological data: the initial date of 426 BC assigned to the destruction of the first temple area in Alai has now been revised to 480 BC based on a reassessment of the destruction within the Archaic-Classical levels. This was originally thought to be associated with the Persian invasion by Xerxes in 480 BC, but the discovery of new evidence in the form of a toppled wall has led to the suggestion that the destruction is probably the result of an earthquake occurring around the same time. The re-dating of the Archaic-Classical finds and of the destruction at Alai throws into question the dating of the proposed seismic destruction at Kyparissi, which is based upon typological comparison of terracottas with those from Alai. Therefore, although the remains at Kyparissi have been identified by the excavators as having been seismically damaged, the dating of the destructive event is not clear, primarily because of the redating of the comparative material at Alai and the possibility of deformation during the 1894 event. Clearly, destruction event horizons that can be correlated between neighbouring sites lend confidence to the interpretation of regionally important earthquake (or other) events rather than local, possibly site-specific and not necessarily seismic, events. Conversely, a lack of correlation between positively identified inter-site destructions should not exclude the possibility of an earthquake, because there are a number of factors that influence the extent to which an earthquake is likely to be preserved in the archaeological record. The archaeological reasons for non-correlation are discussed below in the context of the Atalanti sites.


41

Site occupation

Dating

Where sites are established, occupied and abandoned at different chronological periods their capacity to retain evidence of damaging seismic events will vary. Alai shows occupation levels dating from the Neolithic to the Byzantine period, with an occupation hiatus of 4000 years at the end of the Neolithic period (c. 54005300BC). Archaeological finds from the Kyparissi area begin in the Late Helladic IIIB-C period (1300-1100 BC), but the actual architectural relics date only from the Archaic and Classical periods. Both sites show evidence of occupation at the time of the earthquake mentioned by Thucydides, which means that a lack of evidence of occupation should not affect potential correlation.

Deriving a date for destruction horizons is extremely difficult because they often contain only items of material culture, i.e. building materials, ceramics including tiles, etc. and not artefacts with actual dates (e.g. inscriptions). The majority of these items are not suitable for absolute dating; however, they can be used for relative dating; however, relative dates often have broad ranges, which can result in apparent correlation. Relative dating of occupation levels in the sites studies in the Atalanti region is based on typological comparisons of terracotta and other ceramics and building styles (Dakoronia 1996). Although the Alai site now has several good 14C dates, these date from the earliest Neolithic levels and pre-date the possible earthquake by nearly 5000 years (Coleman, pers. comm.). Clearly, the dating issue affects the correlation in this case study, as, although the dating of the Archaic-Classical levels in Alai has recently been revised, there are no 14C dates for these periods, and the dates for the damage of the Kyparissi stoa still rely on comparative evidence.

Identification Seismic damage may be overlooked by excavators unfamiliar with archaeoseismic criteria, and consequently not recorded in the site reports. Kyparissi was excavated by the regional Ephoreia (regional archaeologist), who is familiar with previous research involving identified earthquake-induced damage at archaeological sites. Similarly, the director at Alai, John Coleman, aware that the area is seismically active, has liaised with the regional Ephoreia on this subject throughout the excavation. Therefore, the lack of identification of seismic damage to archaeological remains is considered unlikely to affect recent studies. However, earlier excavations carried out at these sites may not record possible earthquake damage because of the relatively recent emergence of archaeoseismology. Occurrence of damage The type and quantity of seismic damage are largely a function of the construction methods employed, the quality of materials used, and their response to any physical tilting and shaking of the ground that accompany sizeable earthquakes (Stiros 1996). To date there are no published accounts of the construction methods and materials used in the varying phases of occupation at either of the sites. However, preliminary field observations indicate that there are similarities in the construction methods and materials used between the sites at corresponding time periods, and therefore the architectural remains are likely to have had a similar receptivity to seismic events hosted by adjacent faults.

Preservation The presence of seismic damage will largely depend on the vulnerability of a site and its remains to seismic damage. However, it may be possible to suggest whether the site is good or bad for preservation, on the basis of the environmental context of the remains. Conclusion In summary, neither the classical literary accounts nor the archaeological data convincingly demonstrate that the Atalanti Fault was responsible for the 426 BC earthquake. In addition to being fragmentary, classical accounts have important inconsistencies, and in particular, the well-cited account by Strabo seems to be highly misleading in terms of the regional extent of the event and its exact date. Although Thucydides and (less convincingly) Diodorus Siculus suggest that a damaging earthquake did strike the Atalanti area in 426 BC, archaeological evidence from occupation sites in the vicinity fails to corroborate this event. In particular, a revision of the chronology at Alai has led to the redating of destruction horizons, originally attributed to 426 BC, to C.480BC. In addition, the timing of the abandonment and subsequent deformation of a stoa at nearby

42


Kyparissi (Opus), widely cited as evidence of rupture along the Atalanti Fault, is poorly constrained and, in the light of the revised chronology at Alai, needs reappraisal. In a broader context, it is important to appreciate that the identification of an earthquake in an archaeological, historical or classical context is used primarily within a narrative as cultural information and rarely extends into understanding the earthquake as a geological phenomenon. For example, Sir Arthur Evans (1928) linked the decline of Minoan civilization to repeated earthquakes on the island of Crete, and Thucydides tells us that 'a great many earthquakes' prevented the invasion of Attica by the Peloponnesians in 426 BC. In effect, for archaeologists, identification is generally part of the conclusion of the investigation. In contrast, earthquake geologists start with the identification made by the archaeologists and historians and seek to define earthquake characteristics, such as seismogenic source and magnitude, with the ultimate goal of including them in a regional seismic catalogue and in turn, in seismic-hazard assessment scenarios. In this paper, the main thrust of the argument is not to seek the seismogenic source of the 426 BC earthquake, whether it be the Atalanti Fault or a neighbouring structure. Instead, this examination of the classical and archaeological datasets has sought to illustrate how much uncertainty accompanies both lines of evidence. The message from this study is that earthquake geologists should in general be circumspect when integrating historical and archaeological evidence. In particular, it is crucial that data reliability and coherence are critically appraised when making entries into and using seismic catalogues. In summary, if archaeoseismic evidence is to contribute significantly to our understanding of earthquake activity, practitioners need to confront the challenge of Charles Richter that 'Ancient accounts of earthquakes do not help us much; they are incomplete, and accuracy is usually sacrificed to make the most of a good story' (Richter 1958; cited by VitaFinzi 1986, p. 8).

Ph. Dakoronia. Geological permits were granted by the Institute of Geology and Mineral Exploration Athens. R. Reinders and an anonymous reviewer are thanked for constructive comments on an earlier version of the paper.

Appendix: The classical texts from the Loeb translations in English

Thucydides, Hi. 89.1-5 Tn the following summer the Peloponnesians and their allies, led by Agis son of Archidamus, king of the Lacedaemonians, advanced as far as the Isthmus with the intention of invading Attica; but a great many earthquakes occurred, causing them to turn back again, and no invasion took place. At about the same time, while the earthquakes prevailed, the sea at Orobiae in Euboea receded from what was then the shoreline, and then coming on in a great wave overran a portion of the city. One part of the flood subsided, but another engulfed the shore, so that what was land before is now sea; and it destroyed of the people as many as could not run up to the high ground in time. In the neighbourhood also of the island of Atalante, which lies off the coast of Opuntian Locris, there was a similar inundation, which carried away a part of the Athenian fort there, and wrecked one of two ships that had been drawn up on the shore. At Peparethos [Skopelos Island] likewise there was a recession of the waters, but no inundation; and there was an earthquake, which threw down a part of the wall as well as the prytaneum and a few other houses. And the cause of such a phenomenon, in my own opinion, was this: at that point where the shock of the earthquake was greatest the sea was driven back, then suddenly returning with increased violence, made the inundation; but without an earthquake is seems to me such a thing would not have happened.'

Diodorus Siculus, xii. 59.1-2

'While the Athenians were busied with these matters, the Lacedaemonians, taking with them This research was supported by a Brunei University the Peloponnesians, pitched camp at the Isthmus postgraduate scholarship. The authors would like to with the intention of invading Attica again; but acknowledge J. Coleman, director of CHELP, for 1997 when great earthquakes took place, they were field work support with unlimited access to project filled with superstitious fear and returned to records, A. Ganas for valuable field assistance and their native lands. So severe in fact were the many thought-provoking discussions, and G. Shipley for advice and expertise on the translations of the shocks in many parts of Greece that the sea ancient texts and the original draft manuscript. actually swept away and destroyed some cities Archaeological permits were obtained from the lying on the coast, while in Locris the strip of Greek Ministry of Culture, with invaluable assistance land forming a peninsula was torn through and from the British School at Athens and support from the island known as Atalante was formed.'


43

gie et premiers resultats. Bulletin de la Societe Geoliguqe de France, XIX(3), 679-684. BRITISH ADMIRALTY (1890) Bathymetric Chart of the 'Demetrius of Callatis, in his account of all the Evvikos Gulf (chart 1556), London. earthquakes that have ever occurred throughout COLEMAN, J. E. 1992a. Excavations at Alai in 1991. all of Greece, says that the greater part of the American Journal of Archaeology 96, 346. Lichades Islands and Cenaeum was engulfed; the 19926. Excavations at Halai 1990-1991. Hesperia, hot springs at Aedepsus and Thermopylae, after 61, 265-277. having ceased to flow for three days, began to DAKORONIA, P. 1988. Archaic ceramics of East Lokris. flow afresh, and those at Aedepsus broke forth Hesperia, 59(1), 175-180. also at another source; at Oreus the wall next to 1993. Homeric towns in East Lokris: problems the sea and about seven hundred of the houses with identification. Hesperia, 62(1), 115-127. 1996. Earthquakes of the Late Helladic III Period collapsed; and as for Echinus and Phalara and (12th century BC) at Kynos (Livanates, central Heracleia in Trachis, not only was a considerable Greece). In: STIROS, S. & JONES, R. (eds) Archaeoportion of them thrown down, but the settleseismology. Fitch Laboratory Occasional Paper, ment of Phalara was overturned, ground and all. 7, 41-44. And, says he, something quite similar happened DAVIDSON, C. 18940. Earthquakes in Greece. Nature, to the people of Lamia and of Larissa; and 50(1279), 7. Scarphia, also was flung up foundations and all, 1894&. M. Papavasilore on the Greek earthquakes and no fewer than seventeen hundred human of April 1894. Nature, 5(133), 67. beings were engulfed, and over half as many EVANS, A. 1928. The Palace of Minos at Knossos. Macmillan & Co. Ltd, London. Thronians; again, a triple-headed wave rose up, FOSSEY, J. M. 1990. The Ancient Topography of one part of which was carried in the direction of Opuntian Lokris. Gieben, Amsterdam. Tarphe and Thronium, another part to Thermopylae, and the rest into the plain as far as GANAS, A. & BUCK, V. 1998. A model for tectonic subsidence of the Allai archaeological site, Lokris, Daphnus in Phocis; fountains of rivers were dried central Greece. Bulletin of the Geological Society up for a number of days, and the Sphercheius of Greece, XXXll(l), 181-187. changed its course and made the roadways navig, ROBERTS, G. P. & MEMOU, T. 1998. Segment able, and the Boagrius was carried down a differboundaries, the 1894 ruptures and strain patterns ent ravine, and also many sections of Alope, along the Atalanti Fault, central Greece. Journal of Geodynamics, 26, 461-486. Cynus and Opus were seriously damaged, and , WADGE, G. & WHITE, K. 1996. Fault segmentaOeum, the castle above Opus, was laid in utter tion and tectonic geomorphology in eastern centruin, and a part of the wall of Elaetia was broken ral Greece from satellite data. In: llth Thematic down, and at Alponus, during the celebration of Conference and Workshop on Applied Geologic the Thesmophoria, twenty-five girls ran up into Remote Sensing, Las Vegas, NV (unpublished). one of the towers at the harbour to get a better GOLDMAN, H. 1940. The Acropolis at Halae. Hesperia, view, the tower fell, and they themselves fell with IX, 381-514. it into the sea. And they say, also, of the Atalanta GUIDOBONI, E., COMASTRI, A. & TRAINA, G. 1989. near Euboea that its middle portions, because Catalogue of Ancient Earthquakes in the Mediterranean Area up to the 10th Century. Compositori, they had been rent asunder, got a ship-canal Rome. through the rent, and that some of the plains MITSOPOULOS, K. 1895. The Lokris Mega-earthquake. were overflowed even as far as twenty stadia, and Greek Gov. Publication, Athens. that a trireme was lifted out of the docks and cast MOUYIARIS, N. K. 1988. Destructive historical earthover the wall.' quakes in n.Euboicos and Maliacos Gulfs - their significance to the evolution of the area. In: MARINOS, P. G. & KOUKIS, G. C. (eds) Proceedings of an International Symposium Organised by References the Greek National Group of IAEG. Balkema, Rotterdam, 1249-1256. ANON. 18940. Earthquakes in Greece. Levant Herald NIKONOV, A. A. 1988. On the methodology of archaeoand Eastern Express. seismic research into historical monuments. In: 18946. Severe earthquakes in Greece. The Times. MARINOS, P. G. & KOUKIS, G. C. (eds) ProceedAMBRASEYS, N. & JACKSON, J. A. 1990. Seismicity and ings of an International Symposium Organised by associated strain of central Greece between 1890 the Greek National Group of IAEG. Balkema, and 1988. Geophysical Journal International, 11, Rotterdam, 1315-132. 663-703. BLEGEN, C. W. 1926. The site of Opus. American PAPAZACHOS, B. & PAPAZACHOU, K. 1989. The Earthquakes of Greece. Ziti, Thessaloniki. Journal of Archaeology (Second Series), XXX, ROBERTS, S. & JACKSON, J. 1991. Active normal 401-404. faulting in central Greece: an overview. In: BOUSQUET, B. & PECHOUX, P. Y. 1977. La seismicite ROBERTS, A. M., YIELDING, G. & FREEMAN, B. du Basin Egeen pendant 1'antiquite. MethodoloStrabo, I. Hi. 20

44


The Geometry of Normal Faults. Geological Society, London, Special Publication, 56, 125-142. SIEBERG, A. 1932. Untersuchungen iiber Erdbeben und Bruchschollenbau im ostlichen Mittelmeergebiet. Denkschriften der medizinsch-naturwissenschaftlichen Gesellschaft zu Jena, 18, 161-273. SKOUPHOS, T. 1894. Die zwei grossen Erdbeben in Lokris am 8/20 und 15/27 April 1894. Zeitschrift Gesellschaft Erdkunde zu Berlin, 24, 409-474. STIROS, S. 19880. Deformations of ancient constructions: implications for the history of sites and seismotectonic research. In: MARINOS, P. G. & KOUKIS, G. C. (eds) Engineering Geology of Ancient Works, Monuments and Historical Sites. Balkema, Rotterdam, 1591-1596. 1988&. Earthquake effects on ancient constructions. In: JONES, R. E. & CATLING, H. W. New Aspects of Archaeological Science in Greece. Fitch Laboratory Occasional Paper, 3, 1-6. 1988c. Archaeology - a tool to study active tectonics. EOS Transactions, American Geophysical Union, 69(50). 1996. Identification of earthquakes from archaeological data: methodology, criteria and limitations. In: STIROS, S. & JONES, R. E. Archaeoseismology. Fitch Laboratory Occasional Paper, 7, 129-152. & DAKORONIA, P. 1989. Rulo storico e identificazione di antichi terremoti nei siti della Grecia. In: GUIDOBONI, E. (ed.) / Terremoti Prima del Mille in Italia e nell' area Mediterra-

nea. Storia Geofisica Ambiente (SGA), Bologna, 422-438. & PIRAZZOLI, P. A. 1995. Palaeoseismic studies in Greece: a review. Quaternary International, 25, 57-63. & RONDOGIANNI, T. 1985. Recent vertical movements across the Atalandi fault zone (Central Greece). Pageoph, 123, 832-848. THE HYDROGRAPHIC OFFICE (1948) Greece-East Coast Vorios Evvoi'kos Kolpos and Approaches to Volos (chart 1556), London ViTA-FiNZi, C. 1986. Recent Earth Movements: an Introduction to Neotectonics. Academic Press, London. WALKER, A. L. & GOLDMAN, H. 1915. Report on excavations at Alai of Locris. American Journal of Archaeology, XIX(4), 418-437. WREN, P. 1996. Archaic Halai. MA thesis, Cornell University, Ithaca, NY.

Ancient works cited The Geography of Strabo, vol i, trans. H. L. Jones (1932) (Loeb Classical Library). Cambridge, Mass.: Harvard UP; London: Heinemann. Diodorus of Sicily, vol i, trans. C. H. Oldfather (1933) (Loeb Classical Library). Cambridge, Mass.: Harvard UP; London: Heinemann. Thucydides, vol iv, trans. C. F. Smith (1938) (Loeb Classical Library). Cambridge, Mass.: Harvard UP; London: Heinemann.

Aims and methods in territorial archaeology: possible clues to a strong fourth-century AD earthquake in the Straits of Messina (southern Italy) EMANUELA GUIDOBONI1, ANNA MUGGIA1 & GIANLUCA VALENSISE2 1

SGA (Storia Geofisica Ambiente), Via Bellombra 24J2, 40136 Bologna, Italy ING (Istituto Nazionale di Geofisica), Via di Vigna Murata 605, 00143 Rome, Italy

2

Abstract: This research was stimulated by the need to extend in time the record of Italy's largest earthquakes, which commonly have repeat times of the same order as the length of the available historical record. As a test case we used the 1908 Straits of Messina earthquake, a large event that geologists assume to recur at intervals of roughly a millennium but whose predecessors are as yet unknown. The 1908 earthquake caused enormous territorial upheaval and left signs in the settlements that are still largely recognizable today. We hypothesized that the Straits of Messina, which were densely populated even in ancient times, may similarly retain evidence of one or more much older 'upheavals' of the settlement network, and that this evidence may be recognized through a careful analysis of archaeological observations. We found evidence that the settled area around the Straits of Messina contracted substantially around the middle of the fourth century AD, when many sites were abandoned or relocated. This contraction can hardly be justified by the then current economic and military setting. Specific archaeological findings within the cities of Messina and Reggio Calabria also suggest a serious decline of the region during the same period. The archaeological hypothesis is in good agreement with the available historical and palaeoseismological evidence and suggests that a large earthquake, perhaps similar to the 1908 event, took place in the area surrounding the Straits of Messina around the middle of the fourth century AD.

One of the most important findings of earthquake studies in Italy over the past decade is that the repeat time of individual large earthquakes is of the order of one or more millennia. Similar results had previously been obtained for many seismic areas in the world, but the frequency of large historical earthquakes in Italy had brought seismologists to the conclusion that Italian earthquakes repeated themselves every few centuries or so. Some Italian cities indeed appear to have been struck several times in their history, but results supplied by palaeoseismology (a recently developed discipline aimed at uncovering and dating any direct surface evidence of large earthquakes of the past) suggest that this is due to multiplicity of destructive seismogenic sources rather than to the frequency with which individual sources generate large earthquakes (see, e.g. Boschi et al. 1994). Individual repeat times of the order of 2000 years, such as in the case of the fault that generated the disastrous 1980 Irpinia earthquake (Pantosti et al 1993), have raised significant concerns in the seismological community as to what exactly we can learn from traditional historical catalogues of seismicity and what is bound to be missed, no matter how good the histori-

cal record is (or appears to be; see Valensise & Guidoboni (1995)). Could seismic archaeology become the answer to the need for a substantially longer record of earthquake effects on the human environment than the few centuries usually granted by conventional historical seismology? The Straits of Messina, an area struck by a catastrophic earthquake on 28 December 1908, serves as a good test site to examine the potential of this new discipline. The Straits of Messina region is known to have been occupied for over two millennia but no clear evidence exists of a predecessor of the 1908 earthquake. Current fault-segmentation models (e.g. Valensise & Pantosti 2000) suggest that the causative fault of the 1908 earthquake is the only seismogenic source capable of inducing the collapse of buildings along the shores of the Straits of Messina, and no large historical earthquake from adjacent sources has ever exceeded intensity VIII-IX in both Messina and Reggio Calabria (Boschi et al. 1995, 1997). Estimating the actual repeat time of a 1908type earthquake, and therefore the true seismic behaviour of the region, has been of critical importance in the assessment of the design for the planned permanent crossing of the Straits, to


46

E. GUIDOBONI, A. MUGGIA & G. VALENSISE

be accomplished in the next decade through the construction of a single-span bridge. Bridge planners have assumed that earthquakes such as that of 1908 may recur roughly every 2000 years, well beyond the expected lifetime of the bridge. Our aim is to explore the possibility that historical and archaeological evidence can be brought forward through which this assumption may be substantiated on much firmer grounds. Seismic archaeology: problems of methodology Seismic archaeology (or archaeoseismology) is a young discipline at the boundary between human sciences and geophysics that has emerged in the past 20 years or so. As with many young disciplines, however, seismic archaeology still lacks an established and satisfactory methodology. On the contrary, the diversity of aims, cultures and approaches involved has led to a situation in which observations often 'involve' archaeology, but in different ways and for different purposes. Archaeology is sometimes used as an independent source, in which case it has its own interpretational framework; sometimes as confirmatory evidence of earthquake effects reported by written sources; and sometimes even as a 'springboard for speculation' about catastrophes, which are preferably of extraordinary dimensions and unknown to written history (a noble sentiment perhaps, but one that is extremely difficult to reconcile with the scientific method). In spite of the diversity of the approaches, there is a growing consensus in the seismological community that archaeological sources can indeed contribute to a better understanding of the earthquake history of a given area. Seismic archaeology is often seen as the only tool that can constrain the repeat time of major earthquake sources in areas where the written record is limited and the geological expression of large active faults is unknown or difficult to decipher; but how this is to be achieved is neither widely known nor generally agreed upon. A close scrutiny of the relevant literature shows two basic approaches to the problem of extracting information about earthquakes from archaeological sources, each with advantages and disadvantages to different degrees. Abandonment-collapse sequence approach In this approach, conclusions are drawn from architectural collapses or damage, usually occur-

ring in an abandonment-collapse sequence. This approach includes any attempt to associate earthquake effects with individual archaeological structures, in most cases with serious dating problems and often with the result of simply confirming knowledge already acquired by other means. Hence the 'discovery' and study of a collapse may in itself be of little significance for seismology although of great importance for the history of the site. Only by situating a collapse within the seismic history of the area can its scientific value be assessed. If evidence of seismic collapses is found in an area about which historical seismology already has available a great deal of information, the finding may not justify devoting human and financial resources to it. Nevertheless, careful analyses of collapses are crucial as they supply useful lists of seismic indicators and may steer future work in an appropriate direction. Combinatory method approach The second approach includes all research that adopts a 'combinatory method', making joint use of written sources and archaeological data. From an epistemological point of view, this approach is dangerous, as it may create a kind of circular argument where written sources are used to support and explain archaeological sources and the latter are used to supply what the written sources do not say. Extreme caution is always necessary in using written sources to date in situ earthquake effects, and general conclusions should never be drawn. Only in the exceptional case of Pompeii can one make direct joint use of archaeological and written sources, because of the fact that the city was buried in ash from Mt Vesuvius 17 years after the earthquake (AD 62), while rebuilding work was still going on. The 'combinatory method' is mostly used by the older generation of Mediterranean-based archaeologists. The basic interpretational factor that it seems to have neglected or underestimated is the dating system, as written sources and archaeological sources are based on two conceptually different time scales resulting from the different nature of their data. A written source may indicate not only the year, month and day on which a historical event took place, but even smaller units of time such as the hour. Conversely, archaeological evidence can only be dated by reference to the stratigraphic sequence within which it was found. Archaeological dating, in other words, is indeed based on an ad annum time scale, but in practice it is a span of years that is bound to be involved. When

AIMS AND METHODS IN TERRITORIAL ARCHAEOLOGY archaeological and historical sources are used in combination, however, the former are dated on the time scale of the latter, which may indicate a day or smaller unit of time. Hence the use of written sources to date archaeologically attested earthquake effects may lead to an interpretational error with serious potential consequences. Case study: the 21 July AD 365 earthquake. The date of an earthquake in written sources has occasionally been imposed on a number of different archaeological collapses actually caused by different earthquakes over a period of years or decades and possibly occurring thousands of miles apart. In this way, groups of such events have been interpreted as a single huge earthquake. For example, a reconstruction of the earthquake of 21 July AD 365 using the 'combinatory method' suggests that the area of destructive effects stretched from Cyprus to western Sicily, and even involved North Africa, making it appear an 'archaeological catastrophe'. This may not be of much interest to archaeologists, but it is to historical seismologists, as recent studies have shown that the main and largest earthquake occurred off the southwestern coast of Crete, and even serious damage reported for Egypt, Syria and Palestine in close temporal association with the 21 July event may have been due to smaller earthquakes triggered by the main shock (for a general review of the written sources as well as of the historical and archaeological debate, see Guidoboni et al (1994, pp. 267-274)). It has also often happened that instead of reassessing scientifically untenable positions, users of the 'combinatory method' have concealed their failure to make correct use of the archaeological dating system behind personal polemics, thereby demonstrating a neglect of basic interpretational notions (for the latest case, see Bernabo Brea (1997)). Case study: the 3 January 1117 earthquake. Another example concerns the earthquake of 3 January 1117 in northern Italy (Guidoboni 1984). From numerous studies of 12th-century Romanesque churches of the 'Lombardo' style (there were more than 360 such churches over the whole of northern Italy), art historians have attributed many changes of style, alterations, extensions and so on to destruction caused by this earthquake. The earthquake was mentioned by many written sources, although with very general descriptions of its effects. In relying solely on the written sources, however, the art historians failed to take into account that, in the first half of the 12th century, churches were often built or extended for reasons of population growth, economic change or prestige. This

47

misuse of the 'combinatory method' has resulted in the supposed area of earthquake-induced destruction being enormously enlarged. The suggestion that the event was much more severe that in reality has serious consequences for the earthquake-hazard estimates of what is normally considered a low-seismicity area. It has required long and patient historical research to reassess the true earthquake effects exclusively based on reliable data (Guidoboni 1984; Guidoboni & Boschi 1989).

Territorial evidence of major earthquakes Both the 'abandonment-collapse sequence' and the 'combinatory method' approaches place serious limitations on the use of archaeological sources for seismological purposes. For this reason, a different approach is explored here, although one that also has its limitations. The aim of this approach is to devise a congruent hypothesis, with the realization that, in our present state of knowledge, faith is essential if substantial certainties are to be achieved in seismic archaeology. The proposed approach is based on the following consideration: historical seismology tells us that an earthquake of magnitude c. 1 (substantially larger earthquakes are unlikely at least in the Mediterranean region) striking an area subject to human occupation causes considerable territorial perturbation, in the form of destruction, abandonment, urban area contraction, population shrinkage and emigration, reconstruction, and general disruption of the natural environment, often stretching over a fairly long period. For example the earthquake that took place in the Straits of Messina on 28 December 1908 (Ms 7.1), the only large event known to have occurred in the area, left a very long-lasting territorial imprint (Boschi et al. 1995, 1997; see also the map of effects, Fig. 1). It should be recalled that until this earthquake occurred, the Straits area had been considered of low seismicity, although Messina and Reggio Calabria had already been seriously damaged by earthquakes from adjacent seismogenic sources, for instance in 1783 (Baratta 1901). The question is posed: would it be possible to draw from ancient and late antique archaeological sources a territorial picture that fits the likely 'territorial perturbation' resulting from a disaster similar to that caused by the 1908 earthquake? Historical research specifically aimed at uncovering evidence for large earthquakes in the Straits area had already been carried out for the period from the seventh to the 13th centuries (Guidoboni & Traina 1996). This research allowed the

Fig. 1. Intensity distribution for the 28 December 1908, Straits of Messina earthquake (data from Boschi et al. (1995, 1997)). The dashed line encircles the area of intensity X and over.

AIMS AND METHODS IN TERRITORIAL ARCHAEOLOGY identification of six previously unknown earthquakes, but none of them had a magnitude comparable with that of the 1908 earthquake. This was not totally unexpected in view of the 1000 years average repeat time estimated by seismologists for 1908-type events (Valensise & Pantosti 1992). A systematic search for territorial traces of a great seismic disaster in the preceding millennium, that is, between the end of the sixth century BC and the fifth century AD, commenced in 1996, and lasted for a year. The findings discussed here provide an initial contribution to the seismological application of already available archaeological data. This new perspective incorporates some relatively novel concepts, such as 'archaeological landscape' and 'context' analysis, which we regard as significant in throwing light on territorial dynamics. The data came from a systematic retrieval of results published in archaeological literature on the Straits of Messina. We first established a complete census and record of published studies, which were then subjected to a careful critical scrutiny. The data from the various archaeological finds were then subdivided into a historically significant sixphase chronology (see the section 'A chronology of territorial dynamics' and the Appendix), which yielded a complete representation of our present state of knowledge at territorial level. This, together with the critical reappraisal of seismic indicators reported in the literature, made it possible to re-examine certain earthquake cases within a more precise conceptual framework. Population estimates based on the compiled data allowed a diachronic analysis of the area to be carried out, with the aim of identifying any evidence for significant perturbations of the habitational dynamics between classical and Hellenistic times and the late antique and Byzantine periods from which many historical records commence. Seismic indicators in the archaeological record: some cautionary remarks In post-processual archaeological theory, the concept of landscape as context or as a palimpsestic container of information about the past has become firmly established. In this usage, 'landscape' may be considered as the compound effect of a sequence of activities and/or events acting on different scenarios (geomorphological, demographic, cultural, political, economic) that need to be separated through space and time. From the archaeological point of view, reconstruction of events is possible only to the extent that 'actions' have left identifiable traces in the

49

ground. As a whole, these traces have in the course of time been affected by nature or humans in such a way as to transform or even remove them completely. The analysis and interpretation of an archaeological site, therefore, involves decoding these palimpsests, and the decoding must itself involve a reconstruction of the events that occurred during transformation of the ancient buried record in question (Schiffer 1987; Leonardi 19926, pp. 13-22). As items of evidence, earthquake indicators typically have the characteristics and ambiguities of all other kinds of archaeological evidence: the connection between the depositional record revealed by excavations and the processes that brought it into being can be hypothesized or inferred, but it cannot be proved in a deterministic way. As a result, there are only a few cases where a seismic hypothesis has been accepted to account for a specific situation in the archaeological record. A common characteristic of these cases is the presence of macro-indicators, such as collapses, in structures that can be described, in a functional and constructional perspective, as 'prestige buildings', such as villas, baths and sacred places of the late Imperial age. In most of these cases, moreover, the earthquake is taken to have occurred after abandonment of the site, for it is easier to find indicators of seismic activity where human reaction to seismic damage has not involved restoring the topography and structure of the site. In this way, however, many pieces of seismic evidence may have been bypassed. For example, data concerning rural settlements and modest buildings regularly escape notice, even though they were characteristic of extra-urban territorial organization in classical Greek and Roman times. The same is true of the various ways in which society reacted to seismic disasters from an economic and recovery point of view. In most of the cases we analysed, the earthquake is taken to be a macroscopic agent affecting individual inhabited buildings, the event responsible for collapses and abandonment. However, earthquake effects are never considered as potentially responsible for disturbing already existing archaeological deposits. This contrasts with Anglo-Saxon archaeological literature, in which the primary effect of an earthquake on the buried archaeological record is recognized in the formation of cracks which may be locally filled with artefacts redeposited by natural processes, usually flowing water (Wood & Johnson 1978; Butzer 1982; Schiffer 1987, pp. 231-233). In stratigraphic terms, this results in (1) the formation of deposits containing artefacts in a secondary location, (2) the horizontal and vertical movement of artefacts, and

50


(3) the creation of possible false functional and chronological associations of deposit finds. These are important matters for archaeology, and only careful stratigraphic analysis will reveal them. But if not identified and appropriately situated in the stratigraphic sequence, they may upset the interpretation of the sequence itself. In the archaeological literature concerning the Straits of Messina, cracks and related archaeological evidence are never mentioned. In terms of the evidence that is reported, we also found a certain failure to classify seismic indicators adequately, even though at first sight they seemed to cover a broad range of stratification phenomena. The situation is in fact a complex one, for however analytical the archaeologist's approach may be, the subdivision of the stratigraphic continuum into taxonomic indicator categories will be too selective. This complexity warrants a few final words of caution. The existence of elevated sites introduces the problem of the downslope displacement of archaeological material as a result of erosion and deposition. Secondary deposits of this kind show the need for a careful analysis of possible earthquake effects on slopes. The same care ought to be applied to interpreting deformation and movement in architectural structures, such as leaning walls and sloping floors, situated on steep slopes or in landslideprone areas. A building's resistance to ground movements is obviously conditioned not only by its own structure but also by the nature of the ground itself: the frequency of structural deformation in monumental buildings (theatres, for example) highlights the problem of lack of solidity in basal strata and unstable nature of the backfill on which they were built. Seismic activity tends to exacerbate any problems of statics that may have already existed. Furthermore, abandonment, collapses and destruction may be related not only to damage caused by war, invasion, or political and social upheavals, but also to ordinary decay as a result of lack of maintenance. In turn, ancient restoration or rebuilding works may in fact be related not only to site reoccupation after military destruction, but also to population increases, changes in the organization of production, and functional modifications. Territorial archaeology in the Straits of Messina area Physiographic and archaeological setting The Straits of Messina area is situated between latitudes 37°40'N and 38°50'N, and its coastal

environs cover about 6600km2 of land. The landscape of the Sicilian side is dominated by the Monti Peloritani, which trend parallel to the coast and consist of Palaeozoic granites and crystalline schists. Their slopes are furrowed at nearly regular intervals by short, deep and steep watercourses, locally termed fiumare. The Messina area has very little agricultural hinterland, yet through the centuries the city developed as a successful maritime and commercial centre as a result of its location and its natural harbour. On the Calabrian side, the territory of Reggio Calabria consists of a series of very rough-edged plateaux corresponding to the top of technically uplifted Mid-Late Pleistocene fan deltas. The embayed form of the Straits' coastline makes it suitable for the establishment of small landing places, of which Bova Marina and Pellaro are examples. The principal lines of communication were, and still are, strongly influenced by this physiography; in antiquity there were coast roads from Messana to Panormo (Palermo) and from Messana to Catania and Syracuse. In spite of its great importance in the ancient (classical) and early Byzantine (late antique) world due to its strategic position in the Mediterranean, the area has not been the object of systematic research in territorial archaeology. Its physical characteristics do not make it particularly suitable for settlement, yet there are sound historical reasons for supposing that in ancient times it was covered by a dense settlement network. Archaeological research is hindered by the fact that the geomorphological setting of the Straits area, especially on the Sicilian side, does not favour the preservation of the buried archaeological record. The landscape of the Monti Peloritani is very active, being characterized by sudden erosion and flooding along the steep watercourses. This has frequently resulted in the removal of upland archaeological deposits and the burial of low-lying deposits, with the city of Messina itself being a case in point. In the first stage of our research we systematically scrutinized all the archaeological literature published since 1876, especially as regards those national and local journals in which accounts of excavations carried out by guardianship bodies appeared, and bibliographies. The search involved 11 titles, each spanning a period ranging from at least 30 to more than 100 years (see Appendix 2). A total of 178 contributions were analysed, and 184 finds recorded and precisely located. We also examined a variety of unpublished material made available to us by the Archaeological Superintendency of Reggio Calabria (for a complete bibliography of studies, see Storia Geofisica Ambiente (1996)).

AIMS AND METHODS IN TERRITORIAL ARCHAEOLOGY Unfortunately, these documents are by their nature discontinuous, in part because they reflect the fact that the archaeological research of the time was concerned with monuments and objects of historical and artistic value, and in part because they reveal the rather 'primitive' nature of fieldwork techniques before present-day stratigraphic excavation came into use. A new aspect of archaeological research that emerges from the most recent publications is worth noting. Since the 1980s, Superintendency reports have recorded a substantial reduction in state subsidies. This has resulted in planned research giving way to 'rescue archaeology', an increase in non-systematic field surveys designed simply to identify the buried record, and a decrease in full-scale excavations. Such a change of strategy has had a positive effect on the establishment of a better overall representation of the territory and of its demographic trends, but it also had negative effects because recent advancements in stratigraphic excavation methods have not been applied. The body of information potentially available over the last decade has therefore been insufficiently analysed. The fact that archaeological investigations have concentrated, now as in the past, on ancient towns and their immediate surroundings, has meant that there are numerous visible 'gaps' in the archaeological maps of any given area alongside dense concentrations of finds, which often simply reflect the intensity of research devoted to that area. In spite of these limitations, we consider that the settlement pattern that emerged from our archaeological 'census' over a long time span was indeed representative of real population trends and may reflect events that had a powerful impact on the area.

A chronology of territorial dynamics: fifth century EC to sixth century AD To emphasize changes in territorial dynamics we adopted a conventional six-phase system for the division of periods (Table 1), each period bounded by a major political and historical transition likely to have a lasting effect on territorial organization. This classification was made necessary by the particular nature of the available documentary material and by the aims of our research, which required a flexible system to be applied both to a regional context and to a long chronological span. Given the poverty of some finds, the frequency of clandestine excavations and the intrinsic nature of archaeological research, a large proportion of the record can be dated only to the Greek or Roman periods generically, or cannot be dated at all.

51

Table 1. Six-phase division of chronology adopted in the paper

Phase

Chronological span Historical reference

Phase 1

510/509-406 BC

Phase 2

406-273/212 BC

Phase3

273/212-27 BC

Phase 4

27BC-AD313

Phase 5

AD 313-535

Phase 6

AD 535 to the end of the sixth century

From the fall of Sybaris to the ascent of Dionysius I of Syracuse From Dionysius I to the Roman conquest The age of the Roman republic The Imperial age, from Augustus to Constantine (Edict of Milan) From Constantine to Justinian The Graeco-Gothic War and the Byzantine rule

By adopting conventional and broad periods, however, we could place all datable finds within appropriate spatial and temporal coordinates. Thus every archaeological record, however imprecise or incomplete, preserved its own historical and contextual value, and in several cases the crude find regained significance after insertion into the regional chronological picture. The final product of these broad interpretations is a series of maps, one for each phase (Fig. 2). These 'phase maps' may prove to be underestimations or overestimations. Underestimation will be due to the fact that, although a site may have survived from one phase into another, it has been 'compressed' into a single phase. Overestimation occurs when an imprecise archaeological find is duplicated and placed in more than one phase. In fact, one may reasonably guess that certain types of find assigned to a generic chronological or cultural horizon (e.g. a small 'Roman' farm or cemetery) are unlikely to persist throughout a chronological span of five or six centuries. By using all available information over the long term we inevitably lose some short- and medium-term chronological, functional and geographical variability for the area in question; however, the only alternative would have been the total elimination of general data.

Functional types and demographic factors On the basis of current excavation literature, for each site the analysed finds were classified

52


Fig. 2. Settlement distribution during Phases 1-6, (a)-(f), respectively. Sites in boldface are mentioned in the text. Dashed line as in Fig. 1.

AIMS AND METHODS IN TERRITORIAL ARCHAEOLOGY

Fig. 2. (continued)

53

54

Fig. 2. (continued]


AIMS AND METHODS IN TERRITORIAL ARCHAEOLOGY according to the following functional types: inscriptions; votive deposits or sacred places; settlements: remains belonging to a villa, farm, village or town; funerary contexts: cemeteries or isolated burials; infrastructures: aqueducts, wells, cisterns, bridges or roads; coins and/or jewellery hoards; sporadic finds or uninterpretable contexts; fortifications. All the finds were located through their geographical coordinates and used to create the maps described in the previous section, with the aim of establishing models and finding the settlement density for the area in question (Leonardi, 19920). This made it immediately possible to examine the general population trends during each phase. When sites are of the same size in different phases, an increase in the overall number of sites may be related to a population increase, whereas a decrease may signify a population decline (Cambi & Terrenato 1994). This condition applies at least for Phases 2, 3 and 4 and partly for Phase 5 during which the population appears to be scattered through the territories of the large towns and gathered around farms and country houses, or in occasional villages. In our relative population estimates we first have taken account of all the classified sites, then considered only those that can be considered loci of demographic attraction: settlements, funerary contexts and fortifications. The other find types provide a reliable indicator both of the extent to which a territory has acquired an infrastructure, and of the degree to which the archaeological record has been broken up in certain morphological or inhabited basins because of natural or anthropic transformations (destructive events). How to progress from a relative to an absolute estimate of demographic dynamics is the subject of lively debate in archaeological circles. Solving this problem requires a series of methodological decisions, each involving a substantial degree of arbitrariness and standardization (Cambi & Terrenato 1994). For example, the application of standard population indexes requires a careful assessment of the characteristics of the area under analysis. In our specific case, it seemed counterproductive to attempt an absolute quantitative approach to the problem of population in the period under consideration. Towns required a different procedure (Muggia 1997). Although macroscopic expansion and contraction phenomena in towns can often be identified, it is very difficult to assess small changes in the use of space. For example, how can we assess changes in house size in Roman times within the insulae, which always keep to Greek dimensions? And how can the number

55

of persons per household be established? Here too our understanding of diachronic variations depends on macroscopic and/or dimensional changes in towns. Quantification of the urban population of Messina and Reggio Calabria also required a series of preliminary calculations. First, we calculated the resident population of towns in Sicily and Magna Graecia outside our sample area and whose urban layout is already known or can be easily worked out. From estimates made for nine towns (Lipari, Megara, Hyblaea, Tindari, Eloro, Casmene, Locri, Camarina and Naxos; see also Muggia (1997)), we deduced certain demographic indices in relation to the various phases, summarized as follows: (a) 300 persons ha"1 for the Greek period, equivalent to about 30m 2 per person (Phases 1 and 2); (b) 150 persons ha~ ! , equivalent to about 70 m2 per person, for the period of the Roman republic, generally considered to be a critical stage in urban and territorial organization (Phase 3); (c) 200 persons ha" l , equivalent to about 50 m2 per person, for Imperial and late antique times (Phases 4 and 5); (d) 100 persons ha"1, equivalent to about 100m2 per person, for the Byzantine period (Phase 6). A reasonable estimate for the population of Messina and Reggio Calabria was obtained using a projection of these standard indices, expressed as the product of population density multiplied by the occupied urban space (Hassan 1981).

Archaeological landscapes and their modifications A systematic survey of archaeological literature for the Straits of Messina area allowed the identification of chronological landscape modifications resulting from human and natural activity. Landscape archaeology is a well-known aspect of the study of territorial and population dynamics, and it is here that the idea of landscape as a palimpsest has become established (Cherry 1983; Fowler 1990; Leonardi 19920). Diachronic landscape surveys are a very important instrument for decoding these palimpsests, especially in crucial areas such as the Straits of Messina, where the interaction between humans and events includes destruction brought about by nature or humans themselves. This interaction may account for population movements,

56


changes in settlement structures, and economic and productive employment. From a strictly archaeological point of view, landscape changes have a direct effect on the survival and disappearance of buried sites. In addition, there are recent landscape changes brought about by humans, such as the introduction of new farming techniques, intensive farming and land reclamation and, especially acute in the Straits area, modern construction (buildings, railways, motorways, etc.). along the coast.

Seismic indicators from excavation reports On the basis of excavation reports, a formal classification of seismic indicators that were recognized by archaeologists was established (Table 2). This classification was then applied to the systematic reappraisal of the archaeological literature. Where excavation reports could be

appropriately reinterpreted, they led to a broadening of the database in terms of area, chronology and function. The reinterpretation of the data gathered on the basis of our preferred characteristics made it possible to bring certain typologically and functionally different sites (i.e. modest settlements, as well as farms and cemeteries) into the overall context of an area in diachronic evolution. This fresh scrutiny of the archaeological record made it possible to increase the number of earthquake indicators by adding type 6 (anomalous upward or downward displacement of habitation levels) and type 7 (absence of any archaeological record for a particular chronological phase) (Table 3). In summary, the research was based on the following criteria: (1) We drew from the available archaeological literature items suggesting possible ancient earthquakes, without recourse to the

Table 2. Seismic indicators from the literature Reference

Indicator

Depositional outcome

1 2

Surface faulting

SchifTer (1987)

Structural failure; fissuring of floors and walls

Lattanzi (1982); Cavalier & Bernabo Brea (1993-1994); Spigo (1993-1994)

3

Collapses, ground movement

Voza (1980-1981, 1984-1985); Cavalier & Bernabo Brea (1993-1994); Spigo (1993-1994)

4

Deformation, disintegration; displacement of the archaeological material

Cavalier & Bernabo Brea (1993-1994); Spigo (1993-1994)

5

Ancient restoration work

Voza (1980-1981, 1984-1985)

Table 3. Seismic indicators from reinterpreted records Indicator

Depositional outcome

Reference

1

Surface faulting

Bonanno (1993-1994)

2

Structure failure; fissuring of floors and walls

Orsi (1899); Ferri (1926)

3

Collapse, ground movements

Orsi (1899); Bernabo Brea (1964-1965); Bonanno (1993-1994)

4

Deformation; disintegration; displacement of the archaeological material

Orsi (19166); Bonanno (1993-1994)

5

Ancient restoration work

Orsi (19166); Lattanzi (1986)

6

Anomalous upward or downward displacement of habitation levels

Bacci & Rizzo (1993-1994); Bernabo Brea (1997)

7

Absence of any archaeological record for a particular chronological phase

Cavalier (1994)

AIMS AND METHODS IN TERRITORIAL ARCHAEOLOGY common game of combining deposit characteristics with the chronological process leading to the formation of the archaeological record. (2) Where possible, we sought out unpublished material, to better reconstruct the exact sequence of archaeological deposits through precise stratigraphic reanalyses. The purpose of this was to distinguish between natural and human actions in deposit formation, as well as to date the deposits (as suggested by Leonardi & Balista (1992)). Unfortunately, this approach increased our knowledge only for Reggio Calabria. (3) We assessed the reliability of archaeological indicators of earthquake effects within the historical, demographic and socioeconomic context of the area over a given period. We believe it is possible to identify traces of earthquakes in the form of 'gaps' within the general picture of a period, such as evident changes in habitational patterns that cannot be explained in terms of local historical dynamics.

Archaeological evidence for past earthquakes Evidence of seismic activity from the archaeological literature We now examine published evidence for large earthquakes that may have involved the towns of Messina and Reggio Calabria. Messina. There is no specific archaeological evidence for earthquake effects at Messina, but a general reference to these was made by Orsi (19160), who attributed the lowering of the land level at the Prefecture excavation to the high seismicity of the area. There is possible evidence for seismic deformation in the possible postdepositional disturbance of the isolato 373 cemetery (Scibona 1984-1985), where skeletons seem to be slightly broken up and shifted from their original position. Evidence for significant lowering of the harbour by settling of the underlying loose deposits is found in the accounts of the 5 February 1783 earthquake in the Gioia Tauro Plain, located 30km northeast of Messina (Boschi et al 1995, 1997). This evidence is indicative of a phenomenon that may have characterized the city during previous but unknown large events, either in the Straits or in neighbouring areas. Reggio Calabria. Unlike Messina, Reggio Calabria does provide archaeological indicators of

57

seismicity at four locations in the city, together with an inscription dated to AD 374 that mentions the rebuilding of the baths owing partly to their age and partly to earthquake damage (Guidoboni et al. 1994). The earliest, but very unconvincing, suggestion of earthquake deformation regards the via Vollaro site (06). Fiorelli (1886a) thought an earthquake could account for two sections of a column being in a collapsed position, but it is also possible that these sections were never actually used. Perhaps more convincing, Fiorelli (18866) interpreted disconnections and collapses in the Belvedere aqueduct (51) as a post-depositional effect resulting from the combined action of earthquakes, damp and agricultural work. A significant advance occurred with the Stazione Lido excavations (site 49), where two emergency campaigns were completed in 19761977 and 1979-1980 (Fig. 3). Unfortunately, when the first campaign was carried out, the stratigraphic sequence had already been altered by the works for the construction of the state railway, and no earthquake was identified with certainty (Ardovino 1978). However, on the basis of the usual indicators, we have inferred a seismic origin for a possible mid-first-century BC collapse, as well as for damage to an apsidal structure and to the buttresses of a wall. The archaeologists who carried out the subsequent 1979-1980 excavation (directed by A. Racheli and summarized by Spadea (1993)) showed greater sensitivity to the problem and identified evidence for damage, collapses and restoration works occurring at different times. The published excavation report dates the earliest evidence of a collapse to the end of the fourth century BC, but earthquake effects are offered as an explanation only as one of the many possible causes of collapse in a containment structure. Some fluvial deposits are also identified as belonging to the same period. The most significant evidence, however, belongs to the late antique period: adjacent to the second-century AD nymphaeum are some mid-third-century AD rustic buildings that are reported to have suffered 'traumatic abandonment' in the early or mid-fourth century AD. The banks of the Torrente S. Lucia are found to have been damaged in this phase, and were subsequently flattened. After the midfourth century, the complex was renovated, with added steps and a colonnade, but shortly afterwards there was a new collapse, which was subsequently buried under sandy deposits. Revised evidence of a tsunami. To account for the presence of a thick sandy deposit within the stratification of the Stazione Lido excavation in

Fig. 3. Map of excavations at Reggio Lido.

