Ancient Earthquakes

Ancient Earthquakes

edited by Manuel Sintubin Department of Earth and Environmental Sciences Katholieke Universiteit Leuven Celestijnenlaan 200E, B-3001 Leuven Belgium Iain S. Stewart School of Geography, Earth and Environmental Sciences University of Plymouth Room 109, Fitzroy, Drake Circus Plymouth, Devon, PL4 8AA UK Tina M. Niemi Department of Geosciences University of Missouri–Kansas City 5100 Rockhill Road Kansas City, Missouri 64110 USA Erhan Altunel Department of Geological Engineering Eskişehir Osmangazi University 26480 Eskişehir Turkey

Special Paper 471 3300 Penrose Place, P.O. Box 9140

Boulder, Colorado 80301-9140, USA

2010

Copyright © 2010, The Geological Society of America (GSA), Inc. All rights reserved. GSA grants permission to individual scientists to make unlimited photocopies of one or more items from this volume for noncommercial purposes advancing science or education, including classroom use. For permission to make photocopies of any item in this volume for other noncommercial, nonprofit purposes, contact The Geological Society of America. Written permission is required from GSA for all other forms of capture or reproduction of any item in the volume including, but not limited to, all types of electronic or digital scanning or other digital or manual transformation of articles or any portion thereof, such as abstracts, into computer-readable and/or transmittable form for personal or corporate use, either noncommercial or commercial, for-profit or otherwise. Send permission requests to GSA Copyright Permissions, 3300 Penrose Place, P.O. Box 9140, Boulder, Colorado 80301-9140, USA. GSA provides this and other forums for the presentation of diverse opinions and positions by scientists worldwide, regardless of their race, citizenship, gender, religion, or political viewpoint. Opinions presented in this publication do not reflect official positions of the Society. Copyright is not claimed on any material prepared wholly by government employees within the scope of their employment. Published by The Geological Society of America, Inc. 3300 Penrose Place, P.O. Box 9140, Boulder, Colorado 80301-9140, USA www.geosociety.org Printed in U.S.A. GSA Books Science Editors: Marion E. Bickford and Donald I. Siegel Library of Congress Cataloging-in-Publication Data Ancient earthquakes / edited by Manuel Sintubin … [et al.]. p. cm. — (Special paper ; 471) Includes bibliographical references. ISBN 978-0-8137-2471-3 (pbk.) 1. Paleoseismology. I. Sintubin, M. QE539.2.P34A53 2010 551.22—dc22 2010036726 Cover, front: At Cape Sounion, the southernmost tip of the Attica peninsula (Greece), a temple is dedicated to Poseidon, the “Earth-Shaker,” god of earthquakes. The displaced drums of the multidrum columns are typically believed to be evidence of ancient earthquakes. The temple was built around 440 B.C. (photograph courtesy of K. Reicherter, Rheinisch-Westfälische Technische Hochschule Aachen, Germany). Back (top to bottom): Examples of earthquake-induced damage. Al Harif aqueduct, Syria (see Sbeinati et al., this volume, Chapter 20) (courtesy of M. Meghraoui, Institut de Physique du Globe, Strasbourg, France); Cnidus, Turkey; Hierapolis, Turkey; Sagalassos, Turkey; Petra, Jordan; Baelo Claudia, Spain (see Grützner et al., this volume, Chapter 12) (photographs courtesy of M. Sintubin, Katholieke Universiteit Leuven, Belgium).

10 9 8 7 6 5 4 3 2 1

Contents

Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . v Manuel Sintubin, Iain S. Stewart, Tina M. Niemi, and Erhan Altunel Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiii Understanding Earthquakes in the Ancient World 1. Dynamic landscapes and human evolution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 Geoffrey C.P. King and Geoffrey N. Bailey 2. Tectonic environments of ancient civilizations: Opportunities for archaeoseismological and anthropological studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21 Eric R. Force and Bruce G. McFadgen Historical Earthquakes and Their Societal Impact 3. The door knockers of Mansurah: Strong shaking in a region of low perceived seismic risk, Sindh, Pakistan . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29 Roger Bilham and Sarosh Lodi 4. San Antonio de Mucuñó, Mérida Andes, Venezuela: Relocation of a doctrine town following the 1674 earthquake . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39 Jaime Laffaille, Franck Audemard M., and Miguel Alvarado 5. New interpretations of the social and material impacts of the 1812 earthquake in Caracas, Venezuela . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47 Rogelio Altez 6. The impact of the 1157 and 1170 Syrian earthquakes on Crusader–Muslim politics and military affairs. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59 Kate Raphael 7. Western Crete: From Captain Spratt to modern archaeoseismology . . . . . . . . . . . . . . . . . . . . . . . 67 Manolis I. Stefanakis 8. Earthquake archaeology in Japan: An overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81 Gina L. Barnes

iii

iv

Contents Commentaries and Perspectives on Archaeoseismological Research 9. Historical earthquake catalogues and archaeological data: Achieving synthesis without circular reasoning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97 John D. Rucker and Tina M. Niemi 10. Historical earthquakes in Srinagar, Kashmir: Clues from the Shiva Temple at Pandrethan . . . 107 Roger Bilham, Bikram Singh Bali, M. Ismail Bhat, and Susan Hough 11. Earthquakes and civilizations of the Indus Valley: A challenge for archaeoseismology . . . . . . . 119 Robert L. Kovach, Kelly Grijalva, and Amos Nur 12. Comparing semiquantitative logic trees for archaeoseismology and paleoseismology: The Baelo Claudia (southern Spain) case study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129 Christoph Grützner, Klaus Reicherter, and Pablo G. Silva 13. Long-term effect of seismic activities on archaeological remains: A test study from Zakynthos, Greece . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145 Melek Tendürüs, Gert Jan van Wijngaarden, and Henk Kars 14. Assessment of seismically induced damage using LIDAR: The ancient city of Pınara (SW Turkey) as a case study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 157 Barış Yerli, Johan ten Veen, Manuel Sintubin, Volkan Karabacak, C. Çağlar Yalçıner, and Erhan Altunel Practices in Archaeoseismology 15. Ancient earthquakes from archaeoseismic evidence during the Visigothic and Islamic periods in the archaeological site of “Tolmo de Minateda” (SE Spain). . . . . . . . . . . . . . . . . . . . 171 M.A. Rodríguez-Pascua, P.G. Silva, V.H. Garduño-Monroy, R. Pérez-López, I. Israde-Alcántara, J.L. Giner-Robles, J.L. Bischoff, and J.P. Calvo 16. Distinguishing damages from two earthquakes—Archaeoseismology of a Crusader castle (Al-Marqab citadel, Syria) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 185 Miklós Kázmér and Balázs Major 17. Archaeoseismological studies at the temple of Amenhotep III, Luxor, Egypt . . . . . . . . . . . . . . . 199 Arkadi Karakhanyan, Ara Avagyan, and Hourig Sourouzian 18. Archaeological evidence for Roman-age faulting in central-northern Sicily: Possible effects of coseismic deformation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223 Giovanni Barreca, Maria Serafina Barbano, Serafina Carbone, and Carmelo Monaco 19. Faulting of the Roman aqueduct of Venafrum (southern Italy): Methods of investigation, results, and seismotectonic implications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 233 Paolo A.C. Galli, Alessandro Giocoli, Jose A. Naso, Sabatino Piscitelli, Enzo Rizzo, Stefania Capini, and Luigi Scaroina 20. Timing of earthquake ruptures at the Al Harif Roman aqueduct (Dead Sea fault, Syria) from archaeoseismology and paleoseismology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 243 Mohamed Reda Sbeinati, Mustapha Meghraoui, Ghada Suleyman, Francisco Gomez, Pieter Grootes, Marie-Josée Nadeau, Haithem Al Najjar, and Riad Al-Ghazzi 21. Offset archaeological relics in the western part of the Büyük Menderes graben (western Turkey) and their tectonic implications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 269 Önder Yönlü, Erhan Altunel, Volkan Karabacak, Serdar Akyüz, and Çağlar Yalçıner

The Geological Society of America Special Paper 471 2010

Preface Manuel Sintubin Department of Earth and Environmental Sciences, Katholieke Universiteit Leuven, Celestijnenlaan 200E, B-3001 Leuven, Belgium Iain S. Stewart School of Geography, Earth and Environmental Sciences, University of Plymouth, Room 109, Fitzroy, Drake Circus, Plymouth, Devon, PL4 8AA, UK Tina M. Niemi Department of Geosciences, University of Missouri–Kansas City, 5100 Rockhill Road, Kansas City, Missouri 64110, USA Erhan Altunel Department of Geological Engineering, Eskişehir Osmangazi University, 26480 Eskişehir, Turkey

Damaging earthquakes typically recur at intervals of centuries to millennia, but the instrumental record extends for no more than a century. To reduce the hazards from earthquakes and prepare proper mitigation plans we need a longer record of earthquakes than can be provided instrumentally. Looking for ancient earthquakes may be the key to the puzzle. We define ancient earthquakes primarily as pre-instrumental earthquakes that can only be identified through indirect evidence in the archaeological or geological record. While the latter is the subject of paleoseismology, the former is the subject of archaeoseismology. Earthquakes that are documented in the historical record (historical seismicity) may be included if they left marks in the archaeological or geological record. The problem seismic-hazard practitioners now face is that the instrumental record is too short and the historical record too incomplete. Historical catalogues record only a tiny proportion of the major earthquakes that have struck a region over centuries and millennia (cf. Ambraseys et al., 2002). That missing population of earthquakes clearly tempers reliable seismic-hazard assessments. The archaeological record, however, can bolster and augment that historical archive. What’s more, in extending the earthquake record beyond written sources, archaeoseismology serves as a bridge between instrumental and historical seismology, on the one hand, and paleoseismology and earthquake geology, on the other hand. Only the integration of all potential evidence of ancient earthquakes will enable a better understanding of the complex earthquake history of a region. Archaeoseismology has the potential to be a legitimate and complementary source of seismic-hazard information. Archaeoseismology thus aims at studying ancient earthquakes through indicators left in the archaeological record, such as destruction layers, structural damage to man-made constructions, cultural piercing features, indications of repairs, abandonment, cultural changes, etc. Archaeology can be used in three ways to help confront the seismic-hazard threat. First, where archaeological relics are displaced they can be used to find active faults, show in which direction faults slipped during the earthquake(s), and establish comparative fault slip rates. Second, archaeological information can date episodes of faulting and shaking. Third, we can search for ancient signs of seismic damage, often related to the ground shaking. The obvious difficulty with

Sintubin, M., Stewart, I.S., Niemi, T.M., and Altunel, E., 2010, Preface, in Sintubin, M., Stewart, I.S., Niemi, T.M., and Altunel, E., eds., Ancient Earthquakes: Geological Society of America Special Paper 471, p. v–xi, doi: 10.1130/2010.2471(00). For permission to copy, contact [email protected]. © 2010 The Geological Society of America. All rights reserved.

v

vi

Preface the last approach is that it is hard to distinguish between damage caused by an earthquake and that caused by another destructive event, such as war or the natural failure of foundations. Typologies of earthquakecharacteristic damage have been proposed but rarely have they been subjected to a critical and systematic analysis. Consequently, these archaeoseismological indicators are accepted by some earthquake scientists and rejected by others. In this volume we collected a series of case studies convincingly illustrating the different ways that the archaeological record can serve seismic-hazard studies. Moreover, the difficulties archaeoseismology face are discussed extensively. This volume is the first publication in the framework of the International Geoscience Programme IGCP 567 “Earthquake Archaeology: Archaeoseismology along the Alpine-Himalayan Seismic Zone” (ees.kuleuven.be/igcp567/). The key element of the program is our contention that archaeological earthquake evidence can make a valuable contribution to long-term seismic-hazard assessment in earthquakeprone regions where there is a long and lasting cultural heritage. Archaeological evidence indeed has the potential to determine earthquake activity over millennial time spans, especially where integrated with historical documents and geological evidence, as demonstrated in several papers in this volume. Archaeoseismology’s greatest challenge—and its foremost attraction—is to integrate the principles and practices of a wide range of sciences, from history, anthropology, archaeology, and sociology, over geology, geomorphology, geophysics, and seismology, to architecture and structural engineering. IGCP 567 seeks to establish an inclusive framework in which such a multidisciplinary approach can take root. It aims to encourage transparent discussions between this wide range of specialists, to promote a common vocabulary and understanding of purpose, and to foster a standardized methodology based on the consensus of those active in the field of archaeoseismology. Innovative archaeoseismological research has emerged in many parts of the globe, but its roots lie in the Eastern Mediterranean and the Middle East. The majority of methodological developments has indeed been grafted on archaeological investigations in this earthquake-prone region atop the Alpine collision zone between Africa and Eurasia, and depended strongly on identifying structural damage to buildings and other cultural remains at specific sites. Therefore, the program’s second objective is to extend the geographical provenance of archaeoseismological studies beyond its traditional territory, eventually pursuing a common archaeoseismological knowledge platform and a shared protocol. Considering the wide range of disciplines in which the authors are active, and the geographical distribution of cases presented, the main objectives of IGCP 567 are clearly reflected in this volume. In this respect, Special Paper 471 forms a new contribution to the ongoing process of refining our research strategies and practices in archaeoseismology. It visualizes the significant progress that persistent research efforts during the last two decades has made possible. This volume thus frames in a series of publications, each reflecting the gradual evolution toward an ever increasing multidisciplinary approach (cf. Caputo and Pavlides, 2008; Galadini et al., 2006; Stiros and Jones, 1996).