AIMS AND METHODS IN TERRITORIAL ARCHAEOLOGY Reggio Calabria, Spadea (1993), perhaps influenced by other archaeological work based on the 'combinatory method', invoked the action of a seismic wave (tsunami) caused by the earthquake of 21 July AD 365, located off the southwestern coast of Crete (see earlier discussion of this event). This hypothesis, however, runs completely counter to the chronological sequence of the site, which can be reconstructed with consistency and precision thanks to the presence of dateable material. Unpublished material made available by the Superintendency (produced by the archaeologist A. Racheli), and a detailed examination of the stratigraphic matrix of the site, shows evidence for various destructive events, which seem to have been amalgamated and somewhat neglected in Spadea's published summary (1993). As we have already indicated, the occupation phase testified by the rustic and/ or artisan buildings can be dated to the first half of the fourth century AD. The structures concerned show poor building techniques and sandy floors. The habitational area in question was abandoned as a result of a sudden collapse, as can be judged from the presence of vases in situ. The collapse is sealed by thin layers of fine, golden, micaceous sand containing floated ceramic material (see Fig. 4). This collapse is possibly contemporary with evident damage found in a large first-century AD masonry embankment, but the damage here has been identified as relating to a complex of earlier structures still in use.

Fig. 4. Byzantine wall from excavations at Reggio Lido.

59

A secondary episode of artificial levelling containing little debris and dating to the second half of the fourth and the early fifth century AD immediately precedes a monumental phase characterized by the construction of a staircase, a colonnade and a drain channel, and associated with mortared floors. Above deposits linked to the destruction of these structures there is a second, more substantial sandy layer coincident with a deposit that Spadea (1993) suggested was brought in by the tsunami of AD 365. This layer of 1.5m thickness of lenticular sand and gravel bodies displaying a random particle-size arrangement, has been found throughout the seaward side of the excavation, but the surviving masonry seems to have acted as a barrier to its landward extension. Spadea's hypothesis is based on the draping of this deposit over one of the many collapses at Stazione Lido, but the very large extent of the area under investigation is itself evidence that the sandy deposits were of artificial origin used as infill for terracing (Ardovino 1978, p. 83). Unfortunately, we lack analyses of these crucial sediments as a result of two drawbacks to the Reggio Lido excavation: (1) the urgency and pressure brought to bear on the excavation because public building site operations were not halted, which resulted in bulldozers and other machinery moving about during the excavation works; (2) the failure to realize the value of this large archaeological area overlooking the

60


Straits, which was never considered for preservation as an archaeological park. These two factors had a decidedly unfortunate effect: not only was there a failure to secure appropriate sampling for sedimentological analyses, but the excavation was followed by the total destruction of the site, such that further investigation is now impossible. Photographic and stratigraphic documentation, of real value in view of its good quality, allows us to make certain new observations regarding, in particular, the lower thin sandy layers overlying the collapsed artisan buildings of about the mid-fourth century, which were missed in previous summaries. Their significance arises from the fact that the anomalous sandy matrix is not found in sediments identified at the site. This thin 'exotic' layer could be evidence of a strong tsunami, which violently flung marine sand onto the inhabited coast and left it there.

A large earthquake in the fourth century AD? So far, we have put together a series of 'convergent' observations concerning anomalies in the urban and territorial setting of the Straits that cannot be explained in the usual terms, along with some specific and significant finds. These together suggest that a strong earthquake was felt in the region around the middle of the fourth century (see also Appendix 1). The key observations and finds are as follows: (1) the Straits settlement network contracted substantially in Phase 5; (2) the inhabited area of Messina also contracted substantially and continued to do so until the beginning of Phase 6 to the extent that the phenomenon has been interpreted as abandonment; (3) third- and fourth-century inscriptions, found within the urban area of Reggio Calabria, were reused at the beginning of the fifth century; (4) tombs are found within the city of Reggio Calabria in Phase 5; (5) an inscription of AD 373 records the collapse of the baths at Reggio Calabria; (6) there exists evidence of collapses in the Reggio Lido locality towards the middle of the fourth century; (7) there exists a possible tsunami deposit along the shoreline of Reggio Lido. These 'indicators' are here examined in more detail. Factor (1) is the substantial contraction of the settlement network. Although the diachronic reconstruction of settlements in the Straits area presented here does not contradict current literature (Coarelli 1981; Guzzo 1981; von Falkenhausen 1982; Bejor 1986; Wilson 1990; Greco 1992), it departs from it as regards the striking decrease in available data from Phase 5 onwards. It is known that Reggio Calabria was

sacked by Alaric in AD 410, and that this military campaign created a certain amount of alarm, the manifestation of which was the mass abandonment of the central and northern coast of the Straits, as well as of inland areas. The case of Sicily, however, is different. The balance of probabilities would suggest an increase in the number of rural sites, but in fact exactly the opposite occurs, with the abandonment of the northern coast and its hinterland, and with surviving sites acquiring particular functions. We know that the Vandals raided Sicily between AD 440 and AD 475, but no evidence of military destruction has come to light in the Straits area. From a historical point of view, Phase 6 opens with the Graeco-Gothic War, which affected an area already in a state of considerable decadence. It might be expected that on both sides of the Straits this would be marked by the progressive abandonment of the coast in favour of mountainous inland areas. This occurs only to a small extent in the area of the Monti Peloritani, where there is evidence of a migration from the coast toward inland areas, with only a few vigorous sites surviving. The transition to Phase 6 is even more striking in Calabria. Here, with the exception of the city of Reggio Calabria itself, there is a total disappearance of all sites, including those that had survived through more than one phase and were structurally and functionally stronger. The process of settlement contraction had already begun in Imperial times, but in Phase 6 it is subject to massive acceleration. The economic model offered so far to account for this phenomenon (von Falkenhausen 1982) is unsatisfactory because even productive and receptive structures which centralized activities in late antique times were abandoned. Structures capable of fulfilling similar functions had been created (e.g. the Pellaro-Lume kilns), but their life had been short. Furthermore, the Phase 5 map (Fig. 2e) clearly shows that there is no direct causal relationship between the abandonment of coastal sites 'in a state of crisis' and the opening up of new potential development areas. Factor (2) is the substantial contraction of the Messina settlement, already documented in Phase 5 and even more marked in Phase 6. On the basis of available data, it is probably no exaggeration to describe this as an almost total abandonment of the urban area. There are indications of this phenomenon at Reggio Calabria as well, although probably to a lesser extent than in Messina. Factor (3) is the striking incidence of a phenomenon that we might call 'migration' of inscriptions. There is evidence for six thirdand fourth-century AD inscriptions at Reggio

AIMS AND METHODS IN TERRITORIAL ARCHAEOLOGY Calabria being reused for building materials or found as erratic elements. At two sites (38 and 61), third- and fourth-century public inscriptions (Fiorelli 1888; Orsi 1922) were reused in fairly close chronological contexts. We interpret such a reuse for private contexts of public inscriptions after only a brief time interval as evidence of an abrupt and substantial decay of both the urban landscape and its administrative structures. Factor (4), the presence of tombs within the urban area of Reggio Calabria in Phase 5 also indicates a substantial contraction of the urban area. An inscription of AD 374, which mentions the collapse of the Reggio Calabria baths as a result of an earthquake (factor (5)), was first published by Putorti (1912) and was used as evidence for earthquake damage in Reggio Calabria (Guidoboni 1989; Guidoboni et al 1994). On the basis of this inscription, Baratta (1936) attributed the collapse of the baths to the earthquake of 21 July AD 365. The last two significant factors are the collapses found at Reggio Lido (factor (6)), dated to about the middle of the fourth century AD, and the associated evidence for a tsunami (factor (7)), which was identified from the sand deposits overlying the collapse.

Reappraisal of literary sources At this point it seemed reasonable to ask: how is it that surviving written records from the late antique period do not contain any explicit reference to this hypothesized seismic disaster? To answer that question and attempt to explain the written sources, we need to recall the main features of the cultural and political situation at the time. The second half of the fourth century was steeped in a fairly extreme climate of ideological conflict brought about by the emperor Julian. He was strongly opposed to Christianity and tried to revive ancient pagan culture. Julian came to power in November 361 and died on 26 June AD 363. His death dashed the hopes of many pagan intellectuals, including the famous rhetorician Libanius, and initiated a difficult period for the empire. Libanius is the author of a famous epitaph dedicated to Julian, probably written immediately after his death in June 363 (in the view of Henry (1985), and Jacques & Bousquet (1984)). A literal translation of the Greek text reads as follows: 'Earth, at least, was duly aware of her loss and has honoured our hero [Julian] with fitting mourning. Like a horse tossing its riders, she has destroyed a great number of

61

cities - many in Palestine, and all those in Libya. The greatest cities of Sicily lie in ruins, as does every city in Greece except one [according to Henry's (1985, p. 48) hypothesis, it is the city of Athens]; Nicaea the lovely is laid low, and our loveliest of cities [Nicomedia] is shaken and can have no confidence in the future' (Lib. Or. 18. 291-293). Libanius appears to have used for his own ideological and rhetorical purposes various earthquakes that had indeed struck the Mediterranean area but he did not bother about the exact chronology of events or the places struck, referring to the latter only in a poetic and metaphorical way. Nevertheless, some of the cities and of the associated earthquakes are easily identified because they are also recorded in other sources: Nicomedia, Macedonia and Pontus (AD 358), Nicaea and Nicomedia (2 December AD 362), Corinth and other Greek cities (between AD 361 and AD 363), Palestine (363) and Libya (mid-fourth century AD) (Guidoboni et al. 1994). All of these earthquakes occurred before Julian's death, and therefore before the AD 365 earthquake, and seem to have been used by Libanius as a prediction. Also, it must not be forgotten that in Libanius' earlier Monodia, mention was made of earthquakes that had shaken the world (Lib. Or. 17. 30). A strong earthquake may therefore have involved Sicily and particularly Messina, which was indeed one of the major cities on the island, in the second half of the fourth century AD, probably between AD 350 and 363 (the year of Julian's death). Libanius' reference to the destruction of Sicilian cities in earthquakes is not the only evidence of Sicily's involvement in strong seismic activity. A passage of the Chronicon of Jerome (c. AD 347-419), who continues that of Eusebius of Caesaria, mentions Sicily as having been struck by the great tsunami of 21 July AD 365. It must be pointed out that Jerome was probably about 18 at the time and living in Rome, though he later lived in Antioch and Constantinople; hence he may well have been able to gain indirect oral evidence of what happened. The date of the tsunami was treated as epoch-making by Christian rhetoricians, who, for strictly ideological purposes, intended to stress the crisis that struck the empire after the death of Julian. It is therefore reasonable to suggest that this date may have attracted other local events to the general description of 'earthquake'. Given the present state of knowledge of the earthquake of 21 July AD 365 we find it very difficult to identify effects on the coast of Sicily that could be seen as even vaguely fitting Jerome's text: There was an earthquake


62

throughout the world, and the sea flowed over the shore, causing suffering to countless peoples in Sicily and many other islands.' We must point out that the literary expression 'throughout the world' seems to have been used by Jerome more to bring together various seismic events that could be placed under the general heading 'earthquake' rather than to indicate a real geographical horizon (even to the world as it was known at that time) within which a single phenomenon occurred. This same literary device is also found in the monastic annals of early and mid-medieval culture, where the earthquake as a sign becomes universal, in spite of its appearance as a local event and a phenomenon circumscribed by the natural world. We think that the universalistic aspect of late antique Mediterranean culture, permeated as it was with values and expectations tending to accentuate moments of ideological crisis, provides a key to these enigmatic references in surviving contemporary sources to a seismic disaster involving Sicily. As for the failure to mention Calabria, on the other side of the Straits, this may well be a case of synecdoche; even today, the 1908 earthquake is often referred to as 'the Messina earthquake', although the number of damaged sites on the Calabrian side is larger than the number of Sicilian sites (see Table 4). It is worth mentioning that earlier research did find possible traces of earthquake activity in mid-fourth-century Sicily and Calabria by simple analysis of written sources (Guidoboni 1989; Guidoboni et al 1994), but these studies concluded nothing about the quality of the phenomena, and the effects were portrayed as separate regional events. For a number of reasons, ancient written sources can indeed preserve the record of even very small earthquakes. What our archaeological research has now done is to provide new and convergent evidence supporting the hypothesis that a strong earthquake, large enough to produce destruction on both sides of the Straits, struck the region between AD 350 and 363. Table 4. Number of sites of Calabria and Sicily that were damaged by the 28 December 1908 earthquake (intensity VII-XI)

MCS intensity

Calabrian sites

Sicilian sites

XI X and X-XI IX and IX-X VIII and VIII-IX VII and VII-VIII Total

10 57 20 110 85 282

1 13 15 61 70 160

Conclusions Previous sections have described in detail and discussed several lines of evidence converging towards a scenario that is typical of a large earthquake. Such evidence includes a substantial contraction of the settlements on both sides of the Straits between the fourth and the beginning of the fifth centuries; the presence of tombs of about the same age within the urban area of Reggio Calabria further supporting the contraction of the city; the existence of third- and fourth-century inscriptions reused for building materials only a few decades after their first emplacement; the presence of an AD 373 inscription referring to the collapse of the Reggio Calabria baths and of archaeological findings suggesting collapses at Reggio Lido around the middle of the fourth century; and the presence of anomalous sandy deposits overlying collapsed fourth-century buildings and suggesting the action of a tsunami wave. Can the 1908 earthquake in the Straits of Messina be taken as the 'modern analogue' of this fourth century disaster? Or, in other words, could the two earthquakes be representative of the activity of the very same seismogenic source? We believe the answer is 'yes'. Modern seismological wisdom suggests that the largest seismogenic sources tend to rupture repeatedly in successive similar-sized or 'characteristic' earthquakes occurring at relatively regular time intervals (Schwartz & Coppersmith 1984). Although the regularity in time of even the largest earthquakes has been since questioned and thoroughly debated, the hypothesis of similarity among subsequent ruptures along the same fault has been successfully tested in many areas of the world. For the specific case of the Straits of Messina, Valensise & Pantosti (1992) have shown that the main landscape features of the Straits (including the sharpness of the Peloritani range, the steepness of the fiumare and the existence of several elevated plateaux) are controlled by tectonic processes that are well explained by the superimposition of continuous land uplift and the pattern of crustal deformation observed following the 1908 earthquake. In other words, the present structure of the Straits is compatible with the repetition of earthquakes similar to that in 1908 as the main local tectonic agent. This finding has two fundamental implications. The first is that if the 1908 earthquake was a 'characteristic earthquake', one can expect any of its predecessors to have produced roughly the same pattern of damage and land modification as that seen at the beginning of this century (see Fig. 1). This is especially important in view

AIMS AND METHODS IN TERRITORIAL ARCHAEOLOGY of the fact that the 1908 event also generated a tsunami that was especially intense around Reggio Calabria. The second implication is that if 1908-type earthquakes dominate the tectonic scene of the Messina Straits, that is to say, if they relieve most of the tectonic stress periodically accumulated in the region, there should be little room left for additional significant sources unless they are located at the edges of the 1908 fault. As discussed in the Introduction, none of the earthquakes known to have occurred within the Straits produced widespread collapses in Reggio Calabria or Messina, and only two of those that occurred outside the Straits (the 5 and 6 February 1783 Calabrian earthquakes) induced limited building collapses around the Straits. The combination of archaeological and seismological evidence therefore suggests that the AD 350-363 earthquake can be taken as a 1500 years older ancestor of the 1908 catastrophe. Such a long interval falls at the upper limit of an existing palaeoseismological estimate (an average repeat time of 1000 (+500, -300) years for repeated 1908-type earthquakes was postulated by Valensise & Pantosti (1992)) and agrees well with average repeat times estimated for other large Italian earthquakes (for a summary see, e.g. Valensise & Pantosti (2000)). Explaining the profound changes in the Straits' territorial dynamics around AD 350-363 as the result of a catastrophic earthquake throws new light on the

63

history of occupation of this important site, supports the hypothesis that the Straits will not experience another catastrophic earthquake at least for several centuries, and highlights the value of archaeoseismology for a better understanding of seismic activity in areas where the candid assessment of the true earthquake potential has become an issue of national importance. Appendix 1: Territorial dynamics in the Straits of Messina area

Extra-urban sites The following is a brief phase-by-phase summary of the territorial scenarios obtained from the archaeological record for extra-urban settlements. The analysis focuses on the areas that would more clearly record the effects of a possible 1908-type earthquake. The two cities of Messina and Reggio Calabria are dealt with separately (see below). A general overview of the development of the settlement network is shown in Fig. 5 for the Sicilian sites and Fig. 6 for the Calabrian sites. Phase 1 (510-406 BC). 59 sites were identified for Phase 1 (Fig. 2a). Although this should be a stable and well-defined horizon in the political and economic organization of the Greek poleis and their respective territories, the associated evidence is not homogeneous on the two sides of the Straits. The only evidence on the Sicilian side is in Messina. The situation is more developed in Calabria, as most of the identified sites

Fig. 5. Trend of settlement distribution for the Sicilian side of the Straits.

64


Fig. 6. Trend of settlement distribution for the Calabrian side of the Straits. (e.g. Fiumara di Muro 16, Scilla 53, Gallina 21, Pellaro 46, Gioia Tauro 22 and 23, Melito di Porto Salvo 32, Condofuri 13, Condofuri Marina 14 and 15, Palizzi 44) are spread along the Tyrrhenian and Ionian coast in a fairly regular way. The sites in question are settlements, cemeteries, uninterpretable contexts and/or sporadic materials, which are nevertheless evidence of more than occasional occupation. Sometimes they are associated with structures of a productive type (clay quarries at Calanna 5; a kiln and a landing place at Occhio di Pellaro 40). There is also evidence of farms as a form of settlement (Scilla 53). Penetration inland takes place along the river valleys up to a distance of about 10 km from the coast. Inland sites tend to be situated on the edges of the plateaux to control access. A fortified site in the chora of Reggio Calabria has also been documented (Serro di Tavola 54). Phase 2 (406-2731212 BC). A total of 89 finds were identified for Phase 2 (Fig. 2b), a period for which evidence exists on both sides of the Straits but with a very disproportionate number of sites. The only documented settlement on the Sicilian side is the city of Messina. Inland we find a hoard (Monforte S. Giorgio 48) and a cemetery (Rometta 68). Conversely, on the Calabrian side we find a significant increase in the number of sites, both on the coast and inland. Compared with the previous phase, we find 23 sites showing continuity of settlement but also at least 20 new sites. The occupation pattern on the Calabrian coast continues to be regular and rather spaced out rather than close-knit, and it now takes on a decidedly more stable form with the continuing presence of settlements (Occhio di Pellaro 55) and cemeteries (Melito di Porto

Salvo 43, Reggio S. Caterina 65 and Reggio Piani di Modena 66) as well as the appearance of sacred places (Saline 74). Inland occupation is denser (Fiumara di Muro 22, Calanna 8 and 9 Vito Superiore 89, S. Salvatore Cataforio 73, Motta S. Giovanni 49 and 50, Condofuri 18 and Gallina 28), as one can tell from the presence of sporadic materials and uninterpretable contexts. The quantitative relationship between the birth, death and survival of sites confirms that the general picture for the fourth century BC is one of developing territorial structures: 44 sites show continuity with Phase 1 15 sites disappear, and 45 new sites are recorded for the whole area. Phase 3 (273/212-27BC). A total of 92 finds were identified for Phase 3 (Fig. 2c). On the Sicilian side evidence exists along the northeastern coast of Sicily at Messina, at Spadafora 80 in the form of a kiln, and at Capo Peloro 10, a settlement with monumental features. The main inland evidence is the cemetery at Monforte S. Giorgio 50. The number of sites increases on the Calabrian side, where 25 new sites appear, 30 survive from the previous phase and 21 have disappeared, making a total of 55 sites. The numerical ratio between birth, death and survival of sites seems to suggest a substantial continuity of territorial structures during the transition from the Greek age to the domination of Rome, which is partially confirmed by the survival of Occhio di Pellaro 60 as a port and production centre. A careful examination of the Fig. 2c, however, shows that the habitational centre of gravity seems to be slipping towards the extreme southern tip of Calabria.

AIMS AND METHODS IN TERRITORIAL ARCHAEOLOGY Territorial structures seem to be developing principally on the coast, where we find new villas (Gioia Tauro 28, Motta S. Giovanni 55) and less prestigious settlements, possibly involved with production (Campo Calabro 9 and Condofuri Marina 20). Inland occupation is substantially unchanged in form and shows signs of continuity, whereas reported new sites are cemeteries and uninterpretable contexts (Motta S. Giovanni 51 and 53), largely situated in the more southerly stretches of the mountain area. Roman domination appears to have given priority to selective restructuring over the development of territorial structures. The overall distribution of cemeteries, however, might suggest that particular aspects of the Greek landscape were being deliberately modified. As far as religious and cultural structures are concerned, for example, there seems to be evidence of this in the abandonment of sacred places at Saline (Phase 2, 74). The general picture for this phase is rather unclear because a reduction in the number of sites was expected, whereas in fact there is a slight increase in the context of a fairly highly structured territory. This reveals a situation of transition, with possible evidence of a nascent negative trend, which, taken together with the selective nature of settlement and the shifting of its centre of gravity, can be related to the already noted reconfiguration of the settlement patterns. Imperial Phase 4 (27BC-313 AD). A total of 87 finds were identified for Phase 4 (Fig. 2d), for which evidence exists in Messina and other sites along both sides of the Straits. The most significant of these (Capo Peloro 16) survives from the preceding phase. Two particular new sites appear: the coastal settlement at Messina-Pistunina 50, which can be identified as a mansio or vicus with agricultural and pastoral structures, and the area of Fiumedinisi 28, situated slightly inland in the middle of a mining area characterized by the presence of remains of late antique farms. The site at Ganzirri 87, where there exist traces of Roman age occupation, may also have come into being during this phase. In contrast, Calabria shows a slight decrease in the overall number of sites to 52. Twenty-one sites from the previous phase have disappeared, but the persistence of 29 sites indicates stable settlement. These longlasting sites, the most outstanding being the villas of Motta S. Giovanni 56 and Gioia Tauro 35, and the functional centre of Occhio di Pellaro 59, are mostly located along the coast. The persistence of inland occupation is deduced from uninterpretable contexts. The appearance of new data-points on the map is related to finds of a settlement type: either villae rusticae with associated cemeteries, mostly situated along the coast (Pellaro 68, Cannitello 15, Taureana 80 and Gioia Tauro 36), or areas typified by small rural settlements. Also, this phase suggests a reduction in inland occupation and an increase of coastal sites, with the exception of a sudden reduction of occupation between Reggio Calabria and Villa S. Giovanni. Late antique Phase 5 (AD313-535). A total of 46 finds were identified for Phase 5 (Fig. 2e), which is characterized by a marked reduction in the number of

65

sites. On the Sicilian side and apart from Messina there is evidence of the survival of the three coastal sites that had come into being in the previous phase (Fiumedinisi 13, Pistunina 27 and Ganzirri 46). In particular, the mansio or vicus of Pistunina enjoys its period of greatest expansion. Its functional importance within local dynamics probably led to the restoration of buildings destroyed by a catastrophic flood around AD 425-450. (Bacci Spigo 1993-1994). In contrast, signs of occupation in the hinterland and on the northern coast now disappear. A similar situation is seen in Calabria, where the sampling of sites is more reliable. Out of 25 sites, 21 show continuity of occupation with the previous phase whereas four are new. Among the new sites, the most significant are the cemetery at Palmi 33 and the kiln at Pellaro—Lume 35, which, however, shows a very short period of activity (until about the middle of the fourth century AD). Inland sites are found to have survived, whereas there is a substantial reduction in coastal settlement involving the whole coastline north of Reggio Calabria, as far as Palmi 33 and Taureana 41. At the same time, however, the basic settlement fabric of villae rusticae and functional centres survives along the southern coast, most of them spanning more than one phase (Melito di Porto Salvo 26, Motta S. Giovanni 30, Taureana 41, Gioia Tauro 21 and Occhio di Pellaro 32: the last in use from Phase 1 onwards). Sometimes, as in the case of Taureana 41, survival is the result of new functions acquired in the course of time: there, the Roman villa, which is associated in the Early Christian period with the cult of S. Fantino, becomes a basilica with cemetery, and is hence an example of how topographic continuity is maintained within a totally different cultural system. It is worth pointing out that there are no traces of earthquake activity in the literature at any of the sites recorded in this phase, not even when the record is reinterpreted, but for four sites (Occhio di Pellaro 32, Pellaro 35, Gioia Tauro 21 and Gallina 17) we do record evidence for abandonment or change of activity. This evidence leads us to hypothesize that an anomalous perturbation of the territorial structure may have occurred in the second half, or perhaps at the end, of the fourth century AD (the archaeological dating system does not allow better accuracy). Phase 6 (AD535 to the end of the sixth century). A total of 27 finds were identified for Phase 6 (Fig. 2f), during which we find significant changes in the dynamics of the territorial system. On the Sicilian side there is an increase in the number of sites in the mountainous area of Rometta Messinese (21, 22 and 23) and Monforte S. Giorgio (15 and 16), which dominates the northern coast and provided no finds at all in previous phases. There is evidence of continuing occupation in the usual sites spanning more than one phase (Pistunina 14, Fiumedinisi 8 and Ganzirri 27). There is a decline in occupation, however, in the agricultural and mining area of Fiumedinisi, as one can tell from the way the archaeological record changes from settlement finds to uninterpretable contexts. Pistunina is confirmed as a site with stable, if'poor', occupation by the contextual presence of settlement and cemetery

66


structures. There is indisputable evidence of stable structures at Ganzirri. By contrast, the Calabrian coast shows a strong contraction of the population: only seven sites are recorded, and 20 disappear from the previous phase. In the Straits area, all we find are a hoard (18) and a new fortification (19), both in the hinterland of Motta S. Giovanni. The ecclesiastical structures of Taureana 26 also survive.

The city of Messina From a historical point of view, the Zankle-Messana site is characterized by uninterrupted occupation from the eithth century BC until today, although no urban structures have survived. Topographical data for the Greek and Roman periods, therefore, have to be drawn from a variety of scattered archaeological evidence. As we already pointed out, archaeological research at Messina is hampered by the geological nature of the site, with its characteristic abundance of thick sandy and gravelly deposits associated with rapid and catastrophic alluviation phenomena. Little research was carried out in the nineteenth century. The 1908 earthquake was, in a way, a missed opportunity for collecting new data, but at least ample and good quality evidence has been collected in recent decades. Topographic pattern and evidence of its diachronic dynamics. The city of Messina is bounded by the Torrente Portalegni to the south and the Camaro to the north, and stands a few metres above sea level on a 1 km long (north to south) sloping platform having a maximum width of 500m (east to west). It is within this morphological unit that the most significant cemetery and settlement finds relating to the period of maximum urban development of Zankle (sixth to early fifth century BC) have been made. As no defensive structures from archaic times have survived, the size of the settlement has to be deduced from materials (mostly sporadic) found within the city fabric. The settlement occupied the alluvial plain to the south of the port and of the S. Ranieri peninsula. It is not possible to establish whether the settlement finds from Via S. Cecilia relate to urban or suburban dwellings; but there may be some evidence as to the southern limit of the archaic city in the huge cemetery along the S. Cosimo stream, 1300m to the south of Via S. Cecilia. Establishing the northern limit of the city is even more difficult. There are some fifth-century materials from isolato 327 in the modern town plan (the site of a Hellenistic cemetery), which may well belong to a funerary context. If so, the archaic city extended no further north than the Hellenistic and Roman cities. At the time of its maximum expansion, probably at the end of the sixth century BC, the site of Zankle covered a semicircular area facing the port and having a diameter of 1500m. In this phase (Phase 1) the city appears to have been made up of nearly east-west oriented rectangular blocks with the main streets parallel to the watercourses. Inhabited areas alternated with free areas, as excavation evidence clearly shows, and sanctuaries were arranged along the edges of the settlement. At this time the urban area probably covered about 80 ha.

In the mid-fifth century BC, the urban area was reduced by half in the east-west direction, as we deduced from the position in present-day Piazza Cairoli of the cemetery, which was to remain in use until late Hellenistic times. The expansion of the cemetery area is followed by the abandonment of a large part of the urban area so far considered, but there are also signs in the Hellenistic city (Phases 2 and 3) that the centre of gravity of the settlement shifted from the southern plain in a northerly and northwesterly direction. The city centre continues to gravitate around the port, though there is no evidence of buildings on the S. Ranieri peninsula. A substantial group of finds indicates that the heights of Montepiselli and the Tirone foothills, which face the southern plain, were inhabited from the end of the fourth or beginning of the third century BC. In late Hellenistic times, however, there is evidence (kilns) that the area between Viale S. Martino and the coast was used for production activities. The Hellenistic cemeteries together follow a sort of curve forming the city's southern boundary, thus allowing us to establish its size at the beginning of the third century BC; but we have no evidence to suggest how its size may have changed during the following period. There is evidence that in classical and Hellenistic times, the western edge of the inhabited area was near isolato 187 (where the Scuola Galatti is situated today) and 108 (Cinema Garden) near Via Martino. Subsurface surveys at the first of these sites have shown the presence of Hellenistic deposits below the level of the Roman inhabited area. At the second site, the modest remains of fourthcentury dwellings looking onto a broad open space suggest that this was the periphery of the ancient inhabited area, and that there was little urbanization. The city thus shrinks to inside the Torrente Portalegni to the south and the Torrente Bocetta to the north, lying squeezed between the sea and the hill slopes. Its area decreases appreciably, perhaps to as little as 40 ha. The reduction in size in Hellenistic times has been related to the Carthaginian destruction of the city in 396 BC; but the process of contraction had in fact already begun, as the Via Cairoli cemetery indicates, and must have occurred for reasons of a more economic nature and because of changes in the functional use of space. Remains of structures are very scanty, and in general equally scanty are finds relating to Imperial and late antique times (Phases 4 and 5), which seem to be concentrated between the Torrente Portalegni and the Teatro Vittorio Emanuele to its north and extend no further than the Torrente Bocetta cemetery. The structure of the Roman city must have been homogeneous, and seems to have been concentrated around the port, with a possible expansion westwards. Late antique (Phase 5) finds prove to be concentrated in the small coastal strip to the west of the port and indicate a further decrease in the size of the inhabited area (which shrinks to about 30 ha), whereas evidence for the Byzantine period (Phase 6) is scattered and insubstantial. During these two phases the number of mapped sites drops from eight to three. Although it is still punctuated by gaps, the archaeological description of urban Messina appears to be reasonably representative of the site's topographical dynamics (Vallet 1958; Scibona 1992, 1993). There is

AIMS AND METHODS IN TERRITORIAL ARCHAEOLOGY

67

Fig. 7. Population trends for the towns of Messina and Reggio Calabria.

therefore particular significance in the almost total lack of evidence for the Byzantine period, which, unless we postulate drastic changes in location strategies (although such changes have not been identified by archaeologists), implies that the contraction of the city must have been extreme. The port. We have little information about coastline changes along the S. Ranieri peninsula. Votive finds indicate sacred places that were frequented over a long period in areas that are still usable. Soundings in Piazza Cavallotti show that in Greek and Roman times the area was covered by a marshy beach without buildings. The rise of the coastline is thus a recent and entirely artificial phenomenon, which began in the Middle Ages, when earth was brought in to reclaim new land and subsequently to build on it (Bacci Spigo 1993-1994). Population: long-term estimates. We have made estimates of the resident population of ancient urban Messina using the criteria set out above. The results are shown in Fig. 7.

The city of Reggio Calabria Unlike Messina, the archaeological inheritance of Reggio Calabria was actively protected in the nineteenth century, and since the 1908 earthquake activity has been systematically and well documented. Recent activity by the Superintendency, however, was concentrated on a small number of excavations, for the museum sector has been given preferential treatment. The net result is an increase in the number of archaeo-

logical finds that are easy to situate topographically but often insufficiently informative for dating purposes. Topographic pattern and evidence of its diachronic dynamics. Thanks to the survival of significant fragments of its late classical defensive works, a reconstruction of the topography of Reggio Calabria can be of much greater help than is the case with Messina. Archaic and classical age (Phase 1) finds of a habitational and religious kind seem to be located mostly in the northern part of the city, near the Rada dei Giunchi, as well as in what may have been an area outside the walls (keeping in mind that the inhabited area in archaic times may have been 22-25 ha). Like the modern city, the Reggio Calabria of Hellenistic times (Phase 2) extended along the coast for about 1 km, and was bounded by the Torrente S. Lucia to the north and by the Torrente Calopinace to the south. If we accept Tropea Barbara's convincing reconstruction (1967), the late fifth-century BC city walls stretched about 750m up into the eastern heights bounding an area of 75 ha, although only about 30 ha of the coastal strip were settled and the rest was an open space. There is some debate, however, about the archaeological evidence (Vallet 1958; Turano 1966). The fact is that we have comparatively few settlement finds for Reggio Calabria, but a very large number of finds relating to religious practices and infrastructures, especially wells and cisterns for the local water supply. The conic section cisterns are a useful clue to the boundaries of the urban area, for they are arranged both within the street system and on the hill slopes around it. They belong to classical times and remained in use until the first century BC, when aqueducts were built. In Roman times (Phases 3 and 4) we notice an increase in settlement structures, i.e. houses and baths

68


(Orsi 1922). There is evidence for expansion beyond the walls both to the north and to the south, probably in the form of large suburban villas rather than as a continuous fabric, and the size of the urban area became about the same as that of the Hellenistic city (about 30 ha). We cannot say that there was any contraction of the settled area in late antique times (Phase 5), for there is homogeneous evidence of villas and baths right along the urban strip, but the appearance of some tombs within the city (15, 39) suggests that the settlement fabric had become less compact. With the arrival of the Byzantine age, the number of finds decreases sharply from 19 to nine. Unlike the case of Messina, evidence available in this phase (Phase 6) enables us to identify an area of denser settlement in Reggio Calabria. The close proximity of a building (61), some sacred places (67, 69 and 52) and some bronze inscriptions (32) suggests that this may have been a public area, perhaps with an archivum with funerary contexts around it. This pole of attraction shows topographical continuity with structures belonging to the late antique phase, and occupies an area of 8 ha. At the northern end of the city, near the Stazione Lido, there are some Byzantine structures of an artisan type (49: fish preparation plant and cemetery), which may indicate the existence of a new type of settlement, organized in separate but neighbouring nuclei. To sum up: the settlement dynamics at Reggio Calabria is different from that of Messina, for we find in the former an archaic town of moderate dimensions (25 ha) which spreads to 30 ha in accordance with an organized town plan. From Hellenistic times to the late antique period the settlement remains constant at least in area (30 ha), if not in terms of the homogeneity of the urban fabric. In the Byzantine phase, from the mid-fourth century onward, the settlement becomes much smaller and acquires a more scattered form. The port. There has been a lively debate over the identification of the Greek and Roman port (a synthesis and bibliography have been given by Tropea Barbaro (1967, pp. 66-83)). The present-day port, to the north of the mouth of the Torrente dell'Annunziata, is an artificial basin dating to 1873. Sediments carried by the Torrente delPAnnunziata cannot have had much effect on the morphology of the Rada dei Giunchi, though they must have produced a moderate advancement of the coastline. On the basis of Thucydides (VI 44, 2-3), many scholars have identified the Rada dei Giunchi as the place where the Athenians disembarked in 415BC. It must be stressed, however, that the present coastal morphology lacks an important point of reference, Punta Calamizzi, a small promontory located to the south of the city, which disappeared into the sea in 1562 (Tegani 1873). If it is true that Punta Calamizzi was cultivated, as local historians maintain, it presumably stretched some way out into the sea, forming a large sheltered inlet together with Rada dei Giunchi. Population: long-term estimates. We have made estimates of the resident population of ancient urban Reggio Calabria using the criteria set out above. The results are shown in Fig. 7.

References ARDOVINO, A. M. 1978, Edifici ellenistici e romani ed assetto territoriale a nord-ovest delle mura di Reggio. Klearchos, 20(77-80), 75-112. BACCI, M. G. & Rizzo, C. 1993-1994. Attivita della Soprintendenza: Taormina. Kokalos, 39-40, 945-951. BACCI SPIGO, M. 1993-1994. Attivita della sezione ai Beni Archeologici della Soprintendenza, B.C.A. di Messina negli anni 1989-1993. Kokalos, 39-40, 923-943. BARATTA, M. 1901.1 ten-emoti d'Italia. Saggio di storia geografia e bibliografia sismica italiana. Bocca Torino (anastatic reprint, Forni Sala Bolognese, 1979). 1936, / terremoti in Italia. R. Pubblicazioni della commissione italiana per lo studio delle grandi calamita, Vol. VI. Accademia Nazionale dei Lincei, Le Monnier, Florence. BEJOR, G. 1986. Gli insediamenti nella Sicilia romana: distribuzione, tipologie e sviluppo da un primo inventario dei dati archeologici. In: GIARDINA, A. (ed.) Societd romana e impero tardoantico, HI. Le merci e gli insediamenti. La terza, Bari, 463-519. BERNABO BREA, L. 1964-1965. Due secoli di studi, scavi e restauri del teatro greco di Tindari. Rivista dell'Istituto Nazionale di Archeologia e Storia dell'Arte, 13-14, 99-144. 1997. Note sul terremoto del 365 d.C. a Lipari e nella Sicilia nord orientale. In: GIARRIZZO, G. (ed.) La Sicilia dei terremoti. Lunga durata e dinamiche sociali. Maimore Catania, 87-97. BONANNO, C. 1993-1994. Scavi e ricerche a Caronia e a S. Marco d'Alunzio, Kokalos, 39-40, 953-986. BOSHI, E., GUIDOBONI, E., FERRARI, G., VALENSISE, G. & GASPERINI, P. (eds) 1995, 1997. Catalogo dei forti terremoti in Italia dal 461 a.C. al 1990. Istituto Nazionale di Geofisica and Storia Geofisica Ambiente (SGA) Bologna and Rome (also on CD-ROM and web site http://storing.ingrm.it/cft/ index.htm). BOSCHI, E., PANTOSTI, D. & VALENSISE, G. 1994. L'identificazione geologica delle faglie sismogenetiche. Le Scienze, 310, 36-46. BUTZER, K. W. 1982. Archaeology as Human Ecology. CUP, Cambridge. CAMBI, F. & TERRENATO, N. 1994. Introduzione all'archeologia dei paesaggi. Nuova Italia Scientifica, Rome. CAVALIER, M. 1994. Panarea. Bibliografia Topografica della Colonizzazione Greca in Italia e nelle hole Tirreniche, 13, 321-329. & BERNABO BREA, L. 1993-1994. Attivita della Soprintendenza: isole Eolie. Kokalos, 39-40, 987-1000. CHERRY, J. 1983. Frogs around the pond: perspectives on current archaeological survey in the Mediterranean region. In: KELLER, D. R. & RUPP, D. W. (eds) Archaeological Survey in the Mediterranean Region. British Archaeological Report (BAR) International Series, 155, 375-416. COARELLI, F. 1981. La Sicilia tra la fine della guerra annibalica e Cicerone. In: GIARDINA, A. &

AIMS AND METHODS IN TERRITORIAL ARCHAEOLOGY SCHIAVONE, A. (eds) Societd romana e produzione schiavistica I. L'Italia: insediamenti e forme economiche. Laterza, Bari, 2-18. FERRI, S. 1926. Gioiosa lonica (Marina). Teatro romano e rinvenimenti varii. Notizie degli Scavi di Antichitd, 1926, 332-338. FIORELLI, G. 18860. Reggio di Calabria. Nota del can., A. M. Di Lorenzo Vicedirettore del Museo di Reggio. Notizie degli Scavi di Antichitd, 1886, 59-64. 1886Z?. Reggio di Calabria. Note del vice direttore del Museo can., A. M. Di Lorenzo. Notizie degli Scavi di Antichitd, 1886, 436-441. 1888. Reggio di Calabria. Avanzi di edificio termale ed epigrafi onorarie latine scoperte in Reggio. Rapporto del vice-direttore del Museo civico can. A. Di Lorenzo. Notizie degli Scavi di Antichitd, 1888, 715-717. FOWLER, P. J. 1990. Site, landscape and context. In: FRANCOVICH, R. & MANACORDA, D. (eds) Lo scavo archeologico dalla diagnosi all'edizione, III ciclo di lezioni sulla ricerca applicata in archeologia (Certosa di Pontignano 1988). All'ln segna del Giglia, Florence, 121-131. GRECO, E. 1992. Archeologia della Magna Grecia. Laterza, Bari. GUIDOBONI, E. 1984. 3 Janvier 1117: le tremblement de terre du Moyen Age roman, aspects des sources. In: HELLY, B. & POLLING, A. (eds) Tremblements de terre histoire et archeologie, IVemes rencontres Internationales d'archeologie et d'histoire d'Antibes. Actes du colloque, 2-4 novembre 1983, Association pour la Promotion et la Diffusion des Connoissance Archeologiques Valbonne, 119-139. & BOSCHI, E. 1989. I grandi terremoti medievali in Italia. Le Scienze, 249, 22-35. & TRAINA, G. 1996. Earthquakes in medieval Sicily. A historical revision (VII-XIII century). Annali di Geofisica, 39, 1201-1225. , COMASTRI, A. & TRAINA, G. 1994. Catalogue of ancient earthquakes in the Mediterranean area up to the 10th century. Istituto Nazionale di Geofisica and Storia Geofisica Ambiente (SGA), Bologna. Guzzo, P. G. 1981. II territorio dei Brutii. In: GIARDINA, A. & SCHIAVONE, A. (eds) Societd romana e produzione schiavistica. I. L'Italia: insediamenti e forme economiche. Academic Press, Bari, 115-136. HASSAN, F. A. 1981. Demographic Archaeology. New York. LATTANZI, E. 1982. Attivita archeologica della Soprintendenza Archeologica della Calabria. In: Megale Hellas. Nome e immagine. Atti del XXI convegno di studi sulla Magna Grecia, Toronto 1981, Istituto per la Storia e 1'archeologia della Magna Grecia, Taranto, 217-236. 1986. L'attivita archeologica in Calabria nel 1985. In: Neapolis, Atti del XXV convegno di studi sulla Magna Grecia, Taranto 1985, 417-430. LEONARDI, G. 19920. Assunzione e analisi dei dati territoriali in funzione della valutazione della diacronia e delle modalita del popolamento. In: BERNARDI, M. (ed.) Archeologia del paesaggio,

69

IV ciclo di lezioni sulla ricerca applicata in archeologia (Certosa di Pontignano 1992). All'ln segna del Griglio, Florence, 25-66. \992b. II deposito archeologico: bacini, processi formativi e trasformativi. In: LEONARDI, G. (ed.) Processi formativi della stratificazione archeologica, Atti del seminario internazionale. 'Formation processes and excavation methods in archaeology: perspectives', Departmento di scienze dell'antictifa Universita degli studi, Padua, 13-47. & BALISTA, C. 1992. Linee di approccio al deposito archeologico. In: LEONARDI, G. (ed.) Processi formativi della stratificazione archeologica. Atti del seminario internazionale 'Formation processes and excavation methods in archaeology: perspectives', Departmento di scienze dell'antictifa Universita degli studi, Padua, 75-99. MUGGIA, A. 1997. L'area di rispetto nelle colonie magno-greche e siceliote. Studio di antropologia della forma urbana. Sellerio, Palermo. ORSI, P. 1899. Buscemi. Sacri spechi con iscrizioni greche, scoperti presso Akrai. Notizie degli Scavi di Antichitd, 24,452-471. \9l6a. Messana. La necropoli romana di S. Placido e di altre scoperte avvenute nel 19101915. Monumenti Antichi pubblicati per cura della Reale Accademia dei Lincei, 24, 121-218. 19166. Nocera Tirinese. Ricerche al Piano della Tirena sede dell'antica Nuceria. Notizie degli Scavi di Antichitd, 1916, 335-362. 1922. Reggio Calabria. Scoperte negli anni dal 1911 al 1921. Notizie degli Scavi di Antichitd, 1922, 151-186. PANTOSTI, D., SCHWARTZ, D. P. & VALENSISE, G. 1993. Paleoseismology along the 1980 Irpinia earthquake fault and implications for earthquake recurrence in the southern Apennines. Journal of Geophysical Research, 98, 6561-6577. PUTORTI, N. 1912. Di un titolo termale scoperto in Reggio Calabria. Rendiconti della Reale Accademia dei Lincei. Classe di scienze morali, storiche e filologiche, serie V, 21, 791-802. SCHIFFER, M. B. 1987. Formation Processes of the Archaeological Record, University of Utah Press, Albuquerque. SCHWARTZ, D. P. & COPPERSMITH, K. J. 1984. Fault behavior and characteristic earthquakes: examples from the Wasatch and San Andreas fault zones, Journal of Geophysical Research, 89, 5681-5698. SCIBONA, M. 1984-1985. Messina: notizia preliminare sulla necropoli romana e sul giacimento preistorico del torrente Boccetta. Kokalos, 30-31, 855-861. 1992. Messina. Bibliografia Topografica della Colonizzazione Greco in Italia e nelle Isole Tineniche, 10, 16-36. 1993. Punti fermi e problemi di topografia antica a Messina: 1966-1986. In: Lo stretto crocevia di culture, Atti del XXVI convegno di studi sulla Magna Grecia, Taranto 1986, 433-458. SPADEA, R. 1993. Le citta dello stretto e il loro territorio. Reggio Calabria. In: Lo stretto crocevia di culture, Atti del XXVI Convegno di Studi sulla Magna Grecia, Taranto 1986, 459-474.

70


SPIGO, U. 1993-1994. Capo d'Orlando: il complesso termale di eta imperiale romana di Bagnoli S. Gregorio. Scavi 1987-1992. Kokalos, 39-40, 1027-1037. STORIA GEOFISICA AMBIENTE (SGA) 1996. Territorial archaeology in the Straits of Messina area between 500 BC and AD 500. Bologna, Rome, Report 158/ 1996. [In Italian.] TEGANI, A. (ed.) 1873. Cronaca del cantore Antonio Tegani [e dei suoi continuatori (1480-1625)], a cura di, A. M. De Lorenzo. Memorie da servire alia storia sacra e civile di Reggio e delle Calabrie, vol. I (parte I fasc. 3). Tipografia Siclari Reggio Calabria, 9-60. TROPEA BARBARO, E. 1967. II muro di cinta occidentale e la topografia di Reggio ellenica, Klearchos, 9, 7-130. TURANO, C. 1966. Carta archeologica di Reggio Calabria del XIX secolo. Klearchos, 8(29-32), 159-180. VALENSISE, G. & GUIDOBONI, E. 1995. Verso nuove strategic di ricerca: zone sismogenetiche silenti o silenzio delle fonti? In: BOSCHI, E., FERRARI, G., GASPERINI, P., GUIDOBONI, E., SMRIGLIO, G. & VALENSISE, G. (eds) Catalogo deiforti terremoti in Italia dal 461 a. C. al 1980. Istituto Nazionale di Geofisica and Storia Geofisica Ambiente (SGA), Bologna and Rome, 112-127. & PANTOSTI, D. 1992. A 125 Kyr long geological record of seismic source repeatability: the Messina Straits (southern Italy) and the 1908 earthquake (Ms 7±). Terra Nova, 4, 472-483. & 2000. Seismogenic faulting, moment release patterns and seismic hazard along the central and southern Apennines and the Calabrian Arc. In: VAI, G. B. & MARTINI, I. P. (eds) Anatomy of a Mountain Chain: the Apennines and Adjacent Mediterranean Basins. Kluwer, Dordrecht.

VALLET, G. 1958. Rhegion et Zancle. Histoire, commerce et civilisation des cites chalcidiennes du detroit de Messine. De Boccard, Paris. VON FALKENHAUSEN, V. 1982.1 Bizantini in Italia. In: PUGLIESE CARRATELLI, G. (ed.) / Bizantini in Italia. Scheiwiller, Milan, 1-136. VOZA, G. 1980-1981. L'attivita della Soprintendenza alle Antichita della Sicilia Orientale. Parte I. Kokalos, 26-27, 674-693. 1984-1985. Attivita nel territorio della Soprintendenza alle Antichita di Siracusa nel quadriennio 1980-1984. Kokalos, 30-31, 657-678. WILSON, R. J. A. 1990. Sicily under the Roman empire. The archaeology of a Roman province. 36BC-AD535. Aris & Phillips, Warminster. WOOD, W. R. & JOHNSON, D. L. 1978. A survey of disturbance processes in archaeological site formation. In: SCHIFFER, M. B. (ed.) Advances in Archaeological Method and Theory I. Academic Press, New York, 315-381.

Archaeological journals subjected to systematic scrutiny Archivio Storico Messinese (1969-1994) Archivio Storico per la Calabria e la Lucania (1931— 1994) Archivio Storico per la Sicilia Orientale (1904-1994) Archivio Storico Siciliano, IV serie (1975-1992) Atti dei Convegni di Studio sulla Magna Grecia (Taranto, 1960-1992) Bibliografia Topografica della Colonizzazione Greca in Italia e nelle hole Tirreniche (1977-1995) Klearchos (1959-1984) Kokalos (1955-1994) Notizie degli Scavi di Antichita (1876-1960) Rivista Storica Calabrese (1893-1994) Sicilia Archeologica (1968-1994)

Santorini (Greece) before the Minoan eruption: a reconstruction of the ring-island, natural resources and clay deposits from the Akrotiri excavation WALTER L. FRIEDRICH, MARIT-SOLVEIG SEIDENKRANTZ & OLE BJ0RSLEV NIELSEN

Department of Earth Sciences, University of Aarhus, DK-8000 Arhus C, Denmark Abstract: Before the catastrophic eruption around 1640BC, Thera, Therasia and Aspronisi formed a ring-shaped island with a sea-flooded caldera in the middle. The so-called PreKameni Island was situated in the centre of the caldera. This reconstruction is based on the study of stromatolites found in eruption products as well as other geological observations. The location of pre-eruption settlements or sites on the present rim of the Santorini caldera seems to support this reconstruction. Many of the rocks and minerals used in the Bronze Age culture are of local origin. Foraminiferal and mineralogical studies enable us to trace the source areas of a clay deposit found in a grave chamber in the Akrotiri excavation. This clay can be used for pottery making. The foraminiferal and mineralogical studies also help identify the natural drainage system and thus the freshwater supply, which may have been an important factor deciding the location of the Bronze Age settlement.

The volcanic island of Santorini (36.40°N, 25.40°E) (Fig. 1) in the Aegean Sea is of considerable interest for both geology and archaeology, as a flourishing Late Bronze Age settlement was buried under the volcanic products of the Minoan eruption. According to a new ice-core dating (Clausen et al. 1997), this eruption, which caused the end of the Minoan settlement on Santorini, occurred around 1640BC. A large buried settlement, located on the Akrotiri Peninsula (Fig. 2), is currently under excavation. The source areas of some of the natural resources (especially clay and water supply) from this excavation will be discussed in this paper. The geography of Santorini before the Minoan eruption

Geological evidence Today Santorini consists of the three older islands (Thera, Therasia and Aspronisi), as well as the two young Kamenis (Palea and Nea Kameni) (Fig. 2). The formation of the Kameni Islands started at 192BC and has continued until modern times. Geological evidence shows, however, that Santorini formed a ring-shaped island

before the Minoan eruption (Fig. 3). Especially significant for this reconstruction were the findings of stromatolitic blocks in the third layer of the products from the Minoan eruption (Fig. 3). Stromatolites are calcareous, globular structures, which were formed by the interactions of algae and bacteria in shallow-marine conditions. They were radiocarbon dated to about 13000 years BC, and their distribution in the northern part of Thera and Therasia made it possible to trace them to their growth area (Friedrich et al 1988; Eriksen et al 1990; Friedrich 1994). Together with other geological observations, this helped answer one of the questions that geologists and archaeologists have asked themselves for a long period: what did Santorini look like before the eruption? It was concluded that the central part of the caldera that existed before the Minoan eruption contained an island, the so-called Pre-Kameni Island (Friedrich et al 1988) (Fig. 3). The location of this island was deduced from the location of the vent of the Minoan eruption and the fact that the first phase of the eruption (the Plinian phase) was not influenced by contact with water. This reconstruction has been confirmed by additional geological observations by Druitt & Francaviglia (1990 1992), who found pumice products of the Minoan eruption plastered on


72

W. L. FRIEDRICH, M.-S. SEIDENKRANTZ & O. B. NIELSEN

Fig. 1. Location of Santorini (marked in black) in the Aegean Sea.

the inner side of the caldera wall underneath the town of Fira (Fig. 2) on Thera and at other localities. Archaeological observations The new idea of the shape of Santorini before the Minoan eruption has also had an influence on the concept of archaeological thinking, especially the question of whether there was access to local occurrence of minerals and ores. Furthermore, the distribution of settlements or sites on the present-day shape of the island could be explained much better using the new reconstruction (Friedrich & Doumas 1990).

tracing natural resources to local areas on Santorini are the following: (1) the talc used to produce white decorations on pottery (Aloupi & Maniatis 1990) occurs at Plaka (Fig. 3) (Friedrich & Doumas 1990; Friedrich 1994); (2) reddish marble used in the production of a stone vase found in the Akrotiri excavation occurs at Echendra, close to Cape Exomiti on Thera (Fig. 3); (3) pigments used in the production of wall paintings are still under investigation, but the sources for the pigments might be the phyllites that occur between Thermia and Plaka (Fig. 3). These localities were presumably accessible by boat before the Minoan eruption.

Clay from the Akrotiri excavation Source areas of materials found in the Akrotiri excavation In some cases, rocks used for the building of houses, as well as minerals, and other natural resources in the Akrotiri excavation could be traced back to their source areas (Fig. 3). This investigation was carried out mainly by Einfalt (19780,&) and supplemented by further observations by Friedrich (1994). Three examples of

During the construction of a Dextion roof in the Akrotiri excavation (Figs 3 and 4), the archaeologists dug a hole, where they placed a pillar for the roof construction. They eventually found a clay deposit on the site of Pillar 17 (Marinatos 1976) (Fig 4). This specific clay has been used for making modern pottery, as demonstrated in 1989 at the opening ceremony at the Third Conference on Thera and the Aegean World', when the leader of the Akrotiri excavation, Professor

SANTORINI BEFORE THE MINOAN ERUPTION

73

Fig. 2. Present geography of the Santorini island group, showing the whole volcanic edifice, including the Kolumbo volcano to the NE.

Christos Doumas, gave a piece of pottery to the organizer, Peter Nomikos. The clay deposit at Pillar 17 is found in the immediate surroundings of the pillar, where it has a thickness of 0.3-2.5m. The clay was also found in a small cave dug out under an ignimbrite (hard, volcanic rock) by the inhabitants of the settlement (Fig. 5). They closed the cave by a stone wall, and archaeological evidence indicates that the cave had the function of a grave chamber (Doumas 1995). Further excavations (Doumas 1996) revealed that the grave was from the Early Cycladic period and it contained seven

marble idols, stone settings and bones. It was not, however, possible to determine whether the bones were of human or animal origin. The clay is found behind the stone wall inside the grave chamber. The stratification and increasing thickness of the clay deposit towards this wall lead to the conclusion that the clay filled the cave at a later time through an opening in the wall. The cause of the infill could be either natural processes, e.g. the influence of wind and water, or human activity. The layering of the sediment, however, suggests that this deposit was formed naturally. This does not conflict with

74


Fig. 3. Geography of Santorini before the Minoan eruption. The reconstruction is based on geological and archaeological evidence (after Friedrich 1994). Black squares mark the location of Bronze Age archaeological sites. Circles of different size show the present-day occurrences of ejected stromatolitic blocks (the larger the circle, the more common the stromatolitic blocks). White dots north of the Pre-Kamenia Island represent the assumed place of growth for stromatolites before the Minoan eruption. Possible sources of rocks and minerals found in the Akrotiri excavation are shown in this reconstructed shape of the Bronze Age island. Some of the information on the resources is from Einfalt (1978&). The interval between the isohypses is 100m.

the observation of the archaeologists (Doumas 1995), who found that the immediate surroundings of the area had been spared in later building, indicating that the Therans knew of the

existence of their ancestor's grave and respected the area as a holy site. The clay samples, which we analysed, contained a piece of charcoal, which has been

SANTORINI BEFORE THE MINOAN ERUPTION

75

Fig. 4. Plan of the Akrotiri excavation showing the location of Pillar 17-

radiocarbon dated using the accelerator mass spectrometry (AMS) method at the AMS Laboratory, University of Aarhus (sample no. AAR-1566). The sample gave an age of 4050 ± 6014C years BP, corresponding to 2570-2510BC

in calendar years (calibrated after Stuiver & Renner (1993)) (see also the discussion below). Geological setting around Pillar 77. Two of us (O.B.N. and W.L.F.) examined the site in 1993.

76


Fig. 5. Sketch of the grave chamber with the clay at Pillar 17.

According to our observations, the roof of the grave chamber is formed by the so-called Cape Riva Ignimbrite, which can be traced from the caldera wall (Friedrich et al. 1977) to the excavation by observations in the field, i.e. in the Potamos Valley. The situation in the Akrotiri excavation is similar to that in the Potamos Valley, where the ignimbrite sheet ends abruptly parallel to the present-day coastline, as a result of coastal and fluviatile erosion. Furthermore, this ignimbrite crops out directly on the south coast, roughly halfway between the Akrotiri excavation and Cape Exomiti (Figs 3 and 6a, b). The source area of the clay. Clay samples taken from inside the grave chamber at Pillar 17 (Fig. 4) in the Akrotiri excavation have been identified as an erosional product of local sediments from the Archangelos-Loumaravi Complex (Figs 3 and 6a, b). The study was based on its content of foraminiferal and volcanic components. The clay sample contains a poor foraminiferal assemblage of shallow- and deep-water species (Fig. 6a), indicating that the clay deposit at Akrotiri is derived from more than one source area. Marine Plio-Pleistocene sediments from the Archangelos-Loumaravi Complex on the

Akrotiri Peninsula have previously been studied by Seidenkrantz & Friedrich (1993). The Foraminifera (Figs 6a and 7) revealed that both shallow- and deep-water marine sediments occur in this particular area. The sediments of the Archangelos-Loumaravi Complex are a mixture of clay, marl and pumice, and were deposited before or during the formation of a volcanic dome. The highest clay and marl content is found in the deeper water (upper epibathyal) deposits at Cape Loumaravi (Fig. 3), whereas the shallower (littoral to inner neritic) deposits at Mt Archangelos and Mt Loumaravi (Fig. 3) have a larger content of pumice. The foraminiferal assemblage from the Pillar 17 clay shows the closest affinity to the assemblages found at Cape Loumaravi (see Seidenkrantz & Friedrich 1993) (Figs 6a and 7) but with an element of the shallow-water or epiphytic species that dominate the assemblage from Mt Archangelos. We can thus partly confirm the idea of Fouque (1879), who investigated prehistoric pottery shards from the Akrotiri area that contain Foraminifera and other marine organisms. He concluded that the clay was of local origin, possibly derived from a shallowwater site between Aspronisi and the Akrotiri Peninsula, which might have been accessible before the Minoan eruption. Vaughan (1990), who studied Early Cycladic wares from the Akrotiri excavation, concluded that part of the material was of local origin. The mineralogical investigations, especially the contents of smectite, cristobalite and zeolites of the clinoptilolite-heulandite series, also indicate that the clay is an erosional product, derived from a volcanic source area. The compositions of the sediments around Pillar 17 and in the cave (Fig. 6b) are rather similar to the marine sediments of Akrotiri Village, which outcrop at the church of Agios Epiphanios. They differ from the deposits below the ignimbrite in the Potamos Valley, in which smectitic clay minerals are almost absent (Fig. 6b). We thus conclude that the Pillar 17 clay is an erosional product of the marine sediments from the Archangelos-Loumaravi updomed area on the Akrotiri Peninsula. The clay was washed out and deposited in depressions, such as observed at Pillar 17. Before the Minoan eruption, the area of outcropping marine Plio-Pleistocene sediments must have been larger, as today major parts of the Akrotiri Peninsula are still covered by products of the eruption. The erosional clay products from these sediments were presumably accessible to the prehistoric population at several localities. The clay of Pillar 17 thus gives us an

Fig. 6. The Akrotiri peninsula. Source area of the clay from Pillar 17. The extent of the Cape Riva Ignimbrite and the freshwater drainage system during the Bronze Age is marked as a white hatched area. Ak, Akrotiri Village; Pot, Potamos Valley. (A) Foraminiferal data. The Foraminifera are divided into categories according in part to wall structure (porcellaneous, agglutinated, hyaline; all benthic habitat) and in part according to ecological requirements (deep- and shallow-water benthic species (only for the hyaline forms), and planktonic species), de, deep-water hyaline, benthic Foraminifera; sh, shallow-water and epiphytic hyaline, benthic Foraminifera; po, porcellaneous benthic Foraminifera; ag, agglutinated benthic Foraminifera; ot, other benthic Foraminifera; pi, planktonic Foraminifera. (B) Mineralogical data. Ca, carbonates; Cl, clinoptilolite; Cr, cristobalite; Fe, feldspar; Qu, quartz; It, illite; Ka, kaolinite; Sm, smectite.

78


Fig. 7. Some typical Foraminifera. Foraminifera are marine organisms that commonly have a size between 0.1 and 1.0mm. Some forms live on the sea floor (benthic); other species live in the upper water masses (planktonic). The illustration shows scanning electron microscope photographs of benthic (upper row) and planktonic (lower row) forms from the Plio-Pleistocene sediments of the Akrotiri Peninsula.

idea of how clay for pottery making could have that, at the time of their investigations, ruins in been obtained in prehistoric times. the Akrotiri valley were visible in several places. There are several possibilities as to how the This means that erosion had already removed prehistoric Therans could have obtained their much of the covering pumice material. The pottery clay. One or several natural outcrops of general erosion by wind and water may have the clay might have existed in the hills around the been intensified by tsunami activity, which has Akrotiri settlement, or the clay may have been hit the Akrotiri area at least twice since the collected in the lowlands from human-made cav- Minoan eruption: first, the tsunami generated ities, such as freshwater cisterns. Cisterns made from the collapse of the roof of the Minoan for animals in rural district act today as clay magma chamber, producing the 'fourth phase' traps and have to be cleaned from time to time. deposits; and, second, the tsunami generated in Our experiments have shown that the bottom connection with the eruption in AD 1650 (Ross sediment from a cistern for animal-use at Plaka 1840; Friedrich 1994) of the Kolumbo volcano, (Fig. 2) resulted in a good pottery clay. located in the northeastern part of the Santorini volcanic edifice (Fig. 2). Contemporaneous eyeErosion (wind, water and tsunamis). The clay witness reports state that water entered, among at Pillar 17 itself may have been deposited when others, the Akrotiri region (Ross 1840). Tsunami the settlement was still inhabited. However, it sediments of this event are also seen in the may also have been deposited after abandon- Potamos Valley, where rounded pebbles with ment of the settlement because of the Minoan marine, calcareous worm tubes Spirorbis were eruption, as erosion has cut its way through the found (Friedrich 1994). During this long period covering pumice deposits and reached the ruins of erosion, water carrying clay material might of the settlement. Reports of the first discovery have reached cavities in the buried settlement of the Akrotiri site by Mamet & Gorceix and deposited the clay. The dating of the above(18700,6) as well as Fouque (1879) mentioned mentioned charcoal found at Pillar 17 (sample

SANTORINI BEFORE THE MINOAN ERUPTION no. AAR-1566) does not necessarily give the age of the sedimentation process, as the charcoal could have been redeposited. The freshwater drainage system near the Akrotiri settlement before the Minoan eruption The south-facing bay with a shallow beach, where boats could be pulled ashore, was an excellent choice for a settlement for the Bronze Age population. The site also took advantage of the red Cape Riva Ignimbrite. This ignimbrite has been radiocarbon dated to approximately 18000 years BP on the basis of charcoal found underneath and within this rock formation (Eriksen et al 1990; Friedrich 1994). The red ignimbrite was deposited in a preexisting erosional valley. It can be traced from the caldera rim to the Akrotiri excavation (Fig. 6a and b) following the outer down-slope of the volcanic edifice with an angle of c. 8° to the south. Before the Minoan eruption, the houses were built both on top of the ignimbrite and below its distal erosional edge. The grave at Pillar 17 was located at this very edge, dug into the soft alluvial material underneath the ignimbrite. The valley in which the ignimbrite is deposited is now the main route for the flow of rainwater, as the ignimbrite prevents the water from penetrating into the ground. The valley was also the route for transport of rainwater prior to the Minoan eruption; it was part of the main drainage system of the Archangelos-Loumaravi Complex, transporting the erosional products, including the clay, from the hills to the Akrotiri settlement (Fig. 6a and b). The Cape Riva Ignimbrite thus had a very significant influence on the course of the drainage system and thus the freshwater supply. It was presumably, together with the natural harbour, among the important factors deciding the location of the Bronze Age settlement. We would like to thank C. G. Doumas, the leader of the Akrotiri excavation, University of Athens, for the kind permission to take samples in the Akrotiri excavation, and M. Arvanitis, Santorini, for his help in the field. We would also like to express our gratitude to B. Hallager. H. Sigala and A. Frang for their help in literature search, and to S. M. Christiansen, J. G. Nielsen, J. G. Petersen, U. Bjerring, T. K. Rasmussen, B. Winsl0v, M. Dybdahl and U. Viskum for technical assistance. The radiocarbon dating performed by J. Heinemeier and N. Rud is gratefully acknowledged. R. Wilson kindly helped us improve the English of the text. The Carlsberg Foundation (WLF) and the Danish Natural Science Foundation (MSS) funded this study.

79

References ALOUPI, E. & MANIATIS, Y. 1990. Investigation of the technology of manufacture of the local LBA Theran pottery: the body and pigment analysis. In: HARDY, D. A. (ed.) Thera and the Aegean World III, Vol. 2. Thera Foundation, London, 459-469. CLAUSEN, H. B., HAMMER, C. U., HVIDBERG, C. S., DAHL-JENSEN, D. & STEFFENSEN, J. P. 1997. A comparison of the volcanic records over the past 4000 years from the GRIP and DYE 3 Greenland ice cores. Journal of Geophysical Research, 102(C12), 25707-26723. DOUMAS, C. G. 1995. Excavation on Thera (1992). Praktika Archaiologikis Etaireias, 176-188 [in Greek]. 1996. Excavation on Thera (1993). Praktika Archaiologikis Etaireias, 176-178 [in Greek]. DRUITT, T. H. & FRANCAVIGLIA, V. 1990. An ancient caldera cliff line at Phira, and its significance for the topography and geology of Pre-Minoan Santorini. In: HARDY, D. A. (ed.) Thera and the Aegean World HI, Vol. 2. Thera Foundation, London, 362-369. & 1992. Caldera formation on Santorini and the physiography of the islands in the late Bronze Age. Bulletin of Volcanology, 54, 484-493. EINFALT, H.-C. 19780. Chemical and mineralogical investigations of sherds from the Akrotiri excavations. In: DOUMAS, C. (ed.) Thera and the Aegean World I. Thera Foundation, London, 459-469. 1978&. Stone materials in ancient Akrotiri a short compilation. In: DOUMAS, C. (ed.) Thera and the Aegean World I. Thera Foundation, London, 529-527. ERIKSEN, U., FRIEDRICH, W. L., BUCHARDT, B., TAUBER, H. & THOMSEN, M. S. 1990. The Stronghyle Caldera: geological, palaeontological and stable isotope evidence from radiocarbon dated stromatolites from Santorini. In: HARDY, D. A. (ed.) Thera and the Aegean World III, Vol. 2. Thera Foundation, London, 139-150. FOUQUE, F. 1879. Santorin et ses Eruptions. Masson, Paris. FRIEDRICH, W. L. 1994. Feuer im Meer - Vulkanismus und die Naturgeschichte der Insel Santorin. Spektrum, Heidelberg. & DOUMAS, C. G. 1990. Was there local access to certain ores/minerals for the Theran people before the Minoan eruption? An addendum. In: HARDY, D. A. (ed.) Thera and the Aegean World III, Vol 1. Thera Foundation, London, 502-503. , ERIKSEN, U., TAUBER, H., HEINEMEIER, J., RUD, N., THOMSEN, M. S. & BUCHARDT, B. 1988. Existence of a water-filled caldera prior to the Minoan eruption of Santorini, Greece. Naturwissenschaften, 75, 567-569. , PICHLER, H. & KUSSMAUL, S. 1977. Quaternary pyroclastics from Santorini, Greece and their significance for the Mediterranean palaeoclimate. Bulletin of the Geological Society of Denmark, 26, 27-39.

80


MAMET, H. & GORCEIX, C. 18700. Archeologie et geologic. Recherches et fouilles. Bulletin de I'Ecole Francaise d'Athenes, 9, 183-191. & 1870&. Archeologie et geologic. Fouilles a Santorin. Bulletin de I'Ecole Francaise d'Athenes, 10, 199-203. MARINATOS, S. 1976. Excavations at Thera VII. Bibliotheke tes en Athenais archaiologikes hetaireios, Athens Ross, L. 1840. Reisen auf den griechischen Inseln des dgdischen Meeres. J. G. Cotta, Stuttgart und Tubingen.

SEIDENKRANTZ, M. S. & FRIEDRICH, W. L. 1993. Santorini, part of the Hellenic Arc: age relationship of its earliest volcanism. Bulletin of the Geological Society of Greece, 28(3), 99-115. STUIVER, M. & RENNER, P. J. 1993. Extended 14C data base and revised CALIB 3.014C age calibration. Radiocarbon, 35, 215-230. VAUGHAN, S. J. 1990. Petrographic analysis of the Early Cycladic wares from Akrotiri, Thera. In: HARDY, D. A. (ed.) Thera and the Aegean World HI, Vol. 1. Thera Foundation, London, 470-487.

The eruption of the Santorini volcano and its effects on Minoan Crete JAN DRIESSEN1 & COLIN F. MACDONALD2 1

Department d'archeologie, Universite Catholique de Louvain, B-1348-Louvain-la-Neuve, Belgium (e-mail: [email protected]) 2 British School at Athens, 52, Souedias Str., Athens 106 76 Abstract: Sometime in the course of the second millennium BC, an earthquake appears to have triggered a massive eruption of the Santorini volcano. The immediate consequences of the earthquake closely followed by the eruption for Cretan society during the Late Minoan I period are rather difficult to characterize, although physical evidence in the form of Theran ash has shown up at an increasing number of sites. Certain features of the archaeological record, taken in isolation, have hardly been noticed in the past. The long-term effects of the eruption, however, have recently become more comprehensible thanks to a reconsideration of old and new archaeological evidence. The combined picture gives the impression of a period of societal stress following these events. Changes in architecture, storage and food production, artisan output, the distribution of prestige items, administrative patterns and ritual manifestations can be pinpointed archaeologically. These may and should be interpreted as disturbances in the political, economic, cult and security-related domains. It is argued that the inability of the Minoan palatial centres to adapt to changing circumstances caused by a double disaster, an earthquake followed by the eruption of Santorini, led to an increase in crisis-related situations, culminating in the widespread fire destructions which brought this palatial phase of Minoan civilization to an end and opened the way for mainland Mycenaean domination of the Aegean.

Iconographic or literary evidence for natural catastrophes during prehistory is rather scarce. A notable exception is the eruption of the Hasan Dag on the east fringe of the Konya plain, depicted on a seventh-millennium BC mural of £atal Huyiik. Even in historical times, catastrophes seldom seem directly to have influenced artistic creation; another exception is provided by some relief sculpture from a house at Pompeii. These show the effects of the AD 62 earthquake, the traces of which were still very visible in the cities destroyed by the eruption of Vesuvius 17 years later; the fatal mountain itself rarely inspired Roman artists. Sometime during the beginning of the Late Bronze Age, the Aegean witnessed one of the largest eruptions in the history of mankind when the volcanic island of Santorini erupted (see Driessen & Macdonald (1997) for references and details). This caused, amongst other things, the disappearance of the flourishing Cycladic town of Akrotiri beneath metres of ash (Fig. 2) We are not entirely sure of the absolute date of this eruption, as both the second half of the 17th century BC and the middle of the 16th century BC have been suggested. Recent dendrochronological evidence from Porsuk in Turkey, for instance, has been used by the supporters of the high date

(Kuniholm et al. 1996), whereas pumice found in the Nile Delta would rather be suggestive of a 16th-century date (Bietak 1996). The discussion on the absolute date of this eruption is still not closed and new evidence is presented regularly; hence we will probably hear more about the Nisyros eruption, closer to the Turkish coast, which is claimed to have occurred in more or less the same period (Liritzis et al. 1996), and during the London conference the Avellino eruption of Vesuvius was also dated to the 18th or 17th century BC. Moreover, the 1996 campaign at Palaikastro, a Minoan site in East Crete, came upon an extensive, 15 cm thick silt layer suggestive of a water event in an archaeological period dating to the end of the Middle Bronze Age, which would comfortably agree with the high date, suggested by dendrochronology, but is clearly earlier than the Santorini eruption (MacGillivray et al. 1998). At present, there appears to be a stalemate or rather an impasse with regard to the absolute date, and people agree to disagree. In the present paper we are concerned only with the relative date of this eruption and with its archaeological effects. Here we are on firmer ground. The settlement of Akrotiri was clearly destroyed when so-called Late Minoan (LM) IA pottery was in use, a Cretan pottery

From\ McGuiRE, W. G., GRIFFITHS, D. R., HANCOCK, P. L. & STEWART, I. S. (eds) The Archaeology of Geological Catastrophes. Geological Society, London, Special Publications, 171, 81-93. 1-86239-062-2/00/S15.00 © The Geological Society of London 2000.

Fig. 1. Map of Crete with sites mentioned in text.

SANTORINI VOLCANO AND MINOAN CRETE

Fig. 2. Minoanized Cycladic settlement of Akrotiri, Thera.

style characterized by a popularity of floral motifs. The style of the imported pottery found at Akrotiri indicates that this event may have happened rather late in the LM IA period. For many years, this eruption was also blamed for a broad horizon of destructions by fire on Crete (Fig. 1) and for the demise of Minoan palatial civilization. In this scenario, the combined effect of earthquakes, volcanic ash and tsunamis on the island should have caused the downfall of the Cretan palace societies and turned them into easy prey for marauding Mycenaeans from the Greek mainland. On Crete, however, the pottery found in the destruction layers is distinctly later than that found at Akrotiri, and is of a style called Late Minoan IB, the main characteristic of which is the preponderance of marine elements such as octopuses, starfish and argonauts. This implies that the annihilation of Akrotiri on Santorini and the destructions on Crete differed by at least a generation or two. Although some researchers have tried to harmonize the chronology of these two events, excavations of the last decade on Crete and the Dodecanese now make this an untenable sce-

83

nario. The sequence has been clarified and can fall in line with the Akrotiri evidence. First, a serious earthquake occurred sometime before the eruption, as there are signs of earthquake damage followed by clearing and rebuilding at Akrotiri. Then followed the eruption with massive ash fall and the destruction of the Akrotiri settlement. On the island of Rhodes, extensive Theran ash layers have been found up to a metre or more thick (Fig. 3). In East Crete, less substantial but still considerable ash layers have been identified, especially at the sites of Mochlos (see, e.g. Soles et al 1995) and Palaikastro. In these settlements, tephra was found associated with LM IA features and stratified below LM IB architectural features. The ash layers usually vary between 5 and 12cm in thickness, which, allowing for compression, dispersal and bioturbation, would imply an original ash carpet of more than 15 cm, sufficient to cause substantial damage to crops, livestock, buildings and water supplies. In most cases, rebuilding activities or repairs follow the ash fall, and many argue that Crete really witnessed its greatest days after the eruption, during the LM IB period.The eruption, at any rate, cannot have been the direct cause of the demise of the Minoan polities at the end of the LM IB period. What, then, was the cause of Minoan downfall and decline? Civil war, mainland Mycenaean conquest and earthquake, or a combination, are some of the most popular explanations. We suggest that the combined effect of an earthquake and the Santorini eruption in LM IA did indeed cause the eventual demise of Minoan civilization in the sense that these natural catastrophes caused serious crises on the island early in the succeeding Late Minoan IB period, producing a snowball effect that culminated in the destruction of the Minoan palace states. Some earth scientists have shown that there is a causal link between intermediate-depth earthquakes and volcanic eruptions and, in the case of Santorini, a delay of 2-5 years between an earthquake and the eruption has been suggested. There is good evidence for a destructive earthquake sometime before the tephra fell at Trianda on Rhodes, and at Mochlos and Palaikastro on Crete. Indeed, most of the Cretan sites, irrespective of the presence or absence of tephra, illustrate minor or major destructions or abandonments during the LM IA period (Fig. 4). This is why we regard it as highly likely that a major earthquake also triggered the Minoan eruption of Santorini, a hypothesis surely capable of scientific judgement in the near future. Unfortunately, there is still very little geomorphological evidence published that deals with the

84

J. DRIESSEN & C. F. MACDONALD

Fig. 3. Map of the south Aegean indicating general direction of the ash fall.

Fig. 4. Late Minoan IA destructions on Crete.

effects of the eruption on Crete. Minuscule quantities of tephra have been identified at about a dozen Cretan sites as well as in the Eastern Aegean, illustrating the path of the ash cloud. However, except for Mochlos, Palaikastro and the Dodecanesian sites, no good ash layers have yet been identified or published, although we expect this only to be a matter of time. Cores, for instance, in swamp areas close to the shore at Malia, have apparently failed to produce evidence for tephra. No data have been published

to our knowledge corroborating the 1981 study by Karsten and Cita indicating a major tsunami in approximately this period, on the basis of sea-bed anomalies, although at the London conference Marinos informed us that such evidence had been identified in 1996 on the coasts of Asia Minor, where tsunami deposits were overlaid by tephra. Indeed, recent tsunami research seems more concerned with elaborate simulation models than with actual field observations (Monaghan et al. 1994). The evidence of Theran

SANTORINI VOLCANO AND MINOAN CRETE pumice distribution is also better left aside as its taphonomy is debatable. However, as it is beyond doubt that a serious eruption did happen, it can be assumed that a tsunami destroyed some ships, damaged northfacing harbour installations (e.g. Poros, Amnissos and Nirou Chani in Central Crete) and salinated lowlying northern coastal areas, rendering them useless for agriculture for some years. That a major tidal wave did strike the coast is proved to us by Macdonald's personal observations of pumice on the hill west of the Villa of the Lilies. He has found pumice at the 15m contour level on the north side of the hill but no higher. Given erosion since the eruption, we might say that the tsunami that lifted the pumice to that height was over 15m in height but possibly less than 32m, i.e. the height of the hill. It should be noted that a tsunami does not carry pumice from the volcano; pumice ejected in the early phases of the eruption would have been carried, floating on the sea, south by currents and winds so that when the tsunami occurred at the Cretan coast, it was already there and was lifted by the wave(s) and deposited on the Amnissos hill. Tephra will also have affected crops and killed some animals, if not humans. Volcanic tremors may have caused yet more damage to buildings, although we should not expect evidence for a pan-Cretan volcanic earthquake. The abandonment and/or partial destruction of many Late Minoan IA sites could be an immediate consequence of the Santorini eruption, although it seems to us preferable to link this to the earthquake before the eruption. It may also be expected that climatological anomalies ensued, for example a 'volcanic winter', again with disastrous effects on agriculture and the economy in general. The so-called 'Storm' stele of pharaoh Ahmose (mid-late 16th century BC) has been cited in connection with this phenomenon, as it recounts a series of devastating storms early in the XVIIIth Dynasty. If the climatic anomalies referred to in the inscription are indeed a result of the eruption, the situation on Crete may have been devastating although it may have left few immediate traces in the archaeological record. Olive trees, vines and other crops may have suffered, a situation aggravated by any ash fall, which would also have been extremely dangerous for animals as it would have abraded their teeth and clogged their digestive system. Water pollution is also likely to have occurred, again with disastrous consequences, and it should be noted in connection with this that several Late Minoan I sites illustrate how some wells went out of use at this time and other new ones were dug, often protected from the elements

85

in some way. It may be recalled that the Tambora eruption in Indonesia in 1815 caused what has been called the 'last great subsistence crisis in European history' (Arnold 1988), when the cooler and wetter summer led to crop failure, disease, social unrest and famine in Western Europe. Lord Byron even recalled these events in his poem entitled Darkness (McGann 1986). It may be wise at this point to recall some anthropological features of 'disasters'. Essentially, a 'disaster is an event that involves a combination of a potentially destructive agent from the natural or technological environment and a population in a socially and technologically produced condition of vulnerability' (Oliver-Smith 1996). This combination leads to damage of the major social organizational elements and physical facilities of a community to such a degree that the essential functions of the society are interrupted or destroyed. This results in individual and group stress and social disorganization of varying degrees of severity. Disasters, therefore, tend to affect most aspects of community life. There can then be little doubt that on the psychological level, the eruption, from its first rumblings to the final explosive event, must have had an affect on the population of Crete. It is interesting, therefore, to note that that one of the effects of the AD 725 eruption of Santorini was to encourage the Emperor Leo III of Byzantium to remove icons from churches, as he assumed the eruption was a sign of divine wrath. It was the final not the only, reason for the onset of the iconoclastic period in the Byzantine world. A more recent eruption is also worthy of note. In the seven months after the eruption of Mount St Helens in 1980, there were increases of 18% in death rate, 21% in emergency room visits, 200% in stress-aggravated illnesses, 235% in mental illnesses, 45% in domestic violence and 37% in aggression, all mainly caused by the ash fall up to 40 miles away from the volcano (Adams & Adams 1984). If we turn to the purely archaeological data of Late Minoan I Crete, it can be argued that there are several features that may be interpreted as disaster induced and that, combined, may have caused Minoan society to begin to disintegrate. Such features include the following: abandonment of settlements; reduction of occupied space within settlements; decrease or absence of new construction and of new settlement; occurrence of crisis architecture (which implies changes of plan and function as well as construction with spolia and the erection of protective features); abandonment of old wells and the digging of new ones; evidence for an increase in conspicuous consumption by elites

86


(visible in architecture and Palatial Style pottery); hoarding of precious metals and objects; concentration of economic resources (food production, artisan items and their storage become closely guarded in the hands of the elite); ritual excesses, victimization and shifts within religious beliefs; a tendency towards political fragmentation and the emergence of new alliances. Several of these features can be described as posttraumatic stress symptoms and some of these are briefly highlighted below. Of particular importance is the fact that, although many new settlements were founded in

LM IA, none can be said to have been established first in LM IB. Moreover, although there is evidence for reconstruction and repair after an LM IA destruction, very few new buildings can be said to have been constructed during LM IB. Almost all of the Cretan settlements show signs of disruption in LM IA, some not necessarily at its very end (e.g. Vathypetro, Galatas). Several were destroyed, mostly by earthquake, some by fire, or else deserted at this point; others were destroyed or damaged again either still within LM IA (e.g. Galatas and Petras) or in the course of LM IB (e.g. Palaikastro), and yet others only

Fig. 5. Vathypetro settlement of LM IA with later additions (hatched).

SANTORINI VOLCANO AND MINOAN CRETE at the very end of this period (e.g Zakros). Of the 54 settlements occupied in LM IA, only 32 definitely remained occupied in LM IB (c. 60%), with this number dropping to about 10 in LM II, after which time it rises again. Hence, there was a gradual reduction in the number of settlements, albeit with a sudden rise in the number of destructions during LM IB. This process of abandonment must be seen against the general background of settlement contraction and a reduction in the area occupied within settlements during the LM I period. What this drop both in the number of settlements and in the area occupied within settlements actually means in terms of population decline has rarely been addressed. This decline can only have been caused by 'prime movers' such as famine, exile, emigration, warfare, disease and natural catastrophe, as these are more likely to account for abrupt and drastic declines in population and settlement numbers. The abandonment or destruction of settlements during the LM I period surely also affected the overland communication on the island (e.g. abandonment of the Palace at Galatas) and must have had serious consequences with respect to internal and international trade and transport (e.g. destruction of Building J/T at Kommos). Santorini, of course, was at the very least a stepping stone for Minoan trade, and the importance of the elimination of the settlement of Akrotiri should not be underestimated. The hypothetical 'Western String' trade route, connecting Crete with the Argolid (via Melos) and Attica (via Kea), received a serious blow and the void may have been eagerly filled by Mycenaeans. Those settlements that were not destroyed or abandoned in LM IA witness a series of changes during the LM IB period, and it is these changes that are particularly informative. Either because of earthquake damage, or for other reasons, many fine mansions lost their former prestigious appearances. Broad entrances were either blocked or made narrower; large dwellings were subdivided; cheaper materials were used for repairs and fine rooms lost their original appeal. Most surprising perhaps is the fact that the access to many structures became much more difficult during LM IB. This loss of prestige and the restriction of access should be translated in terms of socio-economic changes. A related feature is the fact that during the mature LM I period a series of enclosures were constructed within towns or around countryhouses. This was either for defensive reasons and/or to keep animals close. In addition, several old wells were given up and new ones, some of monumental size, were dug. Most peculiar, however, was the change of room functions.

87

Large public rooms were subdivided and modified into areas for storage or food production. The clearest examples may be briefly mentioned. The fine mansion of Vathypetro (Fig. 5) suffered earthquake damage in LM IA and was then patched up, given an enclosure wall and most of its rooms were modified into storage and agricultural production areas. The mansions at Tylissos seem to have had their entrance systems entirely modified before their LM IB destruction and gave pride of place to storage; the same occurred at Nirou Chani. The fine mansion at Xeri Kara suffered earthquake destruction and was changed into a storage and production unit with the loss of its original architectural prestige; the same happened at Nerokourou and Amnissos. After an LM IA earthquake, the centre of Minoan Palaikastro was not rebuilt but reserved for an large enclosed space in which two new wells were dug, one of which seems to have been for public use. Part of the central area of the Gournia town was also abandoned after its LM IA destruction and other houses were subdivided afterwards to house more people. The major building (J/T) at Kommos with its fine colonnades on either side of a large court was largely abandoned during LM IA, but 'squatters' used the ruins for metal and pottery production thereafter, perhaps still within LM IA. The mansion at Zou, easy of access in LM IA, was enclosured on all sides and access made very difficult. The enclosure probably served to protect animals and a pottery production unit; the same appears to have happened at the Makryghialos mansion. The Villa Reale at Ayia Triada was subdivided some time before its LM IB destruction and its final functions appear especially to have included storage and metal and pottery production. Several of its finer rooms were changed into storage areas. There are many more examples, which continue to stress the same phenomena. The Minoan palaces show a similar pattern. The newly discovered palace at Galatas suffered some damage in LM IA, followed by a 'squatter'-like occupation concentrating on foodprocessing; the palace was then abandoned and later destroyed by an earthquake. LM IB is so far only represented by a destruction in a building outside the palace to the north. Another, smaller palace, recently found at Petras (Fig. 6) had a large open central court and state rooms in LM IA, which were then reduced in size and the rooms used as storage areas. The extent of disruption and change in LM IA at the palaces of Knossos, Phaistos and Malia has not been sufficiently understood, nor the fact that there is

88


Fig. 6. Palatial building or palace at Petras, Siteia.

little evidence for them to continue as fully functional units in LM IB. They have in common that their access systems were modified (Fig. 7), probably after a LM IA destruction; thereafter, occupation appears in a limited scale only, on the basis of the small amount of LM IB ceramic evidence. Access is also restricted at the palaces of Zakros and Gournia during LM IB. In addition, in the Phaistos palace, a pottery production unit was installed on the east and a shrine placed in the West Wing, which was only accessible from outside the palace. The Malia palace also seems to have been largely abandoned before its destruction, but here too a shrine was placed in the South Wing, only accessible from outside. It is possible that Malia's so-called 'silos' to the southwest reflect a storage

area added to the palace only at a very late stage and with the specific purpose of serving the community rather than the palace. The palaces at Malia, Zakros and Phaistos have several blocked doorways, changed circulation patterns and fine rooms changed into storage areas. There are other related features such as the use of cheap building materials, special water provisions, more workshops and enclosed areas. We believe that all these examples illustrate some kind of crisis architecture, an adaptation forced by changing socio-economic conditions. The restrictions at exits and entrances imply a safety concern, and the other changes all seem to relate to making room for either agricultural or artisan production units in existing structures and adding storage rooms. The changes in room

SANTORINI VOLCANO AND MINOAN CRETE

89

Fig. 7. Restricted access after LM IA at Knossos, Gournia, Phaistos, Zakros and Malia.

function illustrate this most eloquently. The increase of storage is thus not a feature accompanying the simultaneous reduction of storage in the palaces at the beginning of the Neopalatial period, as some have claimed (Moody 1987), but a later feature, as some of the palace storage areas were created by modifying the original plan of the buildings. We may ask, therefore, whether local authorities were forced to increase storage capacities because the palaces were unable to allieviate the problems of food shortage by themselves? If so, it not only tells us something about the conditions of the palaces but also illustrates a new phenomenon in LM IB, namely decentralization of power. Indeed, the construction of community storage systems such as the 'silos' at Malia, the Bastione and other areas at Ayia Triada and elsewhere obviously increased the power basis of the local elites. In cult also there are many new features in the mature LM IA and LM IB period, which may perhaps be interpreted as disaster related and, therefore, represent forms of 'crisis cult'. Several, century-old cult forms, such as peak sanctuaries and the so-called lustral basins, were almost entirely given up in the course of this LM IA period. The main 'peak sanctuary' of Central Crete on Mount luktas is the only sanctuary of this kind to yield any evidence of use during LM IB, and so slight is that evidence that it may merely have been visited a few times; it was certainly not functioning as a major cult centre at this time, a phenomenon even clearer

at the other peak sanctuaries of Crete, which are entirely lacking in evidence for use at this time. The formally planned sanctuary of Kato Symi (Fig. 8) appears to have suffered earthquake destruction in LM I; thereafter, although it was rebuilt, its earlier grandeur with processional roadway and central podium was replaced with new constructions of a less impressive and formal nature. This change of plan may well reflect a radical change in worship at the site and a fall-off in palatial visitations. One of the new forms of cult in LM IB is the community shrine. We have mentioned the shrines in the palaces of Phaistos and Malia, which were only accessible from outside the palace and can, therefore, be described as public. At Palaikastro, some rooms of Building 5 (Fig. 9) were blocked off from the rest of the mansion and this part of the building was given an ashlar facade and a separate imposing entrance. The great statue of a young Minoan, made of ivory, gold and other materials (chryselephantine) was found in pieces scattered here and in the public area outside the house. Similar places of community worship can be found at other settlements: House B2 on Mochlos with its two Pillar Crypts, the Shrine on Pseira with its monumental fresco of a seated goddess in relief and its court, and the Gournia Shrine forming the end of a blind alley. That cults were brought into the settlements is further illustrated by the discovery of bronze figurines and inscribed Linear A stone vessels, votive objects previously found only in rural shrines.

90


Fig. 8. Formal arrangement of the Kato Symi sanctuary with later modifications (hatched).

Fig. 9. Part of Building 5 at Palaikastro converted into a community shrine in Late Minoan IB.

We may add here that there are examples in which volcanic pumice was included in ritual offerings; for example, in a foundation deposit for architectural alterations beneath a threshold of the mansion at Nirou Chani. Horns of consecration of this period have been found at Palaikastro, Zakros, Amnissos and Petras.

New forms of cult appear making greater use of sacred symbols and here the new Marine Style, the ceramic hallmark of the LM IB period, should be stressed. Its sudden appearance demands a rational explanation. Marine style pottery is definitely a distinct elite style, a 'palatial' style. This is demonstrated not only

SANTORINI VOLCANO AND MINOAN CRETE by the high quality of the product but also by its specific find places. It appears to be the demonstration par excellence of elevated status and group affiliation. We suggest that its production also filled a sudden gap within the corpus of propaganda-laden objets d'art in other media such as stone and ivory. However, its cultic associations, we feel, almost certainly derive directly from the 'Santorini experience', displaying a new awareness of the power of the sea, as both the eruption, visible from Crete, and the ensuing tsunami will have had a profound effect on the Minoan psyche, following so closely on the heels of the earthquake that caused the eruption in the first place. There may even have been a certain feeling that the true origins of earthquakes lay within the sea itself. Palatial ceramic styles, of which the Marine Style is one aspect, bring us to changes in arts and crafts during the LM I period. It is largely in this area that it is possible to argue for & floruit of Neopalatial Crete in the LM IB period, as the greatest corpus of fine, valuable objects comes from LM IB destruction levels. A priori, the link between fine objets d'art and prosperity is often a tenuous one. For example, using very broad and crude comparisons, Switzerland is a very prosperous country yet, except for cuckoo clocks, it has rarely produced major works of art, whereas the Italian renaissance was an unstable and war-stricken period during which numerous masterpieces were produced. Minoan workshops had been scattered throughout the settlements in LM I A, whereas, in LM IB, they become prominent features of palatial and major buildings or replace fine houses; for example, the kilns west of the Stratigraphical Museum at Knossos. It seems that specialization in arts and crafts begins to occur in previously prestigious surroundings and may be under tighter control by the elites. Under this heading, we return to Palatial LM IB styles of pottery, which were not only

91

under the control of the palatial elite but also served their purposes, perhaps even in the dissemination of new religious aspects to proclaim that religious control remained a palatial preserve. That said, we must still ask to which period should the manufacture of the finest and most valuable objects be assigned. It is possible to argue that the floruit of production of elite objects, with the notable exception of ceramics, was during LM IA, on the basis of the amazing Minoan wealth of the LM IA-LH I Shaft Graves at Mycenae and the finest objects from Akrotiri on Thera. It is interesting to note that not a single stone vase rhyton was complete at the time of the LM IB destructions. This is not to say that all production ceased in LM IB, but rather there was a decline in the production of valuable objects, heirlooms became more valued, even when damaged, and Palatial Style ceramics were brought in to fill the gap and press changes in cult emphasis. The great fresco works of Knossos belong to the 'Great Rebuilding' in LM IA and not to LM IB. Certain LM IB exceptions appear to challenge our hypothesis, notably, the monumental frescoes from Room 14 in the Villa Reale at Ayia Triada and the relief fresco of a seated goddess in the Pseira Shrine. However, these appear to us to reinforce the decentralization of Crete during LM IB, when local elites sought to bolster their power base through palatial emulation. Abandonment and destruction deposits of the mature LM IA and LM IB periods are also informative on a different level. It has been noted before (Georgiou 1979) (Fig. 10) that there are a surprising number of bronze hoards or hidden collections of valuables in Late Minoan I contexts compared with earlier and later periods on Crete: daggers hidden beneath pithoi, copper ingots walled-up at Ayia Triada, three bronze hoards of tools and weapons at Gournia,

Fig. 10. Distribution of bronzes in Late Minoan I contexts on Crete.

92


Fig. 11. Distribution of LM I administrative documents on Crete. bronze basins hidden in several places at Mochlos, weapons hidden both in the Malia palace and some of the houses, weapons and bronzes hidden at Palaikastro and Zakros, and bronze vases and tools hidden in several houses destroyed in mature LM IA at Knossos. These hoards seem to reflect an unsettled period, which had already begun before the end of LM IA. We have mentioned that many settlements increased their storage and artisan-production capacities by LM IB. One of the consequences of this decentralization would have been an increase of local administration. It is not surprising, therefore, that there is a proliferation of administrative documents throughout Crete in LM IB, both of Linear A tablets and sealings (Fig. 11). Surprisingly, the settlements with palaces yielded few or no administrative documents. This contrasts with the later picture of Linear B and argues for a notable degree of decentralization during LM I instead of the later single integrated economic system with the palaces as capitals. We argue that the local centralization of storage systems, industrial units and administrative records are all part of a single phenomenon, namely, the rise of powerful local eonomic systems because of failure within the earlier central, palatial administrations. On a wider level, LM IB is also the period during which the Minoans lose their influence in the Cycladic and Dodecanesian islands. All of this began with the eruption of Santorini and the destruction of the Minoan Cycladic base, Akrotiri. Conclusion The following historical reconstruction seems to us to be the most appropriate model, based on our interpretation of the evidence. The archaeological data suggest that there was a severe

economic dislocation in Crete, triggered by the Santorini eruption, and that this dislocation gradually worsened as LM I progressed. Moreover, a combination of a general feeling of uncertainty caused by the eruption and its accompanying effects, the earlier destructions caused by earthquake, and the need to rebuild and re-establish normal economic life, may have presaged the end of the LM IA centralized sociopolitical and economic system and created the suspicion that the existing palatial elites had lost their divine support. Because of problems with food production and distribution, the existing network disintegrated, resulting in a decentralization or fragmentation of the political landscape, which went hand in hand with an increase in elitist power and competition. In all probability, the latent tension between this and the demonstrable tendency to exclude those not of the group (by hoarding, segregation measures and storage) would have eventually led to conflict. That this did occur is indicated by the enormous number of conflagrations and selective destructions of houses, villas and palaces in the LM IB period. In such crises, there is a widespread tendency to explain disasters in terms of the sins of the people and one often looks for scapegoats. Although disasters can be somewhat even-handed in distributive destruction, affecting all levels of society in various ways, the recovery process is not at all egalitarian. Conflict in the post-emergency or rehabilitation period of a disaster tends to arise in two areas, namely, the allocation of blame and the allocation of resources for rehabilitation. The changes in architecture, production, storage, cult, and arts and crafts discussed above may well all have been triggered by the situation of conflict and tension existing after the eruption. Famine would have led to strife and years of internal unrest, population movement and religious

SANTORINI VOLCANO AND MINOAN CRETE upheaval. This is why we feel justified in assuming that the Santorini eruption affected almost every aspect of society, producing responses that, in all probability, created a chain reaction leading to fundamental changes. Many of the features highlighted resulted from post-eruption stress. During LM IB, Crete's main preoccupation seems to have become the protection of food production, storage, livestock and industrial production through local centralization. Indeed, it appears that the old nodal points, the palaces of LM IA, were not capable of offering an adequate response to the immediate problems. This inadequacy resulted in secondary centres being allowed (or, perhaps assuming) much greater regional control, which eventually led to a political fragmentation. The Santorini eruption is here given the role of a precipitant or catalyst, triggering an entire series of changes that culminated in Crete being absorbed to a greater or lesser extent into the Mycenaean, and thereafter, the Greek world. References ADAMS, P. R. & ADAMS, G. R. 1984. Mount Saint Helens ash fall. Evidence for disaster stress reaction. American Psychologist, 39, 252-263. ARNOLD, D. 1988. Famine, Social Crisis and Historical Change. Oxford, 30. BIETAK, M. 1996. Avaris. The Capital of the Hyksos. Recent Excavations at Tell el-Dab'a. British Museum Publications, London, 77-78.