Preface UNDERSTANDING EARTHQUAKES IN THE ANCIENT WORLD The first two papers in Ancient Earthquakes reflect on the way tectonically active environments may have influenced man’s own prehistory. These thought-provoking papers give a wider, more theoretical context to a volume that is primarily focused on the effects of single earthquake events. In the paper by King and Bailey the authors argue that the geological instability in tectonically active regions may very well be an environmental driver for human development by creating more stable environmental conditions for sustained human settlement. This paper does not focus on the occasional disruptive earthquake events but on the modifications of regional topography resulting from the cumulative effect of earthquake activity over centuries to millennia. The refreshing hypothesis is convincingly illustrated by the early hominid settings in different parts of the East African Rift. This idea is further developed in the second paper. Force and McFadgen see a robust spatial correlation between ancient civilizations and active plate boundaries in the Alpine-Himalayan seismic zone, in particular with respect to the outline of trading routes. They also point out that a correlation can be supposed between increased seismicity and accelerated cultural change, which may define a new research question challenging practitioners of archaeoseismology. HISTORICAL EARTHQUAKES AND THEIR SOCIETAL IMPACT In this section four papers focus on historically known earthquakes to look for clues to assess their societal impact. These papers show that each earthquake, independent of its magnitude and frequency, provokes different societal responses, largely depending on the political, social, and economic context. The latter will eventually determine whether or not an earthquake disaster leads to the decline of a society. Earthquakes in themselves are incapable of causing the collapse of a community, let alone a civilization. These historical papers clearly show that any reference to neocatastrophism (e.g., Evans, 1928; Marinatos, 1939; Nur, 2008; Schaeffer, 1948) that has classically plagued archaeoseismology (cf. Ambraseys, 2005; Kovach and Nur, 2006) should be omitted. Bilham and Lodi argue that the discovery of decorated door knockers beneath a collapsed wall in Mansurah, the eighth-century capital of the Sindh province in Pakistan, supports an earthquake hypothesis to explain the destruction and abandonment of the capital in ca. 980 A.D. Such indications for strong shaking in a region of low perceived seismicity make a case for establishing a record of long-term seismicity using the 5 millennia of archaeological remains in Pakistan. In the paper by Laffaille et al. the impact of a historical earthquake in 1674 in northern Venezuela to a seventeenth-century Spanish settlement is analyzed. By balancing historical data with geological, geomorphological, and paleoseismological data, earthquake effects on a microscale are demonstrated. Earthquaketriggered landslides and mudflows eventually caused a relocation of the settlement over a few hundred meters. Altez, on the other hand, presents a detailed historical analysis of the very destructive earthquake that struck Caracas, Venezuela, on 26 March 1812. Altez uses documentary evidence to show that this earthquake had a significant impact on the Venezuelan society, primarily because the earthquake struck in a context of political, social, and economic turmoil. In this vulnerable historical context the quality of the build environment has deteriorated, biasing post-factum intensity estimates. Also the number of earthquake victims was overestimated because of the inclusion of victims of societal turmoil (e.g., war, famine). This social and political context is even more obvious in the historical assessment by Raphael of two devastating earthquakes that struck the Levant in a time span of less than 20 years in the twelfth century. Raphael nicely illustrates that the political and military balance between the Crusader Kingdom of Jerusalem and the Muslim Sultanate of Syria played a crucial role in the way both earthquakes influenced the regional political and military affairs. From these historical-anthropological approaches a clear lesson can be learned by earth scientists primarily focusing on the physical aspects of earthquake events. An earthquake disaster always seems to have a significant social component to it. The historical context should therefore be taken into account before any conclusions are made with respect to the societal effects of ancient earthquakes that are evidenced in the archaeological record. This section is rounded off with two papers reflecting on the history of archaeoseismology. The paper by Stefanakis is paying tribute to Captain T.A.B. Spratt (1811–1888) as the nineteenth-century forerunner of

vii

viii

Preface archaeoseismology in western Crete, Greece. His observations of coastal uplift in western Crete, as well as in the ancient harbors of Phalasarna and Kissamos, are still in line with the current knowledge with respect to the seismotectonics of western Crete. The paper by Barnes illustrates how the development of an archaeoseismological methodology largely depends on the particular regional context. Japanese earthquake archaeology differs greatly from the standard Mediterranean archaeoseismology, primarily because of the nature of the evidence. While in the Mediterranean approach focus has been put on the structural damage to buildings and other archaeological remains, secondary phenomena, such as liquefaction, landslides, and surface cracking, within a more territorial context are the main features used in the Japanese approach. COMMENTARIES AND PERSPECTIVES ON ARCHAEOSEISMOLOGICAL RESEARCH The first paper in this section on archaeoseismological research, by Rucker and Niemi, comments on one of the major problems that is encountered in archaeoseismology. When confronting historical earthquake catalogues with archaeological and geological evidence—all inherently incomplete—it should be clear that an ancient earthquake is not necessarily present in all records. A forced correlation will inevitably lead to circular reasoning. Rucker and Niemi state that the latter can only be avoided by independent supporting evidence and a clear assessment of the level of uncertainty of an earthquake hypothesis. Only then can archaeological earthquake evidence—objectively revealing unknown ancient earthquakes—provide useful data for seismic hazard assessment and mitigation. Also the paper by Bilham et al. holds a cautionary commentary on earthquake hypotheses that are postulated solely based on structural damage to buildings. By means of the fortuitous existence of photographs taken of the same place from the same viewpoint over time, the authors were able to use the tenth-century Shiva Temple at Pandrethan as a strong-motion seismometer in an attempt to elucidate the earthquake history of the Kashmir valley in the past millennium. Their analysis shows that an earthquake hypothesis cannot be retained to explain the damage recorded on the series of photographs of the temple taken at different times during the past 200 years. The paper by Kovach et al. focuses on the Harappan civilization in the Indus Valley region (Pakistan, India). Their survey clearly shows that this part of the Alpine-Himalayan seismic zone has great potential for archaeoseismology. But the archaeoseismological research is confronted with the challenges to incorporate secondary earthquake effects, such as changes in the fluvial systems and coastal elevation. The three following papers in this section present the results of different novel approaches, clearly offering new perspectives for future archaeoseismological research. These papers frame in our collective search for shared protocols and standardized methodologies. Grützner et al. use their extensive archaeological, geomorphological, and geological database in and around the archaeological site of Baelo Claudia—a Roman city along the southwestern Iberian coast—to compare the logic tree approach for paleoseismology, developed by Atakan et al. (2000), and for archaeoseismology, designed by Sintubin and Stewart (2008) (the latter is also applied in the paper by RodríguezPascua et al.). Both approaches try to express semiquantitatively the level of confidence with respect to an earthquake hypothesis. This comparison reveals a number of strengths and weaknesses of both approaches. In their paper Tendürüs et al. tackle another interesting problem of archaeoseismology. It concerns the question of whether or not the spatial distribution of archaeological remains and their preservation conditions can be correlated to seismic activity. The Tendürüs et al. modeling approach calculates the cumulative effect of continuing seismic activity for a period of 100 years and expresses it as cumulative peak ground acceleration. With respect to the build environment the focus is shifted from the effects of single earthquake events to a continuing deterioration. The model is applied to the island of Zakynthos in western Greece. It results in a promising correlation between the distribution and preservation of archaeological remains and seismic activity. Finally, Yerli et al. explore the applicability of ground-based LIDAR (Light Detection and Ranging) as a new technique in our pursuit of a more quantitative archaeoseismology. They have chosen the Roman theater in the ancient city of Pınara in southwestern Turkey as a case study. Ancient earthquakes have indeed been evidenced at this archaeological site. The detailed LIDAR mapping reveals a particular damage pattern (e.g., back-tilting of seating rows), that would otherwise pass by unnoticed. Yerli et al. suggest that the Roman theater may have recorded fault-block rotation related to activity during Roman times on a nearby fault.

Preface PRACTICES IN ARCHAEOSEISMOLOGY This section consists of seven papers illustrating the diversity of practices in archaeoseismology. Each archaeological site has unique characteristics, related to both the archaeological and seismotectonic context, so that a specific approach is required each time. These papers illustrate how archaeological sites may serve as seismoscopes recording ancient—to date unknown—earthquakes. The first three cases concern archaeological evidence for damage caused by earthquake-related ground shaking. The paper by Rodríguez-Pascua et al. focuses on the ancient settlement of El Tolmo de Minateda in southeastern Spain. This site has been occupied for ca. 3800 years, from the Late Bronze age to the Visigoth and Islamic periods. This site is situated in an active intraplate region characterized by a very low frequency of major seismic events. Archaeological sites are thus a primary tool to discover major ancient earthquakes, contributing to a correct assessment of the seismic hazard in the region. The integration of archaeological evidence with unique geomorphological evidence—rock falls including Visigoth carved tombs—reveals at least two major seismic events in the seventh to tenth century A.D. Kázmér and Major discuss in their paper earthquake hypotheses for typical structural damage and repairs in the well-preserved thirteenth-century fortification of the Al-Marqab citadel in Syria. They conclude that the damages observed can be associated with major ground shaking caused by at least two separate earthquake events. The major damage is attributed to the 1202 earthquake, suggesting a local intensity of VIII–IX at the site of the citadel. The latter is higher than assumed before, inferring an increase of calculated magnitude for that particular historical earthquake. In the paper by Karakhanyan et al. the results of an extensive study of the damage patterns on the worldfamous Colossi of Memnon in the temple of Amenhotep III at Luxor in Egypt are presented. An earthquake hypothesis is developed based primarily on structural damage characteristics, but furthermore supported by evidence of massive liquefaction exposed by paleoseismological trenching. The authors’ findings are seemingly in contradiction to the lack of any clear earthquake account in 3500 years of papyri and epigraphic sources, with the exception of the 27 B.C. earthquake described by Strabo. The earthquake evidence cannot, however, be correlated with Strabo’s earthquake, but is indicative of a major earthquake that struck the region of Luxor between 1200 and 900 B.C. The widespread destruction of the temple suggests a nearby—to date unknown—active fault, capable of producing major—potentially hazardous—earthquakes. The following four papers deal with displaced archaeological remains that serve as cultural piercing points from which to derive fault displacement, identify individual surface-rupturing seismic events, and estimate long-term slip rates. The cases illustrated in the paper by Barreca et al. are found in Late Roman sites in central-northern Sicily, Italy, a region characterized by moderate instrumental and historical seismicity. The archaeological evidence consists of a ruptured votive niche, the prevalence of pottery pieces and coins dated to the fourth century, a sudden decrease in human activity by the end of the fourth century, and an atypical Late Roman grave. This evidence is correlated to the historical 361 A.D. earthquake in central Sicily. Barreca et al. consider their results as a first step in better constraining the epicentral area of this historical earthquake. Aqueducts are very interesting man-made structures when it comes to identifying capable faults and deriving slip rates, primarily because of their linear nature extending over long distances across a seismic landscape and their particular slope, guaranteeing sufficient hydrological head. The use of aqueducts as archaeoseismological tools is convincingly demonstrated in the following two papers. Galli et al. present the results of a geological, geophysical, and geodetic survey of a first-century B.C. aqueduct in southern Italy. The cumulative displacement infers at least two surface-rupturing seismic events after the construction of the aqueduct. No historical record exists for either event. The Al Harif Roman aqueduct in Syria is a unique archaeoseismological site, because it crosses the Dead Sea transform fault—i.e., the plate boundary between the African and Arabian plate. The aqueduct exhibits a left-lateral offset of more than 13 m since Roman times. Moreover, calcareous deposits associated with the overflowing of carbonate-saturated waters can be used as an extra dating tool in the reconstruction of the earthquake history. Sbeinati et al. combine the extensive archaeological and paleoseismological evidence with the specific dating evidence from the tufa deposits to reconstruct the succession of surfacerupturing events between 63 B.C. and 1170 A.D., the last faulting event affecting this segment of the Dead Sea transform.

ix

x

Preface Finally, the paper by Yönlü et al. discusses displaced relics in the ancient city of Priene and the deformation of an Ottoman bridge, both situated along the northern margin of the active Büyük Menderes graben in southwestern Turkey. Detailed field observations and LIDAR mapping show a significant right-lateral offset on top of the expected normal component of the boundary fault system of the graben. This volume gives a nice sample of our ongoing efforts with respect to the search for ancient earthquakes. It shows the diversity of approaches and the wide range of disciplines involved, but also the potential of archaeoseismology with respect to a better understanding of the earthquake history in many places around the world that are threatened by seismic hazards. We hope this volume offers a taste of the complexity with which archaeoseismologists are confronted, from interpreting structural damage patterns and secondary earthquake phenomena on archaeological sites to assessing the historical, political, and social context in which the ancient earthquake occurred. In this respect, we hope to have been able to arouse an interest for archaeoseismology in the broader community of earth scientists, seismologists, historians, and archaeologists, one of the primary aims of the International Geoscience Programme IGCP 567. 13 September 2010

REFERENCES CITED Altez, R., 2010, this volume, New interpretations of the social and material impacts of the 1812 earthquake in Caracas, Venezuela, in Sintubin, M., Stewart, I.S., Niemi, T.M., and Altunel, E., eds., Ancient Earthquakes: Geological Society of America Special Paper 471, doi: 10.1130/2010.2471(05). Ambraseys, N.N., 2005, Archaeoseismology and neocatastrophism: Seismological Research Letters, v. 76, no. 5, p. 560–564, doi:10.1785/gssrl.76.5.560. Ambraseys, N.N., Jackson, J.A., and Melville, C.P., 2002, Historical Seismicity and Tectonics: The Case of the Eastern Mediterranean and the Middle East, in Lee, W.H.K., Kanamori, H., Jennings, P.C., and Kisslinger, C., eds., International Handbook of Earthquake & Engineering Seismology: International Geophysics Series 81A: Academic Press, Amsterdam, p. 747–763. Atakan, K., Midzi, V., Moreno Toiran, B., Vanneste, K., Camelbeeck, T., and Meghraoui, M., 2000, Seismic hazard in regions of present day low seismic activity: Uncertainties in the paleoseismic investigations along the Bree Fault Scarp (Roer Graben, Belgium): Soil Dynamics and Earthquake Engineering, v. 20, p. 415–427, doi: 10.1016/S0267 -7261(00)00081-6. Barnes, G.L., 2010, this volume, Earthquake archaeology in Japan: An overview, in Sintubin, M., Stewart, I.S., Niemi, T.M., and Altunel, E., eds., Ancient Earthquakes: Geological Society of America Special Paper 471, doi: 10.1130/2010.2471(08). Barreca, G., Barbano, M.S., Carbone, S., and Monaco, C., 2010, this volume, Archaeological evidence for Romanage faulting in central-northern Sicily: Possible effects of coseismic deformation, in Sintubin, M., Stewart, I.S., Niemi, T.M., and Altunel, E., eds., Ancient Earthquakes: Geological Society of America Special Paper 471, doi: 10.1130/2010.2471(18). Bilham, R., and Lodi, S., 2010, this volume, The door knockers of Mansurah: Strong shaking in a region of low perceived seismic risk, Sindh, Pakistan, in Sintubin, M., Stewart, I.S., Niemi, T.M., and Altunel, E., eds., Ancient Earthquakes: Geological Society of America Special Paper 471, doi: 10.1130/2010.2471(03). Bilham, R., Singh, B., Bhat, I., and Hough, S., 2010, this volume, Historical earthquakes in Srinagar, Kashmir: Clues from the Shiva Temple at Pandrethan, in Sintubin, M.,

Stewart, I.S., Niemi, T.M., and Altunel, E., eds., Ancient Earthquakes: Geological Society of America Special Paper 471, doi: 10.1130/2010.2471(10). Caputo, R., and Pavlides, S.B., 2008, Earthquake Geology: Methods and Applications: Tectonophysics, v. 453, 296 p. Evans, A., 1928, The Palace of Minos, part II: London, 844 p. Force, E.R., and McFadgen, B.G., 2010, this volume, Tectonic environments of ancient civilizations: Opportunities for archaeoseismological and anthropological studies, in Sintubin, M., Stewart, I.S., Niemi, T.M., and Altunel, E., eds., Ancient Earthquakes: Geological Society of America Special Paper 471, doi: 10.1130/2010.2471(02). Galadini, F., Hinzen, K.-G., and Stiros, S.C., 2006, Archaeoseismology at the beginning of the 21st century: Journal of Seismology, v. 10. Galli, P.A.C., Giocoli, A., Naso, J.A., Piscitelli, S., Rizzo, E., Capini, S., and Scaroina, L., 2010, this volume, Faulting of the Roman aqueduct of Venafrum (southern Italy): Methods of investigation, results, and seismotectonic implications, in Sintubin, M., Stewart, I.S., Niemi, T.M., and Altunel, E., eds., Ancient Earthquakes: Geological Society of America Special Paper 471, doi: 10.1130/2010.2471(19). Grützner, C., Reicherter, K., and Silva, P.G., 2010, this volume, Comparing semiquantitative logic trees for archaeoseismology and paleoseismology: The Baelo Claudia (southern Spain) case study, in Sintubin, M., Stewart, I.S., Niemi, T.M., and Altunel, E., eds., Ancient Earthquakes: Geological Society of America Special Paper 471, doi: 10.1130/2010.2471(12). Karakhanyan, A., Avagyan, A., and Sourouzian, H., 2010, this volume, Archaeoseismological studies at the temple of Amenhotep III, Luxor, Egypt, in Sintubin, M., Stewart, I.S., Niemi, T.M., and Altunel, E., eds., Ancient Earthquakes: Geological Society of America Special Paper 471, doi: 10.1130/2010.2471(17). Kázmér, M., and Major, B., 2010, this volume, Distinguishing damages from two earthquakes—Archaeoseismology of a Crusader castle (Al-Marqab citadel, Syria), in Sintubin, M., Stewart, I.S., Niemi, T.M., and Altunel, E., eds., Ancient Earthquakes: Geological Society of America Special Paper 471, doi: 10.1130/2010.2471(16). King, G.C.P., and Bailey, G.N., 2010, this volume, Dynamic landscapes and human evolution, in Sintubin, M., Stewart, I.S., Niemi, T.M., and Altunel, E., eds., Ancient Earthquakes: Geological Society of America Special Paper 471, doi: 10.1130/2010.2471(01).

Preface Kovach, R.L., and Nur, A., 2006, Earthquakes and archeology: Neocatastrophism or science?: Eos (Transactions, American Geophysical Union), v. 87, p. 317–318. Kovach, R.L., Grijalva, K., and Nur, A., 2010, this volume, Earthquakes and civilizations of the Indus Valley: A challenge for archaeoseismology, in Sintubin, M., Stewart, I.S., Niemi, T.M., and Altunel, E., eds., Ancient Earthquakes: Geological Society of America Special Paper 471, doi: 10.1130/2010.2471(11). Laffaille, J., Audemard M., F., and Alvarado, M., 2010, this volume, San Antonio de Mucuñó, Mérida Andes, Venezuela: Relocation of a doctrine town following the 1674 earthquake, in Sintubin, M., Stewart, I.S., Niemi, T.M., and Altunel, E., eds., Ancient Earthquakes: Geological Society of America Special Paper 471, doi: 10.1130/2010.2471(04). Marinatos, S., 1939, The volcanic destruction of Minoan Crete: Antiquity, v. 13, p. 415–439. Nur, A., 2008, Apocalypse: Earthquakes, Archaeology, and the Wrath of God: Princeton, New Jersey, Princeton University Press, 309 p. Raphael, K., 2010, this volume, The impact of the 1157 and 1170 Syrian earthquakes on Crusader-Muslim politics and military affairs, in Sintubin, M., Stewart, I.S., Niemi, T.M., and Altunel, E., eds., Ancient Earthquakes: Geological Society of America Special Paper 471, doi: 10.1130/2010.2471(06). Rodríguez-Pascua, M.A., Silva, P.G., Garduño-Monroy, V.H., Pérez-López, R., Israde-Alcántara, I., Giner-Robles, J.L., Bischoff, J.L., and Calvo, J.P., 2010, this volume, Ancient earthquakes from archaeoseismic evidence during the Visigothic and Islamic periods in the archaeological site of “Tolmo de Minateda” (SE Spain), in Sintubin, M., Stewart, I.S., Niemi, T.M., and Altunel, E., eds., Ancient Earthquakes: Geological Society of America Special Paper 471, doi: 10.1130/2010.2471(15). Rucker, J.D., and Niemi, T.M., 2010, this volume, Historical earthquake catalogues and archaeological data: Achieving synthesis without circular reasoning, in Sintubin, M., Stewart, I.S., Niemi, T.M., and Altunel, E., eds., Ancient Earthquakes: Geological Society of America Special Paper 471, doi: 10.1130/2010.2471(09). Sbeinati, M.R., Meghraoui, M., Suleyman, G., Gomez, F., Grootes, P., Nadeau, M.-J., Al Najjar, H., and Al-Ghazzi, R., 2010, this volume, Timing of earthquake ruptures at the

xi

Al Harif Roman aqueduct (Dead Sea fault, Syria) from archaeoseismology and paleoseismology, in Sintubin, M., Stewart, I.S., Niemi, T.M., and Altunel, E., eds., Ancient Earthquakes: Geological Society of America Special Paper 471, doi: 10.1130/2010.2471(20). Schaeffer, C.F.A., 1948, Stratigraphie Comparée et Chronologie de l’Asie Occidentale: London, Oxford University Press. Sintubin, M., and Stewart, I.S., 2008, A logical methodology for archaeoseismology: A proof of concept at the archaeological site of Sagalassos, southwest Turkey: Bulletin of the Seismological Society of America, v. 98, no. 5, p. 2209– 2230, doi:10.1785/0120070178. Stefanakis, M.I., 2010, this volume, Western Crete: From Captain Spratt to modern archaeoseismology, in Sintubin, M., Stewart, I.S., Niemi, T.M., and Altunel, E., eds., Ancient Earthquakes: Geological Society of America Special Paper 471, doi: 10.1130/2010.2471(07). Stiros, S.C., and Jones, R.E., 1996, Archaeoseismology, in Whitbread, I.K., ed., Fitch Laboratory Occasional Paper: Athens, Institute of Geology & Mineral Exploration & The British School at Athens, p. 268. Tendürüs, M., van Wijngaarden, G.J., and Kars, H., 2010, this volume, Long-term effect of seismic activities on archaeological remains: A test study from Zakynthos, Greece, in Sintubin, M., Stewart, I.S., Niemi, T.M., and Altunel, E., eds., Ancient Earthquakes: Geological Society of America Special Paper 471, doi: 10.1130/2010.2471(13). Yerli, B., ten Veen, J., Sintubin, M., Karabacak, V., Yalçıner, C.Ç., and Altunel, E., 2010, this volume, Assessment of seismically induced damage using LIDAR: The ancient city of Pınara (SW Turkey) as a case study, in Sintubin, M., Stewart, I.S., Niemi, T.M., and Altunel, E., eds., Ancient Earthquakes: Geological Society of America Special Paper 471, doi: 10.1130/2010.2471(14). Yönlü, Ö., Altunel, E., Karabacak, V., Akyüz, S., and Yalçıner, Ç., 2010, this volume, Offset archaeological relics in the western part of the Büyük Menderes graben (western Turkey) and their tectonic implications, in Sintubin, M., Stewart, I.S., Niemi, T.M., and Altunel, E., eds., Ancient Earthquakes: Geological Society of America Special Paper 471, doi: 10.1130/2010.2471(21). MANUSCRIPT ACCEPTED BY THE SOCIETY 19 MAY 2010

Printed in the USA

Acknowledgments

The editors acknowledge the reviewers for their contribution to the quality of this volume. The reviewers were: A. Agnon (Hebrew University, Israel), H.S. Akyiiz (Istanbul Technical University, Turkey), N. Ambra­ seys (Imperial College London, UK), K. Atakan (University of Bergen, Norway), F. Audemard (Fundacion Venezolana de Investigaciones Sismologicas, Venezuela), B. Batten (Oberlin University, Tokyo, Japan), C. Beck (Universite de Savoie, France), Z. 6 ka

grassland

site are believed to be related to tectonic activity, and the wider region is seismically active (Douglas, 2006; Kuman et al., 1999). The later site of Boomplaas clearly takes advantage of a favorable environment created by ongoing activity. In other words, all inland sites are associated in some way with tectonically created and maintained features. For the present, coastal sites are excluded because of the complications associated with sea-level variation and coastal change, and we turn to a closer examination of some of the key inland sites next. Makapan Valley The Makapan Valley contains a series of caves dating to various periods within the Pliocene-Pleistocene and into historic

times, such as the Buffalo Cave, Cave of Hearths, and the famous Makapansgat Limeworks (Fig. 7). From the Limeworks Cave Member 3 and 4 breccias (dated at 3.2–2.7 Ma), 27 fossil specimens of Australopithecus africanus have been recovered (Tobias, 2000). Hominins and cercopithecine monkeys were accumulated by various predators, including hyenas, birds of prey, and carnivores (Reed, 1997). The later deposit, Member 5 contains a small assemblage of mammalian fossils but no hominin specimens. A three-dimensional view highlights the topographic complexity of the Makapan Valley region (Fig. 11A), and faults and other geomorphological features are interpreted in Figures 11B and 11C, with close-up detail shown in Figure 12. In the south (upper part of Figs. 11A and 11B), there is a large plain, now

Dynamic landscapes and human evolution

11

C

A

D

foreground scale

~500 m

B

ve wi ry r th ece crater lake ve nt rti fa ca ul D l s ts ca C rp s photographs E F

narrowing valley

smaller plain

E

Figure 10. Close-up of the region indicated in Figure 9. (A) Modified Google image showing a crater lake, volcanic lavas, and fault scarps in the vicinity of the Gablaytu volcano. (B) Interpretation of features shown in A, highlighting the Gablaytu crater lake, a narrow valley confined by lava flows and faults, and associated plains of varying size. (C–F) Photographs of features indicated in B, showing, respectively, major fault scarps (C), crater lake (D), and valleys bounded by relatively impassable lava flows and fault scarps (E) and (F).

F

large plain

exploited for agriculture, which is down-dropped with respect to the mountains in the foreground by an active fault system identified as fault Alpha (α). A contemporary feature of this plain is the Nylsvlei wetland, which has long been considered to be possibly related to continued tectonic activity (McCarthy and Hancox, 2000; Wagner, 1927). The active fault scarp probably associated with the creation and maintenance of the wetland forms a geomorphic feature shown in Figure 11C. A second fault, Beta (β), can be identified with the River Nyl running close to its base. The asymmetry of the valley could indicate continued tectonic activity or simply a shift of the river course to the west by sediment that reaches the valley from the east (left). A third fault, Gamma (γ), passes close to the site. It shows unequivocal evidence for ongoing activity (Fig. 12) and is responsible for uplift and consequent down-cutting and sedimentation close to the site, i.e., conditions corresponding to the model outlined in Figure 5B.