93

DRIESSEN, J. & MACDONALD, C. F. 1997. The Troubled Island. Minoan Crete before and after the Santorini Eruption. Aegaeum, 17. GEORGIOU, H. 1979. The Late Minoan I Destruction of Crete. Metal Groups and Stratigraphic Considerations. Los Angeles. KUNIHOLM, P. I., KROMER, B., MANNING, S. W., NEWTON, M., LATINI, C. E. & BRUCE, M. J. 1996. Anatolian tree rings and the absolute chronology of the Eastern Mediterranean, 2220-71 SBC. Nature, 381, 780-783. LIRITZIS, I., MICHAEL, C. & GALLOWAY, R. B. 1996. A significant Aegean volcanic eruption during the second millennium BC revealed by theromoluminescence dating. Geoarchaeology, 11, 361-371. MACGlLLIVRAY, J. A., SACKETT, L. H. & DRIESSEN, J.

1998. Excavations at Palaikastro, 1994 and 1996. Annual of the British School at Athens, 93,221-268. McGANN, J. J. (ed.) 1986. Lord Byron. The Complete Poetical Works IV. Oxford, 41. MONAGHAN, J. J., BlCKNESS, P. J. & HUMBLE, R. J.

1994. Volcanoes, tsunamis and the demise of the Minoans. Physica D, 77, 217-228. MOODY, J. 1987. The Minoan palaces as a prestige artifact. In: HAGG, R. & MARINATOS, N. (eds) The Function of the Minoan Palaces. Svenska Institutet i Athen, Stockholm, 240-241. OLIVER-SMITH, A. 1996. Anthropological research on hazards and disasters. Annual Review of Anthropology, 25, 303-328. SOLES, J. S., TAYLOR, S. R. & VITALIANO, C. J. 1995. Tephra samples from Mochlos and their chronological implications for Neopalatial Crete. Archaeometry, 37, 385-393.


Late Minoan IB marine ware, the marine environment of the Aegean, and the Bronze Age eruption of the Thera volcano PETER BICKNELL Department of Classics and Archaeology, Monash University, Clayton, Vic. 3168, Australia Abstract: Late Minoan IB fine ware pottery includes a number of decorative styles. The most spectacular of these is characterized by motifs, hitherto only rarely deployed by Cretan vase painters, drawn from the marine world. Late Minoan IB marine ware turns up in ritual contexts, which include human sacrifice. The pottery style is likely to reflect, then, not simply a vagary of secular fashion, but a circumstance or circumstances requiring far-reaching religious attention. It is proposed that Late Minoan IB marine ware and the cult activities in which it was deployed were a response to negative effects of the Late Bronze Age Thera eruption on the marine environment of the Aegean.

At a point during the Late Bronze Age falling within the long reign, c. 1500-1450 BC, of the Egyptian Pharaoh Thutmose III (Wachsmann 1987, pp. 127-130) Mycenean invaders from the Greek mainland established control over Minoan Crete. Their appearance in the island is reflected in various ways, not least by the decorative style, motifs and shapes (in particular, the so-called Ephyraean goblet and squat alabastron) of Late Minoan II pottery (Betancourt 1985, p. 159). Late Minoan IPs ceramic predecessors, Late Minoan IA and IB, are still essentially Minoan in spirit and design. Although the content of the former is relatively dull and circumscribed, the artistic quality of some vessels, of which a group found in room C58 at Gournia (for location of sites mentioned in this paper, see Fig. 1) is a case especially in point, is extremely high. The Late Minoan IA vocabulary includes heraldic motifs such as the double axe and bull's head, others drawn from the plant world, and various abstract spirals, wheels, ripples and meanders (Popham 1967; Betancourt 1985: pp. 128-133). Late Minoan IB, the last ware produced while the Minoans were still masters of Crete, is far more complex. There are two overarching categories (Betancourt 1985, pp. 137-158). The first is the so-called standard tradition, relatively unexciting and uninnovative, which continues at the outset to exploit Late Minoan IA motifs. The second comprises stunning fine ware vessels that come in multiple concurrent varieties, each dominated by a different decorative mode. The principal modes are the floral, marine

and geometric-abstract. To judge from frequent incorporation of marine co-motifs and submotifs into other modes (in the case of one such hybrid, from Phaistos, for example, the base features a marine motif whereas on the body abstract rosettes and zigzags alternate), the marine mode was the most significant (Mountjoy 1984). Exemplars are certainly more abundant than those of its floral and geometric-abstract counterparts. The period, no more at most than a single generation (Popham 1990), during which Late Minoan IB pottery was in vogue turned out to be one of major disaster for the Minoans. Almost every significant site on Crete was terminally or temporarily destroyed, with violent conflagrations the normal accompaniment of devastation. Palace complexes such as those at Phaistos, Mallia and Kato Zakro, prosperous townships such as Gournia and Palaikastro, and rural mansions, with those at Sklavokambos, Nirou Khani and Makrygialos typical examples, were equally affected (Page 1970, pp. 1-12; Betancourt 1985, pp. 133-139). Almost, but not quite, exceptionally the largest palace complex, that at Knossos, survived relatively unscathed to become, in due course, the administrative centre of the Myceneans. Many of the structures around it, however, were incinerated like entire sites elsewhere (Wall et al 1986). Convincing explanation of the Late Minoan IB holocaust in Crete remains the greatest of the challenges confronting the historian of the Bronze Age Aegean. Significant involvement of the Late Bronze Age eruption of the Thera

From: McGuiRE, W. G., GRIFFITHS, D. R., HANCOCK, P. L. & STEWART, I. S. (eds) The Archaeology of Geological Catastrophes. Geological Society, London, Special Publications, 171, 95-103. 1-86239-062-2/OO/S 15.00 © The Geological Society of London 2000.

96

P. BICKNELL

Fig. 1. Location of sites mentioned in text.

volcano has been routinely discounted (Doumas 1983, pp. 144-147; Barber 1987, pp. 221-222; Dickinson 1994, p. 304), as the stratification on Crete of distal ashfall from the paroxysm put it beyond doubt that the event occurred while Late Minoan IA pottery was still in use (Warren 1991; Soles et al 1995; Soles & Davaras 1996). At first sight attractive attempts to inculpate rampaging Mycenean invaders have been devastatingly criticized on various grounds by Doro Levi (see Page 1970, p. 12) and others (Dickinson 1994, p. 304), and are not comfortably compatible with the fact that at least a decade may separate the Late Minoan IB destructions at Phaistos and peripheral Knossos from those at Palaikastro and Kato Zakro (Downey & Tarling 1984). It is as likely as not that multiple causes contributed to Minoan decline. In what follows, a new entry point into a labyrinth of problems is attempted by way of the motifs and context of Late Minoan IB marine style pottery. Although not immediately and overtly implicated, as once envisaged (Marinatos 1939), in the Late Minoan IB vicissitude of Crete, the Thera volcano, it will emerge, may have made at least one significant indirect contribution to conditions that paved the way for Mycenean takeover.

Late Minoan IB marine ware; motifs and context By 1984, when Elizabeth Mountjoy published a corpus of all Late Minoan IB marine ware vessels, together with marine-floral and marinegeometric-abstract hybrids, known to her, 19 sites in Crete were represented. Subsequent discoveries have brought to light further exemplars at both the same and additional sites, such as Mochlos and Poros 'Knossos' port. Possibly manufactured in a travelling workshop based on the Knossos palace (Betancourt 1985, p. 140), marine style vessels were not only distributed throughout Crete but found their way to Minoan outliers such as Kastri, Kythera and Rhodian Trianda and beyond to Egypt and the Levant. Several of the Late Minoan IB motifs were imitated or adapted in an often extravagant and baroque manner in the Cylades islands and on the Greek mainland, and, to return to Crete, an artistically degenerating selection is part of the vocabulary of Late Minoan II ware and its successors, all produced during the period of Mycenean domination. Late Minoan IB marine ware comes in a wide range of pleasing shapes and features sea scenes usually, but always underwater. Main

MINOAN MARINE WARE AND THE THERA ERUPTION and sub-motifs are often densely crowded provoking diagnosis of horror vacui on the part of artists concerned (Betancourt 1985, p. 145). An initial appearance of naturalism turns out on closer inspection to be a mirage. Counterfactual stylization of some motifs is the rule and by way of more fundamental departure from realism, the core Minoan heraldic emblem of the double axe and an enigmatic stellate object (see Fig. 2) occasionally intrude amongst marine fauna, flora and rocks. Also, it may be necessary to acknowledge instances of disregard of scale. Two classes of marine motif proper can be distinguished, main and background filling. The latter includes rocks, sometimes covered with a minute cellular pattern that may be intended to evoke coral (Fig. 2), water bubbles, seaweed of different types and a species of sea urchin (Fig. 2, 4 and 5) probably to be identified as the rockdwelling Paracentrotus lividus, whose 3cm long spines protrude from a test some 6cm in diameter. If, with Mountjoy (1984), we exclude fish depictions on isolated sherds as post Late Minoan IB innovations, there are four main motifs. First, and less common than its counterparts, is the common dolphin, Delphinus delphis (Fig. 2, 3). Much more frequent are two cephalopods and a mollusc of the gastropod class with whorled shell. Despite the single rather than double row of suckers on the tentacles, the octopus that appears, for example, on vessels from Gournia and Palaikastro (Fig. 2, 4 and 5) must be identified as Octopus vulgaris. Configuration in general and absence of webbing between the tentacles at their base in particular rule out Eledone moschata and Eledone cirrosa. The other cephalopod (Fig. 2, 1) is the so-called paper nautilus, Argonautica argo. The standard representation of this pelagic, nocturnal and usually submarine creature, with three arms emerging from a closed region of a distorted shell, is remote indeed from realism. The painters concerned either failed to understand, or, prompted by aesthetic considerations, chose to ignore, the true relationship between the cephalopod and its shell, and also the manner of its motion. The gastropod (Fig. 2, 4 and 5) has been variously identified. Influenced by the spiny protrusions, Marinates (Marinates & Hirmer 1960, p. 76) confidently identified it as one of the purple dye, producing murices, either Murex trunculus or Murex brandaris, both of which attain a maximum height of 8-9 cm. If Marinatos is right, the sharply demarcated and tapering secondary whorls, not a feature of either of the two species, must be regarded as yet another example of departure from realism.

97

Strikingly, and remarkably, the late fifth century die-cutters of Tyre, eventual eastern Mediterranean chief centre of the purple industry, depicted one of the two dye-yielding murices (probably M. trunculus, given that vast dumps of expended shells of this species lie close to Tyre's remains) similarly modified on the reverse of the shekel. Taking considerations of relative scale more seriously, Mountjoy (1984) pronounced the gastropod a triton without discussion. Gill (1985) conceived of insinuation of the trumpet triton, either Charonia variegata or Charonia nodifera, both of which are capable of reaching a length well above 40 cm, but actual representation, for aesthetic reasons, of the more distinctive, if smaller (up to 18cm in length), ranella, Ranella gigantea. Much earlier, Bosanquet (1904) and Sir Arthur Evans (1928, pp.306 and 316) tried to have it both ways by opting for a deliberately contrived conflation of murex and triton. In different ways both gastropods have a high Minoan profile. Evidence is abundant that the triton played a prominent role in religious ritual. On a lentoid seal from the Idaian cave a priestess is depicted holding a trumpet shell aloft with her left hand (Evans 1935, p. 344). Actual shells have been found in shrines and other cult areas at Knossos, Mallia, Palaikastro, Phaistos, Pseira, Pyrgos and Kato Zakro (Reese 1990). Minoan exploitation of the murex for dye production had a long history. Large-scale production during the protopalatial period is attested by substantial dumps of processed shells on the offshore island of Kouphonisi (Bosanquet 1904; Stieglitz 1994). Eventually, the centre of gravity of the industry shifted to the neighbourhood of Palaikastro, whose again impressive shell dumps represent a time-span extending from the protopalatial period through to the neopalatial Late Minoan IB period (Hood 1971, p. 94; Reese 1987). Neopalatial purple manufacture on a smaller scale is reflected, for example, by plentiful crushed Murex trunculus fragments in the remains of the coastal mansion at Makrygialos (Reese 1987). Final choice between triton and murex for identification of the Late Minoan IB marine ware gastropod motif is not easy, but the absence of extended pointed protrusions from Charonia nodifera and Ranella gigantea, and the crowded multiple molluscs depicted on some rhyta from Palaikastro tip the scales, to my mind, in favour of the murex. As Aristotle (History of Animals, 5. 15) was first to record, murices gather together in large numbers in spring. Tritons are consistently more solitary. Before the neopalatial period, the context of Minoan representations of marine creatures and

98

P. BICKNELL

Fig. 2. Late Minoan IB vessels: 1, provenance unknown; 2 and 5, Palaikastro; 3, Pseira; 4, Gournia.

of marine paraphernalia in general is unmistakably cultic. The religious association of the triton shell, noted above, is one case in point. In a religious sanctuary uncovered at Anemospilia pebbles from the seashore were placed upon a stepped altar (Sakellarakis & SapounaSakellaraki 1991). Protopalatial faience models of paper nautilus shells and other sea creatures, together with hundreds of actual seashells, colourfully painted, were found in repositories in the Knossos palace in the company of representations of a female deity (Evans 1921,

p. 520). Expectation of a similar setting for Late Minoan IB vessels decorated with marine motifs is born out by the results of a seminal study by Mountjoy published in 1985. In the paper concerned, Mountjoy noted four separate associations at Knossos of Late Minoan IB marine ware with areas clearly used for ritual purposes. She went on to record two further such juxtapositions at Palaikastro and single ones at Archanes close to Knossos, Gournia, Nirou Khani, Pseira, Pyrgos, Tylissos and Kato Zakro. To these sites we can probably

MINOAN MARINE WARE AND THE THERA ERUPTION add building B2 recently excavated at Mochlos (Soles & Davaras 1996). At Knossos one of the associations is especially dramatic. Remains of five Late Minoan IB marine ware vessels were discovered in the 1980s in the debris of an incinerated structure, one of the casualties of the Late Minoan IB period of disaster, that has been labelled the House of the Children's Bones. Here, on what is by far the most coherent and plausible interpretation of the evidence overall (Wall et al 1986), after ritual sacrifice and consumption by priests and votaries of portions of their flesh, stripped, disarticulated bones of three or four young children were deposited in a small basement room inaccessible to predators. The proximity of the isolated skull of a child to ritual appurtenances, including Late Minoan IB marine style vases, in the west wing of the palace complex at Kato Zakro (Platon 1971, p. 120) and a similar concatenation (here the skull is that of a young woman) in house B2 at Mochlos (Soles & Davaras, 1996) suggest that human sacrifice during the Late Minoan IB period was not confined to a single site. If Nilsson (1950, pp. 194-235) was right in insisting upon a sacrificial ambience for the double axe that pervades Minoan iconography, its occasional appearance amid the underwater scenery depicted on Late Minoan IB marine ware is readily explained. The provenance of two examples is the House of the Children's Bones (Mountjoy 1984). The Late Bronze Age Thera eruption and the marine environment of the Aegean In the case of the relatively mass representational medium of vase painting, Minoan deployment of motifs drawn from the marine world is for long both rare and sporadic. Occasional vessels featuring tunny, stylized octopuses or dolphins turn up in Middle Minoan II and III ware, the first entirely protopalatial, the latter extending into the new palace period. Within the decorative repertoire of the subsequent Late Minoan IA pottery, produced, in the opinion of Popham (1990) over a time-span approaching 75 years, we encounter not a single instance of a motif drawn from the coastal or open sea. Such prolonged dearth makes the proliferation and profusion of marine motifs in the dominant variety of Late Minoan IB all the more startling. The most conspicuous valency of Late Minoan IB marine ware, to reiterate, is religious. Rhyta and other types of vessel displaying sea creatures and marine environmental features were employed in, or in association with, ritual acts

99

that included human sacrifice. This particular conjunction is as startling as the sudden prominence of marine motifs. Provided one follows Hughes (1991, pp. 13-17) in discounting premature claims (Sakellarakis & Sapouna-Sakellaraki 1981, 1991) about discoveries in the early 1980s at Anemospilia, there is no evidence for earlier Minoan recourse to this kind of offering. Other things being equal, one would expect the decorative motifs of vessels chosen for ritual purposes to reflect the ambience and purpose of the rites concerned. Late Minoan IB marine ware, it ought to follow, was employed in ceremonies somehow connected with marine creatures and their habitat. These ceremonies, performed with the intent of influencing the deities who presided over the sea and all within it, included the sacrifice of human beings. To account for unprecedentedly intense cultic preoccupation with the sea, preoccupation sufficiently urgent to dictate recourse to the ultimate expedient in the ancient world for placating supposedly affronted gods, we are compelled to envisage at the outset of the period during which Late Minoan IB pottery was produced some profound and catastrophic disturbance of the local marine environment. The context and stratigraphy of ash fall-out at the Mirabello Bay site of Mochlos put it beyond reasonable doubt that the Late Bronze Age eruption of the Thera volcano took place at the very end of the Late Minoan IA pottery phase. Having made this point, Soles and his coinvestigators went on to note without further elaboration that it was not uncommon for natural catastrophes in the Aegean to be associated with changes in pottery style (Soles et al. 1995). From the religiously conditioned perspective of early peoples, natural catastrophes reflect extreme displeasure of the god or gods who preside over the sphere in which disaster takes place. Given the close juxtaposition that Soleg and his colleagues emphasized, it is difficult to resist the temptation to bring into close connection the Thera paroxysm and the marine upheaval to which Late IB marine style pottery and human sacrifice were religious responses. The first explicit attempt, as far as I am aware, to associate Late Minoan IB marine ware with the Thera eruption is that of Wilson in a popular book published in 1985. After observing that addition of marine motifs to the floral and heraldic ones that were the staple of Late Minoan IA pottery is striking, Wilson (1985, pp. 139-141) proceeded to draw his readers' attention to the description of the Late Roman imperial historian Ammianus Marcellinus (c. AD 325-395) of the great Mediterranean

100

P. BICKNELL

seismogenic tsunami of AD 365. In Ammianus' words (26. 10. 15; I have modified the translation Wilson uses): The sea with its swelling waves was driven back and withdrew from the land, so that ... many kinds of marine creatures were revealed in the mud and numerous ships were stranded Many people roamed about without qualms in the shallow remainder of the water in order to gather fish and similar creatures with their bare hands. Then the roaring sea ... gathered up and crossing the now seething former shallows, hurled itself on islands and broad stretches of mainland and levelled countless structures — The great volume of water killed many thousands of people.' Have presented this account, Wilson suggested that the Minoans introduced depictions of sea life on to their pottery in the wake of devastation of Crete's northern coast by devastating tsunamis propagated in the course of the Thera eruption. The giant waves were interpreted by their victims as punishment for some offence and the motifs of Late Minoan IB marine ware were a contribution to expiation. The main problem with this scenario is that there is considerable doubt as to whether any significant Thera-generated tsunamis were directed towards Crete. Formidably deterrent to belief in tsunami intrusion anywhere along the island's northern coast are a seemingly ubiquitous absence of marine-bed boulders deposited above storm tide levels (Dominey-Howes 1996, pp. 45-93) and lack of any convincing indication in the archaeological record of any Minoan site of any effects that could be ascribed unequivocally, or even with a reasonable degree of plausibility, to tsunami damage. It is true that Francaviglia (1990) reported recovery of tsunami-emplaced pumice at Amnisos from the hillslope, 10-15m high, behind a villa excavated by Marinates, but caution, I believe is called for. Francaviglia's claim appears to be in conflict with the representations of Marinates himself, who reported finding pumice only in the villa itself close to the sea (Dominey-Howes 1996, p. 87), and early in 1997 I failed to detect any trace of pumice on the hill despite persistent searching. In a nutshell, although it is just conceivable that evidence of large-scale Late Bronze Age tsunami incursion into Crete is still to emerge (by way, for example, of telltale sediment in cores extracted from coastal locales), current indications are that Minoan observation of a temporarily denuded sea bed, demolition by inundation of major sites in northern Crete, and longer-term detrimental effects, such as salination, of huge volcanogenic waves are unlikely to have provided the stimulus for Late Minoan IB pottery's marine motifs.

Wilson's approach ruled out, it becomes necessary to look for different connections between the Thera eruption and adverse conditions of the Aegean marine environment that dictated a dramatic change of Minoan pottery style. Further consideration of the motifs of Late Minoan IB marine ware may be a helpful preliminary to establishing what these might have been. Commenting on marine motifs in general in Minoan representational art and recognizing their basic cultic valency, Gill (1985) emphasized that they are likely to have been selected for practical rather than aesthetic reasons. The purpose of choice will have been to ensure continuity of harvests from the sea by depiction of those creatures most useful to human beings, and of the animal attendants or symbols of the god or gods who presided over that aspect of Minoan life. All four main sea creature motifs of Late Minoan IB marine ware together with the background filling sea urchin are eminently compatible in one way or another with their profile. A curious idea, strangely shared by many, that the Minoans avoided culinary use of products from the sea is comprehensively refuted by data assembled by Guest-Papamanoli (1983) and Powell (1996). There is little reason to doubt, it emerges, that the common octopus and the sea urchin Paracentrotus lividus, would have been protein sources as significant for the Minoans as they were for Crete's Greek occupants of the Classical period (Thompson 1947, pp.72 and 208). Although the Classical Greeks by and large refrained from hunting and eating cetaceans, a cylinder seal depicting dolphins hung from poles, like fish, to dry, indicates that Delphinus delphis contributed to the diet of their Minoan predecessors (Gill 1985). The murex provided food in Classical times (Thompson 1947, p. 218), and in this respect too the Minoans certainly anticipated those who came after them. Detritus from the so-called unexplored mansion at Knossos, a purely residential structure certainly unconnected with dye processing, included a few shells of Murex trunculus alongside those of other edible species and Paracentrotus tests (Popham 1984, pp. 246-256). Any dietary significance of the murex, however, would have been outweighed for the Minoans by its major importance in the purple industry. Given the evident extent of operations at Kouphonisi and then in the Palaikastro area, the dye produced ought to have been well in excess of local requirements. In all probability, purple was exported by the Minoans in large quantities. To conclude this conspectus with the paper

MINOAN MARINE WARE AND THE THERA ERUPTION nautilus, although it certainly could have been eaten by the Minoans, its comparative rarity rules out its having been a dietary staple. The pseudo-shell's function as cult object emerged in the previous section and it is possible, in addition, that specimens were used locally and exported elsewhere as decorative items. All but one of the sea creatures depicted on Late Minoan IB marine vessels, it can be said with confidence, are likely to have contributed significantly, or at least not negligibly to the Minoans' food supply. In particular, as well as being edible the murex was the source of a luxury product that may have been a key export item. Given such ambience, it is logical to seek to bring into connection with the pottery style and the conditions that determined its motifs indisputable effects or by-products of the Late Bronze Age Thera eruption inevitably deleterious to marine fauna. Prime candidates are floating pumice and distal ash fall. To begin with the former, the first paroxysmic phase of the Thera eruption resulted in the emplacement of pumice metres thick over the whole of the convulsed island. As in similar circumstances at Krakatoa in 1883, massive amounts will have been deposited into the surrounding sea. Again as in the case of the Krakatoa eruption, both discrete lumps of pumice of various sizes and rafts of accumulated pieces, some coherent and thick enough to support persons prepared to walk on them, will have drifted in all directions, ending up in coastal areas in some instances and remaining afloat in open water for years in others (Bullard 1976, p. 46; Simkin & Fiske 1983, p. 15, 91-95, 200-201 and 211). As well as presenting a major obstruction and hazard to fishing vessels, such seaborne tephra can be directly detrimental to a variety of marine life. In the wake of the Alaskan Katmai-Novarupta eruption of 1912, for example, floating mats of pumice frightened halibut from fishing grounds around Kodiak Island (Erskine 1962, pp. 204-208). Pumice projected into the sea during the 1952 Barcena (Mexico) eruption had more extensive effects on the marine environment, with intertidal fishes and marine invertebrates seriously depleted as a result of the material's abrasive action (Brattstrom 1963). Copious Thera-derived pumice should have been similarly deterrent to fish in the open sea near Crete and injurious to denizens, vertebrate and invertebrate, of coastal waters, especially, but far from only, to the north of the island. Given a natural temptation, stressed above, to treat projections as facts, it would be welcome to be presented with direct indication of Minoan

101

awareness of and concern about floating Thera pumice. It is just possible that such evidence exists. The third vessel depicted in the illustration is one of a pair of identical pear-shaped, Late Minoan IB marine ware rhyta discovered in the remains of a Minoan township on the offshore island of Pseira in the Mirabello Bay area of northern Crete. The dolphins, the main motif, are surrounded by a reticulate pattern which, as it does not cover the cetaceans themselves, cannot represent an entrapping net. In fact, the pattern, as Morgan (1988, p. 35) and others have explained, is a Minoan artistic convention for representation of the surface of the sea. Normally the modules of the sea connoting mesh are either empty, or contain a single central circle, a central or sub-central curved line, or a couple of unobtrusive dashes. Net elements packed with dots are unique for Minoan Crete and, as far as I am aware, paralleled only in parts of the complex, rococo ornamentation of vessels from Ayia Irini and Phylakopi in the Cyclades, and Vaphio on the Greek mainland, all three of which reflect imitation of a variety of Minoan models (Bosanquet 1904; Mountjoy 1984). The dot-clogged sea cells of the Pseira vases, I hesitantly suggest, are not an aberrant symptom of a painter's abhorrence of empty space, but by way of illustration of or allusion to a sea cluttered in the immediate aftermath of the Thera eruption with floating pumice, a circumstance leading to an absence of dolphins, an important Minoan food source. If such is the case, the representations on the two rhyta are apotropaic: a condition is depicted that it is hoped the deity or deities concerned will consent to bring to an end. In contrast, the majority of Late Minoan IB vases display underwater scenes in which cepholopods, gastropods and echinoderms flourish in evidently clear water. Here, by way of converse sympathetic magic, conditions are portrayed that have ceased to be the case and that the gods are solicited to restore. Distal Thera ash fall into the sea, as well as pumice floating on its surface, would also have been capable of a substantial contribution to marine environmental degradation sufficient to evoke extreme religious countermeasures. As Blong (1984, p. 338) observed, tephra falls of 3cm or less into water have been associated with the mortality of its inhabitants, with various specific circumstances such as particle size, heavy metal constituents or attached aerosols influencing the extent and nature of effects, which may be subtle. Only one link in a food chain, Blong went on to emphasize, need be affected in order for there to be detrimental consequences for an economically significant

102

P. BICKNELL

population. Relatively insignificant ash fall into the Bay of Naples after the 1906 Vesuvius eruption caused massive destruction of invertebrates and temporary disappearance of fish from the area (Bianco 1906). Minor ash fall again, from Hekla in 1947, was sufficient to drive fish away from coastal waters affected (Blong 1984, p. 337). At Mochlos in three places distal Thera ash up to 10cm in depth has been exposed during recent excavations. In all cases the tephra represents airborne deposit rather than being the result of human clearance or construction activity (Soles et al 1995). Before settling, the ash blanket may have been deeper than at present. It follows that at least 10cm of ash, probably more, fell into the sea to Crete's north. Dolphins, it is worth stressing, given that they were part of the Minoan diet, are especially sensitive and vulnerable to marine pollutants. Their immune system suppressed, they become sick, unable to feed and highly susceptible to virus infection. In the second half of 1990 an area of the Mediterranean between North Africa and Spain was severely contaminated as a result of industrial pollution. Poisoned by heavy metals and virally infected, close to 6000 dolphins died in the space of 3 months. No other species was palpably affected (Pastor 1991, pp. 60 and 128-129). Any significant deoxygenation of water stemming from the presence of pumice or ash would be especially deleterious to life in the sea. A possibility to bear in mind is that of accumulation of nutrients in Cretan and other Aegean coastal waters as a result of progressive mineral release. Given eventual action of sunlight on the nutrient-saturated sea, the stage is set for the syndrome (now familiar in the Adriatic as a result of deposition into the sea of artificial fertilizers leached from coastal soil) of runaway phytoplankton blooms, subsequent crash of the inflated algal population, and mass marine mortality as a result of oxygen-reducing bacterial decomposition. To proceed a little further, and more adventurously, down this track, there are some episodes of bloom formation, so-called red tides, for example, that kill fish and other sea creatures not only by eventual deoxygenation of their immediate environment, but also, earlier and more directly, through the release of virulent neurotoxins. Toxin-permeated phytoplankton is consumed in the first instance by bivalves and gastropods, whose metabolic and feeding rates are adversely affected so that large numbers die prematurely. The poisons simultaneously spread upward through the marine food chain in increasing concentration as shellfish are ingested by higher-order molluscs such as cepha-

lopods, which are in turn eaten by vertebrate predators including cetaceans and fish. The earliest description of a red tide (Exodus 8. 19-25) relates to an outbreak in the Nile Delta in the Late Bronze Age. The local fish population was exterminated by a combination, presumably, of poisoning and oxygen starvation. Land-based casualties of toxic blooms include, of course, human consumers of shellfish and other tainted fauna. Fatal poisoning of some Minoans by tainted molluscs, dolphin or fish would have dictated temporary abstention from sea food, whatever the shortage of other foodstuffs, and provided further stimulus to gruesome religious expiation. Conclusion Drawing to a close, I hasten to underline the speculative nature of the content of the final paragraph of the previous section. One is not compelled to think in terms of exotic algal blooms and deadly toxins, any more than to insist that huge tsunamis ravaged northern Crete, to bring the Late Bronze Age eruption of the Thera volcano into connection with a serious disturbance of the marine environment of the eastern Aegean. Prolonged adverse conditions as a result of profusion of floating pumice and copious distal ash fall will have been more than sufficient to elicit an extreme religious response on the part of the Minoans and to play a significant role in the enigmatic collapse of their civilization, which set the scene for eventual Mycenean domination of the whole of Late Bronze Age Crete. References BARBER, R. L. N. 1987. The Cyclades in the Bronze Age. Duckworth, London. BETANCOURT, P. P. 1985. The History of Minoan Pottery. Princeton University Press, Princeton, NJ. BIANCO, S. L. 1906. Azione della pioggia di cenere caduta durante 1'eruzione del Vesuvio dell' Aprile 1906 sugli animali marini. Mitteilungen Zool. Stat. Neapel. 18, 73-104. BLONG, R. J. 1984. Volcanic Hazards. Academic Press, Sydney, NSW. BOSANQUET, R. C. 1904. Some 'Late Minoan' vases found in Greece. Journal of Hellenic Studies, 24, 317-329. BRATTSTROM, B. H. 1963. Barcena Volcano 1952 - its effect on the fauna and flora of San Benedicto Island, Mexico. Proceedings, 10th Pacific Science Congress 1961. Bishop Museum Press, 499-524. BULLARD, F. M. 1976. Volcanoes of the Earth. University of Texas, Austin.

MINOAN MARINE WARE AND THE THERA ERUPTION DICKINSON, O. 1994. The Aegean Bronze Age. Cambridge University Press, Cambridge. DOMINEY-HOWES, D. 1996. The geomorphology and sedimentology of five tsunamis in the Aegean Sea region, Greece. PhD thesis, Coventry University. DOUMAS, C. G. 1983. Thera; Pompeii of the Ancient Aegean. Thames and Hudson, London. DOWNEY, W. S. & TARLING, D. H. 1984. Archaeomagnetic dating of Santorini volcanic eruptions and fired destruction levels of Late Minoan civilisation. Nature, 309, 519-523. ERSKINE, W. F. 1962. Katmai. Abelard-Schuman, London. EVANS, A. J. 1921 1928 and 1935. The Palace of Minos, Vols. I II and IV. Macmillan, London. FRANCAVIGLIA, V. 1990. Sea-borne pumice deposits of archaeological interest on Aegean and eastern Mediterranean beaches. In: HARDY, D. A. & RENFREW, A. C. (eds) Thera and the Aegean World HI, Vol. 3. Thera Foundation, London, 127-134. GILL, M. A. V. 1985. Some observations on representations of marine animals in Minoan art, and their identification. Bulletin de Correspondance Hellenique, Supplement XI, 64-81. GUEST-PAPAMANOLI, A. 1983. Peche et pecheurs minoens. In: KRZYSZKOWSKA, O. & NIXON, L. (eds) Minoan Society. Bristol Classical Press, Bristol, 101-108. HOOD, M. S. 1971. The Minoans. Thames and Hudson, London. HUGHES, D. D. 1991. Human Sacrifice in Greece. Routledge, London. MARINATOS, S. 1939. The volcanic destruction of Minoan Crete. Antiquity, 13, 425-439. & HIRMER, M. 1960. Crete and Mycenae. Thames and Hudson, London. MORGAN, L. 1988. The Miniature Wall Paintings of Thera. Cambridge University Press, Cambridge. MOUNTJOY, P. A. 1984. The marine style pottery of LM I B/LH II A: towards a corpus. Annual of the British School at Athens, 79, 161-218. 1985. Ritual associations for LM I B marine style vessels. Bulletin de Correspondance Hellenique, Supplement XI, 231-242. NILSSON, M. P. 1950. The Minoan-Mycenean Religion and its Survival in Greek Religion. Kungl. Humanistiska Vetenskapssamfundet, Lund. PAGE, D. L. 1970. The Santorini Volcano and the Desolation of Minoan Crete. Society for the Promotion of Hellenic Studies, London. PASTOR, X. 1991. The Mediterranean. Greenpeace Books, Collins and Brown, London.

103

PECK, R. (trans.) 1970. Aristole: The History of Animals, Harvard University Press, Cambridge, Mass. PLATON, N. 1971. Zahros: the Discovery of a Lost Palace of Ancient Crete. Scribner, New York. POPHAM, M. R. 1967. Late Minoan pottery, a summary. Annual of the British School at Athens, 62, 337-351. 1984. The Minoan Unexplored Mansion at Knossos. Thames and Hudson, London. 1990. Pottery styles and chronology. In: HARDY, D. A. & RENFREW, A. C. (eds) Thera and the Aegean World III, Vol. 3. Thera Foundation, London, 27-28. POWELL, J. 1996. Fishing in the Prehistoric Aegean. Paul Astroms Forlag, Jonsered. REESE, D. S. 1987 Palaikastro shells and Bronze Age purple dye production in the Mediterranean basin. Annual of the British School of Athens, 82, 201-206. 1990. Triton shells from east Mediterranean sanctuaries and graves. Journal of Prehistoric Religion, 3-4, 7-14. ROLFE, J. C. (trans.) 1940. Ammianus Marcellinus: The History, Harvard University Press, Cambridge, Mass. SAKELLARAKIS, J. A. & SAPOUNA-SAKELLARAKI, E. 1981. Drama of death in a Minoan temple. National Geographic, 166, 205-222. & 1991. Archanes. Ekdotike Athenan, Athens. SIMKIN, T. & FISKE, R. S. 1983. Krahatau 1883. Smithsonian Institution Press, Washington, DC. SOLES, J. S. & DAVARAS, C. 1996. Excavations at Mochlos 1992-1993. Hesperia, 65, 175-230. , TAYLOR, S. R. & VITALIANO, C. J. 1995. Tephra samples from Mochlos and their chronological implication for neopalatial Crete. Archaeometry, 37, 385-393. STIEGLITZ, R. R. 1994. The Minoan origin of purple. Biblical Archaeologist, 57, 47-54. THOMPSON, D'A. W. 1947. A Glossary of Greek Fishes. Oxford University Press, Oxford. WACHSMANN, S. 1987. Aegeans in the Theban Tombs. Uitgeverij Peeters, Leuven. WALL, S. M., MUSGRAVE, J. H. & WARREN, P. M. 1986. Human bones from a Late Minoan I B house at Knossos. Annual of the British School at Athens, 81, 333-388. WARREN, P. M. 1991. The Minoan civilisation of Crete and the volcano of Thera. Journal of the Ancient Chronological Forum, 4, 29-39. WILSON, I. 1985. The Exodus Enigma. Weidenfeld and Nicolson, London.


Ground-penetrating radar mapping of Minoan volcanic deposits and the Late Bronze Age palaeotopography, Thera, Greece JAMES K. RUSSELL1 & MARK V. STASIUK2 1

Igneous Petrology Laboratory, Department of Earth and Ocean Sciences, University of British Columbia, Vancouver, B.C. V6T 1Z4, Canada (e-mail: [email protected]) 2 Environmental Science Division, Institute of Environmental and Biological Sciences, Lancaster University, Lancaster LAI 4YQ, UK Abstract: The Late Bronze Age (LBA) eruption of Santorini volcano deposited ash over most of the eastern Mediterranean, distributed thick deposits of pyroclastic material over the local landscape, and instantly buried the Minoan-aged living surface of these islands. Ground-penetrating radar (GPR) studies of the LBA volcanic deposits on Thera have allowed us to establish the thickness of individual pyroclastic units, to trace units laterally, and to establish facies variations in areas where the deposits are unexposed. GPR data are presented for two sites: Site A is a survey over LBA volcanic deposits exposed in the Phira quarry, immediately south of the town of Phira, and Site B is a 550m survey of the LBA deposits underlying the Akrotiri peninsula immediately north and south of the Akrotiri archaeological excavation. These traverses show that GPR can define structures as deep as 18-20m (velocity 0.1 mns" 1 ) and can accurately map the thicknesses of the LBA volcanic deposits from the caldera wall to Thera's southern coast. Furthermore, our best datasets suggest that the Phase 1 fall deposits can be differentiated from the Phase 2-4 deposits, and that GPR can clearly image the interface between the volcanic deposits and the underlying LBA living surface. Future GPR surveys could be used to delineate palaeotopographic lows and valleys associated with LBA streams or drainages or, used in combination with geological mapping, could refine the position and nature of LBA shorelines.

The present-day landscape of the Santorini islands results directly from the Late Bronze Age (LBA) or Minoan eruption of Santorini volcano. The eruption enlarged the central caldera to its present 70-80 km2 and produced a series of rhyodacitic pyroclastic deposits that blanket most of the islands (Bond & Sparks 1976; Heiken & McCoy 1984; Druitt 1990). The deposits bury the Minoan living surface to depths of tens of metres; in several instances, the pyroclastic deposits have buried several important Minoan settlements (e.g. Marinatos 1939; Sparks 1979; McCoy & Heiken 1994), the most notable of which is that near Akrotiri on the south coast of Thera. In this paper we explore the potential of ground-penetrating radar (GPR) for subsurface mapping of the LBA pyroclastic deposits on Thera. GPR data are presented for two sites (Fig. 1). Site A is a calibration survey over LBA volcanic deposits exposed in the Phira quarry, immediately south of the town of Phira, and Site

B is a regional 550m survey of the LBA deposits underlying the Akrotiri peninsula immediately north and south of the Akrotiri archaeological excavation. Our original intent was to use GPR to map volcanic stratigraphy. Specifically, we aimed to map volcanic deposits in regions that had little vertical exposure and to trace individual pyroclastic units within the subsurface, This endeavour is relevant in that the Minoan deposits are relatively undissected and, except along the crater wall or the coastlines, are poorly exposed. Stratigraphic measurements such as these can provide hard data with which to constrain ideas on the eruptive and depositional processes attending the LBA eruption. Our results also directly benefit archaeological investigations. With GPR, we are clearly able to image the interface between the LBA volcanic deposits and the underlying Minoan living surface (e.g. Vaughan 1986; McCoy et al. 1992; Camerlynck et al. 1994; Goodman 1994). Radar surveys, therefore, can show the total thickness

From: McGuiRE, W. G., GRIFFITHS, D. R., HANCOCK, P. L. & STEWART, I. S. (eds) The Archaeology of Geological Catastrophes. Geological Society, London, Special Publications, 171, 105-121. 1-86239-062-2/00/ $15.00 © The Geological Society of London 2000.

106

J. K. RUSSELL & M. V. STASIUK

Fig. 1. Physiographic map of Santorini showing the locations of surveys in relation to major geographical features. Survey sites include the Phira quarry (Site A) and a traverse across the Akrotiri peninsula (Site B). Digital elevation model cartography is courtesy of Tim Druitt.

of volcanic material overlying specific sites or horizons, can map palaeotopography in the subsurface, and can map the position and nature of the ancient Minoan shorelines (e.g. Heiken et al. 1990; Druitt & Francaviglia 1992; Forsyth 1996; Friedrich et al. 2000). The Minoan volcanic deposits The Santorini islands include Thera, Therassia, Aspronisi and the younger Palaea Kameni and Nea Kameni (Fig. 1). Thera, Therassia and Aspronisi are remnants of the LBA island that was partially destroyed and heavily modified by the cataclysmic 3600 years BP (e.g. Aitken 1988) Plinian eruption of Santorini volcano. These islands form a broken ring around the nowflooded caldera, produced by a series of collapse events associated with the Minoan and previous voluminous eruptions (Heiken & McCoy 1984; Druitt & Francaviglia 1992; Forsyth 1996). The Kameni islands are substantially younger than the LBA eruption (e.g. Fytikas et al. 1990) and are, in fact, the emergent portions of a large (2km 3 ) submarine dacitic intracaldera shield volcano. There have been at least nine subaerial eruptions between 197BC and 1950 (Fytikas et al. 1990; Druitt et al. 1996).

The Minoan eruption of Santorini produced over 36 km3 of volcanic material. Previous workers have established a stratigraphy of four mappable pyroclastic units (Phases 1-4). Below is a summary of the eruption sequence and a description of the resulting pyroclastic deposits based on the work of Bond & Sparks (1976), Heiken & McCoy (1984) and Druitt (1990), and summarized by Druitt et al. (1999). The eruption began with phreatic and phreatomagmatic explosions from a vent near present-day Nea Kameni and showered SE Thera with about 106m3 of ash (Heiken & McCoy 1984; Doumas et al. 1997). The Plinian phase of the main eruption (Phase 1) commenced at the same vent and the column attained a height of 36km (Sigurdsson et al. 1990). Phase 1 fall deposits represent c. 2 km3 of magma and occur over all of the Santorini islands to a maximum depth of 6 m. Several hours into the Plinian eruption, sea water gained access to the vent, causing violent phreatomagmatic explosions and the production of surge deposits (Phase 2). The surge deposits are laminated, cross-bedded and poorly sorted, and are themselves overlain by massive, poorly sorted deposits up to 35m thick (Phase 3). The Phase 3 deposits are thought to represent lowtemperature, fluid-rich pyroclastic flows caused by high-intensity phreatomagmatic explosions. Phase 4 deposits are dominantly fine-grained ignimbrite and are thought to derive from hightemperature pyroclastic flows; they are distributed mainly around the outer coasts of Thera, Therassia and Aspronisi. The Phase 4 deposits are fan-shaped in cross-section, in that they thicken with distance from the caldera wall (up to 40 m). One of the most notable features of the Phase 4 deposits is the abundance of lenses and layers of lithic breccias enclosed within the fine-grained ignimbrite. Lastly, overlying the entire eruption sequence, there are alluvial deposits associated with flash floods presumed to have occurred shortly after the Plinian eruption. In the vicinity of the Akrotiri excavation site, these alluvial gravel and boulder deposits have eroded and covered the Phase 4 ignimbrite. Ground-penetrating radar GPR is used extensively for imaging the shallow subsurface; the technique uses electromagnetic (EM) waves in the Megahertz (MHz) frequency range to image subsurface variations in electrical properties. Conceptually, it is the EM analogue of reflection seismology (e.g. Ardon 1985); the fundamental principles of the technique have been well described by Annan & Davis (1977) and Davis & Annan (1989).

GPR STUDIES OF VOLCANIC DEPOSITS ON THERA GPR has been used effectively in the fields of glaciology (e.g. Clarke & Cross 1989), geotechnical engineering (e.g. Ardon 1985; Holloway et al. 1986), environmental geophysics (Knoll et al 1991; Rea et al 1994) and, recently, in archaeology (e.g. Vaughan 1986; Goodman 1994; Camerlynck et al 1994; Marco et al 1997) as a complement to other geophysical surveys (e.g. Williams & Cronkite-Price 1995). The development of higher-power transmitters (e.g. 1000 V), lower-frequency antennae (e.g.

Stacks

Traces

SI(m)

Length (m)

A A B B B B

Phira quarry Phira quarry Akrotiri Akrotiri Akrotiri Akrotiri

CMP COR COR COR COR COR

CMP1 MK1E

512 512 900 512 512 512

64 256 256 128 128 128

54 283 80 200 160 192

0.2 0.5 1.0 1.0 1.0 1.0

141 79 199 159 191

MK4A MK4B MK4C MK4D

CMP, common-mid-point survey; COR, common-offset reflection survey.

108


For deeper reflectors, however, our ability to image is severely impeded because primary signals generally become weaker with depth, resulting in a low signal to noise ratio. In such instances, gain cannot be used to enhance the

primary signals without increasing the noise. We found that when we operated within lineof-sight of Profitis Ilias the background radar signal masked all reflectors at depths greater than c. 8 m.

Fig. 2. Volcanic stratigraphy of LBA deposits exposed in Phira quarry, (a) Field photograph of quarry bench that was used for 141 m GPR survey. Numbers denote positions of measured stratigraphic columns. (b) Schematic cross-section of volcanic stratigraphy shown in field photograph (after Milner 1997). (c) Stratigraphic columns for measured sections of LBA volcanic deposits (see text and Fig. 4).

GPR STUDIES OF VOLCANIC DEPOSITS ON THERA To address this interference we applied a lowpass filter to selectively eliminate the highfrequency noise associated with the background radar signal. Specifically, we chose a cutoff of 10% of the Nyquist frequency or 62.5 MHz and passed only lower-frequency information. All radar data shown in this paper had a lowpass filter applied.

Site A: Phira Quarry Immediately south of the town of Phira is an abandoned quarry situated on the eastern lip of the caldera. The quarry exposes several hundred metres of Minoan volcanic deposits (Bond & Sparks 1976; Druitt & Francaviglia 1992) lying on top of older volcanic rocks that comprise the ancient Late Bronze Age (Minoan) living surface. The LBA volcanic deposits are all pyroclastic in origin and, in this location, are proximal in nature. We chose to run a GPR survey for calibration purposes along the top of one of the quarry benches (Fig. 2a). Stratigraphic sections were measured at six locations along the front face of the bench. The locations of the sections are marked in each of Fig. 2a, 2b, and 2c. Detailed, and more general, descriptions of these deposits are available from a number of sources (Bond & Sparks 1976; Heiken & McCoy 1984; Druitt et al. 1999). The top of the bench is cut into Phase 3 pyroclastic deposits and the survey line is underlain by pyroclastic flow (Phase 3), pyroclastic surge (Phase 2), and fall deposit (Phase 1). The pyroclastic deposits are themselves underlain by well-indurated pre-Minoan volcanic rocks, including a basaltic-andesite lava breccia and well-stratified, finer-grained tuffs. Figure 2b is a sketch of the field photograph shown in Fig. 2a, and shows the distribution of the individual pyroclastic units in cross-section. The six measured stratigraphic sections are plotted in Fig. 2c. The main features of this stratigraphic section include the following: (1) the contact between the LBA volcanic deposits and the basement dips slightly to the east and lies between 9 and 14m beneath the surface of the bench. (2) Overlying the pre-Minoan volcanic deposits and the Minoan living surface is a wellsorted, massive, 6m thick fall deposit (Phase 1) comprising pumice lapilli of 1-20 cm. The largest blocks are c. 0.2m. The only structure observable in this unit is a crude coarsening upwards of pumice clasts.

109

(3) Phase 2 pyroclastic surge is 4-6 m thick, poorly sorted, moderately well laminated, and comprises ash and lapilli with rare block-sized lithic fragments. Laminations dip c. 15° to the east and merge asymptotically with the base of the unit. Many layers show prominent cross-laminations. (4) Phase 3 pyroclastic flow is massive, poorly sorted, and contains large (0.5m to >2m) blocks of lithic fragments. Phase 3 contains weak interlayers of pumice beds and has an upper lithic-rich zone. The apparent dip of this crude stratification is c. 5-10° east. (5) As a package, the pyroclastic flow and surge are crudely bedded, form sets of prograding lobes or beds, and dip slightly to the east.

Common-mid-point (CMP) analysis Two-way travel times are easily converted to depths wherever the velocity of the radar in the deposit is known. The radar velocities or the dielectric properties of these deposits are generally not known a priori (e.g. Russell & Stasiuk 1997) and, therefore, an important first task is to estimate the radar velocity of the target deposit. For a well-exposed section, as found in the Phira quarry (Fig. 2a), there are two ways to do so. A relatively simple way to constrain the velocity is by selecting a range of velocities and finding the one that gives the best 'depth' match between stratigraphic features in the deposit and corresponding reflections in the GPR profile. This method works well where the deposits are homogeneous and contain prominent stratigraphic markers that are also geophysically distinct. In the Phira quarry section, we found that a velocity of O.lmns" 1 produced a good correspondence between the measured depth to prominent reflectors (e.g. Minoan surface; Fig. 2) and the apparent depth in the radargram (e.g. Fig. 4, see below). The average radar velocity for these deposits has also been estimated with a field experiment called a CMP survey (Table 1), the results of which are summarized in Fig. 3. Figure 3a shows the GPR data plotted as distance between antennae and energy returned as a function of time. These data are displayed after application of a lowpass filter and with a moderate gain (SEC). Figure 3b shows the results of the CMP analysis in which numerous repeated traces are averaged (stacked) to produce single traces and the results are shown as a function of model velocities (0.01-0.3 mns" 1 ). The output record is time v. velocity. The appropriate velocity of the deposit is recognized by the trace with the largest

110


Fig. 3. Results of common-midpoint (CMP) survey run in Phira quarry, used to estimate a value for average radar velocity for LBA volcanic deposits. The CMP data are shown in (a) and the analysis of CMP data is plotted in (b) (see text).

amplitudes at depth. In this case (ignoring the direct ground- and air-wave data), the velocity appears to be constrained to between 0.09 and O.llmns^ 1 . We elected to use a mean value of 0.1 mns" 1 based on this survey, which is consistent with the 'stratigraphic' estimate of average radar velocity.