The immediate vicinity of the site is characterized by a gorge and associated steep cliffs and rough terrain; these features would have afforded important opportunities for safety and security. The river and the fertile, well-watered, sedimentary plains would have offered good foraging areas nearby. If, as Anton et al. (2002) argued, the foraging range of australopithecines was as small as 38 ha (essentially the area within ~500 m of a given point in the landscape), then this localized combination of rough terrain and productive resources would have been a key feature of their local environment. A viable breeding population would of course require a larger territory, so that the wider area around the Makapan Valley would also be important to long-term viability. The wider area reveals a combination of smaller plains, large open plains, and wetlands within ranging distance of the valley itself. In addition, a variety of valley conditions ranging from dry to marshy would have offered a range of habitats. These features are consistent with on-site indicators suggesting high biodiversity

12

King and Bailey

A Site

~ 5 km

N

B

Nylsvlei wet land large plain

several fault branches down dropped

Site

fault α

A

up down

fault γ back-til te

d

medium plain small plain 1

small plain 2

C

do wn

up

fault β

Figure 11. Region around the Makapan Valley, Northern Province, South Africa. (A) Landsat thematic mapper image (bands 2, 4, and 7 are red, green, and blue, respectively) draped over exaggerated digital elevation data to give a three-dimensional (3-D) effect. A vertical exaggeration of between 5 and 10 times the standard elevation is used for oblique images, which enhances the visual interpretation of high-lying versus low-lying regions. The effect of this exaggeration gives an impression similar to that of a land-based observer viewing the topography. (B) Faults α, β, and γ are identified. Recent activity of α is partly demonstrated by the presence of the Nylsvlei wetland, which results from a perturbed river (McCarthy et al., 2004). (C) A fault scarp typical of repeated earthquakes is associated with fault α. Fault β may be active, since the river is displaced to the west side of the valley, but river displacement could also be due to sediment sources coming from the east. Fault γ close to the site is clearly active (see caption to Fig. 12).

24.8°S, 28.43°E fault scarp

fault scarp

and diverse habitat conditions within close range of the site, including both wetland (C3) and dry-land (C4) indicators (Cadman and Rayner, 1989; McKee, 1999; Reed, 1997; Sponheimer and Lee-Thorp, 1999; Vrba, 1982). Taung The lime-mining quarry at Taung was the site of the discovery of the type specimen of Australopithecus africanus in 1924 (Dart, 1925), though dating based on faunal correlations suggests that it is younger than other australopithecine sites, with an age of ca. 2.6–2.4 Ma (McKee, 1993; Partridge, 2000; Tobias, 2000).

Much of the original hominin-bearing tufa deposit was destroyed by mining processes before systematic excavation could be undertaken, so that on-site paleoenvironmental data are lacking, and interpretation depends solely on interpretation of landscape features (Fig. 13). Two faults on either side of the Taung region create a rift valley (graben) with uplifted, drier flanks on each side and a downdropped, sedimented plain in the center (Fig. 13). As can be seen in the foreground, this fertile, sedimented plain is today being used for agriculture. The faulting has down-dropped the valley, causing rivers to cut into the uplifted valley sides. These faults

Dynamic landscapes and human evolution

13

A

Site

lt

α fau

small plain 2

small plain 1 foreground scale

~1 km

B

View from site to small plain 1

C

View from site up valley

crosscut earlier geological structures and are not controlled by them. Two rivers cut into the rift flanks, with downcutting in the vicinity of the Taung site. All these features indicate essentially similar conditions to those in the Makapan Valley, corresponding to the model outlined in Figure 5B, with varied habitats near the site and a complex topography affording opportunities for protection in the immediate vicinity and for monitoring of resources in the wider landscape. Sterkfontein The deposits at Sterkfontein make up seven members, of which two are well studied: Member 4 (ca. 2.8–2.4 Ma), with Australopithecus africanus fossils, and Member 5 (ca. 2.5– 1.4 Ma), with a succession of Oldowan and Acheulean stone-tool industries as well as two later species of hominins, namely, early Homo and Paranthropus (Kuman and Clarke, 2000). Faunal remains indicate a mixture of grassland and woodland species in Member 4, and fossilized wood fragments indicate the presence of gallery forest and tropical understory shrubs in the near vicinity of the site (Bamford, 1999). In Member 5 times, all faunal

Figure 12. Close-up of Makapan Valley. (A) Closer view of the topography modified from Google Earth image. Fault γ is indicated by arrows. Red arrows indicate steepening of the base of the slope. A yellow arrow indicates a “wine glass,” a valley that narrows toward the fault. A blue arrow indicates a spur that has been truncated by the fault. Together, these features are unambiguous evidence of active faulting. Small fertile plains result from sediment back-filling of earlier features as a result of the tectonically modified drainage. (B) Sediment-filled valley resulting from down-dropping on fault γ. (C) Steep topography in the Makapan Valley.

indicators suggest generally more open conditions but with some persisting woodland. The topographic setting of the site (Figs. 14A and 14B) shows evidence for faulting that crosscuts the mapped geological structures and has disturbed the profile of the adjacent river to create areas of sedimentation and downcutting of several meters, most probably due to continued movement (as in Fig. 5B). However, there are no clear earthquake fault scarps as at Makapansgat and Taung, so that continuing activity cannot be unequivocally established as yet. In the region to the north, the rivers are deeply incised, again suggesting activity. As at Makapansgat, these features are consistent with the presence of environmental signals in the on-site evidence indicating a combination of open grassland and more wooded habitats. A project is currently under way to improve the seismic network coverage in the area and to measure erosion and river downcutting rates using cosmogenic dating. Boomplaas Boomplaas cave has an important sequence of deposits dating from ca. 70 ka onward and includes stone tools from the

Up U p

Up

Site

Down

activ

e

cu down

tting

ps car lt s Fau

sedimented plain

~10 km

N

Figure 13. Region around the Taung Valley. Landsat thematic mapper image (bands 2, 4, and 7 are red, green, and blue, respectively) draped over exaggerated digital elevation data to give a three-dimensional (3-D) effect. An earthquake fault scarp on the east side of the valley is indicated by white arrows. The fault crosscuts geological features and varies in altitude and thus cannot be a river terrace or other erosional or depositional feature. Uplifting of the eastern flank has caused rivers to incise, and the down-dropped valley has been filled with sediment. Faulting may also be associated with the western side of the valley, causing the incised valley that hosted the breccia with the hominin fossil. The down-dropped sedimented basin is today used for agriculture. The presumed faults at depth are shown as gray dashed lines.

A Up Down

50 km B

B

10 km gorge

Elevation (m)

A

1450

Sterkfontein

deposition

1400

undis 1350

downcutting

turbe

d pro

gorge riv

file

er pro file

~ 25 km A

B

Figure 14. Evidence of tectonic activity in the Sterkfontein region. (A) Shaded relief map of the Sterkfontein “Cradle of Humankind” area (yellow rectangle) and the region to the east. White arrows indicate an east-west fault. This substantial feature extends for >150 km, offsets the morphology (up to the north, down to the south), and crosscuts earlier geological structures. The shaded map uses SRTM 3 data with the light source located at N45°W at an elevation of 5°. (B) Landsat thematic mapper image (bands 2, 4, and 7 are red, green, and blue, respectively) with relief shading based on digital elevation model (DEM; ~10 m resolution) derived from Stereo SPOT images. The fault identified in A passes to the north of the Sterkfontein site (dashed gray line) and apparently controls the position of the Blaaubank River, which follows the northern side of the valley. As it flows to the east, the river passes through a gorge (several meters deep). The inset shows a profile of the river (from SPOT DEM), which is downcutting in the gorge and aggrading above it. Such a profile is commonly associated with active faulting.

Dynamic landscapes and human evolution Middle Stone Age (MSA) industries of Still Bay and the succeeding Howiesons Poort, Later Stone Age material, and evidence of sheep pastoralism (Deacon, 1995; Deacon et al., 1978; Henshilwood, 2010). The cave itself has an open aspect overlooking a valley filled with sediment, and Deacon (1979) suggested that the site was well placed to intercept migratory animals. From a topographic perspective, the site is located on the boundary between a down-dropped valley and an uplifted scarp (Fig. 15). The valley area has been down-dropped by the fault to the north, and this has caused valleys in the earlier topography to become partly buried, resulting locally in highly fertile sediment-filled valleys. The cave site itself was formed in the earlier topography and is

15

not a consequence of the system that is now active. Fault scarps like those shown in Figure 15 can be found elsewhere in the cape region and may well partly control coastal sites. DISCUSSION Our review of key South African fossil and archaeological sites shows that there is considerable evidence for dynamic landscape changes resulting from tectonic activity over the time span during which the sites were formed, and that this evidence has not previously been recognized. The changes are not as rapid or as dramatic as in the African Rift, but they are changes that

A Up Down

fault

Site Sedimented Cango valley Downcutting

0

kilometers 5

10

B

Site

Sedimented valley

Figure 15. Evidence for tectonic activity in the Boomplaas region. (A) Landsat thematic mapper image (bands 2, 4, and 7 are red, green, and blue, respectively). White arrows indicate two faults. Yellow arrows indicate the sediment-filled Cango valley. Red arrows indicate downcutting. (B) View of cave site showing the Cango valley in foreground. The valley is subject to intermittent flooding and provides rich farmland. Earthquakes are commonly felt in this region.

16

King and Bailey

would have been advantageous for human occupation. They would have created repeated disturbance of drainage networks to sustain localized wetlands with fertile resources and water supplies, and formed or maintained complex local topography affording diversity of resource zones in close proximity and tactical advantages in hiding from predators or accessing mobile prey. The relatively subdued nature of tectonic activity compared to the Ethiopian Rift, which at first sight appears to pose a difficulty for the hypothesis of a close association between tectonically disturbed landscapes and early hominin settlement, turns out to be an advantage, allowing a direct evaluation of the relationship between tectonic structures and human activity without resorting to an analog approach. Three issues for discussion arise from this review. A first issue is the problem of bias in the distribution of known archaeological and fossil sites resulting from differential preservation and visibility of material. Geological processes in tectonically active areas tend to generate higher rates of sedimentation, which bury and therefore preserve fossil and archaeological material, and high rates of subsequent disruption and erosion, which then expose to view deposits formed at a much earlier period in time. Thus, it can be argued that the correlation between early archaeological and fossil sites and areas of tectonic activity is purely coincidental and is the result of greater visibility and exposure of material in such environments compared with environments where tectonic activity is very low or nonexistent. A similar argument is sometimes made about caves and rock shelters. These tend to be easily visible targets for archaeological investigations and often provide a protective environment for the accumulation of sediments and the preservation of archaeological material in stratified sequences. Hence, the argument can be proposed that the distribution of prehistoric settlement patterns derived from cave and rock-shelter deposits is unrepresentative and biased by conditions of geological visibility, a proposition pertinent to our South African region, where many key sites are in caves. Taken to its logical conclusion, this argument would require us to suppose that the distribution of prehistoric archaeological and fossil materials in space and time is solely a function of geological processes affecting visibility, that tectonically active areas of Africa have no particular significance for human evolution, and that actually other regions of Africa or elsewhere were equally important, if not more so, despite the absence of evidence in favor of such a proposition. Such an argument would be simplistic. We doubt that the concentration of finds in East and South Africa is wholly unrepresentative or can tell us nothing about the environmental conditions in which early human populations prospered. Other areas with early sites are sometimes claimed to lack the tectonic activity or the topographic features to which we have drawn the readers’ attention here. A notable case in point is the early finds of hominin fossils in the Chad region (Brunet et al., 1995). However this region, though far from the East African Rift, is one of the most tectonically active areas of sub-Saharan Africa outside the rift (Burke, 1996).

Factors of differential visibility resulting from geological processes are not trivial, and we do not discount them. However, they can be addressed in a variety of ways. In the case of caves and rock shelters, not all that were available for use contain evidence of human activity. Of those that do, some clearly show evidence of more activity than others. Some regions with available caves and rock shelters clearly show greater concentrations of evidence than others. Open-air sites can be targeted to provide a control and are often found once they are sought out. A similar approach can be employed in relation to tectonic factors. Moreover, it is not necessarily the case that archaeological materials from very early periods will be invisible in areas that are not tectonically active because these areas are likely to have smoothed surfaces that are subject neither to accumulation of obscuring sediment nor to erosion, and artifacts once deposited are likely to remain in place for many tens or hundreds of millennia. Granted, surface artifacts are more difficult to date than stratified material and likely to comprise only stone tools, but if such areas were attractive at an early period, we might expect to find concentrations of distinctive stone tools characteristic of Lower Paleolithic industries. Many such areas exist in the African Rift sensu lato, including the now-uplifted flanks of the rift, which would have been available for occupation at a relatively early stage in the Pleistocene, but little archaeological evidence of human activity was recorded in such areas until much later periods of human development and, in many cases, not until the expansion of pastoralist societies in the Holocene. A second issue concerns the variability in rates of landscape change resulting from different levels of tectonic activity and the long-term evolutionary implications of variable activity. The South African region as a whole clearly differs in its general rate of activity compared with the most active parts of the East African Rift. Even within South Africa, there appears to be variation between the sites and regions we have discussed, and there are additional sites that we have not included in our review where tectonically informed studies have yet to be carried out. Nevertheless, at a general level our results suggest that even quite modest rates of tectonic activity are likely to generate the sorts of topographic features we have described, and therefore are likely to be advantageous for human settlement, even in regions with few or no earthquakes, or relatively small ones within the lifetime of a human individual. A critical variable in this equation between rates of tectonic activity and creation of rough landscapes is the rate of erosion. In regions with relatively soft rocks and active forces of erosion, modest rates of tectonic activity may be insufficient to offset the smoothing effects of erosion, and the long-term trend will be toward a flat topography lacking the advantages for human activity that we have described. Conversely, areas with very hard rocks may preserve the rough and complex topographic features created by occasional tectonic disruption for longer periods despite generally low rates of tectonic activity. This is certainly a contributing factor to the topography of the areas we have described in South Africa, where the rock formations are metamorphosed and

Dynamic landscapes and human evolution mostly were formed earlier than the Cenozoic (Hartzer, 1998). These rocks are very hard and resistant to erosion, and they tend to maintain vertical cliffs and fissures rather than degrading to the rounded and flattened features that result from erosion of softer rock formations. There is, then, a continuum of topographic conditions. At one extreme, regions have so little tectonic activity, and rock formations liable to erosion, such that the resulting topography is likely to be generally smooth. Such plains environments offer little diversity of resources or complexity of topography offering tactical advantage, except at their margins or in localized areas where erosion has created some topographic relief. From this perspective, it is no surprise that extensive plains environments such as the Asian steppes or the Great Plains of North America show limited evidence of human occupation in prehistory until the adoption of the horse as a riding animal. Equally at the other extreme, very active tectonics may have consequences that are as much destructive as constructive, at least at a local or subregional scale, resulting in geographical displacement of favorable areas, and destruction of once-fertile basins. The implications of such variation for evolutionary trends, particularly with respect to the general pattern of hominin evolution, remain to be worked out, but it is worth noting that Reynolds (2007) related the higher rate of species turnover amongst large mammals in East Africa in comparison with South Africa to the higher rates of tectonic activity in the East African Rift. The selective impact of dynamic topography may take different forms, depending on the scale of the landscape changes involved. One possibility is that populations of large mammals are isolated by insurmountable topographic barriers, resulting in genetic divergence and speciation. Another possibility is that rough topography, by sustaining favorable environmental conditions during climatic downturns, helps to maintain higher levels of population than would otherwise have been the case, and hence maintains a larger pool of genetic variability that can provide the basis for later adaptation and evolutionary change. Finally, a tectonically active and complex topography may select for greater locomotory and cognitive adaptability able to cope with changeable environmental conditions, for example, bipedal movement suitable for moving through rough terrain and climbing rock barriers, and cognitive abilities able to take tactical advantage of complex topography in tracking mobile animal prey or avoiding predators. A critical factor in exploring further these possibilities is careful reconstruction of topographic conditions at a variety of scales and, in particular, better dating of rates and periodicities of tectonic movement, earthquake repeat times, and volcanic eruptions. Satellite imagery along with a new generation of cosmogenic dating techniques will play an important role in conjunction with existing methods of dating, mapping, and stratigraphic interpretation. Without a tectonically informed reconstruction of local landscape conditions as they existed during the periods in question, claims that very early archaeological sites or hominin fossils typically occur in regions lacking rough topography or tectonic

17

activity cannot be sustained. Issues of differential preservation cannot be discounted and almost certainly add an extra layer of variability that needs to be addressed alongside other factors. More studies are needed at a regional scale of the type that we have described here, involving a systematic program of dating of geological surfaces and deposits alongside systematic surveys for archaeological and fossil sites. ACKNOWLEDGMENTS This article is a contribution to the United Nations Educational, Scientific and Cultural Organization–funded International Geoscience Programme IGCP 567, “Earthquake Archaeology: Archaeoseismology along the Alpine-Himalayan Seismic Zone.” We acknowledge funding from NERC, UK (grant NE/ A516937/1) as part of its EFCHED (Environmental Factors in Human Evolution and Dispersal Programme). Funding for the South African work came from the France–South Africa !Khur project. It is IPGP (Institute de Physique du Globe de Paris) contribution number 3034. Satellite data © CNES 2007, distribution Spot Image S.A., was used for some figures. REFERENCES CITED Ambrose, S.H., 1998, Late Pleistocene human population bottlenecks, volcanic winter and differentiation of modern humans: Journal of Human Evolution, v. 34, p. 623–651, doi: 10.1006/jhev.1998.0219. Anton, S.C., Leonard, W.R., and Robertson, M.L., 2002, An ecomorphological model of the initial hominid dispersal from Africa: Journal of Human Evolution, v. 43, p. 773–785, doi: 10.1006/jhev.2002.0602. Ayele, A., Jacques, E., Kassim, M., Kidane, T., Omar, A., Tait, S., Nercessian, A., de Chabalier, J.-B., and King, G.C.P., 2007, The volcano-seismic crisis in Afar, Ethiopia, starting September 2005: Earth and Planetary Science Letters, v. 255, p. 177–187, doi: 10.1016/j.epsl.2006.12.014. Bailey, G.N., 1997, Klithi: Palaeolithic Settlement and Quaternary Landscapes in Northwest Greece: Cambridge, UK, McDonald Institute Monographs. Bailey, G.N., King, G.C.P., and Sturdy, D., 1993, Active tectonics and land-use strategies: A Paleolithic example from northwest Greece: Antiquity, v. 67, p. 292–312. Bailey, G.N., King, G.C.P., and Manighetti, I., 2000, Tectonics, volcanism, landscape structure and human evolution in the African Rift, in Bailey, G., Charles, R., and Winder, N., eds., Human Ecodynamics: Proceedings of the Association for Environmental Archaeology Conference 1998: Oxford, Oxbow, p. 31–46. Bailey, G.N., Reynolds, S., and King, G.C.P., 2010, Landscapes of human evolution: Models and methods of tectonic geomorphology and the reconstruction of hominin landscapes: Journal of Human Evolution, doi: 10.1016/j.jhevol.2010.01.004 (in press). Bamford, M., 1999, Pliocene fossil woods from early hominid cave deposit, Sterkfontain, South Africa: South African Journal of Science, v. 95, no. 5, p. 231–237. Berger, L.R., de Ruiter, D.J., Churchill, S.E., Schmid, P., Carlson, K.J., Dirks, P.H.G.M., and Kibii, J.M., 2010, Australopithecus sediba: A new species of Homo-like Australopith from South Africa: Science, v. 328, no. 5975, p. 195–204, doi: 10.1126/science.1184944. Bondevik, S., Svendsen, J.I., Johnsen, G., Mangerud, J., and Kaland, P.E., 1997, The Storegga tsunami along the Norwegian coast, its age and runup: Boreas, v. 26, p. 29–53. Brunet, M., Beauvilain, A., Coppens, Y., Heintz, E., Moutaye, A.H.E., and Pilbeam, D., 1995, The first australopithecine 2,500 kilometres west of the Rift Valley (Chad): Nature, v. 378, p. 273–275, doi: 10.1038/378273a0. Burke, K., 1996, The African plate: South African Journal of Geology, v. 99, no. 4, p. 341–409.

18

King and Bailey

Cadman, A., and Rayner, R.J., 1989, Climatic change and the appearance of Australopithecus africanus in the Makapansgat sediments: Journal of Human Evolution, v. 18, no. 2, p. 107–113, doi: 10.1016/0047-2484(89)90065-1. Dakaris, S.I., Higgs, E.S., and Hey, R.W., 1964, The climate, environment and industries of Stone Age Greece, Part I: Proceedings of the Prehistoric Society, v. 30, p. 199–244. Dart, R.A., 1925, Australopithecus africanus: The ape-man of South Africa: Nature, v. 115, p. 195–197, doi: 10.1038/115195a0. Deacon, H.J., 1979, Excavations at Boomplaas Cave: A sequence through the Upper Pleistocene and Holocene in South Africa: World Archaeology, v. 10, no. 3, p. 241–257. Deacon, H.J., 1995, Two late Pleistocene-Holocene archaeological depositories from the southern Cape, South Africa: South African Archaeological Bulletin, v. 50, no. 162, p. 121–131, doi: 10.2307/3889061. Deacon, H.J., Deacon, J., Brooker, M., and Wilson, M.L., 1978, The evidence for herding at Boomplaas Cave in the southern Cape, South Africa: South African Archaeological Bulletin, v. 33, p. 39–65, doi: 10.2307/3888251. de Heinzelin, J., Clark, J.D., White, T., Hart, W., Renne, P., WoldeGabriel, G., Beyene, Y., and Vrba, E., 1999, Environment and behavior of 2.5million-year-old Bouri hominids: Science, v. 284, p. 625–629, doi: 10.1126/ science.284.5414.625. Dirks, P.H.G.M., Kibii, J.M., Kuhn, B.F., Steininger, C., Churchill, S.E., Kramers, J.D., Pickering, R., Farber, D.L., Mériaux, A.-S., Herries, A.I.R., King, G.C.P., and Berger, L.R., 2010, Geological setting and age of Australopithecus sediba from southern Africa: Science, v. 328, no. 5975, p. 205–208, doi: 10.1126/science.1184950. Douglas, R.M., 2006, Formation of the Florisbad spring and fossil site—An alternative hypothesis: Journal of Archaeological Science, v. 33, p. 696– 706, doi: 10.1016/j.jas.2005.10.005. Gasse, F., 2001, Paleoclimate: Hydrological changes in Africa: Science, v. 292, no. 5525, p. 2259–2260, doi: 10.1126/science.1061940. Hartzer, F.J., 1998, A Stratigraphic Table of the SADC Countries: Praetoria, Council for Geoscience (South Africa), published on behalf of SADC by the Council for Geoscience, 3 sheets. Henshilwood, C.S., 2010, The ‘Upper Paleolithic’ of southern Africa: The Still Bay and Howiesons Poort, in Reynolds, S.C., and Gallagher, A., eds., African Genesis: Perspectives on Hominin Evolution: Johannesburg, Witwatersrand University Press (in press). Higgs, E.S., and Vita-Finzi, C., 1966, The climate, environment and industries of Stone Age Greece, Part II: Proceedings of the Prehistoric Society, v. 32, p. 1–29. Higgs, E.S., Vita-Finzi, C., Harris, D.R., and Fagg, E.E., 1967, The climate, environment and industries of Stone Age Greece. Part III: Proceedings of the Prehistoric Society, v. 33, p. 1–29. Johanson, D., Taieb, M., and Coppens, Y., 1982, Pliocene hominids from the Hadar Formation, Ethiopia (1973–1977): Stratigraphic, chronologic and palaeoenvironmental contexts, with notes on hominid morphology and systematics: American Journal of Physical Anthropology, v. 57, p. 373– 402, doi: 10.1002/ajpa.1330570402. Kalb, J., Oswald, E.B., Tebedge, S., Mebrate, A., Tola, E., and Peak, D., 1982a, Fossil mammals and artefacts from the Middle Awash Valley, Ethiopia: Nature, v. 298, p. 17–25, doi: 10.1038/298017a0. Kalb, J.E., Jolly, C.J., Mebrate, A., Tebedge, S., Smart, C., Oswald, E.B., Cramer, D., Whitehead, P., Wood, C.B., Conroy, C.G., Adefris, T., Sperling, L., and Kana, B., 1982b, Fossil mammals and artefacts from the Middle Awash Valley, Ethiopia: Nature, v. 298, p. 25–29, doi: 10.1038/298025a0. King, G.C.P., and Bailey, G.N., 1985, The palaeoenvironment of some archaeological sites in Greece: The influence of accumulated uplift in a seismically active region: Proceedings of the Prehistory Society, v. 51, p. 273–282. King, G.C.P., and Bailey, G.N., 2006, Tectonics and human evolution: Antiquity, v. 80, p. 265–286. King, G.C.P., and Vita-Finzi, C., 1981, Active folding in the Algerian earthquake of 10 October 1980: Nature, v. 292, no. 5818, p. 22–26, doi: 10 .1038/292022a0. King, G.C.P., and Yielding, G., 1984, The evolution of a thrust fault system: Processes of rupture initiation, propagation and termination in the 1980 El Asnam (Algeria) earthquake: Geophysical Journal of the Royal Astronomical Society, v. 73, p. 605–639. King, G.C.P., Tselentis, A., Gomberg, J., Molnar, P., Roecker, S.W., Sinvhal, H., Soufleris, C., and Stock, J.M., 1983, Microearthquake seismicity and active tectonics of northwestern Greece: Earth and Planetary Science Letters, v. 66, p. 279–288, doi: 10.1016/0012-821X(83)90141-3.