GPR survey results from Phira quarry Our survey of LBA volcanic deposits in the Phira quarry was 141m long, used a 0.5m sample spacing, and comprised 283 traces. The main

purpose of this survey was to calibrate the geophysical results against a well-characterized section of the Minoan pyroclastic deposits. Figure 4 shows the survey results and compares them with the stratigraphic sections prepared for the face of the quarry bench (e.g. Fig. 2a-c). Figure 4a is a radargram for the survey shown at 2.5 times vertical exaggeration. The radar data are shown after application of a dewow and the lowpass (10%) filter, and have been enhanced with a moderate SEC gain. Figure 4b is a line diagram drawn for the radargram and showing the position and shape of the

Fig. 4. Results of GPR survey of Phira quarry section, (a) Radargram (looking south); caldera rim is on the western end of the traverse, (b) Line drawing of same section showing prominent reflectors within upper (Phase 2 and 3) pyroclastic flows (bold lines), structures within underlying pre-LBA units (dashed lines labelled B) and a reflection that possibly derives from the quarry wall or other out-of-plane features (R). The shaded region denotes the upper and lower surfaces of the Plinian fall deposit as interpreted from the geophysical data.

GPR STUDIES OF VOLCANIC DEPOSITS ON THERA 111

112


prominent geophysical reflectors. Superimposed on the line diagram are lithological columns representing the measured stratigraphic sections as well as a shaded region representing the distribution of the Phase 1 fall deposit as interpreted from the geophysical data. The radar profile (Fig. 4a) shows information collected for two-way travel times of 0-400 ns; this range corresponds to a maximum depth of just over 20m (v^O.l mns" 1 ). The radargram shows strong reflectors down to 200-250 ns and weaker, but distinct, reflectors down to 350400ns. Furthermore, there is no apparent ringing (echoing of EM waves) at these depths, corroborating our decision to use a 3 m separation. Our interpretation of geological features is based on identifying continuous reflectors or horizons of reflectors that show the same polarity (black v. white). Reflectors that represent the same geological interface should show the same polarity. There appear to be three distinct packages or regions in the radargram. At times between 200 and 350ns, the radargram shows a number of continuous but weak reflectors. The reflectors (dashed lines labelled B in Fig. 4b) are variable in attitude: they dip to the east (0-40 m), or are concave upwards (40-100m), or dip to the west (110-140 m). These are 'basement' reflectors and represent geological structures in the pre-Minoan stratigraphy. The concave-upwards structures are particularly evident in Fig. 2a and b. The signals are weak relative to the more shallow parts of the radargram, but they are constant in character and on this basis are reliable. The upper portion of the radargram (0-150 ns) also has a very distinctive pattern. This stratigraphic package shows abundant strong discontinuous dipping reflectors (see Fig. 4a and b). The package, taken as a whole, represents the Phase 2 and 3 pyroclastic surge and flow deposits overlying the Phase 1 fall deposit. The most distinctive feature of this part of the radar profile is the abundant choppy, eastward-dipping reflectors seen right across the panel. This feature mirrors the characteristic laminated structure of the surge and flow deposits (Fig. 2a-c). The last zone shown in Fig. 4a is best displayed between horizontal positions 10 and 110m and at times of 130 and 190ns. This part of the radargram is characterized by little or no internal reflection; amplitudes are generally low (little black) and the reflectors that exist tend to be flat and broken-up. The pattern, where viewed from the side, is somewhat reminiscent of cross-hatching. We interpret this zone as indicative of the pyroclastic fall. The fall is homogeneous, massive, and has virtually no

internal structure. These properties are consistent with the relatively featureless character of the corresponding portions of the radar section. Our interpretation of the distribution of the Phase 1 fall, based on the radar data, is shown as a shaded region in Fig. 4b. The correlation between our interpretation and the actual distribution (see observed lithological columns) is not perfect but there are pronounced similarities in thickness, dip and position. Our interpretation of the radargram is somewhat hampered by the presence of a very strong horizontal reflector, which occurs on the eastern end of the traverse at times of 190-210ns. This reflection, labelled R in Fig. 4b, results from interaction of the EM waves with either the face of the cliff to the quarry bench or the surface of a road located at the base of the cliff-face. The GPR survey was run within a few metres of the face of the bench to optimize the correlations between the radar data and the geological features measured on the face of the quarry bench. The radar beam, however, appears not to be sufficiently focused to avoid generating these artefacts.

Site B: Akrotiri Peninsula Akrotiri peninsula comprises the southern end of Thera (Fig. 1); at its narrowest the peninsula is just over 1 km in width (Fig. 5). The LBA pyroclastic deposits form a continuous sheet across the peninsula and vertical exposures are more or less limited to the caldera wall or the southern coastline. We ran a regional-scale GPR survey across the peninsula (Table 1) in two segments as shown in Fig. 5. The traverse is oriented radially to the vent region of the 3600 years BP eruption. The position and orientation of the traverse were chosen so that the survey could sample more proximal deposits near the caldera wall (Fig. 5, no. 1), intersect the Akrotiri archaeological site (Fig. 5, no. 2), and finish at the southern shoreline, where there are more distal pyroclastic deposits (Fig. 5, no. 3). The primary aim of the regional survey was to map the distribution of LBA volcanic deposits in the subsurface between points of known stratigraphy (Fig. 5, nos 1 and 3). Mapping of young volcanic deposits commonly relies on trust: stratigraphy derived from a few well-exposed sections is used to extrapolate across regions with little or no stratigraphic exposure. Volcanic stratigraphy is notoriously complex, however, and this practice can be very

GPR STUDIES OF VOLCANIC DEPOSITS ON THERA

113

Fig. 5. Detail location map for GPR survey of Akrotiri peninsula showing locations of individual segments of survey line with respect to: (1) stratigraphic section measured on rim of caldera, (2) archaeological excavation site, and (3) stratigraphic column measured on south coast of peninsula. dangerous without substantial information from areas between sections. Regional-scale GPR surveys can be used to test and strengthen these extrapolations. In addition to elucidating the distribution of the LBA volcanic deposits, our survey has the ability to define and trace the palaeotopography of the Minoan living surface. In this survey, the results should be of import to researchers working on the Akrotiri excavation site because our results define the thickness of the LBA volcanic deposits and provide physical descriptions of the hinterland to the Minoan townsite. Potentially, the survey can also recover the ancient coastline.

Volcanic stratigraphy of Akrotiri peninsula Our geophysical survey was constrained at two locations where we measured the stratigraphy of LBA volcanic deposits. We measured a complete section through the LBA deposits exposed in the caldera wall on the northern margin of Akrotiri peninsula. The location of section II-1 is shown in Fig. 5 (no. 1). Another section (ML-2) was measured on the southern margin of the peninsula just south of the archaeological excavation (Fig. 5, no. 3). The southern section comprises over 30m of LBA pyroclastic material and

clearly represents more distal deposits that were accreted to Thera during the 3600BP eruption. Both stratigraphic sections are summarized in Fig. 6. At the caldera wall there are 12m of LBA volcanic deposits resting unconformably on the pre-Minoan surface. In this location the basement comprises Cape Riva ignimbrite (Druitt 1985; Druitt & Francaviglia 1992). Immediately above the Cape Riva formation is 2 m of Phase 1 fall deposit. Phase 2 surge overlies the fall; it is strongly laminated and contains abundant redoxidized lapilli. The upper 7 m of the section is made of poorly laminated Phase 3 pyroclastic flow (c. 5 m) capped by just over 2 m of Phase 4 ignimbrite. The Phase 4 ignimbrite comprises a tan-coloured ash matrix containing abundant lithic fragments. The unit shows a crude stratification defined by concentrations of block-sized dense lava lithic fragments. The concentrations of lithic fragments form discrete horizons, which show substantial topography and are commonly channel-like in form. The top of the section is more or less even in elevation with the north end of the GPR traverse (e.g. MK4D, Table 1). The stratigraphic sequence measured at the southern coastline (ML-2; Fig. 6) is substantially different. There is a depth of over 30m of exposed LBA volcanic deposits and, although the section is continuous to sea level, the

114


Fig. 6. A comparison of LBA volcanic stratigraphy measured for the northern edge of the Akrotiri peninsula and situated on the south rim of the caldera (no. 1: Fig. 5), and for the southern shoreline of the Akrotiri peninsula and immediately south of the archaeological excavation site (no. 3: Fig. 5). Sections compare thicknesses of pyroclastic fall and flow (Pf) deposits and show distribution of lag breccias (Bx).

GPR STUDIES OF VOLCANIC DEPOSITS ON THERA pre-Minoan surface is not exposed. Phase 1 and 2 deposits are absent from the section. The base of the section is made up of slightly over 3 m of Phase 3 pyroclastic flow; the unit is massive and made up of lithic and pumice lapilli in a fine-grained ash matrix. Overlying Phase 3 and making up most of the section (>24m) is Phase 4 ignimbrite. The majority of the unit is massive, fine-grained, poorly sorted material with lenses and lithic-rich layers or domains. These layers mainly comprise dense lava clasts, which range in size from lapilli to blocks. The layers can be continuous and define a slight (25m), combined with their proximity to the partially excavated Minoan settlement (which establishes a relative elevation for the Minoan living surface), suggests a significant elevation drop (20-30 m) over a narrow horizontal distance (0.95; 'marginally similar' is 0.93-0.94.

THE BROOKS RIVER ARCHAEOLOGICAL DISTRICT and the underlying deposit II-C. Probably most of the contamination occurred by biological activity or frost churning, as the amount of inter-tephra sediment between most of the tephra deposits at our sample sites is minimal (Fig. 5). To try to confirm resampling of Dumond B at a fourth locality we sampled a deposit between the base of 1912 tephra and the obvious correlative of Dumond C (as well as the correlative of C 'Ash C'). The sample 'Ash B(?)' is highly similar to III-C1 (correlative of Dumond C), whereas the upper part of 'Ash C' is similar to III-C3 and the base of 'Ash C' is highly similar to III-C2 and III-C3. Thus, as at Site II, 'Ash B(?)' at the fourth site is either the reworked top of the underlying Ash C or is a closely succeeding deposit of the same magma as formed Ash C. Because the presence of a separate tephra deposit between Dumond C and the 1912 Katmai tephra has been unequivocally shown at some archaeological sites (e.g. Dumond 1981, fig. 6.23; Harritt 1988, table 1), we do not dispute the existence of Dumond B. However, we concur with Nowak (1968, p. 33) that Dumond B is 'rarely distinct enough to be easily recognized'. Most of the identifiable components in the eight analysed BRAD deposits are highly similar to pyroclasts from Aniakchak Crater, 160km southwest of the BRAD (Fig. 1). Aniakchak has had the largest number of major tephra-forming eruptions during Holocene time of any volcano we have studied on the Alaska Peninsula: at a minimum, six during late Holocene time, nine during a period of intense activity that culminated with caldera formation at 3460 a BP (Miller & Smith 1987), four to six just before the onset of the caldera-forming activity, and several more during early Holocene time. Prevailing winds to 30 000 ft above the Alaska Peninsula are northeasterly (Fig. 3), so an abundance of Aniakchak deposits in the BRAD is not surprising. BRAD deposits II-D, II-I and II-J have major components that cannot be assigned to specific sources, and other deposits have minor components of unknown sources (Table 5). These unknown components may have originated at the Katmai-group volcanoes: if their sources

255

were more distant volcanoes, at least some coarse-grained correlatives should have been identified nearer the sources. In any case, few of these deposits of unknown sources are similar to any Site 16 glasses, which suggests that Site 16 deposits represent yet additional eruptions of Katmai volcanoes. It is readily understandable why Nowak (1968) and Dumond (1979) concluded that, with the exception of Dumond C (and, of course, the 1912 tephra), the BRAD tephra deposits are not reliably distinguishable from one another. Each is a mixture of multiple components. Whatever distinguishing characteristics the individual components may have had initially, they have surely lost by mixing.

Correlations of tephra deposits among BRAD sites A requirement of a National Historic Landmark is that, to preclude loss or damage of resources, all ground disturbances be preceded by competent archaeological assessment. Thus, as Katmai Park managers strive to maintain adequate visitor facilities in the face of growing Park visitation, the frequency of archaeological assessment of small areas has increased. Such limited compliance assessments do not have the scope of a larger investigation that might provide a critical stratigraphic context or that might yield datable organic materials. The ability to identify a specific tephra deposit, especially if its age were known, would be highly useful in these compliance studies. To test the potential applicability of tephra correlations to archaeological research at the BRAD, we sampled and analysed 10 'unknown' tephra deposits from four study sites (Fig. 6). One is the drainfield site east of Brooks Lake on a 19m beach-ridge terrace (site IV, Fig. 2). The other three sites are located within 100m of one another, on emergent 13.5-14.5m beach ridges of Naknek Lake near Ranger Headquarters (Sites V, VI and VII, Fig. 2). Analytical results for these samples are not reported because the question is only whether these samples correlate with other BRAD

Fig. 5. Holocene deposits at three sites (see Fig. 2 for locations) in the Brooks River Archaeological District, (a) Measured section at Site II on a c. 7000 year old, 22m emergent terrace. Capital letters to the right refer to analysed samples discussed in this paper, (b) Holocene tephra deposits at Site I (where they are better exposed than at Site II). The ofT-white, sandy material beneath modern sod at the top is 20cm of 1912 Katmai tephra. Labels are tephra deposits 'C, D, F, G, I, J'. (c) Exposure of Ash C (beneath the thumb) along the Brooks River (Site III). The stratification of Ash C into a pale yellow-grey base, a greenish grey centre and a greyish brown top should be noted. The three strata were separately sampled as Cl, C2 and C3 (Table 4). One-half metre of poorly sorted material beneath Ash C is cultural refuse of a pre-Ash C occupation; Ash C is overlain by a thin soil, which in turn is overlain by post-Ash-C refuse of a younger occupation. The 1912 Katmai tephra extends from top of fingers to top of bank.

Table 4. Major-element compositions of glass in tephra deposits from Sites II (all except C) and III (C) in the Brooks River Archaeological District

Bl

B2

B3

B4

B5

B6

CM

Cl-2

Cl-3

Cl-4

Na2O MgO A12O3 Si02 K2O CaO Ti02 MnO FeOT

3.98(2.8) 0.26(12) 12.2(2.0) 75.7(1.0) 2.85(2.6) 1.26(7.7) 0.32(18) 0.05 1.61(6.4)

3.83(3.7) 0.14(24) 11.9(2.1) 76.7(1.0) 3.20(2.8) 0.89(20) 0.20(42) 0.04 1.17(11)

3.92 0.35 12.6 74.9 2.72 1.53 0.33 0.04 1.78

4.67(3.5) 1.09(5.4) 14.7(0.8) 66.6(0.3) 2.84(2.0) 3.12(0.8) 0.97(4.5) 0.19 4.69(4.8)

4.49 2.09 14.1 62.4 2.26 4.18 1.23 0.23 7.17

4.26 2.74 15.4 58.6 1.55 6.34 1.07 0.18 7.40

4.11(2.0) 2.74(4.9) 16.2(1.8) 58.7(1.1) 1.67(4.3) 6.18(3.6) 1.18(11) 0.19 7.56(4.7)

4.45(3.5) 2.29(16) 15.5(3.6) 61.1(0.8) 2.00(8.6) 4.92(11) 1.05(22) 0.19 7.22(4.4)

4.42(4.8) 1.69(6.0) 15.3(3.1) 63.3(1.4) 2.28(4.3) 4.08(3.9) 1.01(14) 0.15 5.91(11)

4.34(7.5) 0.89(2.7) 15.0(4.2) 67.4(1.7) 3.11(16) 2.52(24) 0.73(16) 0.16 3.67(10)

4.05(3.7) 2.77(8.4) 16.2(2.2) 58.1(2.0) 1.59(7.5) 6.37(5.9) 1.11(13) 0.19 7.70(6.1)

Total

98.2(16)

98.1(5)

98.2

98.9(3)

98.2

97.5

98.5(3)

98.7(5)

98.1(5)

97.8(4)

98.1(18)

C2-2

C3-1

C3-2

C3-3

C3-4

Dl

Na2O MgO A1203 Si02 K2O CaO Ti02 MnO FeOT

4.62 1.23 16.0 66.2 2.67 3.38 0.71 0.17 4.51

4.15(3.9) 2.74(8.2) 16.3(1.4) 58.8(1.8) 1.59(8.9) 6.24(5.4) 1 .12(6.3) 0.19 7.53(8.4)

4.83(2.7) 0.81(6.6) 15.2(1.8) 68.8(0.5) 2.69(3.0) 2.54(5.3) 0.70(13) 0.16 3.36(5.6)

3.88 2.92 15.7 55.7 1.47 6.6 1.20 0.20 8.19

4.74 1.85 15.7 63.0 1.82 4.55 0.89 0.21 5.67

4.61(10) 0.55(19) 14.4(4.8) 70.7(2.4) 2.70(9.5) 2.00(18) 0.53(19) 0.09 2.48(13)

Total

99.6

98.7(9)

99.1(7)

95.9

98.4

98.1(15)

D2 4.07 0.16 12.5 78.3 3.11 1.07 0.26 0.06 1.28 100.8

C2-1

D3

D4

El

E2

E3

E4

4.48 1.64 15.6 61.4 2.18 4.43 1.19 0.20 7.22

3.92 0.34 12.7 74.3 2.81 1.73 0.43 0.06 2.09

4.59(11) 0.57(20) 14.2(4.8) 70.9(1.2) 2.46(11) 2.23(17) 0.56(21) 0.10 2.62(12)

4.71(4.5) 1.10(5.6) 15.8(0.4) 67.2(0.5) 2.45(3.8) 3.21(0.5) 0.76(9.4) 0.16 3.81(1.4)

4.27(3.7) 2.85(5.0) 16.4(0.6) 58.1(1.5) 1.50(1.7) 6.27(3.7) 1.34(8.0) 0.21 7.02(6.1)

4.56 1.36 15.4 65.1 2.55 3.58 0.98 0.20 4.89

98.4

98.4

98.2(7)

99.2(4)

98.0(3)

98.7

E5

Fl

F2

F3

F4

F5

F6

¥1

Gl

G2

G3

HI

H2

Na2O MgO A1203 Si02 K20 CaO Ti02 MnO FeOT

3.84 0.38 12.8 74.6 2.85 1.80 0.45 0.05 2.00

4.17(3.6) 2.70(6.6) 15.8(1.8) 57.2(1.7) 1.58(4.2) 6.11(5.8) 1.35(5.3) 0.22 8.55(7.0)

4.31(5.4) 2.16(13) 16.3(6.6) 59.6(0.1) 1.79(14) 5.56(9.3) 1.19(8.7) 0.18 7.26(9.0)

4.27(7.0) 1.89(13) 15.1(4.0) 61.0(1.0) 2.18(2.8) 4.69(3.7) 1.31(7.1) 0.16 7.70(12)

4.35 1.22 14.5 63.6 2.69 3.61 1.15 0.15 6.65

4.16 1.48 13.5 69.4 2.15 3.26 0.53 0.10 3.62

4.02(2.6) 0.58(6.2) 13.5(0.7) 71.3(1.4) 2.41 (0.7) 2.37(4.9) 0.50(20) 0.06 2.68(6.2)

3.57 0.40 12.9 74.5 2.85 1.76 0.41 0.05 1.90

4.76(2.5) 0.39(9.7) 14.5(1.4) 70.8(0.7) 3.25(5.3) 1.54(5.1) 0.38(23) 0.12 2.42(3.6)

3.76(3.6) 0.30(16) 12.5(1.4) 75.1(0.6) 2.73(19) 1.52(15) 0.32(21) 0.03 1.65(8.1)

4.79 1.20 15.3 66.3 2.55 3.20 1.00 0.15 4.45

4.66(5.7) 1.44(22) 15.5(1.2) 63.9(1.6) 2.40(7.6) 3.84(3.7) 1.04(10) 0.15 4.98(12)

4.89 0.78 14.7 69.2 3.01 2.20 0.76 0.15 3.46

Total

98.8

97.7(4)

98.4(3)

98.3(4)

97.9

98.3

97.4(3)

98.3

98.2(7)

97.9(5)

98.9

97.9(7)99.2

H3

H4

11

12

13

Jl

J2

J3

J4

J5

J6

J7

Na2O MgO A1203 Si02 K2O CaO Ti02 MnO FeOT

3.95(9.0) 0.65(15) 13.3(5.4) 72.7(1.1) 2.52(9.1) 2.27(15) 0.54(12) 0.08 2.73 (9.7)

4.67 0.93 15.5 65.5 2.80 3.29 1.09 0.15 4.49

4.17(0.1) 0.22(5.2) 11.8(0.6) 75.8(0.9) 2.95(1.4) 1.07(1.6) 0.32(17) 0.04 1.64(4.3)

4.32(3.3) 0.67(12) 13.6(2.0) 71.0(1.3) 2.37(2.2) 2.44(9.5) 0.57(10) 0.08 2.93(5.5)

4.16(8.9) 0.48(11) 12.9(1.1) 73.2(0.9) 2.62(9.7) 2.06(16) 0.51(8.2) 0.06 2.34(7.2)

5.12 0.84 16.0 64.0 2.28 4.28 0.83 0.08 4.78

4.46(5.6) 0.66(8.8) 13.6(2.2) 70.6(1.0) 2.35(3.0) 2.48(7.2) 0.58(11) 0.09 2.73(7.3)

3.19(15) 0.06 12.6(7.8) 73.8(3.3) 4.91(16) 0.83 (60) 0.18 0.01 0.60(20)

4.36 0.54 13.3 72.1 2.52 2.13 0.58 0.05 2.47

4.01 0.12 12.0 77.0 3.24 0.82 0.30 0.03 1.44

3.42 0.15 11.4 74.2 3.76 0.89 0.22 0.03 1.15

4.06 0.35 12.5 73.1 2.90 1.85 0.65 0.03 2.62

Total

98.7(4)

98.4

98.0(3)

98.0(10)

98.3(4)

98.2

97.6(13)

96.1(3)

98.1

99.0

95.2

98.1

Details same as in explanation for Table 2 except that mafic-phenocryst ratios are not reported because all samples are mixtures of multiple components, and standard deviations are not reported for components having fewer than 3 analytical points.

258

J. R. RIEHLE ET AL.

Table 5. Comparison of samples from the Alaska Peninsula with samples of Brooks River tephra deposits at Sites II and III, based on similarity of glass compositions Deposit B (historical): Bl and B3 are similar to 1912 dacitic Katmai tephra and B2 is similar to 1912 rhyolitic tephra. B3 is also highly similar to Deposit G2. B4 (3 points) and B6 (2 points) are similar to components of upper Holocene deposits south of the Katmai region that are correlated with Aniakchak Crater; B6 is also highly similar to underlying Deposit Cl. We conclude that B at Site II is chiefly 1912 tephra; there is also a minor amount of tephra similar to that of Aniakchak Crater but whether this is a separate tephra fall similar to underlying Deposit C or is contamination by Deposit C is unknown. Deposit C (about 400 years old): The major component of each of 3 subsamples (Cl-1, C2-1 and C3-1) is highly similar to many samples of upper Holocene deposits of Aniakchak Crater. The other components of Cl, C2 and C3 are similar to some components of these, or other, Aniakchak deposits. We conclude that the 3 layers of deposit C represent 3 closely succeeding, late prehistoric deposits of Aniakchak Crater. Deposit D (900-1200 years old): Dl and D2 are similar to no other samples (except underlying El). D3 is marginally similar to a few deposits of several volcanoes on the southern Alaska Peninsula. D4 is similar to several, upper Holocene Katmai-area deposits and to Dumond E5 and F7. We conclude that deposit D is a Katmai-area deposit of unknown source. Deposit E (1200-1800 years old): El is similar only to overlying Dl. E2, E3 and E4 are similar to middle and late Holocene deposits of Aniakchak Crater. E5 and D4 are similar to a few local deposits. We conclude that the two deposits represent a slightly younger tephra of local origin (D) and a slightly older deposit of Aniakchak Crater (E). Deposit F (2100-3100 years old): Fl is highly similar to several samples of the mid-Holocene, caldera-forming eruptions of Aniakchak Crater, as well as other, lower and upper Holocene Aniakchak deposits. F2 is similar to a few deposits of Aniakchak Crater of a range of ages. F3 is highly similar to two middle Holocene deposits of Aniakchak Crater. F4 and F5, both one-shard analyses, are only marginally similar to a few other samples. F6 is similar to only one, lower Holocene deposit from elsewhere in the Katmai region and is marginally similar to the 1912 andesitic fallout. F7 is similar to two other, upper Holocene deposits in the Katmai region and to Dumond E5 and D4. We conclude the deposit is largely or entirely tephra of the 3460 year BP, caldera-forming eruptions of Aniakchak Crater or a closely succeeding Aniakchak eruption. There may also be a minor amount of locally derived tephra (F4-F7; but only F6 consists of more than two analytical points). Deposit G (3800-4000 years old): Gl is similar to a few middle Holocene deposits of Aniakchak Crater. G2 is similar only to 1912 dacite and to a few other Brooks River samples. G3 is marginally similar to a few mid- and upper Holocene deposits of Aniakchak Crater. We conclude the deposit is a mixture of Aniakchak and locally derived tephra. Deposit H (3800-4400 years old): HI is highly similar to some middle Holocene, pre-caldera-forming deposits of Aniakchak Crater and (or) of Black Peak (about 4000 yrep). H2 is similar to some middle and upper Holocene deposits of Aniakchak Crater. H3 is similar to some middle Holocene, Katmai-region deposits including F6. H4 is marginally similar to only 3 other samples. We conclude that deposit H is mainly middle Holocene, Aniakchak tephra and possibly a minor amount of a local tephra. Deposit I (roughly 5500 years old): II is marginally similar to 1912 dacitic tephra, to 16-F1, and to overlying B2 and Dl. 12 is similar to the Lethe tephra, to several middle Holocene deposits in the Katmai region, and to two other, upper Holocene deposits at Brooks River, as well as to overlying H3 and underlying J2.13 is marginally similar to only 16-E and to Dumond J4. We conclude that this is a middle Holocene deposit of local origin. Deposit J (roughly 6500 years old): Jl and J3 are similar to no other samples. J2 is highly similar to the Lethe tephra, to a few middle Holocene, Katmai-region deposits, and to Dumond 12. J4 is marginally similar to 1912 andesitic tephra, to a lower Holocene deposit 80km north of Katmai, and to overlying Dl, El, H3 and 13. J5 is marginally similar to only overlying B2 and J6 is marginally similar to 16-K1. We conclude that this is a lower Holocene deposit of local origin. 'Similar' means a similarity coefficient >0.95; 'marginally similar' is 0.93-0.94. Ages refer to the original Dumond deposits (1979) (see Table 1), which may or may not correlate with samples from Sites II and III.

samples (analyses available from J. Riehle). Of the 10 'unknown' samples, only three, VI-B, VI-C and V-H, have s.c. values >0.95 with one another (Table 6). Two of these samples are succeeding deposits at Site VI that clearly cannot be the same deposit. The poor degree of correlation among these closely adjacent sites

means that not all tephra falls are reliably preserved at every BRAD site, as a result of local, minor unconformities in combination with the thin, fine-grained nature of the deposits. Moreover, only V-C1, V-H, VI-B1 and VI-C1 are similar to a major component of a Site II deposit, Thus, these 10 'unknown' deposits include at

Table 6. Similarity coefficients for pairs of some upper Holocene tephra deposits from the Brooks River Archaeological District, Alaska IV-D

IV-D IV-E1 IV-E2 IV-F IV-H VII-C1 VII-C2 VII-E VI-B1 VI-B2 VI-C1 VI-C2 V-C1 V-C2 V-H

IV-E1

IV-E2

IV-F

IV-H

VII-C1

VII-C2

VII-E

VI-B1

VI-B2

VI-C1

VI-C2

V-C1

V-C2

V-H

0.96

0.97

0.96 0.97

Only values >0.95 are included because lower values typically preclude correlation as the same deposit. (To find the coefficients of a particular sample with every other sample, first read down the column for that sample until the dash is intersected, then continue along the row to the right. For locations of sample sites, see Fig. 2; for stratigraphic settings, see Fig. 6.) Samples that are similar to the original Dumond deposits (Table 4) are listed below because the table is not large enough to include all components of these deposits. Samples '16' are from Site 16 in the Valley of Ten Thousand Smokes (Table 2). IV-D, similar to II-E5 and II-F7 (neither is a major component); IV-E1, marginally similar only to II-C1-3; IV-E2, no similarity coefficient >0.94 with any other sample; IV-F, similar to a number of Aniakchak deposits, including highly similar to some upper Holocene deposits; IV-H, highly similar to some middle Holocene Aniakchak deposits; VII-C1, similar to some Aniakchak deposits and to 16A1; VII-C2, similar to two, upper Holocene Aniakchak deposits; VII-E, similar to two, lower Holocene deposits to the south of the BRAD; VI-B2, marginally similar only to 16H; VI-B1, highly similar to Lethe tephra, 11-12 and II-J2, and two lower Holocene deposits from the Katmai region (see V-H); VI-C1, same as VI-B1; also similar to 16G3; VI-C2, similar to several middle Holocene Aniakchak deposits and to II-H1; V-C1, similar to several lower Holocene Aniakchak deposits and to II-C1-3; V-C2, similar to two, upper Holocene Aniakchak deposits; V-H, similar to Lethe tephra, to some lower Holocene Katmai-region deposits, and to 11-12 and J2; correlation with 11-12 or J2 or with the lower Holocene deposits is clearly impossible, because of the much greater age of these deposits (>5500 years BP) compared with the age of emergence at site V (>2500 years BP).

260

J. R. RIEHLE ET AL.

Fig. 6. Upper Holocene deposits at four sites of archaeological investigations, Brooks River Archaeological District; locations are plotted in Fig. 2. The age at the base of sections 5, 6 and 7 is not known with certainty, but based on the site elevations within 4 m of the present surface of Naknek Lake, cannot be more than about 3000 years old (Dumond 1981, fig. 2.4). Letters to the right of each section are labelled samples discussed in the text.

least six deposits in addition to the original Dumond deposits.

Biological effects of the 1912 Katmai eruption The record of prehistoric occupation at Brooks River affords an opportunity to study the longterm response of humans to tephra falls. Dumond (1979) noted the lack of correlation of periods of abandonment of the BRAD with times of tephra eruptions, and concluded that the volumes of these tephra falls were insufficient to have had a significant impact on humans there.

The 1912 Katmai eruption provides a useful model to illustrate the relationship between thickness of tephra fall and the significance of its impact on humans, animals and vegetation. The 1912 Katmai eruption comprised three main eruptive pulses that occurred over a period of 60 h, beginning on 6 June (Hildreth 1983). Winds to high levels were mainly east directed during this time, and the main axis of the tephra deposit extends over northern Kodiak and Afognak Islands (Fig. 7; Fierstein & Hildreth 1992). The ashflow in the VTTS was emplaced during the first eruptive pulse; some of the apparent ashfall material within 10km of the VTTS is actually surge deposits that were swept

THE BROOKS RIVER ARCHAEOLOGICAL DISTRICT

261

Fig. 7. Thickness of tephra deposited during 1912 eruption of Novarupta (Katmai) volcanoes, and representative effects on humans, animals and plants related to tephra thickness.

from the moving ashflow, rather than being true ashfall (that is, carried aloft by the uprising eruption column). Both the ashflow and the ashfall material were erupted from the Novarupta vent at the head of the VTTS, not from Katmai Caldera, which did, however, collapse as a result of withdrawal of support because of eruption of magma from Novarupta vent (Curtis 1968; Hildreth 1983, see fig. 3). Humans directly affected by the eruption were located mainly in four areas: Katmai village, 30 km southeast of Novarupta (Fig. 8); a pair of settlements on the Ukak and Savonoski Rivers of which one was sited near the foot of the ashflow deposit, 20km northwest of the vent; Douglas village, 80km to the east; and Kodiak village, 180 km east (Fig. 7). Most of the Douglas and Katmai village inhabitants were at summer fishing camps on Kaflia Bay. The few Katmai villagers left behind fled in fear early on 6 June as the frequency of premonitory earthquakes increased. They were in bidarkas (kayaks) at Cape Kubugakli when the eruption occurred.

Remarkably, no known deaths were directly attributable to the eruption (Martin 1913; Griggs 1922) despite its size, the largest of the twentieth century (estimated Volcano Explosivity Index of six; Simkin et al 1981), and the proximity of people to the vent. There are several reasons for the lack of fatalities. First, none were close enough to be burned by contact with tephra or by the ashflow or its surge, unlike victims 20-30 km from Mount St Helens in 1980 (Rosenbaum & Waitt 1981). Individuals fleeing Savonoski and Katmai villages reported heat from falling tephra, but they were able to continue their journeys until they were beyond the tephra fall (Griggs 1922, p. 17). Second, although there was flooding on drainages heading in the vicinity of Mount Katmai or the VTTS (Griggs 1922; Hildreth 1983), flood volumes were insufficient to damage either Savonoski village or Katmai village. Third, individuals in the zone of heaviest tephra fall were rescued before serious dehydration or starvation. People at Kaflia Bay were rescued by the

262

J. R. RIEHLE ET AL.

Fig. 8. Katmai Village 3 years after the 1912 eruption. Because of the thick primary tephra fall (nearly 1 m at this site only 30 km from the source) and to subsequent instability of the Katmai River channel, the site was abandoned. (Photograph by G. C. Martin, in Griggs 1922.)

steamer Redondo 3 days after the end of the eruption (Revenue-Cutter Service 1913, p. 125), and those from Savonoski village paddled their bidarkas for 2 days to Naknek, beyond significant tephra fall 60km to the west (Fig. 7). The effects of the tephra on animals and vegetation, of major importance to a subsistence culture, depend critically on thickness of tephra and the particular species. Virtually all animals either died or left areas of heaviest tephra fall on the Alaska Peninsula; Griggs (1922) observed bear and fox tracks on the beach of Katmai Bay in 1915, but the only signs of animal life in the upper reaches of the Katmai River valley were occasional birds. By 1919, however, nesting birds, mice and ground squirrels in the same area had increased notably (Griggs 1922, p. 164). Cattle at Kodiak, where 20cm of tephra fell, survived the eruption but then were removed from the island until pastures had fully revegetated 2 years later (Griggs 1922, p. 44). In general, mammals in the areas of heaviest tephra on Kodiak Island were not seriously affected except for malnutrition (Evermann 1914), although smaller mammals may have suffered more heavily than larger ones (Erskine 1962). Caribou on the Alaska Peninsula abraded their teeth on volcanic ash to the point of starvation after a relatively minor eruption of Aniakchak Crater in 1931; many new-born calves were lost as the

herd migrated from the area of ashfall (Trowbridge 1976). Probably the same occurred in areas of even light tephra on the Alaska Peninsula after the 1912 eruption. Caribou graze on low-standing mosses and vegetation whereas moose browse on shrubs and young trees, which are more easily cleared of tephra by wind and rain. Effects of tephra on salmon are complicated because different species (mainly coho, sockeye and pinks in this part of Alaska) have different life cycles and spawning ages, and require different types of spawning beds. Salmon were just beginning to enter streams on the Peninsula and in the Kodiak Islands at the time of the eruption. Those in areas of more than about 10cm tephra either suffocated in tephra-laden waters or returned to the sea, from which they periodically attempted to re-enter the streams (Evermann 1914). Streams in areas of lighter tephra generally cleared in time to permit late spawners to enter in 1912, whereas those in areas of heavy tephra on the Peninsula were still unsuitable for spawning because of eroding tephra and unstable beds as much as 5 years after the eruption (Griggs 1922, p. 161). Young salmon spend their first few months to 3 years in fresh water; those in deeper streams, or sockeyes in lakes at the heads of streams, had better odds of survival than those in shallow


263

Fig. 9. In areas of less than about 25 cm of tephra fall, some species such as this Devil's Club survived by extending from old root systems to the surface of the tephra. The new roots developing just below the new ground surface should be noted. (Photograph by R. F. Griggs, in Griggs 1922.)

streams or in areas of heavier tephra. Survival depends partly on the food supply, which also affects nonmigratory species such as trout; some streams on western Afognak Island (Fig. 7) were devoid of fish food a year after the eruption (Evermann 1914). The impact of the eruption on the salmon resource was not fully realized, however, until several years after the eruption. Sockeye returns to the Kodiak Islands began to decline in 1915 (pink salmon spawn after 2 years, sockeyes and coho, after 3-6 years) and continued to decline until 1920 (Eicher & Rounsefell 1957). Thus, immediately after the eruption, salmon were still

available as a food resource although opportunities to take them from streams in areas of 20cm tephra or more were probably limited. Additionally, marine shellfish were killed in significant numbers in areas of heavy tephra, and even cod were reported to have left traditional grounds near Kodiak Island (Evermann 1914, p. 62). The details of vegetative recovery, recorded by botanist Robert Griggs (1922), are as varied as was the impact on salmon. In areas of heavy tephra, tree limbs were broken, grasses and shrubs were buried, and landslides and floods excavated or deeply buried trees growing in

264

J. R. RIEHLE ET AL.

Fig. 10. Willows were able to develop new (adventitious) roots at the surface of the tephra, even when buried to depths of 2m. Here, the roots were exposed by subsequent erosion of the tephra. (Photograph by D. B. Church, in Griggs 1922.)

floodplains and at the foot of steep slopes. On Kodiak Island, grasses and shrubs had recovered within 2-3 years; recovery was not as much by reseeding as by sending up shoots from existing root systems (Fig. 9). Some plants survived as much as 3 years of burial (Griggs 1922, p. 51). Some willows (Fig. 10) survived even deep burial by sending out adventitious roots just below the new ground surface. Experiments confirmed that the tephra had few available nutrients but would support plant growth if properly fertilized. A striking demonstration of this characteristic was provided by a slow-growing tree from a Kodiak bog, which in the first few years after the eruption more than doubled its diameter (Griggs 1922, p. 53). The benefit of the tephra to the tree was due not to fertilization, but instead to the temporary suppression of grasses at the site, which competed for available nutrients. Summary and conclusions (1) Because of its proximity to the active Aleutian volcanic arc, the Brooks River Archaeological District has experienced numerous

tephra falls throughout the past 4500 years of human occupation. At least 15 prehistoric, Holocene tephra deposits, the oldest about 6500 years, can be recognized. (2) One deposit (Dumond C about 400 years old) is sufficiently thick, distinctive and reliably preserved to be useful as a marker bed. The other deposits, however, are thin and fine grained, and either are not preserved at every locality or are less readily distinguishable in the field on the basis of their megascopic characteristics. (3) Microprobe analyses of glass separates show that most of the deposits are chemically heterogeneous, either the result of eruption of mixed or heterogeneous magma as occurred during the 1912 eruption, or because of mixing of succeeding tephra deposits by biological activity and frost action. (4) About half of the tephra deposits originated at Aniakchak Crater, 160km to the southwest and one of the most frequently active volcanoes on the Alaska Peninsula. Other deposits consist mainly of glass that is either similar only to known deposits of Katmai volcanoes, or is unique to the Katmai region. Because of the heterogeneous nature of these tephra deposits


and the lack of widespread marker deposits, specific ages and sources for most of the prehistoric Katmai deposits cannot be identified at this time. (5) Floral and faunal recovery from the effects of the 1912 eruption of Katmai (Novarupta) volcano varied in detail, depending on the thickness of the tephra and on species-specific characteristics. Thus, effects of volcanic eruptions on ancient subsistence populations may be expected to have varied from case to case. In general, however, recovery was rapid in areas that had less than 10cm of ashfall, which explains the lack of correlation between ashfalls and periods of abandonment of the Brooks River site. References BORCHARDT, G. A., ARUSCAVAGE, P. J., & MILLARD, H. T., JR 1972. Correlation of the Bishop ash, a Pleistocene marker bed, using instrumental neutron activation analysis. Journal of Sedimentary Petrology, 42, 301-306. CARMICHAEL, I. S. E. & MACDONALD, A. 1961. The geochemistry of some natural acid glasses from the North Atlantic Tertiary volcanic province. Geochimica et Cosmochimica Acta, 25, 189-222. CURTIS, G. H. 1968, The stratigraphy of the ejecta from the 1912 eruption of Mount Katmai and Novarupta, Alaska. In: COATS, R. R., HAY, R. L. & ANDERSON, C. A. (eds) Studies in Volcanology. Geological Society of America, Memoir, 116, 153-210. DOWNES, H. 1985. Evidence for magma heterogeneity in the White River Ash (Yukon Territory). Canadian Journal of Earth Sciences, 22, 929-934. DUMOND, D. E. 1964. Archaeological Survey in Katmai National Monument, Alaska 1963. Report to the National Park Service, Western Region. Department of Anthropology, University of Oregon. 1979. People and pumice on the Alaska Peninsula. In: SHEETS, P. D. & GRAYSON, D. K. (eds) Volcanic Activity and Human Ecology. Academic Press, New York, 373-392. 1981. Archeology on the Alaska Peninsula: the Naknek Region, 1960-1975. University of Oregon, Anthropological Papers, 21. EICHER, G. J., JR & ROUNSEFELL, G. A. 1957. Effects of lake fertilization by volcanic activity on abundance of salmon. Limnology and Oceanography, 2, 70-78. ERSKINE, W. F. 1962. Katmai. Abelard-Schuman, London. EVERMANN, B. W. 1914. Alaska Fisheries and Fur Industries in 1913. Report of the US Commissioner of Fisheries. US Government Printing Office, Washington, DC. FIERSTEIN, J. & HILDRETH, W. 1992. The plinian eruptions of 1912 at Novarupta, Katmai National Park, Alaska. Bulletin of Volcanology, 54,646-684.

265

GRIGGS, R. F. 1922. The Valley of Ten Thousand Smokes. National Geographic Society, Washington, DC. HARRITT, R. K. 1988. The Late Prehistory of Brooks River, Alaska. University of Oregon, Anthropological Papers, 38. HILDRETH, W. 1983. The compositionally zoned eruption of 1912 in the Valley of Ten Thousand Smokes, Katmai National Park, Alaska. Journal of Volcanology and Geothermal Research, 18, 1-56. MARTIN, G. C. 1913. The recent eruption of Katmai volcano in Alaska. National Geographic Magazine, 24, 131-181. MILLER, T. P. & SMITH, R. L. 1987. Late Quaternary caldera forming eruptions in the eastern Aleutian arc, Alaska. Geology, 15, 434-438. NOWAK, M. 1968. Archaeological dating by means of volcanic ash strata. PhD dissertation, University of Oregon. PINNEY, D. S. 1993. Late Quaternary facial and volcanic stratigraphy near Windy Creek, Katmai National Park, Alaska. MSc thesis, University of Alaska, Fairbanks. & BEGET, J. E. 1991. Late Pleistocene volcanic deposits near the Valley of Ten Thousand Smokes, Katmai National Park, Alaska. In: REGER, R. D. (ed.) Short Notes on Alaskan Geology 1991. Alaska Division of Geological and Geophysical Surveys, Professional Report, 111, 45-54. REVENUE-CUTTER SERVICE 1913. Annual Report for the Fiscal Year Ended June 30, 1912. US Government Printing Office, Washington, DC. RIEHLE, J. R., MANN, D. H., PETEET, D. M., ENGSTROM, D. R., BREW, D. A. & MEYER, C. E. 1992. The Mount Edgecumbe tephra deposits, a marker horizon in southeastern Alaska near the Pleistocene-Holocene boundary. Quaternary Research, 37, 183-202. , WAITT, R. B., JR, MEYER, C. E. & CALK, L. C. 1998. The age of formation of Kaguyak Caldera, eastern Aleutian arc, Alaska, estimated by tephrochronology. In: GRAY, J. & RIEHLE, J. R. (eds) Geologic Studies in Alaska by the US Geological Survey, 1996. US Geological Survey, Professional Paper, 1595. ROSENBAUM, J. G. & WAITT, R. B., JR 1981. Summary of eyewitness accounts of the May 18 eruption. In: LIPMAN, P. W. & MULLINEAUX, D. R. (eds) The 1980 Eruptions of Mount St Helens, Washington. US Geological Survey, Professional Paper, 1250, 53-68. SARNA-WOJCICKI, A. M., WAITT, R. B., JR, WOODWARD, M. J., SHIPLEY, S. & RIVERA, J. 1981. Premagmatic ash erupted from March 27 through May 14, 1980 - extent, mass, volume, and composition. In: LIPMAN, P. W. & MULLINEAUX, D. R. (eds) The 1980 Eruptions of Mount St Helens, Washington. US Geological Survey Professional Paper, 1250, 569-576. SlMKIN, T., SlEBERT, L., McCLELLAND, L., BRIDGE, D.,

NEWHALL, C. & LATTER, J. H. 1981. Volcanoes of the World. Smithsonian Institution, Washington, DC.

266

J. R. RIEHLE ET AL.

SMITH, D. G. W. & WESTGATE, J. A. 1969. Electron probe technique for characterizing pyroclastic deposits. Earth and Planetary Science Letters, 5, 313-319. STEEN-MclNTYRE, V. 1977. A Manual for Tephrochronology. Colorado School of Mines, Golden, CO. TROWBRIDGE, T. 1976. Aniakchak Crater. In: RENNICK, & PERRY (eds) Alaska's Volcanoes: Northern Link in the Ring of Fire. Alaska Geographic, 4, 70-73. WESTGATE, J. A. & GORTON, M. P. 1981. Correlation techniques in tephra studies, In: SELF, S. &

SPARKS, R. S. J. (eds) Tephra Studies. D. Riedel, Dordrecht, 73-94. WIESNETH, D. W. & EICHELBERGER, J. C. 1996. Vapor phase crystallization in rhyolite lava from Novarupta dome, Katmai National Park, Alaska. EOS Transactions, American Geophysical Union, 77, 770. WILCOX, R. E. 1959. Some effects of recent volcanic ash falls with especial reference to Alaska. US Geological Survey Bulletin, 1028-N, 409-476.

Endemic stress, farming communities and the influence of Icelandic volcanic eruptions in the Scottish Highlands R. A. DODGSHON1, D. D. GILBERTSON2 & J. P. GRATTAN1 1

Institute of Geography and Earth Sciences, University of Wales Aberystwyth, Aberystwyth SY23 3DB, UK (e-mail: [email protected]) 2 The Nene Centre for Research, University College Northampton, Northampton NN2 7AH, UK Abstract: This paper explores present understanding of the possible impacts that volcanic eruptions in Iceland might have had upon the environments and traditional farming systems of the Highlands and Islands of Scotland, before 'the Clearances' of the late 18th and 19th centuries AD. It reconstructs both the nature of the impacts and the character of the risks that might have been faced by subsistence communities within the historical period from such Icelandic volcanic eruptions, and as such serves to redirect a research emphasis that has previously been principally focused upon the European Bronze Age. The study also emphasizes that it is inadequate to envisage the impacts from volcanic aerosols as threats to the community to be understood solely along a continuum of environmental hazards. For example, in the historical period before the Clearances, the wider social, political and economic contexts of subsistence economies affected can be shown to have raised or lowered the thresholds at which environmental risks became real or were turned into subsistence crises. In times past, as now, the capacity of people to cope with such environmental vicissitudes would have varied according to a complex of pre-disposing factors, their recent experiences, attitudes and perceptions, political and social relationships, health and wellbeing (especially their susceptibility to respiratory problems), economy, education and memory, and general inventiveness and resilience. Unlike much earlier research, which focused upon the European Late Bronze Age and emphasized global climatic change and its regional-scale consequences, this account of more recent times emphasizes the small scale, the importance of local pre-disposition and contingency, and hence the likely patchiness and indeterminacy of consequences on the ground of distant volcanic eruptions. The paper concludes that in the historical past, for a variety of environmental, agricultural, social and political reasons, some communities in the Highlands and Islands would have already been typically at risk of a subsistence crisis one year in every four or five. Hence a particular group of people could have been at notable further risk if a significant quantity of volcanically derived noxious and toxic materials had fallen upon them. As a result, for both habitats and human populations in historical times, the consequences of an Icelandic volcanic eruption are likely to have varied from place to place and from time to time. This analysis also suggests that it is difficult to envisage that any postulated region-wide abandonment of settlement in the British Isles might be attributable, directly or indirectly, to the distal impacts of volcanic eruptions in Iceland.

This paper explores the ability of traditional farming systems in the Highlands and Islands of Scotland, before'the Clearances'of the late 18th and early 19th centuries, to absorb and buffer the community from the stresses that might have been introduced as a result of the eruption of volcanoes in Iceland. The magnitude of the potential problem posed by such external events is difficult to estimate. Over the last decade, hypotheses describing the possible impacts have focused upon induced climatic deterioration and

the impacts of phytotoxic compounds both adsorbed on tephra and in solution in acid precipitation. The roles of prevailing synoptic meteorological situations, as well as the inherent susceptibility of the affected habitats, have also been explored. It would be wrong to see these hazards to traditional subsistence communities as something to be described and measured solely along a continuum of environmental hazards, perhaps ranging from the local and relatively minor to the regional and severe.