King, G.C.P., Sturdy, D., and Whitney, J., 1993, The landscape geometry and active tectonics of N.W. Greece: Geological Society of America Bulletin, v. 105, p. 137–161, doi: 10.1130/0016-7606(1993)1052.3.CO;2. King, G.C.P., Bailey, G., and Sturdy, D., 1994, Active tectonics and human survival strategies: Journal of Geophysical Research, v. 99, no. B10, p. 20,063–20,078, doi: 10.1029/94JB00280. King, G.C.P., Sturdy, D., and Bailey, G.N., 1997, The tectonic background to the Epirus landscape, in Bailey, G.N., ed., Klithi: Paleolithic Settlement and Quaternary Landscapes in Northwest Greece: Cambridge, UK, McDonald Institute Monographs, p. 541–558. Kuman, K., and Clarke, R.J., 2000, Stratigraphy, artefact industries and hominid associations for Sterkfontein, Member 5: Journal of Human Evolution, v. 38, no. 6, p. 827–847, doi: 10.1006/jhev.1999.0392. Kuman, K., Inbar, M., and Clarke, R.J., 1999, Palaeoenvironments and cultural sequence of the Florisbad Middle Stone Age hominid site, South Africa: Journal of Archaeological Science, v. 26, p. 1409–1425, doi: 10.1006/jasc .1999.0439. Manighetti, I., Tapponnier, P., Gillot, P.Y., Courtillot, V., Jacques, E., Ruegg, J.-C., and King, G.C.P., 1998, Propagation of rifting along the ArabiaSomalia plate boundary into Afar: Journal of Geophysical Research, v. 103, p. 4947–4974, doi: 10.1029/97JB02758. Manighetti, I., Tapponnier, P., Courtillot, V., Gallet, Y., Jacques, E., and Gillot, P.Y., 2001, Strain transfer between disconnected, propagating rifts in Afar: Journal of Geophysical Research, v. 106, p. 13,613–13,665, doi: 10.1029/2000JB900454. Maslin, M.A., and Christensen, B., 2007, Tectonics, orbital forcing, global climate change, and human evolution in Africa: Introduction to the African paleoclimate special volume: Journal of Human Evolution, v. 53, no. 5, p. 443–464, doi: 10.1016/j.jhevol.2007.06.005. McCarthy, T.S., and Hancox, P.J., 2000, Wetlands, in Partridge, T.C., and Maud, R.R., eds., The Cenozoic of Southern Africa: Oxford, Oxford University Press, p. 145–161. McCarthy, T.S., Tooth, S., Brandt, D., Marren, P., and Hancox, P.J., 2004, On the origin of the Nylsvlei wetland, in Geoscience Africa 2004 Conference Extended Abstracts: Johannesburg, University of Witwatersrand. McKee, J.K., 1993, The faunal age of the Taung hominid deposit: Journal of Human Evolution, v. 25, no. 5, p. 363–367, doi: 10.1006/jhev.1993.1055. McKee, J.K., 1999, The autocatalytic nature of hominid evolution in African Plio-Pleistocene environments, in Bromage, T.G., and Schrenk, F., eds., African Biogeography, Climate Change and Human Evolution: Oxford, Oxford University Press, p. 115–137. Meghraoui, M., and Doumaz, F., 1996, Earthquake-induced flooding and paleoseismicity of the El Asnam (Algeria) fault-related fold: Journal of Geophysical Research, v. 101, p. 17,617–17,644. Mitchell, P., 2002, The Archaeology of Southern Africa: Cambridge, UK, Cambridge University Press, 515 p. Partridge, T.C., 2000, Hominid-bearing cave and tufa deposits, in Partridge, T.C., and Maud, R.R., eds., The Cenozoic of Southern Africa: Oxford, Oxford University Press, p. 100–125. Philip, H., and Meghraoui, M., 1983, Structural analysis and interpretation of the surface deformation of the El Asnam earthquake of October 10, 1980: Tectonics, v. 2, no. 1, p. 17–49, doi: 10.1029/TC002i001p00017. Radosevich, S.C., Retallack, G.J., and Taieb, M., 1992, Reassessment of the paleoenvironment and preservation of hominid fossils from Hadar, Ethiopia: American Journal of Physical Anthropology, v. 87, p. 15–27, doi: 10.1002/ajpa.1330870103. Reed, K.E., 1997, Early hominid evolution and ecological change through the African Plio-Pleistocene: Journal of Human Evolution, v. 32, no. 2–3, p. 289–322, doi: 10.1006/jhev.1996.0106. Reynolds, S.C., 2007, Mammalian body size changes and Plio-Pleistocene environmental shifts: Implications for understanding hominin evolution in eastern and southern Africa: Journal of Human Evolution, v. 53, no. 5, p. 528–548. Rowland, J.V., Baker, E., Ebinger, C.J., Keir, D., Kidane, T., Biggs, J., Hayward, N., and Wright, T.J., 2007, Fault growth at a nascent slow-spreading ridge: 2005 Dabbahu rifting episode, Afar: Geophysical Journal International, v. 171, no. 3, p. 1–21, doi: 10.1111/j.1365-246X.2007.03584.x. Saunders, I., Brandt, M., Steyn, J., Roblin, D., and Kijko, A., 2008, The South African National Seismograph Network: Seismological Research Letters, v. 79, no. 2, p. 203–209, doi: 10.1785/gssrl.79.2.203. Semaw, S., Rogers, M.J., Qaude, J., Renne, P.R., Butler, R.F., DominguezRodrigo, M., Stout, D., Hart, W.S., Pickering, T., and Simpson, S.W.,

Dynamic landscapes and human evolution 2003, 2.6-million-year-old stone tools and associated bones from OGS-6 and OGS-7, Gona, Afar, Ethiopia: Journal of Human Evolution, v. 45, no. 2, p. 169–177, doi: 10.1016/S0047-2484(03)00093-9. Semaw, S., Simpson, S.W., Quade, J., Renne, P.R., Butler, R.F., McIntosh, W.C., Levin, N., Dominguez-Rodrigo, M., and Rogers, M.J., 2005, Early Pliocene hominids from Gona, Ethiopia: Nature, v. 433, no. 7023, p. 301– 305, doi: 10.1038/nature03177. Sponheimer, M., and Lee-Thorp, J.A., 1999, Isotopic evidence for the diet of an early hominid, Australopithecus africanus: Science, v. 283, no. 5400, p. 368, doi: 10.1126/science.283.5400.368. Stein, R., and King, G.C.P., 1984, Seismic potential revealed by surface folding: The 1983 Coalinga, California, Earthquake: Science, v. 224, p. 869– 872, doi: 10.1126/science.224.4651.869. Tobias, P.V., 2000, The fossil hominids, in Partridge, T.C., and Maud, R.R., eds., The Cenozoic of Southern Africa: Oxford, Oxford University Press, p. 252–276. Vita-Finzi, C., 1969, The Mediterranean Valleys: Cambridge, UK, Cambridge University Press, 140 p. Vrba, E.S., 1982, Biostratigraphy and chronology, based particularly on Bovidae, of southern hominid associated assemblages, in de Lumley, H., and

19

de Lumley, M.-A., eds., Makapansgat, Sterkfontein, Taung, Kromdraai, Swartkrans; also Elandsfontein (Saldanha), Broken Hill (now Kabwe) and Cave of Hearths: Premier Congrès International de Paléontologie Humaine, p. 707–752. Wagner, P.A., 1927, The geology of the northeastern part of the Springbok Flats and surrounding country: An explanation of Sheet 17 (Springbok Flats): Geological Survey of South Africa. White, T.D., Asfaw, B., DeGusta, D., Gilbert, H., Richards, G.D., Suwa, G., and Howell, F.C., 2003, Pleistocene Homo sapiens from Middle Awash, Ethiopia: Nature, v. 423, no. 6941, p. 742–747. WoldeGabriel, G., White, T.D., Suwa, G., Renne, P.R., de Heinzelin, J., Hart, W.K., and Heiken, G., 1994, Ecological and temporal placement of early Pliocene hominids at Aramis, Ethiopia: Nature, v. 371, p. 330–333, doi: 10.1038/371330a0. WoldeGabriel, G., Haile-Selassie, Y., Renne, P.R., Hart, W.K., Ambrose, S.H., Asfaw, B., Heiken, G., and White, T., 2001, Geology and palaeontology of the late Miocene Middle Awash Valley, Afar rift, Ethiopia: Nature, v. 412, p. 175–178, doi: 10.1038/35084058. MANUSCRIPT ACCEPTED BY THE SOCIETY 19 MAY 2010

Printed in the USA

The Geological Society of America Special Paper 471 2010

Tectonic environments of ancient civilizations: Opportunities for archaeoseismological and anthropological studies Eric R. Force* Department of Geosciences, University of Arizona, Tucson, Arizona 85721, USA Bruce G. McFadgen* School of Maori Studies, Victoria University of Wellington, Wellington 6140, New Zealand

ABSTRACT The close spatial relation between ancient civilizations and active tectonic boundaries is robust in the Eastern Hemisphere but counterintuitive given the seismic disadvantages it implies. Explanations for the observation remain debatable, and no single explanation seems sufficient. Some possibly important factors are unrelated to seismicity, e.g., the influence of tectonism on local water resources and on resource diversity. When examined on finer spatial scales, the relation is still robust. A quantifiable influence of tectonism on civilization locations even along Mediterranean shores is suggested by their distribution. The stronger links of tectonism with derivative civilizations suggest a role of ancient trade connections. Several clues point to cultural response as an important ingredient in the dynamics resulting in the spatial relation. These are: correlation between static character and location of civilizations relative to tectonic locus; archaeologic and historic records of accelerated cultural (especially religious) change following tectonic events; and evidence that the spatial relation evolves through time via trade goods and routes. Archaeoseismology is in a key position to provide additional clues to this paradoxical relation in at least three ways: (1) providing detail on evolving societal response; (2) determining the most pertinent tectonic styles; and (3) determining the role of seismicity in Neolithic cultures that eventually became civilizations.

INTRODUCTION

ture in this direction presently supports (Bailey et al., 1993; King et al., 1994; Trifonov and Karakhanian, 2004; Force, 2008). The tectonic boundaries that belong in this pattern are mostly convergent to transcurrent ones, associated with the southern margin of the Eurasian plate. The civilizations are those conventionally regarded as the greatest of antiquity (they could be defined as those that score eight or higher in the ten original criteria of Childe, 1950). It is worth recalling that conventional criteria for greatness in a civilization have valued cities, monumental (preservable) architecture, and (preserved and deciphered)

The papers in this volume offer abundant evidence of the destructive seismic environment of many ancient sites. Yet, the highest civilizations to which these sites belong are clustered along the very seismogenic tectonic boundaries that give rise to the documented destruction (Fig. 1). This counterintuitive pattern requires further attention, much more than a slender litera*E-mails: [email protected]; [email protected].

Force, E.R., and McFadgen, B.G., 2010, Tectonic environments of ancient civilizations: Opportunities for archaeoseismological and anthropological studies, in Sintubin, M., Stewart, I.S., Niemi, T.M., and Altunel, E., eds., Ancient Earthquakes: Geological Society of America Special Paper 471, p. 21–28, doi: 10.1130/2010.2471(02). For permission to copy, contact [email protected]. © 2010 The Geological Society of America. All rights reserved.

21

22

Force and McFadgen

writing—and anthropologists these days are equally interested in other aspects of cultural complexity. However, it is this particular assemblage of traits that corresponds so closely with tectonic boundaries, regardless of the semantic label that is attached. Quantification of the relation was attempted for the Eastern Hemisphere by Force (2008) based on probabilistic comparison to random distribution. Measurements were taken from originating sites (to avoid the problem of imperial sprawl) of 13 civilizations to the nearest tectonic plate boundary as conventionally mapped, and these resulted in an average distance of 75 km, with two prominent exceptions (Egypt and China; the influence of tectonism on development of the latter is appreciable but not treated here). This distance can be converted to a polygon averaging 150 km wide (civilizations could be on either side of a tectonic boundary) along the total on-land length of the boundaries, and the probability was calculated for 11 of 13 civilizations finding themselves in this tectonically defined polygon, assuming random distribution. These probabilities were calculated for two different assumptions of available land areas, and both the included civilizations and tectonic boundary locations (especially where plate boundaries are partitioned) were varied to provide a sensitivity analysis. The calculations were necessarily approximate, but the calculated probabilities are so miniscule that random distribution can be rejected. The conclusion that ancient civilizations seem to be preferentially located near active tectonic boundaries seems robust. The observed pattern is a simple one, but many mechanisms and dynamics seem possible. All of them are inherently untestable in real time, and this limits the rigor with which linkages can be proven. However, we can constrain the factors that are most consistent with our information. It is possible that the importance

of tectonism acts via some other variable that is more obviously required for civilization to become established and thrive, such as climate, soils, water, and transport potential. Volcanism appears not to be of systematic importance, since few of the ancient civilizations are near Holocene volcanic centers (except in the Western Hemisphere, an intriguing relation outside the purview of this volume). Quaternary volcanism, however, did provide important soil and building material assets in Italy. The civilizations differ so much from each other in their apparent environmental characters that it is difficult to say which of them might be related to tectonism (Table 1). Climatecontrolled vegetation varies greatly and may have changed since antiquity, but latitudes ranging from ~27° to 42° have not (the difference between Tampa and Boston). Relation to type of water resource varies similarly. The relation to slightly varying styles of active tectonism is more commonly shared. So far, it is not possible to say for certain whether seismic activity per se is important in the relation, or whether some other aspect of position near an active tectonic boundary is involved. For example, the relation shown in Figure 1 is clearly more closely related to plate boundaries than to seismic risk, especially in Iran, Tibet, western China, etc. One positive effect that is independent of seismicity is local water resources, which can be enhanced by active tectonism. If fractures in fault gouge are abundant and randomly oriented, one set will be held open by active stress, permitting greater permeability of active faults than old inactive ones (Hickman et al., in Force, 2008). Along tectonic boundaries, the prevalence of malarial paleopathology in those Neolithic villages that were involved in transitions to civilization (locations from Maisels, 1999) is consistent with this being an important factor.

TABLE 1. VARIATIONS AMONG RIVERBORNE WATER SUPPLY, CLIMATIC VARIABLES, AND TECTONIC MICRO-ENVIRONMENT OF ORIGINAL SITES OF SOME ANCIENT CIVILIZATIONS Civilization Rivers Climate Tectonic microLatitude Vegetation environment Roman

M

41°50′N

1

Etruscan

S

42°10′N

1

E+S

Gree k

S

1

E > S + Tr

1

E > S + Tr

E+S

Mycenaean

S

38°N 37°50′N

Minoan

S

34°40′N

1

E+S

S&M

31°50′–33°20′N

1, 2

Tr > E

SW Asian Assyrian

L

36°30′N

2

Th + Tr

Mesopotamian

L

2

Th + Tr

Persian

S

31°N 30°–32°20′N

2, 3, 4

Th + Tr

Indus

L

27°20′N

5, 7

Th + Tr

Aryan India

L

29°30′N

5, 6

Th

Egyptian

L

29°50′N

7, 8

N.A.

Chinese

L

34°40′N

9

N.A.

Notes: Rivers: S—small, M—mid-size, and L—large. Vegetation: 1—Mediterranean scrub; 2—short-grass steppe; 3— conifer forest; 4—mountain vegetation; 5—dry tropical forest; 6—dry tropical scrub; 7—desert; 8—floodplain; 9—mixed and broadleaf forest. Tectonic environment: E—extensional; S—subduction-related; Tr—transcurrent; Th—thrust. N.A.— not applicable.

Tectonic environments of ancient civilizations

23

N Eurasian plate

2 1

H-T plate 3

4

13

7 5 6

9 8

12

11 10

Arabian plate

African plate

I n d o - Australian plate 0

1000

2000

km

H-T = Hellenic-Turkish plate

Figure 1. Locations of originating sites of 13 prominent ancient civilizations relative to various aspects of the southern boundary of the Eurasian plate (after Force, 2008). Civilizations (and sites) shown are 1—Roman (Rome), 2—Etruscan (Tarquinii-Veii), 3—Greek (Corinth) and Mycenaean (Mycenae), 4—Minoan (Knossos-Phaestos), 5 and 6—West Asian (Tyre and Jerusalem), 7—Assyrian (Ninevah), 8—Mesopotamian (Ur-Uruk), 9—Persian (Susa-Pasargadae), 10—Indus (Mohenjodaro), 11—Aryan India (Hastinapura), 12—Egyptian (Memphis), and 13—Chinese (Zhengzhou).