268

R. A. DODGSHON, D. D. GILBERTSON & J. P. GRATTAN

Traditional societies trying to maintain a subsistence economy within the difficult, often acidified and unproductive habitats of the north and west of Britain, would always have faced considerable risk. Detailed examination of the social, political and economic contexts of the traditional farm and township economies presented in this paper demonstrates that their social contexts could serve to raise or to lower the threshold at which risks became real and were turned into subsistence crises. Although this paper focuses upon the historical period from the advent of the Scottish clan system in the 12th and 13th centuries AD down to the time of the Clearances at about 1820, the intellectual starting point for this particular approach lies elsewhere, in the coming together over the last decade of several previously distinct and sometimes contentious areas of scholarship: volcanic eruptions, induced global climatic change, and postulated abandonment of upland settlement in the Late Bronze Age in parts of upland Britain; the development of tephrochronology within the UK; reports of the impacts of actual eruptions on historical communities in Iceland; investigations of acid precipitation and more general air pollution upon acidified ecosystems; holistic investigations of documentary sources concerning societies and their environments. Volcanic activity, induced climatic deterioration and the abandonment of upland settlement A variety of environmental and human mechanisms have been advanced to explain the numerous remains of medieval and post-medieval settlement that exist in upland and moorland Britain. Non-catastrophic explanations include land-holding and manorial policies (Austin 1985; for Dartmoor), climatic change (Beresford 1981; Dartmoor), crop failure as a result of persistent bad weather (Parry 1975, 1978, 1981, 1985; Parry & Carter 1985; for the Lammermuir Hills), as well as famine, disease, the Black Death, the Dissolution of the Monasteries, civil strife and invasion (Bell & Walker 1992). It is perhaps surprising to learn of hypotheses that suggest that volcanic eruptions in Iceland may have brought about injurious events to the habitats and populations of Scotland or Ireland, far distant downwind, across the breadth of the eastern Atlantic. Such possibilities are even more unexpected given that the known Holocene eruptions of Icelandic volcanoes were not particularly large, when viewed on a global scale (see Simkin & Siebert 1994).

Nevertheless, several hypotheses exploring these ideas have been advanced over the past decade, and the possible impacts of volcanic eruptions in Iceland (especially Hekla 3 and 4) upon the prehistoric, rather than historical, subsistence communities and environments of the north and west of Britain have been major subjects of archaeological, palaeoecological, palaeoclimatological and stratigraphic research (Burgess 1984, 1985, 1989). These ideas have benefited from the discovery in Scotland over the last decade of tephra from several Icelandic volcanic eruptions. This has provided impetus for seeking further volcanic catastrophic mechanisms by which to seek to explain postulated change in the archaeological and historical records (e.g. Hammer et al. 1980; Dugmore 1989; Blackford et al 1992; Dugmore & Newton 1992, 1996; Dugmore et al. 1992, 1995, 1996; Pilcher & Hall 1992, 1996; Hall et al 1993, 1994; Edwards et al 1994; Grattan & Gilbertson 1994; Gilbertson 1995; Pilcher et al 1995; Grattan et al 1996; Lowe & Turney 1997). At present, no less than 20 Holocene tephras from Icelandic eruptions, of both historical and prehistoric date, are known to have been deposited in northwest Europe and have the potential to form significant stratigraphic markers (Dugmore et al 1996). It is clear that some may have been associated with impacts on the ground that are detectable in the palaeoecological records (Blackford et al 1992). For example, environmental disruption following Icelandic eruptions has been proposed to explain decadal-length reductions in tree-growth increment sequence preserved in the Irish bog-oak record (see Pilcher et al 1984; Baillie 1988, 1989; Baillie & Munro 1988). Some, but not all, of this work was prompted by a proposal by Burgess (1989) that an apparent widespread abandonment of Late Bronze Age settlement on moorland and upland sites in Scotland and elsewhere at the end of the second millennium BC (see also Parker Pearson (1993)) was associated with catastrophic climatic change generated by an eruption, termed Hekla 3, at c. 1159BC, of the Icelandic volcano Hekla. It is difficult to assert with real confidence that any such widespread and synchronous abandonment of settlement did in fact take place in upland Scotland in particular, or upland Britain in general. For example, with the aid of a series of theoretical models, Grattan et al. (1999a) suggested in a paper presented to the Geological Society of London in 1994, that a widespread, prehistoric abandonment of uplands and moorlands in Britain was not a likely consequence of the eruption of either the Hekla 3 or 4 event. Starting from the archaeological information,

ICELANDIC VOLCANIC ERUPTIONS IN THE SCOTTISH HIGHLANDS Young & Simmonds (1996) showed that the evidence of abandonment in northern Britain was not compelling. In their important review of marginality and the nature of prehistoric settlement in the north of England they questioned the extent to which there actually was widespread and sudden abandonment of uplands and whether or not it was 'instantaneous'. They have emphasized that on innumerable occasions people have coped with very difficult situations because they chose not to leave their homes, or that in other situations, communities under pressure have often been supported by the wider and more distant communities in the society of which they form part. Young & Simmonds also indicated that many, perhaps misleading, assumptions lie in the use of the term 'marginal' in the descriptions and analysis of past peoples, economies and habitats of the north and west of the British Isles. More recently, Buckland et al. (1997) called into question the association of Hekla 3 with the putative Bronze Age settlement abandonment. They pointed out the problems in linking this eruption with particular ice-core acidity peaks, or narrow rings in the Irish bog-oak chronology. Starting from a broader perspective, Renfrew (1990) cautioned: 'it is necessary to recognize and discount the common tendency among archaeologists and historians to assume a causal link between distant and widely-spaced events of which they may have knowledge. An eruption here, a destruction there, a plague somewhere else - all are too easily linked in hasty surmise.' Neither has it proved easy to detect reliable, independent and corroborative palaeo-environmental evidence of the various postulated environmental processes and impacts that have been conjectured as agents of settlement change. Indeed, sometimes, the conclusions published concerning the ecological impact of just one ash layer on broadly similar peat-land habitats can be surprisingly diverse. For example, the report of a clear deleterious biological impact by the fall of volcanic ash of Hekla 4 (c. 2395-2279 BC; Pilcher et al. 1995) on pine forests in northeast Scotland given by Blackford et al. (1992) was followed by a statement of there being no such relationship in the north of Ireland (Hall et al. 1994), and even later studies in the north of Scotland further complicate the issue (Charman et al. 1995). Obviously, these differences may reflect genuine differences in impact, rather than point to inconsistencies and deficiencies in the evidence. These results also emphasize the need for a clearer understanding of the mechanisms through which the distal fall-out of volcanic eruptions may produce recognizable impacts

269

upon the landscapes of the British Isles. One mechanism through which volcanic fall-out from Icelandic eruptions may have affected habitats in the north and west of Britain, which has been explored in detail recently, focuses upon local, small-scale variations in habitat susceptibility and the environmental focusing of pollutants by the synoptic meteorological circulation (Grattan & Charman 1994; Grattan & Gilbertson 1994; Grattan & Pyatt 1994; Grattan et al. 1996, 1998). Particular reference has been made to document-based studies of the eruption and distant impacts of aerosols and fall-out from the Laki fissure eruption of AD 1783. This mechanism is explored in detail below.

Impacts upon people and their environment: issues of scale and susceptibility Global climatic changes? Initially, global climatic change produced by the eruption Hekla 3 was suspected to be the mechanism that promoted the volcanically generated equivalent of a nuclear winter for the Late Bronze Age communities in Scotland (Burgess 1989). For highland societies presumed to be living at the margins of sustainable subsistence farming in these remote and difficult environments, this sudden, but extended climatic deterioration, was suspected to have triggered complete collapse. In fact, there are no theoretical or empirical grounds for assuming that the Hekla 3 eruption, nor indeed any eruption that has occurred during the Holocene period was capable of causing such a phenomenon (see Mass & Portman 1989). For example, the massive eruption of Mt Pinatubo, in 1991 (McCormick et al. 1995; Robock & Mao 1995), resulted in a mean global surface air temperature reduction of 0.5°C, which is well within one standard deviation of the temperature variation that occurs normally on a year-to-year basis. It is possible that the prevailing synoptic atmospheric circulation over northwest Europe could have resulted in either an increase or decrease of this effect over northwest Britain (Kelly et al. 1984; Lamb 1992). However, given the variability of weather of northwest Britain, it is unlikely that a volcanically induced atmospheric circulation fluctuation, from any Icelandic volcanic eruption of Holocene time, would have been prominent. It is unlikely that, in themselves, climatic fluctuations generated by Icelandic volcanic eruptions would have been

270


sufficient to have brought about episodes of catastrophic social distress or environmental catastrophe in the subsistence-based communities of Britain in the historical past. Nor is it likely that such climatic fluctuations would have been of sufficient magnitude to account for the palaeo-environmental phenomena that have been tentatively associated with various Icelandic volcanic eruptions (Baillie 1988; Blackford et al 1992; Charman et al 1995).

Accounts of the impact of volcanic aerosols and tephra in Iceland and Europe Global climate change is not the only mechanism that has been invoked to explain settlement abandonment and environmental change in the wake of Icelandic volcanic eruptions. Across Europe, the environmental impacts of volatile compounds released during the Laki fissure eruption in AD 1783 were particularly damaging and provide clues to the manner in which past eruptions may have affected habitats and people. Impacts upon the European subsistence communities living immediately adjacent to these Icelandic eruptions are especially revealing. For example, detailed reading of the accounts by Thorarinsson (1967, 1979), Andresson (1984), Blong (1984) or Ogilvie (1986) provides convincing descriptions of both the direct and indirect downwind effects of noxious and toxic volcanically produced volatile materials and gases upon the subsistence farmers and fisherfolk of that island in the recent historical period. Many of the medical and veterinary symptoms and problems reported in Iceland after the 1783 eruption of the Laki fissure are recognizable as the results of the deposition, inhalation or ingestion of sulphuric, hydrochloric and hydrofluoric acids, affecting cows, sheep, crops, fodder and grazing, as well as people, their well-being, economies and societies (Thorarinsson 1971, 1979, 1981; Devine et al. 1984; Petursson et al. 1984; Palais & Sigurdsson 1989; Thordarson & Self 1993). Of critical importance for understanding the possible impacts of Icelandic volcanic eruptions on the British Isles, these accounts also reveal how the problems caused by one eruption in one area can spread through many aspects of the life of the affected communities and then be communicated to distant coastal communities by the emigration of distressed people. People fleeing the catastrophe caused the lands of the interior of Iceland to become desolate, whereas the coastal communities were impoverished as their

general resource base became overburdened by the influx of these people. The contamination of pastures obviously continued for a number of years, and it is evident that some crops were more susceptible than others to this natural pollution. Even so, these grim events should not automatically be assumed to have affected, or been communicated, throughout all the country of Iceland, or to have been the same from one eruption to the next. Thorarinsson (1971) showed that the vast majority of farms that were permanently abandoned after the eruption of Hekla in AD 1104 were found within 25km of the volcano, although one site 60 km distant was never resettled. He was also able to use their depth of burial in tephra as a more general measure of the severity of impact of the eruption, and related this aspect to the probability that the site would be reoccupied. For example, for the six Icelandic eruptions he studied since the AD 1104 eruption, the general pattern was that a burial in 10cm of tephra caused abandonment for up to 1 year, 15cm for a period of 1-5 years, and 20cm for periods of some decades. In northern and western areas of the British Isles the tephra isopachs are so fine that they cannot even be measured in millimetres. Caution must therefore be exercised before one accepts the assumption that any tephra fall in Britain was the sole cause for extensive and persistent settlement abandonment. Even the most severe of the Icelandic experiences would suggest that human responses to such a tephra-fall situation in the British Isles is likely to have been far more complex than a simple abandonment of property. Apart from the impacts of measurable tephra fall, volcanic gases may also have a marked effect upon the environment. Of key importance to the argument examined here is the fact that in 1783, volcanic gases emitted by the eruption in Iceland were transported to Europe, where they caused considerable respiratory distress to susceptible people and damage to susceptible crops, trees and fish (Grattan & Brayshay 1995; Grattan et al. 1998; and Fig. 1). Very detailed descriptions of severe acid damage to vegetation, insects, people and property have been left by a number of scientists, of which the recently rediscovered records of two Dutchmen, Brugmans (1787) and Van Swinden (1786), deserve particular emphasis. In mainland Europe the volcanic gases were manifested as a 'dry fog', 'acid fog' or 'sulphurous fog'. For example, Brugmans (1787) noted: 'On many days after the 24th June, in both the town of Groningen and countryside there was a strong, persistent fog... the fog was very dense and accompanied by a

ICELANDIC VOLCANIC ERUPTIONS IN THE SCOTTISH HIGHLANDS

271

Fig. 1. The location of currently known reports of environmental damage to plants and crops across Europe attributable to noxious or toxic materials originating from the Laki fissure eruptions in Iceland of 1783, and volcanoes in southern Italy (Camuffo & Enzi 1995) and perhaps east Germany (Grattan et al. this volume).

272


very strong smell of sulphur... many people in the open air experienced an uncomfortable pressure, headaches and experienced a difficulty breathing exactly like that encountered when the air is full of burning sulphur, asthmatics suffered to an even greater degree: On the morning of the 25th the land offered an aspect of severe desolation, the green colour of the plants had disappeared and everywhere the leaves were dry,... some were covered in spots, others changed gradually while some leaves dried up completely Another noticeable change was that in a moment the colour could change from

green to brown, black, grey or white. Afterwards a great quantity of leaves fell.' This account is typical of many that were written at the time and it is now clear from them that long-distance transport and deposition of materials from the Laki fissure eruption in 1783 to the European mainland took place and had significant deleterious effects upon the health of both susceptible crops and people across much of Europe (Grattan & Charman 1994; Grattan & Gilbertson 1994; Grattan & Brayshay 1995; Grattan et al. 1998). Further detail concerning the pollution event that occurred in AD 1793

Fig. 2. The concentration of volcanic gases over Europe in AD 1783. (Synoptic chart adapted from Kington (1988).)

ICELANDIC VOLCANIC ERUPTIONS IN THE SCOTTISH HIGHLANDS has been given by Camuffo & Enzi (1995) and Stothers (1996). It is evident from the example of the AD 1783 Laki fissure eruption that toxic and poisonous aerosols from Iceland were capable of causing modest levels of damage to animals, crops and people far downwind across the Atlantic in northwestern Europe when concentrated by stable air conditions (Fig. 2). Present biogeographical knowledge suggests that, in general, such impacts are likely to have been most evident in the more acidified ecosystems of the British Isles (Gorham 1987; Skiba et al 1989; Bull 1991; Smiths al. 1993).

The social context of risk and subsistence: the example of the Highlands and Western Islands It is therefore against a background of minor, probably undetectable, climatic fluctuations, minor tephra falls and probably patchy impacts of acid rain or fog, that we must assess the risk posed to the pre-'Clearances' subsistence economies of the Highlands and Western Islands by past Icelandic volcanic eruptions. Nevertheless, tephra falls and acid fogs and rain are documented for the historical period and it is therefore against a background of Scottish society in these regions during the historical period that the impact of volcanic eruptions is now considered. Trying to maintain a subsistence economy within a difficult, often marginal environment involved considerable risk for all traditional societies, were they in Iceland or in the north and west of Britain. It is wrong to see these risks as something to be measured solely along a continuum of environmental hazards, ranging from the local or minor to regional or severe. The wider social context of a subsistence economy could serve to raise or lower the threshold at which risks became real or were turned into subsistence crises. Attempts to understand more fully the character and scale of the difficulty faced by past communities with subsistence economies when presented with influxes of noxious or toxic volcanic materials in the study region can be substantially aided by a detailed consideration of the social, political and economical characteristics of an example. It is particularly fortunate in this context that the farm or township economy of the communities that lived in the western Highlands and Islands of Scotland over the late medieval and early modern periods are relatively well known through

273

recent research (Dodgshon 1993, 1994). By any standards, these traditional Highland communities lived in an endemically risk-laden world, but their subsistence crises had as much to do with their socio-political situation as with their natural environment and its perturbations. Three particular factors contributed to this heightened vulnerability.

The political context of subsistence crises The traditional pre-Clearance township in the Western Highlands and Islands was organized for self-sufficiency, with only a marginal involvement in marketing and other onward distribution of products. Inevitably, this sort of relatively closed, subsistence economy, or what some would call a 'natural economy', was more at risk of recurrent subsistence crises than more openmarket integrated economies that could exchange or trade away their problems with other regions. The crises that confronted such an economy were rarely self-correcting as a result of grain or meal flowing from areas of surplus to areas of deficit via regional markets or other exchange networks. However, this traditional Highland and Hebridean society was also a kin-based society organized around the power of clan chiefs. As with chiefly societies elsewhere, what can be called the chiefly economy provided local communities with a form of social storage and an insurance against risk (climatic, ecological, volcanic, or immediately human, social or political). Communities paid landowners large quantities of food (oatmeal, bear (= barley), cattle, sheep, cheese, whisky, fish, etc.) either as a form of rent or via obligations of hospitality (known as coid-oidche). In times of surplus, chiefs consumed such renders through the maintenance of household or fighting men, by extravagant displays of feasting, and, in the case of items of value such as cattle, by using them to sustain alliances and fosterage deals. In times of crisis, meanwhile, chiefs could use what had accumulated in the girnal houses, or food stores, to offset local deficits. This two-way flow of food was imbued with an instrumental value, securing the power of a chief over his kinsmen. As one source put it, the Steward of the Southern Isles was a great man simply because of the food which he had gathered in (Martin 1703, p. 98). The geological metaphor of MacLeod's Tables, opposite Dunvegan Castle, served a similar purpose for the MacLeod of MacLeod, proclaiming the scale of his hospitality. One sideeffect was that inter-clan feuding became

274


focused around food. In character, feuding involved not only constant warring between clans by the fighting men of a clan, men who were supported by the food renders gathered in by a chief, but also, the routine destruction of a rival's food base: a case of fighting with food over food (Dodgshon 1995, pp. 99-109). The ideological overlay to the basic use value of food, however, had begun to change by the start of the 17th century. Legislation such as the Statutes of lona of 1609 controlled the behaviour of chiefs, banning them from maintaining large households and from excessive feasting, both prime uses of food gathered in as rent (Register of the Privy Council X, 1891, pp. 773781). In response, landlords began to market the renders gathered in as rent, or to convert them into cash payments, forcing tenants to market them instead. However, for the majority of farmers, especially those in the far west of the region, marketing involved livestock whereas their basic subsistence continued to depend upon arable. Indeed, all the signs are that dependence on arable increased, as local populations increased and as rents previously based on a range of farm produce now became loaded onto livestock, leaving more arable output for consumption. In these changing circumstances, the risks facing highland communities must have increased. On the one hand, traditional riskaversion strategies based upon chiefly networks of support during subsistence crises were weakened as food rents were converted to cash and more grain was consumed on the farm. More and more, chiefly support took the form of rent rebates rather than food handouts. Only exceptionally is it possible to find landowners buying in meal to offset crises of subsistence. On the other hand, market supply networks worked their way into the region only slowly. In fact, marketing brought new problems, as the more remote islands and even the more distant parts of the mainland began to suffer the downside of a market-based economy, the diseconomies of location. Smaller islands that had maintained large populations and equally large arable sectors, such as Pabbay, Taransay and Ensay off the southern and southwestern corner of Harris, were bound to suffer once production strategies were affected by prevailing market prices.

The marginality of production A second reason why the western Highlands and IsJands were risk-laden before 1820 lies in the low average levels of farm output. Where available data enable us to analyse returns on

seed, it is clear that most parts had only modest returns. Across the region, oats (the basic subsistence crop) averaged returns of 2-3 times on seed, whereas bear yielded 3-4 times, although locally there were islands like Tiree, where returns were lower. Systematic data collected for every township on the island during a 4 year period in the mid-1760s suggests yields of 1:2.2 for oats and 1:3.5 for bear (Dodgshon 1993). Although the island was reported as suffering from soil exhaustion by the mid-18th century, the figures were probably typical of many less fertile Hebridean islands by this point. To a degree, the widespread use of the spade or cashcrom for cultivation provided a small yield bonus in return for a heavy investment of labour compared with the plough. However, it still left communities with only modest surpluses after grain (approximately one-third of the total output) had been paid as rent and seed (again one-third) had been removed. In these circumstances, communities would have experienced crises more frequently, as even modest swings in climate or adverse weather patterns dug deep into what were narrow margins of subsistence. It is this lack of a comfortable annual surplus, coupled with the problems of grain storage in a physical climate that was increasingly hostile, that lay behind the subsistence problems of the region. As Dr Johnson put it during a visit to Mull, 'the consequences of a bad season here is not scarcity, but emptiness, and they whose plenty was barely a supply of natural or present need, when that slender stock fails, must perish with hunger' (Johnson 1971, p. 137). It was often said about the clan system in the western Highlands and Islands that it cultivated men more than land. This was another way of saying that landowners had a vested interest in populating their estates with men who could farm and fight, and therefore tried to absorb population growth back onto the land. This had the effect not just of expanding a clan's food base for its chief, but also of spreading subsistence across a wider range of environments. The outcome was that many environments that would not normally have been cultivated became settled, but at a risk. The extensive cultivation of river floodplain deposits, sandy soils or machair, steep slopes, high and exposed ground can all be documented, but so too can the environmental damage and subsistence crises that followed storms, flooding and the unexpected occurrence of unseasonable cold or wet weather. The scale of this damage can be reconstructed through contemporary sources. In an environment in which cultivable land was scarce, the cultivation of gravelly river floodplain deposits

ICELANDIC VOLCANIC ERUPTIONS IN THE SCOTTISH HIGHLANDS was always likely to be an attractive option. In terms of possible volcanic impacts on habitats, their soils were likely to be relatively nutrient enriched and buffered. The problem was that pushing cultivation close to river banks increased the risk of flooding and erosion during storms. Many communities must have taken a calculated view of the risks, accepting that (as in Glen Shira) where they grew oats one year, they fished for salmon another. Some estates tried to control the risks through restrictions on the cropping of land close to rivers and by compelling tenants to share in the task of building retaining walls. However, such controls could not have been effective because mid-18th century reports suggest that large amounts of arable had already been eroded in areas such as Coigach. Comparable problems of environmental stress were caused by over-ploughing and over-grazing of machair shell-sands in the Hebrides and the northern and western shorelines of the mainland. This environment was perhaps the most effectively buffered habitat in the region against the impact of volcanically derived acid volatiles. The scale of the erosion problem that affected these habitats is defined by a mid-18th century survey of Tiree. The survey reports over 1000 acres of land lost by sand-blows, whose impact reportedly owed as much to the pressure created by ploughing, grazing and the harvesting of plants for thatch, dyeing, etc., as to the power of storms (Scottish Record Office, Edinburgh, RHP 8826/2, General description of the Island of Tiree, by James Turnbull). Although not understating the ability of storms to destroy machair without any trigger from human activity, the latter undoubtedly accentuated the risks. Certainly, contemporary attempts to control the situation saw it as being as much a human as an environmental problem. The damage caused by the storms of 1697 was all the greater where it acted on fragile areas that had been under intensive cultivation. In the case of the former, its impact can be monitored through rentals for MacLeod of MacLeod's estate. The estate included among its possessions a cluster of islands to the south and east of Harris, whose land use included substantial amounts of machair pasture. Rentals drawn up at the close of the 17th century make it clear that islands such as Pabbay lost substantial amounts of arable (one estimate puts the figure at 300 acres) in the storms of 1697, the island being reassessed downwards from a capacity of 16 pennylands to only 10 (Dunvegan Castle, MacLeod of MacLeod Papers, 2/487/15, Rental of Harris 1698; 2/487/18, Rental of Harris 1703). The same storm devastated arable on the

275

western side of Bernaray, as well as Illeray on Baleshare, North Uist, with extensive parts of the sandy arable that lay to the north side of Illeray being swept away and other parts being overwhelmed by freshly blown sand (McKay 1980, p. 13). Later hurricane-force storms, such as the one in February 1749, were reported by Dr James Walker (1764; see McKay 1980) as having caused extensive damage to the arable on Barra, especially on the Eoligarry isthmus (Gilbertson et al 1996), now adjacent to the island's airstrip, with settlements having to be moved. The problem caused by its 'extensive Fields of blowing sand' that 'turn and wheel, and move over the Country, in very hurtfull way' was seen as endemic on the island thanks largely to its 'forbidding' climate and sandy soils, and created an annual deficit in grain for the island (Walker 1764; see McKay 1980, pp. 186-187). For townships on South Uist, there was a widespread temptation to harvest shell-sand from the seaward side of the township and to add it to the peatier soils inland. Thesum effect was that large areas of machair along the west side of the island were regularly destabilized.

Frequency of subsistence crises The frequency of subsistence crises, primarily understood in terms of climatic fluctuations in the Highlands and Islands, can only be appreciated when the wider human context is taken into account. This frequency can be established in a number of ways. Historical accounts provide generalized reference to the years when harvests were bad and starvation widespread across the country as a whole. From such sources, for instance, it is known that between 1550 and 1600, there occurred 17 bad harvests (i.e. 1550-1552, 1560, 1562-1563, 1568, 15711572, 1585-1587 and 1594-1596), and more occurred between 1630 and 1660 (i.e. 1622-1623, 1634-1635, 1640, 1649-1651 and 1655-1656) (Lythe 1960, pp. 16-23; Parry 1978, p. 162). Although it has been argued that poor harvests became less frequent after 1660 thanks to an improvement in yields, this was less true of the Highlands. In the latter, bad harvests were a feature of the late 1670s and of the so-called King William's lean or ill years, 1695-1702. Crop failures followed by subsistence crises are also reported for 1709, 1740-1741, 1745-1747, 1756, 1760-1761, 1778, 1782-1783, 1796 and 1799-1800. The coincidence of periods of famine with Icelandic volcanic eruptions is rare

276


Table 1. The incidence of famine in the Highlands of Scotland and volcanic eruptions 1550-1800 Period of famine

Total volcanic eruptions

Total volcanic eruptions VEI> 3

Total Icelandic volcanic eruptions

1550-1552 1560 1562-1563 1568 1571-1572 1585-1587 1594-1596 1622-1623 1634-1635 1640 1649-1651 1655-1656 1695-1707 1709 1740-1741 1745-1747 1756 1760-1761 1778 1782-1783 1796 1799-1800

20 6 5 1 4 15 10 5 7 9 21 6 30 6 6 8 1 12 6 15 9 20

0 0

1

0 0 0 0 0 0 0 0 0 0 0 1 5 0 0 1 0 0 0

1

0 0

1

0 0 2 1 1 0 2 1 1 1 0 1 0 0

1

2 0

1

(Table 1) and in all cases the eruptions follow rather than cause the initial distress. The opportunity for a more systematic understanding of how often Highland subsistence economies may have been stressed by fluxes in output is provided by available prices for basic foodstuffs. Data on oatmeal prices in Perthshire between 1630 and 1820 provide an excellent long-run series. They show that prices experienced sharp upward surges approximately once a decade throughout the 17th and for much of the 18th century, with at least 14 price peaks being identifiable between 1630 and 1780. By the late 18th century, trends are complicated by the general but strong inflection of prices that took place during the Napoleonic Wars, 17931815, but even during the Wars, the general price is broken by sudden and powerful surges in price (based on http://www.ex.ac.uk/~ajgibson/Scotdata/prices/fi). Of course, not all these sudden price jumps can be attributed to poor harvests. Other factors were involved, but marked yearto-year shifts in farm output are likely to have been at the root of such movements. Estate documents provide an obvious source of data for such subsistence crises. What they add to our perspective is the importance of the scale of the problem. On the one hand, region-

wide crises are well depicted through rentals and estate correspondence. The succession of poor harvests known as King William's lean or ill years, 1695-1702, have been shown to have killed many people in northeast Scotland, but there is also good reason to believe that many perished further west in the Highlands and Islands. Certainly Martin (1703) reported that they killed many of the poorer landholders and cottars in the Hebrides. Detailed rental data show that those rents actually collected fell dramatically, with arrears adding up to two-thirds of what was supposed to have been paid. Close examination of the available information concerning why this calamity occurred indicates that whole townships, such as Kelsay and Kilneave, lay waste for 1 or 2 years, while the crisis worked itself out. Others, like Upper Stincha, lay partially waste (Cawdor Castle, Campbell of Cawdor Papers Bundle 21, Rent of Hay 1703-1707). To the east, on the mainland of Argyll, subsistence problems were experienced during these crisis years by the inland townships of Glenorchy, with the Breadalbane estate having to arrange for meals to be distributed amongst the townships (Scottish Record Office, Campbell of Barcaldine Muniments, GDI70/629). Equally widespread in its effects was the subsistence crisis of the mid17408, 1744-1747. In Mull, Cregeen found that one-quarter of all the Duke of Argyll's farms on Mull were partially or wholly left to waste in 1747 because of poor harvests, as were many of the farms on the mainland around Loch Awe (Cregeen 1969, p. 127). To the northwest, correspondence amongst the Macleod of Macleod MSS also makes it clear that this same subsistence crisis led to famine on Skye. One letter written in 1745 talks of 'most' of the inhabitants of Trumpen More and Beg (Vatternish) being at risk of perishing 'for want of Bread' if relief was not provided within a few days (Dunvegan Castle, MacLeod of MacLeod Papers, 4/151, Letter 12 May 1745). The exceptionally poor season of 1722 also created famine problems for the Macleod of Macleod estate, with 'most part of the inhabitants' in Duirinish Waternish having 'nothing to eat or sow in yr Ground' (Dunvegan Castle, MacLeod of MacLeod Papers, 4/304/1, Letter 28 April 1772). In another letter, it was described as a 'universal' crisis, tenants being 'without Sowing without Bread to support Nature, without money or credit'. It goes on to say that 'hundreds will starve' (Dunvegan Castle, MacLeod of MacLeod Papers, 4/304/2). The widespread collapse of farm output during the poor harvest of 1782, and the loss of stock during the poor winter of 1782, shortly before the eruption of the Laki fissure

ICELANDIC VOLCANIC ERUPTIONS IN THE SCOTTISH HIGHLANDS the following summer, are particularly well documented. Indeed, their devastation in the Highlands prompted a Government Commission and led to the formation of the Highlands and Islands Agricultural Society with the express intention of improving the region out of such subsistence crises. Reports to the Commission suggested that the failure of the 1782 crop was brought about by a combination of a very poor summer for crop growth, followed inevitably by a very late harvest and then, soon after the harvest had begun, heavy snow fell. To compound matters, the following spring and summer were stormy and wet (Handley 1963, p. 72). At present, the significance of the impact of acid volatiles from Laki upon the highlands is largely unknown, beyond a comment of uncertain significance in Geikie (1893) that in 1783 so much volcanic ash fell in parts of Caithness that it was remembered as 'the year of the ashie'. Especially revealing are the reports from northern Scotland which suggest that many farmers and their families abandoned their farms and 'were forced to beg or perish' (Scottish Record Office, Edinburgh, E746.86, Letter from the Minister of Fodderty, 1783). Though the Commission provides evidence of grain and pease being shipped into the region, what stands out from the Commission's evidence and other sources is the extent to which one season's failure could precipitate such a severe subsistence crisis, and the inadequacy of local markets at this time to provide market-led solutions. In addition to providing information on largescale crises, these estate documents indicate that this region was afflicted by numerous local crises. This was in the nature of its natural environment, with storms or sudden downpours devastating crops in one glen, but not in another, or devastating crops on exposed farms but not on more sheltered farms. A survey of rental and estate accounts provides ample evidence for this sort of local disaster. In 1733, for example, eight tenants in the township of Barbreck Lochou petitioned the Earl of Breadalbane, 'showing their losses by a storm and asking for allowance, according to the custom of the country' (Scottish Record Office, Edinburgh, GDI 12/10/1/3/56). Likewise, in 1793, tenants on Seil and Luing, near Oban, reportedly lost half their crops through storms. As a result of the combination of major regional crises and intense local crises, the region is readily seen as risk laden. One reliable estimate for the 1760s, put the risk of a poor harvest as averaging one year in four (i.e. Burl's Letters (1754); see Jamieson 1876, Vol. 1, p. 158). Given the tight margin on

277

yields, such a frequency of risk must have easily turned to crisis when poor harvests followed each other. Conclusions Scottish Highland society during the historical period clearly operated in a continuum of risk, and it is within this continuum that the potential impact of volcanic ejecta must be considered. As a result of the particular conditions of the eruption and the synoptic meteorological situation, toxic and noxious volcanic acid gases and volatiles from Iceland would have been introduced into northern Britain in a patchy, discontinuous manner to habitats, people and communities that also varied greatly in their susceptibility to these inputs. It is evident that during the historical period, subsistence farming in the north and west of Scotland was an enterprise fraught with risks. These risks were environmental, social, political and economic, and were buffered by a complex support system and social obligations, which operated across wide areas. These features can be seen to have occurred, with varying degrees of success, throughout the historical period. In times of subsistence crisis, whole townships were abandoned, albeit temporarily. It is important to note that many people died of famine in the Highlands and Islands of Scotland without the injurious external influences of volcanic materials from Iceland. It can be seen above that famine was frequent and often severe, and that it only occasionally coincided, as in 1783, with the eruption of Icelandic volcanoes (Table 1). It is, however, telling to observe that the Laki fissure eruption of 1783 coincided with a famine that had already been taking place in Scotland for a year and that the contemporary documents concerning Scotland read by the authors give no mention of the role of volcanic gases or ashfall in either exacerbating the famine or its consequences; this despite the widely reported preoccupation across Europe with respiratory problems, plant and animal damage and spectacular atmospheric effects, now associated with this protracted volcanice ruption (Grattan & Charman 1994; Grattan & Brayshay 1995; Grattan et al. 1998). However, although this analysis suggests that explanations of widespread settlement and palaeoecological change in the British Isles that depend only upon Icelandic volcanic eruptions appear inadequate, it is clear that within the continuum of risk described above, Icelandic volcanic eruptions may on occasion have played

278


a role in intensifying subsistence crises in this region in the historical, and perhaps in the prehistoric, period.

References ANDRESSON, S. H. 1984. Athstoth Einokunarverslunarinnar. In: GUNNLAUGSSON, G. A., GUDBERGSSON, G. M., THORARINSSON, S., RAFFNSON, S. & EINARSSON, T. (eds) Skdftareldar 1783-1784. Mai Og Menning, Reykjavik [in Icelandic with English summaries]. AUSTIN, D. 1985. Dartmoor and the upland village of the South-West of England. In: HOOKE, D. (ed.) Medieval Villages: a Review of Current Work. Oxford University Committee for Archaeology Monographs, 5, 71-79. BAILLIE, M. G. L. 1988. Irish oaks record volcanic dust veils drama. Archaeology in Ireland, 2, 71-74. 1989. Hekla 3: how big was it? Endeavour, 13, 78-81. & MUNRO, M. 1988. Irish tree-rings, Santorini and volcanic dust veils. Nature, 332, 344-346. BELL, M. & WALKER, M. J. C. 1992. Late Quaternary Environmental Change: Physical and Human Perspectives. Longman, Harlow. BERESFORD, G. 1981. Climatic change and its effects upon the settlement and desertion of medieval villages in Britain. In: DELANO SMITH, C. & PARRY, M. L. (eds) Consequences of Climatic Change. University of Nottingham, Department of Geography, Nottingham, 30-39. BLACKFORD, J. J., EDWARDS, K. J., DUGMORE, A. J., COOK, G. T. & BUCKLAND, P. C. 1992. Icelandic volcanic ash and the mid-Holocene pine pollen decline in northern Scotland. Holocene, 2, 260-265. BLONG, R. J. 1984. Volcanic Hazards. Academic Press, London. BRUGMANS, S. J., 1787. Natuurkundige verhandeling over een zwavelagtigen nevel den 24 Juni 1783 innde provincie van stad en lande en naburige landen \vaargenomen. (A physical treatise on a sulphuric smog as observed on the 24th of July 1783 in the province of Groningen and neighbouring countries.) Leyden. BUCKLAND, P. C., DUGMORE, A. J. & EDWARDS, K. J. 1997. Bronze Age myths? Volcanic activity and human response in the Mediterranean and North Atlantic regions. Antiquity, 71, 581-593. BULL, K. 1991. Critical load maps for the U.K. NERC News, July, 31-32. BURGESS, C. B. 1984. The prehistoric settlement of Northumberland. In: BURGESS, C. B. & MIKET, R. (eds) Between and Beyond the Walls: Essays on the Prehistory of North Britain in Honour of George Jobey. Edinburgh University Press, Edinburgh, 126-175. 1985. Prehistoric climate and upland settlement. In: SPRATT, D. & BURGESS, C. (eds) Upland Settlement in Britain. British Archaeological Reports, 143, 195-230.

1989. Volcanoes, catastrophe and the global crisis of the late second millennium, BC. Current Archaeology, 111, 325-329. CAMUFFO, D., & ENZI, S. 1995. Impacts of clouds of volcanic aerosols in Italy during the last 7 centuries. Natural Hazards, 11, 135-161. CHARM AN, D. J., GRATTAN, J. P., WEST, S. & KELLY, A. 1995. Environmental response to tephra deposition in the Strath of Kildonan, northern Scotland. Journal of Archaeological Science, 22, 799-809. CREGEEN, E. R. 1969. The Tacksmen and their successes. A study of tenurial re-organisation on Mull, Morven and Tiree in the early eighteenth century. Scottish Studies, 13, 93-144. DEVINE, J. D., SIGURDSSON, H., DAVIS, A. N. & SELF, S. 1984. Estimates of sulfur and chlorine yield to the atmosphere from volcanic eruptions and potential climatic effects. Journal of Geophysical Research, 89, 6309-6325. DODGSHON, R. A. 1993. Strategies of farming in the western Highlands and Islands of Scotland prior to crofting and the clearances. Economic History Review, XLVI, 679-701. 1994. Budgeting for survival: nutrient flow and traditional Highland farming. In: FOSTER, S. & SMOUT, H. C. (eds) The History of Soils and Field Systems. Scottish Cultural Press, Aberdeen, 83-93. 1995. Modelling chiefdoms in the Scottish Highlands and Islands prior to the '45. In: ARNOLD, B. & GIBSON, B. (eds) Celtic Chiefdom: Celtic State. New Directions in Archaeology Series. Cambridge University Press, Cambridge, 99-109. DUGMORE, A. J. 1989. Icelandic volcanic ash in Scotland. Scottish Geographical Magazine, 105, 168-172. & NEWTON, A. J. 1992. Thin tephra layers in peat revealed by X-radiography. Journal of Archaeological Science, 19, 163-170. & 1996. Ideas and evidence from tephra. In: GILBERTSON, D. D., KENT, M. & GRATTAN, J. P. (eds) The Outer Hebrides: the Last 14,000 Years. Sheffield Academic Press, Sheffield, 45-50. , LARSEN, G. & NEWTON, A. J. 1995. Seven tephra isochrones in Scotland. Holocene, 5, 257-266. , NEWTON, A. J., EDWARDS, K. J., LARSEN, G., BLACKFORD, J. J. & COOK, G. T. 1996. Longdistance marker horizons from small-scale eruptions: British tephra deposits from the AD 1510 eruption of Hekla, Iceland. Journal of Quaternary Science, 11, 511-516. , & SUGDEN, D. E. 1992. Geochemical stability of fine-grained tephra in Iceland and Scotland. Journal of Quaternary Science, 1, 173-183. EDWARDS, K. J., BUCKLAND, P. C., BLACKFORD, J. J., DUGMORE, A. J. & SADLER, J. P. 1994. The impact of tephra: proximal and distal studies of Icelandic eruptions. Muncher Geographische Abhandlungen, B12, 108-126. GEIKIE, A. 1893. Text-Book of Geology. Macmillan, London. GILBERTSON, D. D. 1995. Studies of lithostratigraphy and lithofacies: a selective review of research

ICELANDIC VOLCANIC ERUPTIONS IN THE SCOTTISH HIGHLANDS developments in the last decade and their applications to geoarchaeology. In: BARHAM, A. J. & MACPHAIL, R. I. (eds) Archaeological Sediments and Soils: Analysis, Interpretation and Management. Institute of Archaeology, University College London, London, 99-144. , GRATTAN, J. P. & SCHWENNINGER, J.-L. 1996. A stratigraphic survey of the Holocene coastal dune and machair sequences. In: GILBERTSON, D. D., KENT, M. & GRATTAN, J. P. (eds) The Outer Hebrides: the Last 14,000 Years. Sheffield Academic Press, Sheffield, 72-101. GORHAM, E. 1987. Group summary report: wetlands. In: HUTCHINSON, T. C. & MEEMA, K. M. (eds) Effects of Atmospheric Pollutants on Forests, Wetlands and Agricultural Systems. Springer, Berlin, 631-636. GRATTAN, J. P. & BRAYSHAY, M. B. 1995. An amazing and portentous summer: environmental and social responses in Britain to the 1783 eruption of an Icelandic volcano. Geographical Journal, 161,125-134. & CHARMAN, D. J. 1994. Non-climatic factors and the environmental impact of volcanic volatiles: implications of the Laki Fissure eruption of AD 1783. Holocene, 4, 101-106. & GILBERTSON, D. D. 1994. Acid-loading from Icelandic tephra falling on acidified ecosystems as a key to understanding archaeological and environmental stress in northern and western Britain. Journal of Archaeological Science, 21, 851-859. & PYATT, F. B. 1994. Acid-damage caused by the Laki Fissure eruption - an historical review. Science of the Total Environment, 151, 241-247. , CHARMAN, D. & GILBERTSON, D. D. 1996. The environmental impact of Icelandic volcanic eruptions: a Hebridean perspective. In: GILBERTSON, D. D., KENT, M. & GRATTAN, J. P. (eds) The Outer Hebrides: the Last 14,000 Years. Sheffield Academic Press, Sheffield, 51-58. , GILBERTSON, D. D. & CHARMAN, D. J. \999a. Modelling the impact of Icelandic volcanic eruptions upon the prehistoric societies of northern and western Britain. In: FIRTH, C. & McGuiRE, W. (eds) Volcanoes in the Quaternary. Geological Society, London, Special Publications, 161, 147-160. , & DILL, A. 1999/7. 'A fire spitting volcano in our dear Germany': documentaty evidence for a low-intensity volcanic eruption of the Gleichberg in 1783.' This volume. HALL, V. A., PILCHER, J. R. & MACCORMAC, F. G. 1993. Tephra-dated lowland landscape history of the north of Ireland, AD 750-1150. New Phytologist, 125, 193-202. , & 1994. Icelandic volcanic ash and the mid-Holocene Scots pine (Pinus sylvestris} decline in the north of Ireland: no correlation. Holocene, 4, 79-83. HAMMER, C. U., CLAUSEN, H. B. & DANSGAARD, W. 1980. Greenland ice sheet evidence of post-glacial volcanism and its climatic impact. Nature, 288, 230-235. HANDLEY, J. E. 1963. The Agricultural Revolution in Scotland. Burns, Glasgow.

279

JAMIESON, R. (ed.) 1876. Burl's Letters from the North of Scotland, first published in 1754. 2 vols. Edinburgh. JOHNSON, S. 1971 edition. A Journey to the Western Islands of Scotland. Yale University Press, New Haven, CT. KELLY, P. M., WIGLEY, T. M. L. & JONES, P. D. 1984. European pressure maps for 1815-16, the time of the eruption of Tambora. Climate Monitor, 13, 76-91. KINGTON, J. A. 1988. The weather for the 1780s over Europe. Cambridge University Press, Cambridge. LAMB, H. H. 1992. First essay at reconstructing the general atmospheric circulation in 1816 and the early nineteenth century. In: HARINGTON, C. R. (ed.) The Year Without a Summer. Canadian Museum of Nature, Ottawa, Ont., 355-357. LOWE, J. J. & TURNEY, C. S. M. 1997. Vedde Ash layer discovered in a small lake basin on the Scottish mainland. Journal of the Geological Society, London, 154, 605-612. LYTHE, S. G. E. 1960. The economy of Scotland in its European setting 1550-1625. Oliver and Boyd, Edinburgh. MARTIN, M. 1703. A Description of the Western Isles of Scotland, reprinted 1934. Mackay, Stirling. MASS, C. F. & PORTMAN, D. A. 1989. Major volcanic eruptions and climate: a critical evaluation. Journal of Climate, 2, 566-593. MCCORMICK, P. M., THOMASON, L. W. & TREPTE, C. R. 1995. Atmospheric effects of the Mt. Pinatubo eruption. Nature, 373, 399-404. McKAY, M. M. (ed) 1980. The Rev. Dr John Walker's Report on the Hebrides of 1764 and 1771. John Donald, Edinburgh. OGILVIE, A. E. J. 1986. The climate of Iceland 17011784. Jokull, 36, 57-73. PALAIS, J. M. & SIGURDSSON, H. 1989. Petrologic evidence of volatile emissions from major historic and pre-historic volcanic eruptions. In: BERGER, A., DICKINSON, R. E. & KIDSON, J. W. (eds) Understanding Climate Change. Geophysical Monograph, American Geophysical Union, 52, 31-53. PARKER PEARSON, M. 1993. Bronze Age Britain. Batsford-English Heritage, London. PARRY, M. L. 1975. Secular climatic change and marginal land. Transactions of the Institute of British Geographers, 64, 1-13. 1978. Climatic Change, Agriculture and Settlement. Dawson, Folkestone. 1981. Evaluating the impact of climatic change. In: DELANO SMITH, C. & PARRY, M. L. (eds) Consequences of Climatic Change. University of Nottingham, Department of Geography, Nottingham, 3-16. 1985. Upland settlement and climatic change: the medieval evidence. In: SPRATT, D. & BURGESS, C. B. (eds) Upland Settlement in Northern Britain. British Archaeological Reports, 143, 195-229. & CARTER, T. R. 1985. The effect of climatic variations on agricultural risk. Climatic Change, 1, 95-100.

280


PETURSSON, G., PALSSON, G. A. & GEORGSSON, G. 1984. Um Eiturahrif. In: GUNNLAUGSSON, G. A., GUDBERGSSON, G. M., THORARINSSON, S., RAFFN-

SON, S. & EINARSSON, T. (eds). Skdftareldar 17831784. Mai Og Menning, Reykjavik [in Icelandic with English summaries]. PILCHER, J. R. & HALL, V. A. 1992. Towards a tephrochronology for the Holocene of the north of Ireland. Holocene, 2, 255-259. & 1996. Tephrochronological studies in northern England. Holocene, 6, 100-105. , BAILLIE, M. G. L., SCHMIDT, B. & BECKER, B. 1984. A 7272-year tree-ring chronology for western Europe. Nature, 312, 150-152. , HALL, V. A. & MACCORMAC, F. G. 1995. Dates of Holocene eruptions from tephra layers in Irish peats. Holocene, 5, 103-110. REGISTER OF THE PRIVY COUNCIL OF SCOTLAND, X, 1613-1616. 1891. H. M. Register House, Edinburgh. RENFREW, C. 1990. Climate and Holocene culture change: some practical problems. Philosophical Transactions of the Royal Society of London, Series A, 330, 657-663. ROBOCK, A. & MAO, J. 1995. The volcanic signal in surface temperature observations. Journal of Climate, 8, 1086-1103. SIMKIN, T., & SIEBERT, L. 1994. Volcanoes of the World, 2nd edn. Geoscience Press, Tucson, AZ. SKIBA, U., CRESSER, M. S., DERWENT, R. G. & FUTTY, D. W. 1989. Peat acidification in Scotland. Nature, 337, 68-69.

SMITH, C. M. S., CRESSER, M. S. & MITCHELL, R. D. J. 1993. Sensitivity to acid deposition of dystrophic peat in Great Britain. Ambio, 22, 22-26. STOTHERS, R. B. 1996. The Great Dry Fog of 1783. Climatic Change, 32, 79-89. THORARINSSON, S. 1967. The Eruption of Hekla 19471948. The Eruptions of Hekla in Historical Times. H. F. Leiftur, Reykjavik. 1971. Damage caused by tephra fall in some big Icelandic eruptions and its relation to the thickness of the tephra layers. In: KALOYEROPOYLOY, A. (ed.) Acta of the First International Scientific Congress on the Volcano Thera, September 1969, 213-236. 1979. On the damage caused by volcanic eruptions with special reference to tephra and gases. In: SHEETS, P. D. & GRAYSON, D. K. (eds) Volcanic Activity and Human Ecology. Academic Press, New York, 125-159. 1981. Greetings from Iceland. Ash-fall and volcanic aerosols in Scandinavia. Geografiska Annaler, 63A, 109-118. THORDARSON & SELF, S. 1993. The Laki [Skaftar Fires] and Grimsvotn eruptions in 1783-85. Bulletin Volcanologique, 55, 233-263. VAN SWINDEN, M. 1786. Observations sur quelques particularites meteorologiques de 1'annee 1783. Memoir es de I'Academic Roy ale des Sciences, Turin, 1784-1785, 113-140. YOUNG, R. & SIMMONDS, T. 1996. Marginality and the nature of later prehistoric upland settlement in the north of England. Landscape History, 17, 5-16.

Comparison and cross-checking of historical, archaeological and geological evidence for the location and type of historical and sub-historical eruptions of multiple-vent oceanic island volcanoes S. J. DAY 1 , J. C. CARRACEDO2, H. GUILLOU3, F. J. PAIS PAIS4, E. RODRIGUEZ BADIOLA5, J. F. B. D. FONSECA6 & S. I. N. HELENO6 1

Benfield Greig Hazard Research Centre, Department of Geological Sciences, University College London, Gower Street, London WC1E 6BT, UK

2

Estacion Volcanologica de Canarias, CSIC, Aptdo. Correos 195, La Laguna 38206, Tenerife, Canary Islands, Spain 3

Centre des Faibles Radioactivites, Laboratoire Mixte 91198 Gif-Sur-Yvette,

4

University of La Laguna, La Laguna, Tenerife, Canary Islands, Spain 5

6

CEA-CNRS,

France

Museo Nacional de Ciencias Naturales, CSIC, Madrid, Spain

Grupo de Fisica da Terra e do Ambiente Instituto Superior Tecnico, Lisbon, Portugal Abstract: Oceanic island volcanoes, like many others, have many small vents scattered over their flanks in addition to, or in place of, large summit vents. These small vents are commonly monogenetic and many eruptions of this type of volcano involve activity at more than one such vent: lines of volcanic cones are often produced by eruptions fed by dykes, and if more than one dyke is emplaced during an eruption these vents can be along different alignments, and many kilometres apart. Identifying which vents were produced in which eruptions is an important problem in reconstructing the development of multiplevent volcanoes. Reconstruction of historical and sub-historical eruptions of two oceanic island volcanoes, the Cumbre Vieja volcano on La Palma in the Canary Islands and Cha das Caldeiras volcano on Fogo, Cape Verde Islands, has indicated that historical eyewitness accounts and archaeological evidence can be extremely valuable adjuncts to detailed geological studies. In contrast, secondary accounts including folk memories and earlier accounts in the scientific literature are commonly inconsistent both with the eyewitness accounts and with the results of detailed geological mapping, archaeological evidence or other historical documents. A common source of error is confusion of the vents of historical eruptions with older but larger or more prominent volcanic vents that lie along the same line of sight as viewed from adjacent settlements or from convenient viewpoints. Examples of this are provided by mis-locations of the vents of the AD 1677 eruption on La Palma and of some of the vents of the AD 1951 eruption on Fogo. Another problem arises when the location or style of eruption on a volcano has changed in early historical time, as has occurred on Fogo. The differences between early historical accounts of eruptions on this volcano and the more detailed accounts of more recent eruptions has led to the discrediting of the former by some researchers, whereas geological studies have supported the early historical accounts.

Most oceanic island volcanoes, and many other dominantly basaltic volcanoes, have a large number of small volcanic vents on their flanks. These vents are typically monogenetic (formed in a single eruption) and feed lava flows that make up the bulk of the volume of the volcano and commonly present the dominant (and certainly most frequent) volcanic hazard. A discrete summit vent

complex is present on some volcanoes of this type: examples include the Teide-Pico Viejo summit complex on Tenerife, Canary Islands (Ancochea etal 1990), and the Pico do Fogo summit cone on Fogo, Cape Verde Islands (see discussion below), In other cases, however, the summit crater complex is a relatively minor feature superimposed on a deeper rift system and largely resulting from


282

S. J. DAY ET AL.

subsidence as a result of dyke injection along the rifts (such as the caldera of Kilauea, in Hawaii; Ryan et al. 1983), and in many cases, such as the Cumbre Vieja volcano of La Palma and the recent volcanic edifice of El Hierro, both in the Canary Islands, there is no summit complex at all (Carracedo 1994, 1996^,6; Carracedo et al. 1997). The summit regions of these volcanoes are instead formed by the intersection of the volcanic rift zones in which the vents are concentrated. These rift zones overlie the swarms of dykes along which the magma erupted at the monogenetic vents ascends to the surface. The positions of the dyke swarms and volcanic rift zones on volcanoes of this type is a function

of their overall structure. It is well established that the orientations of dykes are closely related to the principal stress directions in the regions through which they propagate. The most notable papers on this subject include those by Anderson (1935), Nakamura (1977), Pollard (1987), Chevalier & Verwoerd (1988), McGuire & Pullen (1989) and Tibaldi (1995). Carracedo (1994, 19960, b) has argued that in many cases oceanic island volcanoes have a triple-rift ('Mercedes Star') structure governed by radial deformation above an inflating subvolcanic magma reservoir. However, Carracedo et al. (\991a,b, in prep.) and Day et al. (1996, in prep.) have shown in a detailed study of the Cumbre Vieja volcano,

Fig. 1. Location map of southern La Palma, showing the distribution of historical and young prehistoric vents, with names and ages indicated.

ERUPTIONS ON MULTIPLE-VENT VOLCANOES La Palma, that this triple-rift structure may be modified on time-scales of thousands of years or less in a process of rift reorganization caused by weakening of the edifice and consequent nearsurface dominance of topographic-gravitational stresses (McGuire & Pullen 1989). Such a process may be an important indicator of volcano instability and a long-term precursor to giant lateral collapse of the edifice concerned. The feeder dyke swarms of the Cumbre Vieja volcano, like those of many other highly active volcanoes, are not exposed. In such cases it is necessary to infer their positions and orientations from the distributions of surface volcanic vents and their elongations and other morphological features, using methods described by Nakamura (1977) and Tibaldi (1995), respectively. An important part of this procedure is to identify vents formed in the same eruptions, so as to correctly infer the sub-surface dyke geometry. Where the vents are close together or overlapping it is possible to show that they are co-eruptive by detailed mapping of the vents and their eruptive products. However, where they are more widely spaced it is generally only possible to show that they are related from historical accounts. In a number of cases, such as those of the 1949 eruption of the Cumbre Vieja (Bonelli Rubio 1950), the 1785 and 1951 eruptions of Fogo (Feijo 1786; Ribeiro 1960; see also below) and the three eruptions of 17041706 in Tenerife (Fritsch & Reiss 1868), vents identified as being active in the same eruption are kilometres or even tens of kilometres apart. In some cases, notably the 1949 Cumbre Vieja eruption and the early 18th century activity on Tenerife, even the vent elongation directions are not the same because the vents are fed by dykes emplaced into distinct stress domains. Analysis and reinterpretation of historical accounts of volcanic eruptions in the light of geological studies is a standard procedure in volcanological and volcanic hazard studies. The historical accounts contribute in particular to an understanding of precursory seismicity, toxic gas emissions and other aspects of the eruptions that leave little or no geological evidence at the surface. They also provide important data on the absolute duration and timing of events within periods of eruptive activity. Recent outstanding examples include the reinterpretation of the AD 79 eruption of Vesuvius (Sigurdsson et al. 1982, 1985). However, these studies typically deal with central-vent volcanoes, in which case there is little doubt about the location of the volcanic vents. In the case of multiple-vent volcanoes, in contrast, vent locations are commonly uncertain

283

and different accounts and the geological evidence may provide conflicting vent locations. In such cases it is necessary to be able to test the reliability of historical accounts and related evidence such as local traditions with regard to such basic features of historical eruptions as the location of the vents involved. It should be noted that if a vent is wrongly identified as belonging to a particular historical eruption all subsequent geological work on that vent, and the volcanic rocks erupted from it, will inevitably give an erroneous impression of that eruption. Incorrect location of vents may also obscure the reconfiguration of volcanic rift zones. In this paper we show how historical accounts of various types, including eyewitness accounts, early maps and local traditions in the settlements around the eruption sites, may be tested using geological and archaeological evidence.

The 1677 eruption of the Cumbre Vieja volcano, La Palma Six out of the 12 historical eruptions in the Canary Islands, in the five centuries since occupation of the islands by the Spanish, have occurred on the Cumbre Vieja volcano of La Palma. This volcano forms the southern third of this island, rises almost 2 km above sea level and has a subaerial volume of some 200km 3 ; it also probably extends well below sea level. All radiometrically dated rocks from the Cumbre Vieja are less than 130 ka old (Carracedo et al. 1996, 1991 a, b; Guillou et al. submitted) and it is at present the most active volcano in the archipelago. Detailed mapping (Carracedo et al. 19976, submitted) has shown that since about 7000 years BP, activity of the Cumbre Vieja has been concentrated along the main N-S ridge of the volcano and from E-W trending en echelon fissure swarms on its western flank. Of the historical eruptions, those of 1646, 1677, 1949 and 1971 involved eruptions from the main ridge, as did a prehistoric eruption c. AD 1480 (Hernandez Pacheco & Vails 1982); whereas west flank fissures were active in 1585, 1712 and 1949 (see Fig. 1 for locations of these eruptions). A number of these eruptions have been poorly located, but in this paper we concentrate upon the 1677 eruption. The 1677 eruption occurred close to the town of Fuencaliente at the southern end of the island: the name of the town alludes to a hot spring on the coast nearby, the Fuente Santa, which before its destruction in the eruption formed the basis of a famous medicinal spa. Local tradition (repeated in a number of previous works

284

S. J. DAY ET AL.

(Hernandez Pacheco & Vails 1982; Romero 1991) as well as in tourist literature) has it that the large volcanic cone of San Antonio formed in this eruption. This cone, one of the two or three largest at present visible in the Cumbre Vieja, is mainly composed of scoria and spatter but has a phreatomagmatic surge deposit of several metres thickness preserved on its rim and western flank (see Fig. 4, below). The existence of this deposit provided the initial

geological clue that the local tradition might be in error. The occurrence of a surge-forming explosive phase in the 1677 eruption would have inflicted considerable damage upon Fuencaliente, less than 1 km away. No such disaster has been recorded. Following the recognition of this anomaly, detailed geological mapping, radiometric dating and a search for eyewitness accounts of the eruption were carried out. The results of this work have been described in

Fig. 2. Map of the pre-1971 geology of southernmost La Palma, showing the various units of the 1677 eruption.

ERUPTIONS ON MULTIPLE-VENT VOLCANOES detail by Carracedo et al (1996); here they are summarized and compared with the results of archaeological investigations carried out independently by the fourth author.

Geological mapping and radiometric dating Geological mapping was carried out partly on the ground and partly by photogeological interpretation of pre-1971 aerial photographs, as a

285

large part of the 1677 lava field was covered by the products of the 1971 Teneguia eruption (Afonso et al. 1974). The results of this work are shown in Fig. 1. The mapping shows that the two youngest pre-1971 vents in this area are a small, NW-SE elongated scoria cone produced by strombolian eruptive activity on the NE side of the San Antonio cone at about 600m above sea level, and a group of spatter vent fissures on the SW side of the San Antonio cone at about 500m

Fig. 3. Views of the 1677 eruption vents and of prehistoric rocks around the San Antonio cone, (a) General view from Fuencaliente of the San Antonio cone and of the upper (northeastern) vent of the 1677 eruption. 'Photo' marks the location of (d). (b) View of the cluster of spatter vents of the 1677 eruption, located on the southwest side of the San Antonio cone. The proximity of these vents to an old sea-cliff should be noted; the coastal exposures of the 1677 lavas lie at the foot of this cliff (see Fig. 2). (c) View from the southwest rim of the interior of the San Antonio crater, with Fuencaliente in the background. The collapse of the southern flank of the 1677 strombolian cone and of the older phreatomagmatic deposits have filled the crater to its present depth, (d) Road cut in the flat ground between the San Antonio cone and Fuencaliente (see (a)), showing strombolian pyroclastic deposits (lapilli and scoria) of the 1677 eruption resting directly on Fuencaliente vents lavas, without intervening pyroclastic deposits from the San Antonio cone, whose rim is less than 200 m from this point.

286

S. J. DAY ET AL.

Fig. 3. (continued}

above sea level (Fig. 3a and b). The latter fed a total of eight discrete lava flows, which together make up the young pre-1971 lava field and lava platform at the coast (Fig. 2). Deposits of the upper, northeastern vent are well exposed on the NE rim of the San Antonio crater, suggesting that they have in part collapsed into it, and clearly overlie weathered phreatomagmatic deposits and other rocks of the old cone (Fig. 3c). There is no evidence to support the interpretation (Hernandez Pacheco & Vails 1982) of young vents on the coast about 2km north west of Fuencaliente as also having been formed in the 1677 eruption. These vents appear to be hornitos or rootless vents in a coastal lava platform produced by lavas erupted from a

group of sub-historical vents upslope from Las Indias: one of these lavas has been dated at 3 ± 2 ka using the K-Ar technique (Carracedo et al \991a, b; Guillou et al submitted). The San Antonio cone can also be shown to be older than a number of other eruptive units in the area. The cone is a substantial topographic feature, rising some 200m above the surrounding terrain on its south and west sides and some 1200m across. A number of younger lavas from vents located within the present-day town of Fuencaliente (Fuencaliente vents lavas in Fig. 2) flowed downhill to the San Antonio cone and were deflected around it; they also overlie a lava erupted from the foot of the San Antonio cone on its western side. They are therefore inferred

ERUPTIONS ON MULTIPLE-VENT VOLCANOES to be younger than it, and this is confirmed by the lack of phreatomagmatic surge deposits lying on top of the Fuencaliente vents lavas (Fig. 3d). These deposits are at present confined to the rim and western flank of the cone (Fig. 4) but their thickness and well-developed low-angle surge cross-bedding indicate the occurrence of a violently explosive phase towards the end of the eruption that formed the San Antonio cone,

287

which would have spread thinner, more distal equivalents of these deposits over a wide area. Neither the San Antonio cone nor the Fuencaliente vents lavas have been dated as yet, but a still younger group of lavas, the Montana del Fuego lavas, have been dated at 4 ± 3 k a (2cr error) by the K-Ar method and by two independent 14C datings, 3255 ±140 a and 3350 ± 50 a (2cr errors) (Carracedo et al. 1991 a, b;

Fig. 4. Phreatomagmatic deposits of the San Antonio cone. (A) Bedded phreatomagmatic deposits perched on the western flank of the San Antonio cone; (B) surge cross-bedding within these deposits.

288

S. J. DAY ET AL.

Guillou et al. submitted). It is therefore likely that the San Antonio cone is substantially more than 3000 years old, consistent with the welldeveloped weathering beneath the younger strombolian deposit (Fig. 3c). The geological evidence therefore indicates that the San Antonio cone is a relatively old feature and the most plausible candidates for the vents of the 1677 eruption are the much smaller vents on the NE and SW sides of the San Antonio cone. In contrast to the San Antonio eruption, which had a violently explosive phase as noted above, the relatively mild effusive and strombolian activity indicated by the geological features of these vents is consistent with the eyewitness accounts of the eruption, as will be shown below. Archaeological evidence Further evidence for the pre-1677 origin of the San Antonio cone comes from archaeological excavations around the cone. The pre-Hispanic

culture of the Canary Islands, associated with a people of North African extraction, the Guanches, disappeared very soon after the occupation of La Palma by the Spanish in the last decade of the 15th century AD. Occurrences of Guanche pottery and other artefacts therefore provide a stratigraphic indicator of a pre-1500 age. Figure 5 shows the distribution of Guanche finds around the San Antonio cone, including pottery; petroglyphs (distinctive carvings on rock faces); post-hole circles and other remains of the foundations of clusters of Guanche dwellings; and artificial caves excavated in soft but cohesive pyroclastic units such as the phreatomagmatic ashes and used as refuges by the Guanches. The distribution of these artefacts and dwelling-sites clearly indicates the pre-Hispanic age of the San Antonio cone and the Fuencaliente vents lavas. The only units that demonstrably overlie Guanche remains are the lava flows from the SW vents of the 1677 eruption.

Fig. 5. Map of the distribution of Guanche archaeological sites and finds around the San Antonio cone.

ERUPTIONS ON MULTIPLE-VENT VOLCANOES

Eye-witness accounts of the eruption and other historical documents The earliest document relevant to the location of the 1677 eruption dates from some 80 years earlier. A map produced in 1586 and accompanying an account of the 1585 eruption at Jedey, some 10km north of Fuencaliente (Torriani 1592), shows a hill named San Antonio just south of Fuencaliente and in approximately the same location as the volcanic cone given that name today. Torriani was a military engineer tasked with construction of fortifications in the island; his skills would have included surveying. Other named topographic features on his map are correctly located, giving some confidence in the accuracy of his positioning of the San Antonio cone. This documentary evidence for the existence of the San Antonio cone before 1585 is consistent with the geological and archaeological evidence that it pre-dates the occupation of La Palma by the Spanish. The contemporary accounts of the 1677 eruption were written by the civil and ecclesiastical authorities of the area and must therefore be interpreted with care. None the less four accounts in particular have survived which can be used in conjunction with geological evidence to reconstruct the course of the eruption: •

•

•

•

Relaciones (Accounts) written by the priest Juan Pinto de Guisla, describing the 1646 (Volcan Martin) and 1677 eruptions of the Cumbre Vieja. A copy of the original, transcribed in 1806, has been given by Lorenzo Rodriguez (1987); translations into German and French were published by Von Buch (1825) and Von Buch (1836), respectively. Gazeta del Ayuntamiento de La Palma (Gazette of the La Palma community council) of 2 December 1677, describing the development of the eruption from 13 to 26 November 1677 and written by an anonymous council official (Anonymous 1677). A copy of the original again has been published by Lorenzo Rodriguez (1987); a translation into French has been published by Webb & Berthelot(1839). Cartas (letters) sent by the chief bailiff of the Spanish Inquisition in La Palma, Antonio Pinto de Guisla, to Bishop B. Garcia Ximenez, on 29 November, 6 and 24 December 1677, and 17 and 27 January 1678. Carta sent by the vicar Melchior Brier to the same Bishop, dated 30 November 1677. The originals of these last two have been lost but copies are preserved in the archives of the Bishop (Anonymous 1678).

289

The single most important piece of information in these accounts describes how the people of Fuencaliente went to the top of Montana del Corral (the name used for Montana San Antonio in that account) to view vents that had opened at its base and that, fortunately after they had returned, a large vent opened in the mountain, which 'had it caught them unawares would have engulfed them in a voracious fire'. This clearly indicates that two vents were involved in the eruption, a lower one at the base of the San Antonio cone and an upper one near its summit. These positions match those of the two young vents on either side of the San Antonio crater and indicate that both were active in the 1677 eruption. The various accounts describe the early phases of the eruption, from its start near sunset on 17 November 1677 until the destruction of the Fuente Santa by lava flows on 26 November, in some detail. The eruption was preceded by intense seismic activity from 13 to 15 November, and emission of 'hot air with a smell of sulphur' from fissures that developed on either side of Montana del Corral/Montana San Antonio from 15 November onwards. These fissures were described as opening in a few minutes to a width 'difficult to jump'. Their opening was accompanied by strong earthquakes, which also caused the collapse of the church tower in Fuencaliente. The earthquakes, and the unusual opening of the fissures some days before to the eruption, may reflect bulging of the steep slopes south of Fuencaliente as a dyke was emplaced beneath them. The 1677 vents lie on an overall trend bearing about 030°, which may indicate the orientation of this dyke at depth, whereas the NW-SE trend of the individual vents (Fig. 2) is parallel to the local topographic contours and indicates the influence of topographicgravitational stresses. Subsequent events are shown in map and schematic form in Figs 6 and 7. The Fuente Santa was destroyed on 26 November, apparently by the southern branch of flow LIII-2 (Fig. 2), and with the loss of this important economic resource interest in the activity declined. It will be noted that two out of the four sets of documents listed above deal solely with the first 2 weeks of the eruption, which continued for a further 2 months afterwards. It appears that once the most serious damage had been done, interest in the eruption declined as no further resources of note were threatened. This is the converse of the situation in the later and much larger 1730-1736 eruption of Lanzarote (Carracedo & Rodriguez Badiola 1991; Carracedo et al 1992), where the poorly

290

S. J. DAY ET AL.

Fig. 6. Maps showing the different stages in the development of the San Antonio area in general and of the 1677 eruption in particular; it should be noted that the distribution of lavas in areas covered by the 1971 Teneguia eruption is based on interpretation of pre-1971 aerial photographs.


291

Fig. 7. Time-line depicting the events of the 1677 eruption, based on the historical eyewitness accounts cited in the text.

documented post-1731 phases of the eruption occurred after complete evacuation of the area of the eruption and an at least partial breakdown of the local authority. Throughout the eruption the upper vent appears to have emitted lapilli scoria and spatter in strombolian style activity, with three pronounced explosions toward the end of the first

stage of activity on 23 November. These explosions were preceded by seismic activity, which may record drainback of magma from the upper vent, as activity at the upper vent ceased for 2 days from 23 to 25 November. Gases (most probably carbon dioxide) emitted from the upper vent seem to have ponded in depressions in the hummocky ground between San Antonio

292

S. J. DAY ET AL.

and Fuencaliente itself: the deaths by asphyxiation of cattle, rabbits, birds and a shepherd were recorded in December 1677. The lower vents erupted a series of lava flows from successive vents slightly offset from one another (Figs 2 and 6), with accumulations of spatter around the vents themselves. In one of the earliest recorded experiments in volcanology, markers were emplaced at known intervals in advance of a lava flow front, and the velocity of its advance was estimated as about 20 m in half an hour, or about 0.01 m s"1; this value is similar to more recent estimates of flow front velocity for basaltic lava flows (Kilburn et al. 1995). There may have been hiatuses or periods of reduced effusion rate at the lower vents, but details are scarce. The last phase of the eruption, producing the L IV-1 and -2 flows shown in Fig. 2, ended abruptly on 21 January 1678.

Why was the San Antonio cone erroneously identified as the vent of the 1677 eruption? Although the detailed mapping and other studies described above were carried out only very recently, the eyewitness accounts of the 1677 eruption and other relevant documents were not lost, and indeed were widely published and translated in the early 19th century, by Von Buch (1825, 1836) in particular. The problem, therefore, is why the main San Antonio cone has come to be identified as the vent of the 1677 eruption in local tradition and both the scientific and popular literature. A hint is provided by the view of the San Antonio cone from the north side of Fuencaliente, shown in Fig. 3a. In this view the northeastern vent of the 1677 eruption lies just in front of the older cone, whereas the lower, southwestern vents are on the opposite side (see also Fig. 2). Thus the 1677 vents and the older San Antonio cone all lie along the same line of sight from Fuencaliente, and furthermore the older cone is the most prominent topographic feature in that direction. It is therefore easy to become confused about what precisely is being pointed out when the position of the eruption is indicated from viewpoints in the vicinity of Fuencaliente. Although this 'line of sight' error could have been corrected on the basis of a detailed examination of the contemporary eyewitness accounts alone, or on the basis of the geological and archaeological evidence alone, we emphasize that the use of both archival and field work together provides a much stronger case for the interpretation of the geology of the San Antonio

cone and its surroundings that we have outlined here, and has removed many ambiguities from the interpretation. The reconfiguration of the Cha das Caldeiras volcano, Cape Verde Islands, within historical time and consequences for interpretation of early historical accounts The Cha das Caldeiras volcano (also known as the Pico do Fogo, but this name is misleading for reasons discussed by Day et al. (submitted)) is on the island of Fogo in the Cape Verde archipelago. It is by far the most active volcano in that archipelago and the only one from which historical eruptions have been recorded. Like the Cumbre Vieja, it is an essentially alkali basaltic volcano, and during the five centuries since first colonization of the Cape Verdes by the Portuguese as many as 26 eruptions have been recorded. Although some of these records are doubtful, it is likely that some minor eruptions may have gone unrecorded in the earlier part of the historical period. This is because organized colonization of the eastern, volcanically active part of the island did not take place until the end of the 18th century, and the remote region within which the summit of the active volcano lies was not occupied until several decades later still. For much of the earlier period, settlement was confined to the vicinity of Sao Felipe in the southwest of the island. The 1995 eruption on Fogo has led to a number of studies at present at various stages of completion (IICT 1997; Heleno da Silva & Fonseca 1999). An interim account of the eruption has been provided by Silveira et al. (1995). Older published works include those by Machado (19650, b), Machado & Torre de Assuncao (1965) and Torre de Assuncao et al. (1967), but perhaps the most detailed work on the island is that by Ribeiro (1960). This provides a detailed description of the geography of the island, as well as an invaluable summary of the historical accounts of volcanic and seismic activity and a detailed description of the 1951 eruption. Many of the eyewitness accounts of early (pre1785) historical eruptions found by Ribeiro were written by sea-captains and other mariners. The first scientific account of an eruption was written by a chemist and natural scientist, Feijo, who wrote a valuable account of the 1785 eruption (Feijo 1786), which is discussed further below. Although he catalogued the earlier accounts, Ribeiro (1960) questioned their accuracy because the descriptions were inconsistent with

ERUPTIONS ON MULTIPLE-VENT VOLCANOES the locations and styles of activity of the more recent and more precisely described eruptions. It is therefore of interest to consider whether the geology of the volcano supports Ribeiro's scepticism or whether it indicates that a change in the style and distribution of activity has genuinely taken place. The more recent activity has involved multiple-vent eruptions and the formation of numerous monogenetic volcanic vents and provides a number of instances where vents have been mis-located and lost entirely for reasons that are also of interest in the context of this paper. The geology of Fogo island and the Cha das Caldeiras volcano The Cape Verde archipelago as a whole is associated with a mantle plume: the best account of its tectonic setting has been provided by White (1989). The present-day location of the plume is inferred from the geoid anomaly and other geophysical data to be in the southwest of the archipelago, close to the position of Fogo island and the adjacent island of Brava, which contains numerous morphologically young volcanic vents and has experienced numerous episodes of seismic unrest in historical time (Heleno da Silva et al 1997) although no eruptions have occurred there. Fogo island is located at the eastern end of a submarine platform connecting it and Brava to the west. It is roughly circular in shape and some 25km across. Limited exposures of basement rocks on both islands indicate that this platform is composed of a pyroxenite-carbonatite intrusive complex and associated volcanic rocks. Built up on this platform and its steep submarine flanks and forming most of the island is a steepsided volcanic edifice, the Monte Amarelo volcano. This edifice is composed of alkaline basic and intermediate lavas (nephelinites, basanites and tephrites), with rare phonolitic lavas and domes. These rocks were erupted from an inferred small summit vent complex and from well-developed volcanic rift zones. These zones correspond to dyke swarms trending c. 030°, 150°, 240° and 300° (the last two possibly forming a single broad east-west-trending rift zone) which are exposed in the near-vertical cliffs, up to 1 km high, that bound an east-facing collapse structure some 8km wide and extending 10 km inland from the coast of the island. This collapse structure removed the summit of the Monte Amarelo volcano, which was originally some 3 km high above sea level: the highest point on the rim of the collapse structure, Ponto Alto

293

do Norte, is at present some 2650m above sea level. Since this collapse the collapse scar has been largely filled in by the growth of a younger volcanic edifice, the Cha das Caldeiras volcano (Fig. 8). The most prominent feature of the Cha das Caldeiras volcano is a steep-sided summit cone, the Pico do Fogo itself, which rises to 2829m above sea level and some 1200m above the general level of the volcano. This cone is located in the centre of the Monte Amarelo collapse scar, with its summit only just over 5 km from the coast: the average slope on this side of the island between the summit and the sea is no less than 28°. The summit cone consists of thick units of volcanic spatter, spatter-fed lavas, coarse scoria and thin-bedded lapilli and yellow phreatomagmatic ash. It has not erupted since 1785 and has since that time undergone considerable erosion by slope failures and rockfalls, particularly in the vicinity of a N-trending zone of active fumaroles on its northern side and in the headwalls of gully systems on its eastern side. The sequences exposed in these slope failure scars are discussed further below in the context of interpretation of the early historical accounts. The growth of this summit cone has isolated the western part of the collapse scar from the sea, and ponding of lavas between the cone and the cliffs of the collapse scar has produced a plain, the Cha das Caldeiras, which is between 1.6 and 1.8km above sea level. 'Cha das Caldeiras' translates as Plain of Craters, referring to the many monogenetic scoria and spatter cones that have formed on it. The locations and vent elongation directions of these vents are shown in Fig. 9. Many of these vents form the up-rift end of inferred volcanic rift zones that extend outside the collapse structure as indicated by the locations (Fig. 8) and elongation trends of young, post-collapse monogenetic scoria and spatter cones and associated lavas on the outer slopes of the old Monte Amarelo volcano. The most active zones trend along approximate bearings of 030°, 150° and 240-270°, broadly corresponding to the rift zones of the Monte Amarelo volcano. However, with the exception of the 1995 eruption to the SW of the summit cone, which took place along the 240°-trending rift system, the most recent eruptions (identified from post1785 eyewitness accounts and the results of reconnaissance geological mapping, which are generally consistent with one another as discussed in the next section) are confined to within the Monte Amarelo collapse structure and are aligned along N-S-trending fissures, which commonly are arranged in en echelon systems. The

294

S. J. DAY ET AL.

Fig. 8. Sketch map showing the principal geological features of Fogo, Cape Verde Islands.

sense of offset within these en echelon sets indicates that the dykes feeding these systems lie along the old 030°- and 150°-trending swarms at depth but rotate as they propagate upwards into conformity with an east-west extensional stress system. This pattern is most easily explained in terms of eastward (seaward) displacement, during eruptions, of a block bounded by the fissure system on the west and strike-slip faults along the

boundaries of the older collapse structure. The implication that the eastern flank of Fogo may have recently entered a phase of instability makes it especially important to determine whether the change in eruptive style implied by the discrepancy between the early historical accounts of eruptions and the more recent activity is real or, as suggested by Ribeiro (1960), a symptom of the unreliability of these prescientific accounts.


295

Table 1. Summary of early historical eruptions ofFogo, based on citations of original sources by Ribeiro (1960) and Machado (1965b); further details are given by those workers Date

Location of vents

Eruptive phenomena

Productions of eruption

1500?

Summit and both flanks

Explosive activity, with sustained paroxysms

Dust, coarse ash and scoria* covering entire island

1564

Summit cone?

Explosive

All of island covered in ash

1580-1585?

Summit cone

Explosive and effusive; 'glowing and giving off flames by day and night'; continuous immense 'flames'; 'rivers of fire' emitted by volcano

Lava flows; probable spatter around summit (note reference to incandescence).

1596

Summit cone?

Explosive

Major ash fall in northeast of island and at sea

1604

Summit and flanks of summit cone

'Flames and sulphurous vapours'

1662 or 1663? (account dated 1664)

Summit and two other vents

Explosive and effusive

1680

Summit?

Major earthquakes; large eruption with great explosion of 'lavas'

1683? (account of 1680 eruption)

Summit

'Great flames'

1689

Summit (and flanks?)

Explosive; 'flames', smoke, sulphurous clouds

Scoria* deposits covering the sea (?)

Explosive; 'smoke' during day, clouds of incandescent sparks at night

Scoria* deposits

'Ash' and rocks ejected from volcano (displacing people from the island)

Large incandescent blocks (like iron slag) rolling down slopes; lava flows entered the sea

Large eruption

1675

1693

Ash fall in west? (Migration of inhabitants to Brava)

1695

Summit?

Explosive; 'fire' visible at night, 'smoke' during day

1697

Summit or upper flanks

Explosive; 'flames' emitted from heights of volcano

1699

Summit

Explosive; thick clouds of 'smoke' visible 50 km away during day; 'flames' at night

1712 (start of 1713 eruption?)

Summit

'Smoke'

1713

Summit

Explosive; 'immense' clouds of 'smoke', visible 90km away on clear days; 'flames' visible at night

Between 1721 and 1725

Summit

Explosive and effusive; large blocks ejected to great height and torrents of 'sulphur' flowing down flanks of cone

Incandescent blocks (spatter), lava flows, cinders (scoria)

1761

?

1769 or 1774

South flank of summit cone

Effusive?

Lavas?

* Described as 'pedra-pomes' (pumice) in original accounts, but as no pumice is found on Fogo this is most likely to refer to highly vesicular basaltic scoria and has been interpreted accordingly.

296

S. J. DAY ET AL.

Early historical accounts of eruptions of the Cha das Caldeiras volcano The eruptions from about 1500 to 1769 and the sources of accounts of them are summarized in Table 1. Table 1 is based upon the versions of these accounts given by Ribeiro (1960) and Machado (19656). Most of these accounts, as noted above, were written by sea-captains and other mariners (including the pirates Anthony Sherley and William Dampier). Ribeiro (1960) failed to find any accounts written by local authorities. The main points that these various accounts reveal can be summarized as follows: •

Eruptions were prolonged and frequent for much of the period in question, although there appear to have been periods of repose or relative quiet from 1500 to 1564 and from 1604 to 1662 or 1663. Most accounts indicate that the eruptions were regarded by the inhabitants of Santiago as normal or frequent occurrences, suggesting that activity was semi-continuous for much of the period. However, strong earthquakes in 1680, associated with a large eruption, led to emigration of some of the inhabitants of Fogo to Brava. Furthermore, the oft-quoted use of the Pico do Fogo as a 'lighthouse' by mariners is based only on the account of the 1662-1663 eruption ('at night it is a mariner's lighthouse by reason of the flames which are thrown without cease from a very high peak') by Andre de Faro (1664; republished in 1945 and quoted by Machado (19656)). It is not clear from this whether de Faro intended to state that the volcano was actually in regular use as a navigation aid or if he sought to draw an analogy with the fires on high headlands that were used as navigator's lights at the time. • Many accounts referred to explosive eruptive activity. Falls of 'cinders' and 'stones' both on the island and at sea around it are frequently mentioned. Incandescent rocks and large blocks are mentioned as being ejected over large areas in the accounts of eruptions in 1662 or 1663 and 1721-1725(7), and the eruption in the period 1580-1585 may Have been similar. As noted above, Ribeiro (1960) questioned the veracity of these accounts, in particular that of the eruption in the period 1721-1725 (by an Englishman resident in the Cape Verdes in that period, Roberts), in part because the style of activity is very different from that of more recent eruptions. In contrast, accounts of eruptions in the period 1675-1713 refer instead to

generation of high eruption columns, described as 'thick clouds' or 'clouds of smoke', visible several tens of kilometres away, indicating a change in the style of activity in that period, the nature of which is discussed further below. During eruptions in this period the glow from incandescent material, generally referred to as 'fire' or 'flames', was visible only at night. The eruption of 1596 is also notable as having produced a widespread fall of ash. • Some accounts also referred to lava flows descending to the coast down the steep eastern slope of the island. In some cases (eruption of 1604) these were considered to originate from multiple vents. In most other cases the sites of vents were not clearly defined, although the eruption of 1769 has been located at Monte Laipo or Monte Lorna on the SE side of the Pico do Fogo, on the basis of a comment in Feijo's account of the 1785 eruption to the effect that the previous eruption was on the south side of Pico do Fogo. There are a number of very young vents in this area that could be the 1769 vent(s) (Fig. 9). • The activity was generally considered to originate from Pico do Fogo and certainly from within the old collapse structure (hence the poor view of the eruptions from Sao Felipe, as noted above). Ribeiro considered it likely that, in fact, most eruptions occurred on the flanks of Pico do Fogo and in the Cha das Caldeiras. He was particularly sceptical of the account of Roberts, which described flows of 'sulphur' descending the cone of Pico like 'torrents of water', and of Roberts' claim that the Pico had not existed before the eruption in the 1721-1725 period. Despite Ribeiro's scepticism regarding the reliability of these early accounts, their nonscientific nature and the genuine lack of information regarding the topography of the interior of the island, it is none the less possible to relate many features of the accounts to the geology of the Cha das Caldeiras volcano and of the Pico do Fogo itself. Perhaps most interesting is the emphasis in many accounts upon more or less explosive activity at the summit of the Pico, with the ejection of large amounts of incandescent rock in many eruptions. Allowing for the cultural background of the observers, these descriptions match the inferred mode of eruption of the welded spatter and (probably spatter-fed) lava units that make up much of the sequences exposed in the gullies on the flanks of the Pico do Fogo summit cone.


297

Fig. 9. Map of vent locations and elongation directions around the Pico do Fogo and Cha das Caldeiras, with dates of identified historical vents.

298

S. J. DAY ET AL.

The sequence in these gullies shows a consistent stratigraphic sequence, which is exposed around the northern and eastern flanks of the volcano. Part of this sequence is shown in Fig. 10. The uppermost part of the sequence is composed of two or more coarse spatter and spatter-fed lava flow units, the uppermost of which is some 20-30 m thick (Fig. 11) and can be traced around the entire summit of Pico do Fogo. These welded units are separated by blocky scoria beds. Below them is about 30m of dark lapilli and yellowish to grey finely laminated ash in a plane-bedded sequence, which in turn overlies further spatter and spatter-fed lava. This sequence can be matched to variations in the descriptions of the historic eruptions that are summarized in Table 1. Accounts of the eruptions of 1662 or 1663 and of 1721-1725 place special emphasis upon the eruption of abundant 'Pedras incandescentes' ('incandescent rocks', interpreted as meaning spatter), 'pedra-pomes' (literally, pumice, but more likely to refer to vesicular scoria, as no true pumice occurs in any of the recent rocks of Fogo) and lava flows from the summit of the volcano. Roberts' account of the eruption that occurred between 1721 and 1725

indicates that this final major summit eruption was especially voluminous: this matches the uppermost spatter and spatter-fed lava unit in the sequence exposed around the summit of the cone. If these uppermost lava and welded spatter units are attributed to a major eruption in 17211725, then the thinner units below may be related to eruptions around 1700, and the finely bedded ash and lapilli unit exposed beneath these on the north and northeastern flanks of the Pico can be attributed to eruptive activity in the period 1675-1680, when the accounts refer more to clouds of smoke, earthquakes and great explosions. The eruption of this largely phreatomagmatic unit would have produced black rather than incandescent eruption columns and widespread ash and lapilli falls. This may have contributed to emigration from the island during the eruption of 1680. The accounts of the 1662-1663 eruption, when the ejection of 'large glowing rocks' is recorded, are consistent with the presence of further spatter and spatter-fed lava units below the ash and lapilli unit (Fig. 10). Alternatively, the finely bedded ash and lapilli unit could correspond to the eruptions of 1500(?) or 1596, notable for the wide distribution of ashfall deposits, but in this case a greater number of

Fig. 10. View of western wall of main gully on north side of Pico do Fogo (see Fig. 9 for location) showing sequence consisting of upper spatter and lava unit, middle finely bedded lapilli and phreatomagmatic ash unit (c. 30m thick), and lower spatter and lava unit. (See text for correlations of these units with historical eruptions.)


299

Fig. 11. Main (>20m thick) spatter and clastogenic lava unit on north flank of Pico do Fogo, close to summit (see Fig. 9 for location), interpreted as having formed in an eruption between 1721 and 1725. This same unit forms the irregular crags all around the summit of Pico do Fogo (Fig. 15).

discrete units of spatter and/or scoria would be expected to occur above it in the sequence beneath the 1721-1725 spatter unit. Whatever the exact dates of formation of these eruptive units the existence of discrete units in the sequence making up the Pico implies a number of similarly discrete eruptions separated by periods of repose. This is consistent with Ribeiro (1960), who interpreted the eyewitness accounts as indicating discrete eruptions, and inconsistent with other interpretations (such as that by Machado, 19650, b) that attach more significance to the description of the Pico as a 'mariner's lighthouse' and consider the early historical activity to have been continuous. In this context it is also to be noted that the welded spatter and spatter-fed lava units that make up much of the Pico would have to be erupted at a very high rate to undergo welding and rheomorphism; such an eruption rate could not be maintained continuously or otherwise the Pico would be far larger than it actually is. The many points of agreement between the early historical accounts and geological features of the sequence forming the Pico do Fogo summit cone, and in particular the match between the variations between accounts of different eruptions and the sequences of different lithological

units, make it clear both that the Pico do Fogo summit cone was the principal site of activity in the early historical period and that the early historical accounts are in fact accurate in many respects. The scepticism of Ribeiro (1960) regarding the veracity of these accounts does not seem to be justified by the geological observations described here.

Recent flank eruptions of the Cha das Caldeiras volcano: more examples of the line-of-sight error The last eruption on Fogo to involve activity at the summit crater of the Pico do Fogo was that of 1785. This eruption, which is also the first eruption of the volcano to be described in detail (Feijo (1786); this account has been transcribed in full by Ribeiro (I960)), was primarily a fissure eruption involving activity at as many as nine vents along the SW-NE-trending volcanic rift zone. Three of these vents were located to the north of the northern boundary of the older collapse structure; the 1785 eruption was also the most recent eruption to involve vents outside this structure. The summit activity appears to

300

S. J. DAY ET AL.

Fig. 12. Sketch map of Fogo island showing the Cha das Caldeiras and the vents of the 1785 eruption, from Feijo (1786). Parts of original key translated from the original.

ERUPTIONS ON MULTIPLE-VENT VOLCANOES have been relatively minor: Feijo describes the generation of a small dark (non-incandescent) eruption column, which deposited fine ash at the start of the eruption. No deposits from this activity are now recognizable at the summit of Pico do Fogo, most probably because any surfaces flat enough for it to have accumulated and been preserved have since been covered by fresh lapilli from more recent eruptions, notably those of 1951 and 1995. Feijo's account included a sketch map and a sketch view of the volcano from the east, reproduced here as Figs 12 and 13. These are sufficiently detailed that the vents of the 1785 eruption can be identified with some confidence (Fig. 9) and the general 030° trend of both vent elongation directions and the overall alignment of vents recognized. A feature of particular interest in the account written by Feijo is the occurrence of vents close to the village of Mosteiros on the northeast coast of the island. A number of flows around Mosteiros still lack significant vegetation, even though they are located on the wet side of the island. Despite the youthful appearance of these flows and the clear evidence for their formation in 1785 from Feijo's account, they have disappeared from most recent literature: they do not, for example, appear in the maps of historical lava flows contained in papers by Machado (I965a,b), Machado & Torre de Assuncao (1965) and Silveira et al (1995). The subsequent eruptions, of 1799, 1816, 1847, 1852 and 1857, all involved relatively small groups of vents, or single vents, mainly to the north of Pico do Fogo (the exception being the 1857 eruption, which occurred southeast of the Pico (Ribeiro (1960; see Fig. 9). Owing to the proximity of the different vents and a thick blanket of lapilli from the 1951 and 1995 eruptions, the identities of the different eruption sites are not entirely clear: particular problems

Fig. 13. Sketch of Fogo in the first stages of the 1785 eruption, viewed from the east; from Feijo (1786).

301

are associated with vents that may date from 1816 or 1847, and the locations of these differ from those presented by Torres et al. (1997). The evidence for the vent locations in Fig. 9 will be discussed in more detail elsewhere (Day et al. in review). It is, however, clear that most of these vents are aligned along a north-south dyke swarm distinct from that feeding the older NEor NNE-trending vents, such as those of the 1785 eruption, which are cross-cut by these younger vents. The interpretations of the sites of the 1951 eruption have varied rather more despite the recent date of this eruption and extensive observations of the later phases of the eruption in particular. Two vents, Monte Preto de Cima on the northwest side of the Pico do Fogo and Monte Orlando to the south (Fig. 9), have been consistently recognized as being active in 1951, and the very fresh appearance of these vents and their products and the fact that the only deposits that overlie them are from the 1995 eruption are entirely consistent with this. However, two other vents appear to have been active in the early phases of the eruption, and recognition of the correct location and even existence of these has been uneven. The first documents relating to the 1951 eruption are an account, and accompanying oblique aerial photographs, by Francisco Mendes, the meteorologist based on the island of Sal at the time. The photographs were taken on the second day of the eruption, 13 June 1951: Mendes' sketches based on these appear here as Fig. 14. Taken in a sequence as the aircraft flew from north to south towards and past the east coast of the island, descending as it did so, the photographs indicate the presence of two very active vents on the south side of the volcano aligned in a broadly southeasterly direction and a less-active vent on the west side of the volcano. This vent appears to be somewhat to the south and west of the position of Monte Preto de Cima, which in the view of Fig. 14c would appear on the Cha das Caldeiras approximately in line with the high point on the rim of the old collapse structure, Ponto Alto do Norte, rather than being hidden from view by the Pico do Fogo. The existence of vents in the western part of the Cha das Caldeiras during the 1951 eruption is confirmed by geological observations (Torres et al. 1997; Day in prep.). The flows from the undoubted 1951 vents at Monte Preto de Cima (Fig. 9) merge with flows that originate from a north-trending line of very small vents to the southwest, on the lower western slopes of Pico do Fogo. It appears that these vents were those

302

S. J. DAY ET AL.

Fig. 14. Sketches by Francisco Mendes, based on his aerial photographs, showing the vents of the 1951 eruption on the afternoon of 13 June 1951 (second day of the eruption), with names of topographic features added. (See text for discussion.) active on the second day of the eruption and that the dyke feeding them subsequently propagated further to the north and opened the Monte Preto de Cima vent. As Monte Preto de Cima is markedly lower than the vents to the southwest, it is likely that activity at the higher vents would have then ended. This must have occurred early in the eruption, before systematic observations of the activity began on 26 June: Monte Preto de Cima was certainly very largely in existence by 28 June. The vents to the southwest are small and have, since 1951, been partly covered by rockfalls from the Pico do Fogo and more recently by lapilli from the 1995 eruption. It is therefore perhaps not entirely surprising that they were not recognized by, amongst others, Ribeiro (1960), although Machado & Torre de Assuncao (1965) and Torre de Assuncao et al. (1967) indicated them as possibly having formed in 1857; this is inconsistent with their merging relationship with the 1951 Monte Preto de Cima flows. The first to correctly identify them as 1951 flows were Silveira et al. (1995). A more remarkable error has arisen with respect to the second 1951 vent, besides Monte Orlando, on the southern side of the Pico do Fogo. The eyewitness accounts by Mendes and others, including the island administrator L. S. Rendall, which were summarized by Ribeiro

(1960) all clearly indicate that this second vent was to the southeast of Monte Orlando on the floor of the Cha das Caldeiras, as in Ribeiro's own account and photographs. Monte Rendall, as the vent was named, ceased its activity well before the end of activity at Monte Orlando and was subsequently partly destroyed by collapse into the extensive Monte Orlando lava flows. It now forms a low hill isolated within the 1951 lava flows (Fig. 15), chiefly distinguished by the presence of a dry fissure system, which also extends well to the south of the cone to the site of a possible third vent, which may have been active in the very early stages of the eruption but is now almost completely buried by the Monte Orlando lavas (Fig. 9). The location and appearance of Monte Rendall has been nevertheless well documented by Ribeiro (1960) and it is therefore remarkable that as early as the mid-1960s, in the papers by Machado (1965#), Machado & Torre de Assuncao (1965) and Torre de Assuncao et al. (1967), the location of Monte Rendall was becoming confused. This situation has persisted to the present, the most recent published case being the paper by Silveira et al. (1995). That study in particular clearly identified 'Monte Rendall' as a vent due east of Monte Orlando and also situated on the lower slopes of Pico do Fogo (Fig. 16; see also Fig. 15). This vent is that named Monte Lantisco on older maps and by Ribeiro (1960). It is markedly older than Monte Orlando: lapilli from Monte Orlando overlie a well-developed weathering surface on the rim and western flank of the vent. In addition, lavas from a vent higher on the flank of Pico do Fogo (behind and slightly to the left of Monte Lantisco in Fig. 15) enter the breached northern crater of Monte Lantisco and curve around its western flank, where they disappear beneath 1951 scoria and lava flows. These lavas are also overlain by boulder screes produced by rockfalls from the slopes of Pico do Fogo. These field relationships are shown in Fig. 16. The flows concerned may well be those that formed in the eruption of 1769 or possibly 1774 (Ribeiro 1960; Torres et al. 1997). The reason for the confusion appears to be the same as that which has caused confusion on La Palma and is apparent from Fig. 15. The only road into the Cha das Caldeiras runs along the foot of the southern boundary cliff. Monte Rendall and Monte Lantisco lie along the same line of sight when viewed from this road, from which the photographs in Fig. 15 were taken. It appears that, as in the case of the 1677 vent and the San Antonio cone on La Palma, a young vent has become confused with an older but


303

Fig. 15. View of the southern vents of the 1951 eruption from the road along the southern side of the Cha das Caldeiras. Monte Orlando at left, Monte Rendall in middle ground with Monte Lantisco behind; Pico do Fogo in background with remnant slabs of last major spatter eruption (1721-1725?; see text) around summit.

more prominent volcanic vent along the same line of sight.

The importance of integrated field (geological and archaeological) and historical studies in reconstructing volcanic eruptions The examples described in this paper provide some important pointers as to how to best carry out studies of historical eruptions of multiplevent volcanoes with a view to assessing volcanic hazards and volcanic structure. Special points to bear in mind are the following: •

Fig. 16. Sketch map of the Monte Orlando-Monte Rendall-Monte Lantisco area on the south flank of Pico do Fogo, showing the field relationships that demonstrate that Monte Lantisco is an early historical or prehistoric vent.

Even basic points such as the location of vents may become confused, especially where they are in proximity to older and more prominent vents, and the 'line-of-sight effect', which appears to have operated in both La Palma and Fogo, may cause confusion. All aspects of the historical accounts should be checked even where these accounts are of recent date (it will be noted that confusion over the identity of Monte Rendall on Fogo occurred within two decades of the eruption in which it formed).