Another proposed factor independent of seismicity is the juxtaposition of different soils and topographies along active faults, especially transcurrent faults (cf. Bailey et al., 1993), providing a greater diversity of resources in the cultural equivalent of a biological edge effect. Trifonov and Karakhanian (2004) suggested the presence of anomalous geochemical fields along active faults, a possibility that seemed remote until de Boer et al. (2001) documented an example at Delphi. ADDITIONAL QUANTIFIABLE LINKS The relation between ancient civilizations and tectonic boundaries seems apparent not only at hemispherical scale but also at finer scales, for example, along the shores of the Medi-

terranean. These shores were certainly important to emerging civilizations there, and some would say this factor predominates. However, if the distribution of the African-Eurasian plate boundary is compared to ancient sites of civilization along opposing shores, a probable influence of tectonism shows through. The tectonic boundary broadly follows the south shore in the western Mediterranean but the northern and north-insular shores in the eastern part (Fig. 1). Ancient sites of civilizations (using a looser definition to allow more cases) follow the same path (Carthage, Syracuse, Rome, Tarquinii-Veii, Corinth, Mycenae, and Knossos-Phaistos). This relation itself looks persuasive, but an additional quantitative test would complement it. We can compare distances of tectonic boundaries versus seashores (of the time, where this is known) for the 11 originating sites of Figure 1.

24

Force and McFadgen

The average seashore distance is ~1.48 times greater than the tectonic-boundary distance, implying a probability of random distribution that is very small but ~75 times greater than that for tectonic boundaries. Even along Mediterranean and Near Eastern shores, civilizations apparently located their originating sites near active tectonic boundaries. Does the relation result in finer-scale vignettes? Perhaps a tectonic influence suggests itself in the Hellenic realm (Fig. 2). Mycenaean and later Greek civilization nucleated in a zone of distributed deformation along the on-land prolongation of the North Anatolian fault and its offsets along the Gulf of Corinth and other extensional structures. Where the tectonic zone passes northward into the Aegean, the spread of each civilization stalled, at Iolkos for the Mycenaeans and on Euboea (Evia) for the Greeks, though earlier, simpler Neolithic villages continued across this boundary into more quiescent parts of Thessaly. Force (2008) attempted to constrain the number and structure of possible solutions by comparing subsets of both the civilizations and their pertinent geologic environments. He showed that the closest relation is between tectonic boundaries and those ancient civilizations generally called derivative, i.e., those that evolved under some influence from more senior civilizations. This relation of civilization subsets suggests an influence of trade (however accomplished).

The propagation of trade routes, from more advanced settlements in the Near East to sites that eventually became great ancient civilizations, mimics tectonic boundaries more closely than one would expect given the availability of other routes. Figure 3 shows such propagation from Bronze Age through early Iron Age times in the Eastern Mediterranean area. Perhaps most impressive is the replication of the “draped” shape of the plate boundary between Cyprus and Crete by trade routes, based on the distribution of Bronze age stone anchors. Other trade routes existed also, of course, extending, for example, to Maikop, Danubian, Hallstatt, Scythian, etc. (cultures in seismically more quiescent north-central Eurasia), but these routes did not produce civilizations (as defined here) until long after the period of antiquity, if ever. If one measures the length of trade routes propagating toward eventual ancient civilizations as shown in Figure 3, ~79% of this length is within 100 km of an active plate boundary. (Routes destined for other sites of production or resource exploitation such as Kanesh do not count in this calculation.) A probability analysis was not attempted, because the appropriate structure for it is unclear, but perhaps it is unnecessary because its conclusion is intuitively obvious. Trade routes to eventual civilizations tended for some reason to follow tectonic boundaries, perhaps because each incremental stage in trade-route propagation was to

N 100 km

X X X

Aegean Sea Plate boundaries including margins of distributed deformation Mycenaean sites damaged 1250 or 1200 B.C. X

Early Geometric sites (ca. 900 B.C.)

Figure 2. Sites of Mycenaean palaces destroyed by earthquakes in 1200 and 1250 B.C. (from Kilian [1996] and other sources) relative to approximate boundaries of distributed deformation along the projection of the North Anatolian fault in the Hellenic realm (from McClusky et al., 2000). The distribution of destroyed Mycenaean palaces, with only a few exceptions, includes all the palaces of that civilization, suggesting not only ancient tectonic activity along this structural trend, but also the localization of palaces along it. The major early Geometric Age sites shown (from Coldstream, 2003) follow the same trend ~300 yr later.

Tectonic environments of ancient civilizations

25

NORTH

BLACK SEA EURASIAN PLATE 42 º

TO MAGNA GRAECIA

40 º

TH ANATOLIAN NOR FAU

KA

TROY

38 º

HELLENIC

AND

NE

LT

SH

TURKISH PLATES

AS

36 º

34 º

SU

R

ARABIAN PLATE

20 º E

U

AFRICAN PLATE 30 º

25 º

40 º

TO INDUS

32 º N

R

MEDITERRANEAN SEA

45 º

ROUTES OF “OLD - ASSYRIAN MERCHANT” (2200 - 1900 B.C.) SUBSEQUENT ADDITIONAL ROUTES OF GEOMETRIC GREECE (900 - 700 B.C.) 0

200

400

600

800

1000 km

Figure 3. Evolution of trade routes in the Near East and Eastern Mediterranean from Bronze Age through early Iron Age times, superposed on plate boundaries. Trade in tin and copper from 2200 to 1900 B.C. is from Kuhrt (1998) and that for Geometric Greece (900–700 B.C.) is from Coldstream (2003), using distribution of late Bronze Age stone anchors in the marine realm from McCaslin (1980, route 2).

a settlement where an expectation of change produced more receptivity to “civilized” goods. This trade-route clue and two others suggest the importance of long-term cultural response in the observed spatial relation. One of these is a relation between cultural character of civilizations and their distance to active tectonic boundaries. Those farthest from these boundaries tended to endure longer times with essentially the same character (Force, 2008, Fig. 2 therein), i.e., they were more static (and perhaps used their building material resources more slowly). A third clue suggesting the part played by cultural response is the link with cultural change, especially religious change, corresponding to tectonic events. Earthquakes in Greece and Cyprus in the fourth century A.D. correspond to changes in predominant religion in Corinth and in Kourion, respectively (Rothaus, 1996; Soren and James, 1988). An earthquake in Sparta ca. 464 B.C. provided the opportunity for a revolution (noted by Thucydides; see de Boer and Sanders, 2005). Earthquakes in Gortyn on Crete separate its Roman-era history into phases (DiVita, 1996). An earlier earthquake in southern Crete separates the Bronze Age archaeology of Kommos into

different modes pre– and post–1700 B.C. (Shaw, 2006). Many other examples (reviewed by de Boer and Sanders, 2005; Nur, 2008) provide abundant evidence of societal change catalyzed by earthquakes. Thus, we have some clues about the structure of a relation between ancient civilizations and active tectonism, but many questions remain, and explanations remain unclear. It therefore seems appropriate to turn to the potential contribution that archaeoseismology can make in explicating the true nature of the relation. This discipline (in cooperation with other archaeologists) is in a key position to provide clues that can be supplied no other way. There are at least three possible avenues of investigation that seem promising; they will now be discussed. ARCHAEOSEISMOLOGY IN DELINEATING AN EVOLUTION OF SOCIETAL RESPONSES Archaeoseismologists have concentrated until now on providing evidence of seismic destruction at ancient sites, and have been very successful at many of them despite many doubts

26

Force and McFadgen

among earlier archaeologists (reviewed by Nur, 2008). Progress in this direction will undoubtedly continue, and it could be harnessed in new ways. First of these is to determine the progression of societal response following destructive events. The work of many archaeologists has shown that habitation layers below and above seismic destruction horizons are different. A variety of causes have been attributed. An especially intriguing example to archaeoseismologists is the increasing sophistication of antiseismic devices through antiquity (cf. Stiros, 1996). We have seen examples of cultural discontinuities that apparently correspond with horizons giving evidence of tectonic activity. Perhaps most remarkable in this regard is the work of the late Klaus Kilian at Mycenaean Tiryns; he showed that archaeologic evidence of earthquakes corresponds temporally in at least three horizons with the emergence of newly dominant pottery styles. He concluded “earthquakes marked the beginning of a new phase and were related to, or even responsible for, changes in the organization and planning of the site” (Kilian, 1996, p. 67). Discontinuities of this sort can be modeled in a number of ways. One end member is represented by a combination of archaeologic and historic evidence about the 1855 earthquake of magnitude 8.2 in the Wellington area of New Zealand, which emptied nearly all kitchen cupboards and smashed the ceramics they contained (Grapes, 2000). The destroyed ceramic population was a mixture of modes that went back several decades, whereas their replacements would have been chosen from what was in vogue at that time (McFadgen and Clough, 2009). The same principle could apply to the architecture of the containing structures and even town layout, as was better demonstrated by the 1848 Marlborough earthquake (Grapes, 2000) and 1931 Napier earthquake of New Zealand. This type of case amplifies change that has already taken place, but in itself would generally not stimulate evolutionary change. To some extent, change is only stylistic. At another extreme, there is cultural change that involves innovation, values, and/or cultural evolution. For example, the 1755 “Lisbon” earthquake set in motion important philosophical changes throughout the Age of Enlightenment world (de Boer and Sanders, 2005). We have seen examples that involve religious change. This cultural-discontinuity model too has historical support. Rozario (2007) reviews evidence of accelerated cultural change initiated by disasters, including its treatment in the economic and psychological literatures. A third end member can be change in habitation patterns as a result of changes in the physical environment of the site, from sea level to landslides, that result from earthquakes (cf. McFadgen, 2007). For all three models, tectonism disrupts societal inertia and thus forces the pace of change. These three end-member models apply in some respects to cultures independent of their complexity. Indeed, it might be useful to better understand the cultural response to tectonism of many different types of cultures, not just the very complex ancient ones addressed here. There is some interest in the anthropological community for this question (Eiselt, 2009).

If a net long-term effect of tectonic activity in antiquity was an evolution toward civilization, we would like to know how the observed shorter-term changes might link up to contribute to this evolution. This will require the attention of archaeoseismologists in conjunction with archaeologist colleagues to determine the responses that are immediate, in closest association with earthquake damage, and those evolutionary changes that occur in response. It would be quite instructive to know at Tiryns, for example, whether characteristic types of pottery or any other cultural material closely followed seismic events but preceded the establishment of longer-lived styles. It would also be interesting to know whether any of these changes themselves form a pattern across a succession of seismic events. ARCHAEOSEISMOLOGY IN ESTABLISHING TYPICAL SEISMIC STYLES OF ANCIENT CIVILIZATIONS A tendency of ancient civilizations of the old world to be located along plate boundaries of transcurrent to convergent type seems clear (Fig. 1). For a number of these civilizations, however, the most pertinent tectonic micro-environment involves extensional faulting in association with such boundaries, as, for example, extension above subduction zones (Table 1). This implies that particular tectonic styles, including seismic styles, are associated with ancient civilizations, and because societal response of some sort must be involved, one would suspect that the styles in question must have characteristic recurrence intervals and/or intensities of events. Recurrence intervals less than human lifetimes (even those of antiquity) seem typical of the pertinent tectonic structures, and the close proximity of originating sites to those structures (Force, 2008) shows that these recurrence intervals did indeed affect civilizing societies. In Assyrian Ninevah, for example, typical recurrence intervals were of sufficient importance and regularity that scribes recorded them as being 21 yr, a figure similar to that derived by Kilian (1996) for Tiryns. Much more could be known about the relations among recurrence, intensity, and the development of civilized society. For example, are there optima in recurrence or intensity that accelerate this development relative to locales with too much activity (leading in extreme cases to abandonment) or too little? Archaeoseismology can potentially establish the chronologies and intensity records upon which the answers must be based. ARCHAEOSEISMOLOGY OF NEOLITHIC SITES In the area of this study, almost all the great ancient civilizations grew from indigenous roots (Aryan India being the one exception), though all those called derivative did so with inputs from adjacent established civilizations. These inputs could have spread to indigenous societies in many directions, but they seem to have preferentially spread instead along tectonic boundaries. It is therefore the influence of tectonism on indigenous prehistoric

Tectonic environments of ancient civilizations societies that needs attention, and the links that lead to civilization must be sought among them. In the region of this volume, this means Neolithic cultures, and some later cultures that retained Neolithic lifestyles as metal trade goods began to arrive (reviewed by Renfrew, 1972; Redford, 1992; Whittle, 1996, for different regions). The case for tectonic influence for some Neolithic (and “Chalcolithic”) societies is stronger than that for the civilizations that evolved from them because those precursors were closer to the tectonic boundaries. This is especially the case for the complex pre–Bronze Age cultures of Mesopotamia and the IndusSaraswati area, as located by Maisels (1999). The first indications of advanced culture arose in the foothills east of the Tigris (at sites like Choga Mami; Oates and Oates, 1976) and west of the Indus (Mehrgarh and Nausharo; Kenoyer, 2000) near the thrust components of partitioned plate boundaries. Some of these in Mesopotamia and the Indus are the Neolithic sites, from which come the previously mentioned evidence of abundantly productive springs associated with active faults. Only when the need for increased irrigation scale arose did the nascent civilizations move away from those boundaries. For the formative stages of these cultures, we of course have no written records; archaeology alone can supply reliable information, and archaeoseismology alone can supply information on the influence of tectonism on these societies. This can then be contrasted with the development of coeval societies in quiescent settings. Since nominally Neolithic societies have survived into historic times in many parts of the world, a complementary approach would be to study effects of tectonism on their lifestyles and history. The processes that influenced cultural change in these societies might contribute to understanding the processes of cultural change that led to civilization elsewhere. McFadgen (2007), for example, showed ties between tectonism and cultural change based on precontact archaeology for the New Zealand Maori. CONCLUSIONS The dynamics of long-term change in cultures subject to active tectonism are ambiguous, but, in contrast, the eventual result seems fairly clear in the area treated in this volume— those cultures that, in addition to having a range of environmental advantages, were subjected to particular varieties of tectonic activity tended to become exceedingly complex (“great ancient civilizations” and some particularly precocious Neolithic precursors). Though the observed spatial relation of ancient civilizations with active tectonism associated with the southern boundary of the Eurasian plate is most apparent at hemispheric scale, it seems valid in some areas at finer scales also. A previously documented closer relation to tectonic boundaries of derivative compared to primary civilizations suggests a relation to trade, and comparison of tectonic boundaries with trade-route propagation