304

•

•

•

•

•

•

S. J. DAY ET AL. In multiple-vent eruptions, smaller vents or those that are active only at the beginning of an eruption may be overlooked in contemporary accounts. Conversely, once the most serious damage has been done (as in the case of the destruction of the Fuente Santa by the 1677 eruption) or the area of the eruption has been completely evacuated or civil authority broken down (as in the case of the 1730-1736 eruption of Lanzarote), interest in the eruption may decline and the later stages of activity be less well documented. Outside observers such as sea-captains, provided they happen to pass by during an eruption, may provide as good descriptions of volcanic activity as local observers. Indeed, if the latter are not directly affected by eruptive activity, they may not record events at all. In contrast, the practice of keeping a ship's log habituates mariners to recording any observations of unusual natural phenomena. A corollary is that the historical archives of seafaring nations may provide as many records of volcanic activity as the archives of the countries in which the activity takes place. Where the style or location (or both, as in the case of the ending of summit activity and reorganization of volcanic rift zones that has taken place on Fogo) of activity changes, this is likely to be reflected in the eyewitness accounts, even when these are written by nonscientists or date from a pre-scientific period. These accounts should not be dismissed because they are inconsistent with more recent and better described activity, but checked against the geological evidence. Archaeological evidence can provide important stratigraphic constraints on the age and timing of activity, especially where the limits of archaeological periods are well defined (through, for example, conquests, socio-economic revolutions or cultural developments of known date that are reflected in preserved artefacts). Cross-checking against geological and archaeological evidence indicates that whereas primary eyewitness accounts, even if written by non- or pre-scientific observers, are accurate within their limits more often than not, local traditions and other secondary accounts (even if written by geologists and other scientists, unless based on detailed fieldwork rather than interpretation of archival evidence or these same traditions) are often in error, as a result of such things as confusion of different vents along the same

line of sight. These errors may develop remarkably quickly. Provided these points are borne in mind, archival and related investigations of historical eruptions of multiple-vent volcanoes can, as noted in the introduction, provide important information on aspects of eruptions that cannot be investigated by means of later fieldwork, most notably the absolute timing and duration of events within the eruptions. There is therefore much potential for collaboration between geologists, archaeologists and historians in the investigation of these relatively small-volume but potentially very significant eruptions. Fieldwork on La Palma was funded by the Spanish DGICYT Research Project PB92-0119, by European Union Environment Programme Project EV5V-CT920170, and the Viceconsejeria de Medio Ambiente of the Canarian Government. Fieldwork on Fogo by S.J.D. was funded by a grant from the Calouste Gulbenkian Foundation to J. Fonseca. We gratefully acknowledge the help of P. N. Perez, mayor of the town of Fuencaliente, who provided old manuscripts, drawings and copies of eyewitness accounts related to the 1677 eruption.

References AFONSO, A., APARICIO, A., HERNANDEZ PACHECO, A. & RODRIGUEZ BADIOLA, E. 1974. Morphology evolution of Teneguia volcano area. Estudios Geologicos, Teneguia special volume 19-26. ANCOCHEA, E., FUSTER, J. M., IBARROLA, E. et al. 1990. Volcanic evolution of the island of Tenerife (Canary Islands) in the light of new K-Ar data. Journal of Volcanology and Geothermal Research, 44, 231-249. ANDERSON, E. M. 1935. The dynamics of the formation of cone sheets, ring dykes and cauldron subsidence. Proceedings of the Royal Society of Edinburgh, 56, 128-157. ANONYMOUS 1677. Gazeta del Ayuntamiento de La Palma, 2 de diciembre de 1677 (Gazette of the City Council of La Palma, 2nd December 1677). (Copy of the original, found in the archive of the Marquis of Guisla Guiselin (Lorenzo Rodriguez 1987).) 1678. Edictos, Cartas, Instrucciones que se han despachado a los benificiados e parrocos en todos este obispado desde principios del ano 66 (1666) en adelante. Archives Parroq. La Concepcion, La Laguna, Arch. B. Garcia Ximenez, libro Q. (Edicts, letters and instructions that were sent to the beneficiaries and parish priests in this bishopric from the first day of 1666 onwards. Parochial Archives of the Archbishop B. Garcia Ximenez, Church of the Holy Conception, La Laguna, volume Q.) BONELLI RUBIO, J. M. 1950. Contribucion al estudio de la erupcion del Nambroque o San Juan (isla de La

ERUPTIONS ON MULTIPLE-VENT VOLCANOES Palma) (Contribution to the study of the eruption of Nambroque or San Juan (island of La Palma). Institute Geografico y Catastral, Madrid. CARRACEDO, J. C. 1994. The Canary Islands: an example of structural control on the growth of large oceanic-island volcanoes. Journal of Volcanology and Geothermal Research, 60, 225-241. 19960. A simple model for the genesis of large gravitational landslide hazards in the Canary Islands. In: McGuiRE, W. J., JONES, A. P. & NEUBERG, J. (eds) Volcano Instability on the Earth and other Planets. Geological Society, London, Special Publications, 110, 125-135. 19966. Morphological and structural evolution of the western Canary Islands: hotspot-induced three-armed rifts or regional tectonic trends? (Reply to comment by MARTI et at.) Journal of Volcanology and Geothermal Research, 72, 151-162. & RODRIGUEZ BADIOLA, E. 1991. Lamarote: La erupcion volcanica de 1730. Cabildo Insular de Lanzarote. , DAY, S. J. & GUILLOU, H. 1997a. Rapid growth and instability of an oceanic island volcano: the Cumbre Vieja rift, La Palma, Canary Islands. Volcanic Activity and the Environment, Symposium of the IAVCEl General Assembly, Puerto Vallarta, Mexico, 19-24 January 1997. , , & GRAVESTOCK, P. J. 19976. Geological map of the Cumbre Vieja Volcano (La Palma, Canary Islands). Viceconsejeria de Medio Ambiente del Gobierno de Canarias, Santa Cruz, Tenerife. , , & RODRIGUEZ BADIOLA, E. 1996. The 1677 eruption of La Palma, Canary Islands. Estudios Geologicos, 52, 345-357. , RODRIGUEZ BADIOLA, E. & SOLER, V. 1992. The 1730-1736 eruption of Lanzarote, Canary Islands: a long, high-magnitude basaltic fissure eruption. Journal of Volcanology and Geothermal Research, 53, 239-250. CHEVALLIER, L. & VERWOERD, W. J. 1988. A numerical model for the mechanical behaviour of intraplate volcanoes. Journal of Geophysical Research, 93, 4182-4199. DAY, S. J., CARRACEDO, J. C. & GUILLOU, H. 1996. Reconfiguration of the Cumbre Vieja volcano, La Palma: advance warning of a giant landslide? High Magnitude Events: Landslides, Floods and Earthquakes, Session of Applied Geoscience Meeting of Geological Society, London, Warwick, 15-18 April 1996. Geological Society, London, pp.xx-xx. DE FARO, A. 1664. Perigrinacao de Andre de Faro a Terra dos Gentios (The Travels of Andre de Faro in Heathen Lands). Republished 1945, Bertrand, Lisbon. FEIJO, J. S., 1786. Memoria sobre a nova irrupcao volcanica do Pico da ilha do Fogo que deve servir de Suplemento a Letra Filosofica No. 5 sobre o mesmo objecto. Archivo Historico Ultramarino, Lisboa, Cabo Verde: Cartografia - mapas nao catalogados (Memoir on the new volcanic eruption of the Pico of the Island of Fogo, supplementary to the 5th

305

Philosophical Letter on the same subject. Overseas Historical Archive, Lisbon: Cape Verde Cartography Section - unserialized maps) (complete transcript, with original diagrams, has been given by Ribeiro (I960)). FRITSCH, K. V. & REISS, W. 1868. Geologische Beschreibung der Insel Tenerife, ein Beitrag zur Kenntnisse Vulkanischer Gebirge. Wuster, Winterthur. HELENO DA SILVA, S. I. N. & FONSECA, J. F. B. D. 1999. A seismological investigation of the Fogo volcano, Cape Verde Islands: preliminary results. Volcanology and Seismology, in press. , FOULGER, G. R., BARROS, I. J., QUERIDO, A., WALKER, A. B. & FONSECA, J. F. B. D. 1997. Seismic activity in Fogo and Brava islands, Cape Verde. Proceedings of the International Congress on the Volcanic Eruption of 1995 in Fogo Island, Cape Verde. IICT, Lisbon, 79-91. HERNANDEZ PACHECO, A. & VALLS, M. C. 1982. The historic eruptions of La Palma island (Canaries Archipelago). Revista Universidad dos Acores, 3, 83-94. IICT 1997. A erupcao vulcanica de 1995 na Ilha do Fogo, Cabo Verde (On the volcanic eruption of 1995 on the island of Fogo, Cape Verde). Institute do Investigacoes Cientificas Tropicales, Lisbon. KlLBURN, C. R. J., PlNKERTON, H. & WlLSON, L.

1995. Forecasting the behaviour of lava flows. In: McGuiRE, W. J., KILBURN, C. R. J. & MURRAY, J. (eds) Monitoring Active Volcanoes. UCL Press, London, 347-368. LORENZO RODRIGUEZ, J. B. 1987. Noticias para la historia de La Palma, tomo I (Notes on the History of La Palma, Vol. I). Inst. Estudios Canaries, Cabildo Insular, La Palma. MACHADO, F. 1965a. Mechanism of Fogo volcano, Cape Verde Islands. Garcia de Orta (Lisboa), 13, 51-56. \965b. Vulcanismo das Islas de Cabo Verde e das Outras Ilhas Atlantidas. (Volcanism of the Cape Verde Islands and other Atlantic Islands.) Junto de Investigacoes do Ultramar, Lisboa, Estudos, Ensaios e Documentos, 117. & TORRE DE ASSUNCAO, C. F. 1965. Carta geologica de Cabo Verde (na escala de 1/100000); noticia explicativa da folha da ilha do Fogo estudos petrograficos. Garcia de Orta (Lisboa), 13, 597-604. McGuiRE, W. J. & PULLEN, A. D. 1989. Location and orientation of eruptive fissures and feeder-dykes at Mount Etna; influence of gravitational and regional tectonic stress regimes. Journal of Volcanology and Geothermal Research, 38, 325—344. NAKAMURA, K. 1977. Volcanoes as possible indicators of stress orientation: principle and proposal. Journal of Volcanology and Geothermal Research, 2, 1-16. POLLARD, D. D. 1987. Elementary fracture mechanics applied to the structural interpretation of dykes. In: HALLS, H. C. & FAHRIG, W. F. (eds) Mafic Dyke Swarms. Geological Association of Canada Special Paper, 34, 5-24.

306

S. J. DAY ET AL.

RIBEIRO, O. 1960. A Ilha do Fogo e as suas erupcoes, 2nd edn (The island of Fogo and its Eruptions). Memorias, serie geografica I. Junta de Investigacoes do Ultramar, Ministerio do Ultramar, Lisbon. ROMERO, C. 1991. Las manifestaciones volcanicas historicas del Archipielago Canario (Historic volcanic activity in the Canarian Archipelago), 2 vols. Consejeria Political Territorial, Santa Cruz de Tenerife. RYAN, M. P., BLEVINS, J. K., OKAMURA, A. T. & KOYANAGI, R. Y. 1983. Magma reservoir subsidence mechanics: theoretical summary and application to Kilauea Volcano, Hawaii. Journal of Geophysical Research, 88, 4147-4181. SIGURDSSON, H., CAREY, S., CORNELL, W. & PESCADORE, T. 1985. The eruption of Vesuvius in AD 79. National Geographic Research, 1, 332-387. , CASHDOLLAR, S. & SPARKS, R. S. J. 1982. The eruption of Vesuvius in AD 79: reconstruction from historical and volcanological evidence. American Journal of Archaeology, 86, 39-51. SILVEIRA, A. B., SERRALHEIRO, A., MARTINS, I. et al. 1995. A erupcao da Cha das Caldeiras (Ilha do Fogo) de 2 de Abril de 1995. Proteccao Civil, 7, 3-14. TIBALDI, A. 1995. Morphology of pyroclastic cones and tectonics. Journal of Geophysical Research, 100, 24521-24535.

TORRE DE ASSUNCAO, C. F., MACHADO, F. & CONCEICAO SILVA, L. 1967. Petrologia e vulcanismo da ilha do Fogo (Cabo Verde). Garcia de Orta (Lisboa), 15, 99-110. TORRES, P. C., MADEIRA, J., SILVA, L. C., BRUM DA SILVEIRA, A., SERRALHEIRO, A. & MOTA GOMES, A. 1997. Carta geologica das erupcoes historicas da Ilha do Fogo: revisao e actualizacao. A erupcao vulcanica de 1995 na Ilha do Fogo, Cabo Verde. IICT, Lisbon, 119-132. TORRIANI, L., 1592. Descripcion e historia del reino de las Islas Canarias (Description and History of the Realm of the Canary Islands). Translation of the original by A. Cioranescu 1978. Editorial Goya. VON BUCH, L. 1825. Physicalische beschreibung der Kanarischen Inseln. Berlin. 1836. Description physique des Isles Canaries. Levrault, Paris. WEBB, B. & BERTHELOT, S. 1839. Histoire Naturelle des lies Canaries, 2 vols. Paris. WHITE, R. S. 1989. Asthenospheric control on magmatism in the ocean basins. In: SAUNDERS, A. D. & NORRY, M. J. (eds) Magmatism in the Ocean Basins. Geological Society, London, Special Publications, 42, 17-27.

'A fire spitting volcano in our dear Germany': documentary evidence for a low-intensity volcanic eruption of the Gleichberg in 1783? J. P. GRATTAN1, D. D. GILBERTSON2 & A. DILL3 1

The University of Wales, Aberystwyth, Institute of Geography and Earth Sciences, Aberystwyth SY23 3DB, UK (e-mail: [email protected]) 2 Nene Centre for Research, University College Northampton, Northampton NN2 7AH, UK 3 Geldenaaksevest 44, B-3000, Leuven, Belgium Abstract: This paper presents documentary evidence suggesting that the most recent volcanic activity in Germany may have occurred just over 200 years ago, rather than the 11 000 years held currently (Ulmener Maar, West Eifel). Several descriptions recounted here suggest that a mountain in Germany, the Gleichberg, may have erupted in the early summer of 1783. The reports of the volcanic event are laden with detail that would tempt the reader to accept them as genuine descriptions of an eruption, had the event been located in an historically active volcanic region. However, there are several reasons to suggest that the report of the Gleichberg eruption was a complex hoax, written to exploit the fear and panic generated by the dry fog present over much of Europe at the same time, which had its origins in the Laki fissure eruption. Geologically, the Gleichberg forms part of the Grabfeld, and is of Tertiary volcanic origin. There are no compelling geological reasons to suggest that this area has been tectonically active in recent times. The convincing detail of the report is used to illustrate the pitfalls waiting for geologists, historians and archaeologists who are using historical documents and folklore to explore the impact of volcanic eruptions upon ancient peoples and environments; time may lend weight to documents and folklore, which in reality may deserve none.

To evaluate the impact that volcanic activity may have had upon past human societies that are being investigated through their archaeological remains, it is necessary to possess accurate and reliable eruption chronologies, as well as detailed knowledge of the volcanic centres that have been active during human history or prehistory. This is not as straightforward as it may seem. Although volcanic eruption chronologies have been dramatically improved in recent years, our knowledge of volcanic eruptions from archaeological or geological evidence beyond the 19th century is often minimal (Simkin & Siebert 1994). Supporting evidence may also be sought from historical documents, but the interpretations of these are also open to challenge. The problems faced by archaeologists and historians working with such sources can be illustrated by a study of the hypothetical 'Gleichberg eruption' in 1783, an event in Germany that was widely reported across Europe by non-scientists. The year 1783 is already well known in volcanological history for the eruption of the Laki fissure in Iceland (Thordarson & Self 1993). In Iceland, the impact of the eruption upon both society and environment was severe (Thorarins-

son 1981), but the Laki fissure eruption also had dramatic impacts downwind, across the eastern Atlantic, upon the distant peoples and environments of Europe. Research using newspapers, private journals and scientific reports has revealed that from mid-June and throughout much of July, a dry acid fog composed of gases emitted by the eruption was distributed across much of Europe (Camuffo & Enzi 1995; Grattan & Brayshay 1995; Stothers 1996), with some regions experiencing higher concentrations of noxious aerosols as a result of the prevailing synoptic meteorological conditions (Grattan & Gilbertson 1994). In some cases, the impacts of the dry acid fog upon societies and environments in Europe were dramatic and have been the subject of extensive research by the authors, who have read hundreds of contemporary newspaper descriptions of the phenomena, as well as many more personal reflections recorded in diaries made at the time (Grattan & Charman 1994; Grattan & Gilbertson 1994; Grattan & Pyatt 1994; Grattan & Brayshay 1995; Grattan et al. 1996, 1999). Amongst all these documentary references to dry sulphurous fogs, crop damage, leaf loss, asthma epidemics and the


308

J. P. GRATTAN, D. D. GILBERTSON & A. DILL

Fig. 1. Location of the Gleichberg and proximal volcanic provinces.

Fig. 2. Geological map of the Gleichberg and surrounding district. Adapted from Von Berthold (1989).

Fig. 3. Geological cross-section of the Gleichberg, Adapted from Von Bernd (1989).

EVIDENCE FOR A VOLCANIC ERUPTION OF THE GLEICHBERG extermination of insects, there are references to eruptive activity on a volcanic mountain in Germany, the Gleichberg (Figs 1-3). These references are puzzling, as there is no other suggestion that the German volcanic centres have been active since the Ulmenar Maar event in the Eifel volcanic field (Hajdas et al. 1995; Zolitschka et al. 1995). This paper explores the nature of the report, considers its veracity and assesses the significance of such accounts for geologists and historians who are exploring the past by 'excavating words' (Weisburd 1985). Descriptions of the 'Gleichberg eruption' In early July 1783, several German newspapers reported the following letter from Hildburghausen: 'A report from Hildburghausen, of 24th June: Here is news from a recent strange natural phenomenon in our area. The Gleichberg, about 2 hours from here, is surely well known to you as a barometer substitute as its periods of smoking always announce subsequent rainfall. Since about Easter smoking has been stronger than ever before and is increasing every day. As a result thick clouds prevail in the whole area between Romhild and Hildburghausen, corresponding to a trip of 8 hours. All the woods in this area are white rather than green, and the whole sky appears to be dominated by pulverised or sublimed limestone. The clouds consist of sulphur killing everything, whereby sun and moon rise and set blood red. Since about 8 days ago terrible frightening thumps have been occurring within the Gleichberg, resembling explosions from cannons, until recently the Gleichberg opened up below thick sulphur clouds, and continuous terrible rumbling and roaring can be heard all around. In all the churches of the area special services are held, and the terrified inhabitants of the surrounding villages have fled as they fear the whole Gleichberg might finally collapse, or cause further misfortune. I shall report on future occurrences, as now we also have a fire spitting volcano in our dear Germany' (translated from Frankfurter Staats Ristretto, 12 July 1783, pp. 108, 456). Interest in the 'event' was so widespread that it was reported by correspondents and appeared in several English newspapers. 'Hildburghausen, July 4th: Mount Gleichberg, situated in our neighbourhood, has since Easter continually thrown out thick sulphurous vapours, and during the last eight days a violent noise has been frequently heard within the mountain ... another opening has since appeared, from which also issues a thick sul-

311

phurous smoke' (The Whitehall Evening Post, 9-12 August 1783). 'Hildburghausen, July 4th: Mount Gleichberg, in our neighbourhood, affords at present a singular and terrible phenomenon; the vapours which continually surround it are increased much, and form a thick mist which extends 8 leagues. This mist, which has destroyed the verdure of our woods, and has been substituted with a whitish tint, is, without doubt, by the scent, formed of sulphurous exhalations ... An aperture is formed, from which arises a very thick sulphureous (sic) smoke, which, with the subterraineous (sic) noise ... gives room to apprehend a new volcano' (The Morning Herald and Daily Advertiser, 12 August 1783). If unchallenged, these descriptions would leave little room to doubt that an eruption, albeit of low intensity, had taken place on the Gleichberg, with the consequences described above. However, another German newspaper, the Meiningen Wochentliche Nachrichten, disputed the report of the eruption. A letter published one week later on 19 July repeated the account published in the Frankfurter Staats Ristretto, but concluded: 'Our eyes do not see anything and our ears do not hear anything of the flight of the frightened inhabitants, the threatened collapse of the Gleichberg ... (nor the) thick sulphur smoke, that has covered the woods with a white blanket.'

The regional context: eruptive phenomena and social unease in Germany and Europe, June-August 1783 The passages translated above from the Frankfurter Staats Ristretto and the Meiningen Wochentliche Nachrichten are the most detailed accounts yet discovered in this study. It is likely that the passages printed in the English newspapers are essentially copies of the piece from the Frankfurter Staats Ristretto, as 18th-century English newspapers relied heavily on the Continental European Press for much of their foreign news (Wiles 1965). In the 6 weeks that followed the report in Frankfurter Staats Ristretto, British and other European newspapers contained hundreds of descriptions of reddened skies, dry fogs, crop damage and tremendous thunderstorms. Present research has indicated that these phenomena were largely the result of aerosols and gases emitted by the Laki fissure eruption in Iceland and transported to Europe by regional air circulation (see Grattan & Gilbertson (1994) and

312


Grattan & Bray shay (1995), for a discussion of the newspaper coverage for this period). It may therefore be possible that the report of the 'Gleichberg eruption' simply attempted to capture and exploit the mood of the moment, which in many cases was one of panic. Diarists, correspondents and newspapers of this time tended to refer to the 'superstitious dread' with which the 'common people' viewed the dry fog and the strange appearance of the sun and moon. That the dry fog from Iceland also reached many areas of Germany is suggested by the following passages extracted from the Meiningen Wo'chentliche Nachrichten. 'Meiningen, July llth: Over almost all areas of Germany, the one fog, or the so-called 'majestic' smoke stretches out. The same has filled our horizon for three weeks. A thunderstorm, which we had here . . . and the following strong rain ... were not able to suppress it' (Meiningen Wochentliche Nachrichten, 12 July). 'Letter from Mannheim, 1 July: Since 17 June a fog has persisted day and night and can certainly be considered an extraordinary phenomenon. The eldest people cannot remember having ever experienced anything similar. This fog comes from the north-east, and is as common in the mountains as it is on the plains. It is also very dry, which is proved by the hygrometer. In the morning and after 7 o'clock at night, the sun has a red colour like glowing iron, and during the day it is very pale, with a stifling heat' (Meiningen Wochentliche Nachrichten, 12 July). These phenomena appear to have triggered such a degree of superstitious panic in the Meiningen area, as well as elsewhere, that a lecture on the nature of fogs was also published to allay the fears of the public: 'Because the sun now has a glowing red colour during sunrise and sunset, ... we will, in response to current superstitions present the following information to calm those of our readers who are afraid' (Meiningen Wochentliche Nachrichten, 12 July). The report of the Gleichberg eruption must therefore be seen against a broader background of Europe-wide phenomena, which sometimes triggered panic and superstition in Britain, France, and Germany, and in the Meiningen region in particular. It is in this context that we must examine whether the original account is likely to be fact or fiction. Geology The GroBer Gleichberg, 50.23°N 10.37°E, is a mountain, 679 m high, of volcanic origin, which

is sited on one of a group of volcanic fissures, known collectively as the 'Grabfeld', which lie to the south of Hildburghausen (Fig. 1) and which are eastern outliers of the German Volcanic System (Hoppe 1974). The fissures have been identified as being sited on Tertiary-early Quaternary, intra-plate volcanic activity (Weber 1955; Kastner 1974; Grumbt & Liitzner 1983; Ziegler 1990). The summits of both the GroBer and Kleiner Gleichberg are formed of basalt, which has to some extent been quarried, and examination of the published geological sources indicates that there is no doubt that the Gleichberge peaks are the result of volcanic activity (Figs 2 and 3). Until recently, the presence on the mountain of a Soviet military base hindered close scientific scrutiny. The area was also in the 'Sperrzone' or protection zone erected along the borders of the former East Germany and to which public access was severely restricted. As a whole, this group of hills has therefore attracted little attention from geologists, especially in comparison with the attention given to the Vogelsberg and Eifel volcanic structures located to the west (Fig. 1). The most recent volcanic activity in Germany in the recent Quaternary is the eruptions associated with the Laacher See tephra, c. 12 000 years ago, and Ulmener Maar, c. 11 000 years ago (Lippolt 1983; Van den Bogaard & Schmincke 1985; Hajdas et al. 1995; Lotter et al 1995; Zolitschka et al 1995), in the Eifel region. These events are widely held to mark the end of active volcanism in the German Volcanic System. Nevertheless, these German volcanoes cannot be considered to be completely inactive. Many hot springs testify to geothermal activity, and the frequent emanations of CO2 and helium isotropy in source waters in the Eifel region are evidence that the younger volcanic regions of Germany cannot be considered entirely extinct (Oxburgh & O'Nions 1987). It is extremely unlikely that this Quaternary activity could in any way be associated with the Tertiary volcanic fissures of the Grabfeld, which are over 2 Ma old, and certainly cannot be used to suggest that the Grabfeld fissures could have erupted in the recent historical past.

Fact or fiction? The detail contained in the original report is superficially convincing, and were it to have originated from an area of acknowledged recent volcanic activity it would probably have been accepted as genuine with only cursory scrutiny. The detail contained in the newspaper article is

EVIDENCE FOR A VOLCANIC ERUPTION OF THE GLEICHBERG remarkable because it suggests that the correspondent had witnessed an actual volcanic eruption and noted a range of associated phenomena, such as emissions of sulphur, tephra and water vapour. Such incidental, but important detail, seems out of place in a deliberate hoax. For instance, the report describes clearly events that can be reasonably interpreted as precursory volcanic activity, which could have eventually culminated in a low-intensity eruptive event of the type reported at the Gleichberg in June 1783. The unknown correspondent also suggested that the mountain was already known for periodic episodes of smoking and, in particular, that it was used locally as a 'barometer substitute'. This property might suggest that water vapour was being released from the ground during periods when the air temperature was cold. It is consistent with a ground surface heated from beneath by geothermal energy. Such an assumption is consistent with observations made in many areas of the world where geothermal heating is an accepted fact of life. Similarly, apparently convincing descriptions of tephra emission are given, with the tree foliage in the area turning from green to white in colour, and accounts of the air being filled with a pulverized limestone. The colour change also suggests the impact of acid deposition from sulphur clouds emitted in an eruption (Wilcox 1959; Caput et al 1978; Lang et al. 1980; Wisniewski 1982). However, it is important to note that similar phenomena were also noted in Britain and the Netherlands on 21-23 June by several reliable witnesses (Cullum 1784; Swinden 1786; Brugmans 1787; White 1789) and appear to be associated with gas emissions from the Laki fissure eruption in Iceland (Grattan & Gilbertson 1994; Stothers, 1996; Grattan et al \999a, b), and there may therefore be a copy-cat element in the Gleichberg account. It might also be argued that the damage to the vegetation observed around the Gleichberg was also due to the deposition of atmospheric acids emitted from the Laki fissure eruption. However, to date, all the known accounts of severe acid damage to vegetation, as opposed to descriptions of the dry fog attributable to the Laki fissure eruption, have come from relatively low-lying coastal districts around the North Sea (Grattan et al. 1999, 1998). It is less certain that the red colour of the sky at sunrise and sunset reported from Hildburghausen were entirely the consequences of the putative Gleichberg eruption, as the skies across Europe were spectacular at this time (Grattan & Brayshay 1995), probably also because of the material erupted from the Laki fissure in Iceland.

313

That the report was given at least shortterm credence is evident in the nature of the original publication. The Frankfurter Staats Ristretto was a serious, not a frivolous, newspaper. The position of the article in the newspaper does not appear to be pandering to taste or popular hysteria. For example, the description of the eruption was placed distinct from, but set amongst, detailed coverage of important domestic and international political and military affairs. Neither have we detected evidence that the actual text is a copy-cat account of events elsewhere, perhaps in southern Italy or Iceland. The date of the Gleichberg 'eruption' may also be important. Published on 12 July, the description of the eruption claims to have been written in Hildburghausen on 24 June 1783. The report may have been written and was certainly published shortly after a dry acid fog appeared in Europe; in Germany on 17 June and in Britain, France and the Netherlands, on 24 June. The intention of the correspondent could have been to exploit the fear and panic that was becoming widespread in Europe from early July as the gases emitted in the Laki fissure eruption obscured the sun, damaged crops and generated breathing difficulties across Europe (Thorarinsson 1981; Grattan & Charman 1994; Grattan & Gilbertson 1994; Grattan & Pyatt 1994; Grattan & Brayshay 1995; Grattan et al. 1996, 1999, 1998; Stothers 1996). It is also sensible to enquire, if the report was a hoax, 'why set it at the Gleichberg?' It is not at all clear that this mountain had been accepted as volcanic in origin at this stage of the 18th century. It should also be remembered that in the 18th and early 19th century, a powerful school of thought contested the volcanic origin of the German basalts, holding instead that they were the result of submarine sedimentation (Geikie 1893; Wagenbreth 1967; Von Bernd 1989). It may be that the Gleichberg was remote enough from Frankfurt, where the report of the 'eruption' was initially published, to prevent detailed editorial scrutiny. It is perhaps wise to sound a note of caution. There is no geological evidence to suggest that the Grabfeld fissures have been active since formed in the Tertiary, and, perhaps more compelling, the Meiningen Wochentliche Nachrichten, which was published much closer to the site of the proposed volcanic activity, announced itself to be unaware of any such event or of its social and environmental consequences. If it was a real event, it would surely have attracted proper scientific attention, and, tellingly, local historians have no record of the 'eruption'. Perhaps the event described was not a volcanic

314


eruption but a forest fire, the impact of which became entangled in the phenomena associated with the gases emitted by the Laki fissure eruption. One such fire in the Mosel valley in the 1990s, was accompanied by descriptions of glowing lava blocks (M. Frechen pers. comm.), which will no doubt confuse researchers in the 2190s. Recent investigations in local museums and discussions with local historians have failed to unearth further confirmation of the Gleichberg event, an outcome suggesting that the original reports described above were either an elaborate hoax or the result of a genuine misunderstanding. Conclusion Although far from conclusive, newspaper articles in Germany suggest that the Gleichberg, a mountain of volcanic origin, near Hildburghausen, may have erupted in 1783. However, a detailed analysis of the text and an understanding of the wider context in which the report was made suggests that the report of the eruption was written to exploit the fear and interest generated by the presence in the atmosphere of a dry fog composed of volcanic gases that originated in Iceland, and is therefore a complex and clever hoax. Field research is needed to test this assertion but available geological data and current geological understanding make it unlikely that the 'eruption' will be confirmed, and the true nature and cause of the 'Gleichberg eruption' will remain a geological mystery. The case of the 'Gleichberg eruption' serves to illustrate some of the difficulties faced by archaeologists, historians and geologists attempting to determine the role and influence of volcanic activity upon ancient peoples, environments and cultures. One cannot rely entirely upon tradition and folklore, nor upon relatively recent detailed documentary material. It is clear that to test the veracity of historical or legendary accounts of volcanic activity one must possess or obtain a detailed understanding of the context in which the passage was written or the legend laid down. The passage of time lends authenticity to written accounts, which may not truly merit serious consideration. The authors are indebted to M. Frechen for detailed and helpful comments on the text, and to the staff of the German Historical Institute, London, in particular C. Freeman, for their assistance in this research. A. Gunst and B. Kulessa provided invaluable assistance in the translation of numerous German texts.

References BRUGMANS, S. J. 1787. Natuurkundige verhandeling over een zwavelagtigen nevel den 24 Juni 1783 inn de provincie Groningen van stad en lande en naburige landen waargenomen. Ley den. CAMUFFO, D. & ENZI, S. 1995. Impacts of clouds of volcanic aerosols in Italy during the last 7 centuries. Natural Hazards, 11, 135-161. CAPUT, C., BELOT, Y., AUCLAIR, D. & DECOURT, N. 1978. Absorption of sulphur dioxide by pine needles leading to acute injury. Environmental Pollution, 16, 3-15. CULLUM, J. 1784. Of a remarkable frost on the 23rd of June, 1783. Philosophical Transactions of the Royal Society. Abridged Volume, 15, 604. GEIKIE, A. 1893. Text-Book of Geology. Macmillan, London. GRATTAN, J. P. & BRAYSHAY, M. B. 1995. An amazing and portentous summer: environmental and social responses in Britain to the 1783 eruption of an Iceland volcano. Geographical Journal, 161(2), 125-134. & CHARMAN, D. J. 1994. Non-climatic factors and the environmental impact of volcanic volatiles: implications of the Laki fissure eruption of AD 1783. Holocene, 4(1), 101-106. & GILBERTSON, D. D. 1994. Acid-loading from Icelandic tephra falling on acidified ecosystems as a key to understanding archaeological and environmental stress in northern and western Britain. Journal of Archaeological Science, 21(6), 851-859. & PYATT, F. B. 1994. Acid damage in Europe caused by the Laki fissure eruption - an historical review. Science of the Total Environment, 151, 241-247. , CHARMAN, D. & GILBERTSON, D. D. 1996. The environmental impact of Icelandic volcanic eruptions: a Hebridean perspective. In: GILBERTSON, D. D., KENT, M. & GRATTAN, J. P. (eds) The Environment of the Outer Hebrides of Scotland: the Last 14000 Years. Sheffield Academic Press, Sheffield, 51-58. , BRAYSHAY, M. & SADLER, J. P. 1998. Modelling the impact of past volcanic gas emissions. Quaternnaire, 9(1), 25-35. , GILBERTSON, D. D. & CHARMAN, D. J. I999a. Modelling the impact of Icelandic volcanic eruptions upon the prehistoric societies of northern and western Britain. In: FIRTH, C. & McGuiRE, W. (eds) Volcanoes in the Quaternary. Geological Society, London, Special Publication, 161, 109-124. GRUMBT, E. & LUTZNER, H. 1983. Saxonian tectonics and basalt volcanism between the Thuringian forest and the fore Rhone. Zeitschrift fur Geologise he Wissenschaften, 943-954. HAJDAS, I., ZOLITSCHKA, B., IVYOCHS, S. D. et al. 1995. Radiocarbon dating of annually laminated sediments from Lake Holzmaar, Germany. Quaternary Science Reviews, 14, 137-143. HOPPE, W. 1974. Geologie von Thuringen. Hermann Haack, Gotha-Leipzig.

EVIDENCE FOR A VOLCANIC ERUPTION OF THE GLEICHBERG KASTNER, H. 1974. Jungtertiarer Vulkanismus. In: HOPPE, W. (ed.) Geologic von Thiiringen. Hermann Haack, Gotha-Leipzig, 782-789. LANG, D. S., HERZFELD, D. & KRUPA, S. V. 1980. Responses of plants to submicron acid aerosols. In: TORIBARA, T. Y., MILLER, M. W. & MORROW, P. E. (eds) Polluted Rain. Plenum, New York. LIPPOLT, H. J. 1983. Distribution of volcanic activity in space and time. In: FUCHS, K. (ed.) Plateau Uplift: the Rhenish Shield - A Case History. Springer, Berlin, 112-120. LOTTER, A. F., BlRKS, H. J. B. & ZOLITSCHKA, B.

1995. Late Glacial pollen and diatom changes in response to 2 different environmental perturbations - volcanic eruption and Younger Dryas cooling. Journal of Palaeolimnology, 14, 23-47 OXBURGH, E. R. & O'NiONS, R. K. 1987. Helium loss, tectonics and the terrestrial heat loss budget. Science, 237, 1583-1587. SIMKIN, T. & SIEBERT, L. 1994. Volcanoes of the World, 2nd edn. Geoscience Press, Tucson, AZ. STOTHERS, R. B. 1996. The Great Dry Fog of 1783. Climatic Change, 32, 79-89. THORARINSSON, S. 1981. Greetings from Iceland: ash falls and volcanic aerosols in Scandinavia. Geografiska Annaler, 63A, 109-118. THORDARSON, TH. & SELF, S. 1993. The Laki [Skaftar Fires] and Grimsvotn eruptions in 1783-85. Bulletin Volcanologique, 55, 233-263. VAN DEN BOGAARD, P. & SCHMINCKE, H.-U. 1985. Lacher See tephra: a widespread isochronous late Quaternary tephra layer in central and northern Europe. Geological Society of America Bulletin, 96, 1554-1571. VAN SWINDEN, M. 1786. Observations sur quelques particularites meteorologiques de Fannee 1783.

315

Memoires de I'Academie Royale des Sciences, Turin, 1784-1785, 113-140. VON BERND, W. B. 1989. Zum geologischen Aufbau der Gleichberglandschaft. Bikourgion, 7-13. 1989/7. Forschungsschichtliche Beziehungen zwischen Geologic und Archaologie am kleinen Gleichberg. Bikourgion, 39-47. VON BERTHOLD, W. 1989. Allgemeiner geologischer Uberlick iiber das Gleichberggebiet. Bikourgion: Zur Geologic des Gleichberggebietes, 14-22. WAGENBRETH, O. 1967. Die Entwicklung des geologischen Weltbildes in den letzten 200 Jahren. Forschungen und Fortschritte, 41, 365-371. WEBER, H. 1955. Einfiihrung in die Geologic Thiiringens. Deutscher Verlag der Wissenschaften, Berlin. WEISBURD, S. 1985. Excavating words - a geological tool. Science News, 127, 91-94. WHITE, G. 1789. The Natural History of Selbourne, reprinted 1977. Penguin, London. WILCOX, R. E. 1959. Some effects of the recent volcanic ash falls with special reference to Alaska. US Geological Survey, Bulletin, 1028-N, 409-476. WILES, R. M. 1965. Freshest Advices: Provincial Newspapers in England. Columbus, OH. WISNIEWSKI, J. 1982. The potential acidity associated with dews, frosts and fogs. Water, Air and Soil Pollution, 17(4), 361-377. ZIEGLER, A. 1990. Geological Atlas of Western and Central Europe. Maatschappij, The Hague. ZOLITSCHKA, B., NEGENDANK, J. F. W. & LOTTERMOSER, B. G. 1995. Sedimentological proof and dating of the Early Holocene volcanic eruption of Ulmener Maar, Germany. Geologische Rundschau, 84, 213-219.


Volcanic soils: their nature and significance for archaeology PETER JAMES1, DAVID CHESTER1 & ANGUS DUNCAN2 1

Department of Geography, University of Liverpool, P.O. Box 147, Liverpool L69 3BX, UK 2 Centre for Volcanic Studies, University of Luton, Luton LU1 3JU, UK Abstract: Whereas previous reviews of volcanic soils are biased in favour of those in tephra, the present paper examines the nature of weathering and pedogenesis in both tephra and lava. The classification of volcanic soils is discussed and examples are described of the response of pedogenesis to variations in climate, drainage, topography, vegetation and type and age of parent material. Archaeological implications considered include the distinctive properties of soils in tephra and the problems these may pose for laboratory analysis, evidence from buried soils, the ages of soils and their rates of development, and the fertility and erosion of volcanic soils.

An important consequence of volcanic activity is the formation of soil in the rocks and sediments that result from volcanic eruptions. It has been estimated that 27% of post-Archaean sedimentary rocks are tephra (including ash, pumice, volcanic bombs, lapilli and scoria (Fisher & Schmincke 1984)) currently covering about 0.84% of the earth's surface (Leamy 1984). Some 80% of this area is potential crop land (Mizota & van Reeuwijk 1989). Lava would appear to be more extensive than tephra, occurring in regions of recent volcanic activity and as Precambrian to Quaternary rocks including continental flood basalts, such as those of the Columbia River Plateau and the Deccan (see Fig. 1). Soil productivity has been one factor accounting for the density of human settlement in many of these areas of volcanic rocks and associated sediments. Despite the great area of land covered by lava, reviews of the literature on volcanic soils are biased strongly in favour of those developed in tephra. The probable reasons for this emphasis are the distinctive character of the products of tephra weathering, the technical challenge they have presented the laboratory analyst, the fact that many soils developed in lava and other rocks have received additions of tephra, and that tephra may form dateable stratigraphic markers, in some cases of considerable extent. Reviews of tephra soils include those by Ugolini & Zasoski (1979), Tan (1984), Wada (1985), Lowe (1986), Mizota & van Reeuwijk (1989) and Shoji et al (1993). Gibbs (1980) and Molloy (1993) discussed many examples of tephra and lava soils in New Zealand. The work by Shoji et al. (1993) is the most detailed review of tephra soils, but is

concerned chiefly with humid temperate regions. Mohr & van Baren (1954) and Mohr et al (1972) described the profile morphology and mineral composition of tephra and lava soils from many locations in the tropics. The collection of papers edited by Fernandez-Caldas & Yaalon (1985) dealt with both tephra and lava soils, but there is no substantial review concentrating on the latter. In contrast to the literature on tephra soils, that on lava soils includes few reports dealing with cool temperate environments. The archaeologist's first, and often most important, impressions of a soil are gained in the field. The general appearance of a soil, its profile morphology, reflects the operation of chemical, biological and physical processes upon parent material. A soil may reveal evidence of past environmental change, whether climatic, geological or human induced. It may reflect its relative age, or it may contain material that will yield a date relating to its age. If the soil contains features resulting from former human activity it is normally possible to distinguish these from natural pedogenic features. In this review we concentrate on the properties and profiles of volcanic soils. A note on classification is essential, as the terms applied to these soils are both many and confusing. We discuss soils in tephra and in lava separately and at some length because, as far as we are aware, this is the first review attempting to cover the full range of volcanic soils. We also consider several pedological aspects that have significance for archaeological research. Some of these are related to fertility, such as the nature and rates of weathering and soil development, and the susceptibility of soils to erosion. The relief of volcanic landforms has

From\ McGuiRE, W. G., GRIFFITHS, D. R., HANCOCK, P. L. & STEWART, I. S. (eds) The Archaeology of Geological Catastrophes. Geological Society, London, Special Publications, 171, 317-338. 1-86239-062-2/00/ $15.00 © The Geological Society of London 2000.

Fig. 1. Global distributions of principal lava flows and tephra deposits (based on Hyndman (1972), Derry (1980) and Chester (1993)).

VOLCANIC SOILS AND ARCHAEOLOGY an important influence on soils and their use. Some soil properties have less obvious implications: the potential of volcanic soils to hold chemical and magnetic signatures of human activity is of interest in geoarchaeological research; there are problems of interpreting the results of certain conventional laboratory procedures when they are applied to tephra soils.

Soil nomenclature For the reader with limited experience of soil science there are many good texts. Most of the pedological terms we use have been explained, for example, by Brady & Weil (1994). Because of the confusing variety of soil nomenclature in the literature, we use the classification, Soil Taxonomy (Soil Survey Staff 1997), where feasible. The scheme, summarized by Brady & Weil (1994), is the most comprehensive that attempts to have global application. To those unfamiliar with this system and its predecessors, the nomenclature will appear strange, yet it serves as a shorthand notation of the essential characteristics of a soil. Where insufficient data make it impossible to classify a soil correctly within Soil Taxonomy we follow the nomenclature used in the works cited, and in many instances traditional names for soils are given, as these are likely to be familiar to many workers in fields other than pedology. We also use the generic terms 'volcanic soils', 'tephra soils' and 'lava soils', which are informal but brief and clear in meaning. In Soil Taxonomy the status of the most distinctive volcanic soils (these being very largely in tephra) has been revised as knowledge of their mineralogy and chemistry has improved with advances in analytical techniques. Before 1990,

319

volcanic soils meeting certain physical and chemical criteria were included in the Andept suborder of the order Inceptisols (soils of limited development) (Soil Survey Staff 1975), but in 1990 they were allocated the highest category, the soil order of Andisols (Soil Survey Staff 1990). These soils, which have the 'andic' properties summarized in Table 1, are largely, but not exclusively, developed in tephra. There are andic or vitrandic (Soil Survey Staff 1997, pp. 23, 135) subgroups in all orders of Soil Taxonomy, with the exception of Vertisols (dark, swelling clays). The early developments in the classification of tephra soils by US workers between 1930 and 1970 have been summarized by Simonson & Rieger (1967). In the classification scheme of FAO-UNESCO (1989), designed for the 1: 500 000 Soil Map of the World, the major soil group, Andosols (a word partly derived from Japanese), is approximately equivalent to the Andisols of Soil Taxonomy. Some national soil classification schemes, such as those for Japan (Shoji et al 1993, p. 95) and New Zealand (Hewitt 1989), include classes for volcanic soils. The importance of tephra in controlling certain soil properties irrespective of climate is reflected in the fact that in FAO-UNESCO and Soil Taxonomy classifications, Andosols and Andisols, respectively, are the only major classes of soils in which the geological origin of the parent material is recognized. This is convenient for workers interested in volcanic soils, but many soils developed in volcanic materials, particularly lava but including tephra and other types of volcaniclastic sediment, do not meet the criteria for inclusion in these classes and so are grouped with other soils typical of the region in which they occur.

Table 1. Chief criteria defining andic soil properties in Soil Taxonomy; modified from Keys to Soil Taxonomy (Soil Survey Staff 1997, pp. 23-24) Most horizons that have andic soil properties consist of mineral soil materials; some consist of organic soil materials but must have 0.40 and, in the 0.02-2.Omm fraction, >30% volcanic glass; or (b) Al plus 0.5 Fe percentages (by ammonium oxalate) totalling >2.0% and, in the 0.02-2.00 mm fraction, >5% volcanic glass; or (c) Al plus 0.5 Fe percentages (by ammonium oxalate) totalling between 0.40-2.0 and, in the 0.02-2.Omm fraction, enough volcanic glass so that the glass percentage, when plotted against the value obtained by adding Al plus 0.5 Fe percentages in the fine-earth fraction, falls within an area defined graphically in Keys to Soil Taxonomy

P. JAMES, D. CHESTER & A. DUNCAN

320

Soils in tephra Volcanic rocks form predominantly from the solidification of silicate melts ranging in composition from 75% (Table 2). The chemical and mineralogical composition of magma erupted from any one centre may vary considerably over time. The data in Table 2 illustrate the variations in chemical composition of lavas and tephra, particularly in their content of silica, iron and the bases, Mg, Ca, Na and K. Tephra and lava may be geochemically very similar, as the data show, and both are composed largely of silicate minerals and glass, but the rapid cooling of tephra results in its being predominantly glassy. Because of the high glass content and fragmental nature of tephra, it weathers rapidly in moist environments. This results in the formation of distinctive secondary materials and in the rapid formation of soil. Lava and tephra may be mixed: ash and lapilli typically accumulate in depressions within the rough surfaces of lava, and ash may be deposited upon developed lava soils. A significant feature common to well-developed soils in

tephra and lava is a high content of clay (data for tephra soils are given in Table 3a).

Weathering of tephra and formation of secondary minerals Tephra occurs to significant depths on the present land surface in areas of active or recently active volcanism (Fig. 1). The major characteristics of tephra soils are determined largely by the nature of their finest and most reactive constituents, the secondary materials produced from the weathering of glass. Under favourable conditions of moisture and temperature, Al, Si and Fe are released by hydrolysis from volcanic glass more rapidly than they are able to crystallize. The secondary materials so produced tend to be non-crystalline, but include poorly ordered (also called 'paracrystalline' or 'short-rangeorder') as well as amorphous materials. They are primarily the hydrous aluminosilicates, allophane and imogolite; the iron hydroxide, ferrihydrite and opaline silica. The secondary

Table 2. Representative geochemical data for lava and tephra

wt% Wt%

IBas

2Bas

3Tra

4 And

5 Dae

6Rhy

7T-Rhy

8T-Bas

Si02 A1203 Fe203 FeO MnO MgO CaO Na20 K2O TiO2 P205+ H20

51.8 14.8 3.9 7.3 0.2 7.1 10.6 2.4 0.7 1.1 0.1 nd nd

45.4 14.7 4.1 9.2 0.2 7.8 10.5 3 1 3 0.4 nd nd

63.7 14.1 2 6 0.3

The archaeology of geological catastrophes

Geological Methods for Archaeology

Catastrophes

Catastrophes!

Catastrophes!

Catastrophes

Catastrophes!

Catastrophes!

The mathematics of natural catastrophes

Mathematics of Natural Catastrophes

A Choice of Catastrophes

A Choice of Catastrophes

The Mapping of Geological Structures (Geological Society of London Handbook)

The Archaeology of Micronesia (Cambridge World Archaeology)

The Archaeology of Micronesia (Cambridge World Archaeology)

The Archaeology of Time

The Archaeology of Weapons

The Archaeology of Shamanism

The Archaeology of Airfields

The Archaeology of Shamanism

The Archaeology of Colorado

The Archaeology of Athens

7 Kinds of Ordinary Catastrophes

Evolutionary Catastrophes: The Science of Mass Extinction

The Archaeology of the Colonized

The Archaeology of the Colonized

The Archaeology of the Caribbean

Geological History of Earth

Geological Landscapes of Britain

Evolutionary catastrophes: the science of mass extinction

The archaeology of geological catastrophes

Geological Methods for Archaeology

Catastrophes

Catastrophes!

Catastrophes!

Catastrophes

Catastrophes!

Catastrophes!

The mathematics of natural catastrophes

Mathematics of Natural Catastrophes

A Choice of Catastrophes

A Choice of Catastrophes

The Mapping of Geological Structures (Geological Society of London Handbook)

The Archaeology of Micronesia (Cambridge World Archaeology)

The Archaeology of Micronesia (Cambridge World Archaeology)

The Archaeology of Time

The Archaeology of Weapons

The Archaeology of Shamanism

The Archaeology of Airfields

The Archaeology of Shamanism

The Archaeology of Colorado

The Archaeology of Athens

7 Kinds of Ordinary Catastrophes

Evolutionary Catastrophes: The Science of Mass Extinction

The Archaeology of the Colonized

The Archaeology of the Colonized

The Archaeology of the Caribbean

Geological History of Earth

Geological Landscapes of Britain

Evolutionary catastrophes: the science of mass extinction

Recommend Documents