27

toward emerging sites of civilization shows some startling resemblances. Two other lines of evidence also suggest that long-term cultural evolution may be directly related to tectonic activity— correlation between static character and location of civilizations relative to tectonic locus, and archaeologic and historic records of accelerated cultural (especially religious) change following tectonic events. If this is the case, anthropological comparisons of “tectonic” and “nontectonic” societies of different sorts seem promising. The existence of any spatial relation between ancient civilizations and active tectonism raises many questions, and most of them remain unresolved. Archaeoseismology is in a unique position to provide important clues in at least three areas: 1. Previous work with seismic destruction horizons has shown corresponding discontinuities in ceramic and/or architectural remains; these tectonic events have forced the pace of change but may or may not reflect cultural discontinuities. Input from the archaeoseismologist is required to differentiate first responses from subsequent ones in order to establish an evolution of changes, and to see if these link up from one event to the next to delineate any kind of long-term influence. 2. The style of societal response may vary with the typical recurrence interval between destructive events and with their intensities. Such variations are implicit in the relation of ancient civilizations to specific tectonic environments—especially extensional faulting within convergent to transcurrent boundaries— and in the typical distances of originating sites to seismogenic structures. The archaeoseismologist is indispensible in compiling the chronologic and intensity records required to establish such links. 3. The region treated in this volume is blessed with ancient historical records in addition to archaeoseismologic evidence. However, the relation probably precedes the historic period; the relation behaves as if certain Neolithic cultures were somehow favored by tectonic factors to become civilizations with histories. The archaeoseismologist is a necessary source of relevant information on the ways in which tectonism influenced Neolithic cultures. ACKNOWLEDGMENTS Force thanks all those who responded to his former website. As presented here, he included thoughts contributed by John Dohrenwend, Daniel Kent, Barbara Mills, Arda Ozacar, Evelyn Roeloffs, David Soren, and Jane Force. McFadgen thanks members of the New Zealand Archaeological Association at the 2009 annual conference for their responses to some of the ideas presented here, and Anne French for useful discussion. We appreciate reviews by Claudio Vita-Finzi and Victoria Hopgood (Buck). Logistic support was provided by Jim Bliss, Jennifer Dodge, and John Birmingham. This article is a contribution to the International Geoscience Programme IGCP 567, “Earthquake Archaeology: Archaeoseismology along the Alpine-Himalayan Seismic Zone.”

28

Force and McFadgen

REFERENCES CITED Bailey, G., King, G., and Sturdy, D., 1993, Active tectonics and land-use strategies: A Paleolithic example from northwest Greece: Antiquity, v. 67, p. 292–303. Childe, V.G., 1950, The urban revolution: The Town Planning Review, v. 21, p. 3–17. Coldstream, J.N., 2003, Geometric Greece (2nd ed.): London, Routledge, 453 p. de Boer, J.Z., and Sanders, D.T., 2005, Earthquakes in Human History: Princeton, New Jersey, Princeton University Press, 278 p. de Boer, J.Z., Hale, J.R., and Chanton, J., 2001, New evidence for the geological origins of the ancient Delphic oracle: Geology, v. 29, p. 707–710, doi: 10.1130/0091-7613(2001)0292.0.CO;2. DiVita, A., 1996, Earthquakes and civil life at Gortyn (Crete), in Stiros, S., and Jones, R.E., eds., Archaeoseismology: Fitch Laboratory Occasional Paper 7 (British School at Athens), p. 45–50. Eiselt, B.S., 2009, Americanist archaeologies: 2008 in review: American Anthropologist, v. 111, p. 137–145, doi: 10.1111/j.1548-1433.2009.01106.x. Force, E.R., 2008, Tectonic environments of ancient civilizations in the Eastern Hemisphere: Geoarchaeology, v. 23, p. 644–653, doi: 10.1002/gea.20235. Grapes, R., 2000, Magnitude Eight Plus—New Zealand’s Biggest Earthquake: Wellington, Victoria University Press, 208 p. Kenoyer, J.M., 2000, Early developments of art, symbol, and technology in the Indus Valley tradition: Indian Archaeological Studies, v. 22, p. 1–18. Kilian, K., 1996, Earthquakes and archaeological context at 13th century BC Tiryns, in Stiros, S., and Jones, R.E., eds., Archaeoseismology: Fitch Laboratory Occasional Paper 7 (British School at Athens), p. 63–68. King, G., Bailey, G., and Sturdy, D., 1994, Active tectonics and human survival strategies: Journal of Geophysical Research–Solid Earth, v. 99, no. B10, p. 20,063–20,078. Kuhrt, A., 1998, The Old Assyrian merchants, in Parkins, H., and Smith, C., eds., Trade, Traders, and the Ancient City: London, Routledge, p. 16–30. Maisels, C.K., 1999, Early Civilizations of the Old World: London, Routledge, 504 p. McCaslin, D.E., 1980, Stone anchors in antiquity: Coastal settlements and maritime trade-routes in the Eastern Mediterranean ca. 1600–1050 BC: Studies in Mediterranean Archaeology, Volume 61: Goteborg, Astroms, 145 p. McClusky, S., Balassanian, S., Barka, A., Demir, C., Ergintav, S., Georgiev, I., Gurkan, O., Hamburger, M., Hurst, K., Kahle, H., Kastens, K., Kekelidze, G., King, R., Kotzev, V., Lenk, O., Mahmoud, S., Mishin, A., Nadariya,

M., Ouzounis, A., Paradissis, D., Peter, Y., Prilepin, M., Reilinger, R., Sanli, I., Seeger, H., Tealeb, A., Toksöz, M.N., and Veis, G., 2000, Global Positioning System constraints on plate kinematics and dynamics in the Eastern Mediterranean and Caucasus: Journal of Geophysical Research, v. 105, no. B3, p. 5695–5719. McFadgen, B., 2007, Hostile Shores—Catastrophic Events in Prehistoric New Zealand and Their Impact on Maori Coastal Communities: Auckland, Auckland University Press, 298 p. McFadgen, B., and Clough, R., 2009, Foreshock-Aftershock: The Archaeology of Chews Lane, Wellington: Auckland, Clough and Associates Report 6, 106 p. Nur, A., with Burgess, D., 2008, Apocalypse: Earthquakes, Archaeology, and the Wrath of God: Princeton, New Jersey, Princeton University Press, 309 p. Oates, J., and Oates, D., 1976, Early irrigation agriculture in Mesopotamia, in Sieveking, G.G., Longworth, I.H., and Wilson, K.E., eds., Problems in Economic and Social Archaeology: London, Duckworth, p. 109–134. Redford, D., 1992, Egypt, Canaan, and Israel in Ancient Times: Princeton, New Jersey, Princeton University Press, 512 p. Renfrew, C., 1972, The Emergence of Civilization; The Cyclades and the Aegean in the Third Millennium BC: London, Methuen, 595 p. Rothaus, R.M., 1996, Earthquakes and temples in Late Antique Corinth, in Stiros, S., and Jones, R.E., eds., Archaeoseismology: Fitch Laboratory Occasional Paper 7 (British School at Athens), p. 105–112. Rozario, K., 2007, The Culture of Calamity: Disaster and the Making of Modern America: Chicago, University of Chicago Press, 313 p. Shaw, J.W., 2006, Kommos: A Minoan Harbor Town and Greek Sanctuary in Southern Crete: Athens, American School of Classical Studies, 171 p. Soren, D., and James, J., 1988, Kourion—The Search for a Lost Roman City: New York, Anchor, 223 p. Stiros, S.C., 1996, Identification of earthquakes from archaeological data: Methodology, criteria, and limitations, in Stiros, S., and Jones, R.E., eds., Archaeoseismology: Fitch Laboratory Occasional Paper 7 (British School at Athens), p. 129–152. Trifonov, V.G., and Karakhanian, A.S., 2004, Active faulting and human environment: Tectonophysics, v. 380, p. 287–294, doi: 10.1016/j.tecto.2003 .09.025. Whittle, A., 1996, Europe in the Neolithic: Cambridge, UK, Cambridge University Press, 460 p. MANUSCRIPT ACCEPTED BY THE SOCIETY 19 MAY 2010

Printed in the USA

The Geological Society of America Special Paper 471 2010

The door knockers of Mansurah: Strong shaking in a region of low perceived seismic risk, Sindh, Pakistan Roger Bilham Cooperative Institute for Research in Environmental Science and Department of Geological Sciences, University of Colorado, Boulder, Colorado 80309-0399, USA Sarosh Lodi Department of Civil Engineering, NED University of Engineering and Technology, Karachi 75270, Pakistan

ABSTRACT Mansurah, the eighth-century Arabic capital of Sindh province, Pakistan, flourished for a mere 200 yr. Its destruction by an earthquake ca. 980 A.D. was first proposed by archaeologists who reported the discovery of crushed skeletons amid dateable coins found among its rubble. An abrupt natural death to the city was challenged by others who noted that the absence of wood or valuables was consistent with the city being sacked and systematically looted. The recent discovery of four decorated door knockers beneath the collapsed walls of one of the largest structures in Mansurah, however, reopens the case for an earthquake, since an invading army would almost certainly have removed them as booty. We suggest that an earthquake not only destroyed the city and its suburbs (intensity ≈ VIII), but resulted in postseismic avulsion of the river on which its citizens depended for agriculture, sanitation, and trade. Since natural levees have been observed in India to collapse in intensity VII shaking, it is unnecessary to invoke coseismic uplift as a requirement for upstream river avulsion. The absence in the past two centuries of large earthquakes in the region has resulted in central Sindh being depicted as a region of low seismic hazard, yet in 1668, in the same province, an earthquake destroyed nearby Samawani and also initiated avulsion of the Indus. A case can be made for reevaluating the five millennia of archaeological ruins in Pakistan to establish a long-term view of seismicity unavailable from the short instrumental record.

INTRODUCTION

historical cities that had been described by Arabic and Mughal historians. Bellasis used the name Brahminabad, but later writers have referred to the ruins as Mansura, al-Mansurah, Bhramanabad, and Bhamanabad, and Cousens (1929) lists a dozen more. It is now known that the ruins described by Bellasis were the ruins of al-Mansurah, the capital of the Arabic province of Sindh established ca. 734 A.D. It is less accepted that Mansurah was constructed on the ruins of the earlier Hindu city of

The ruins at Mansurah (25.882°N, 68.777°E) were brought to the attention of the archaeological community in a series of reports published in Bombay and London magazines in the mid-nineteenth century (Bellasis 1857a, 1857b; Sykes, 1857a, 1857b). At the time of writing, and for many years thereafter, the numerous ruined sites in the region were identified with different

Bilham, R., and Lodi, S., 2010, The door knockers of Mansurah: Strong shaking in a region of low perceived seismic risk, Sindh, Pakistan, in Sintubin, M., Stewart, I.S., Niemi, T.M., and Altunel, E., eds., Ancient Earthquakes: Geological Society of America Special Paper 471, p. 29–37, doi: 10.1130/2010.2471(03). For permission to copy, contact [email protected]. © 2010 The Geological Society of America. All rights reserved.

29

30

Bilham and Lodi

Bhamanabad (Farooq, 1986). The evolving discussion concerning the nomenclature of the site and nearby urban centers can be followed in Elliot (1867), Cunningham (1871), Haig (1874, 1894), Raverty (1893), Cousens (1905, 1929), Panhwar (1983a, 1983b), Wheeler (1992), Hodīvālā (1939), Farooq (1986), and most recently Khan (1990). To avoid confusion, and since the identification of the city with earlier ones is irrelevant to the present discussion, we shall refer to the ruins as Mansurah. The vast area of rubble (2 km × 1.5 km) and the apparently orderly layout of the city led Bellasis to characterize Mansurah as the “Pompeii of the East,” a city frozen in time with many of its citizens interred within its ruins. The ruins are located 30 km east of the present path of the Indus, 60 km NNE of Hyderabad and 16 km ESE of Shahdadpur, Pakistan (Fig. 1). They consist of heaps of fragmented bricks separated by rubble-clogged ancient streets in orthogonal rows on a gently undulating area elevated above the meander of an abandoned river (Figs. 2 and 3). Mansurah was a planned city. Some of the streets were 65 m wide and were paved with bricks usually laid on edge (longest and narrowest dimension down), and underlain with wastewater drains (Farooq, 1986). In places, wastewater and sewage were led via conduits to terracotta-lined soakaway pits (Khan, 1990). The ruins are encircled by the remains of a 3-m-wide fortified wall with a perimeter length of 6.4 km, which some accounts (Elliot, 1867) describe as being surmounted by 1200 bastions— clearly an exaggeration if these ornamented the perimeter of the main city. A more conservative estimate is provided by Abul Fazl writing in the sixteenth century (Jarrett, 1891), who enumerated the number of ruined bastions as 140, spaced ~50 m apart (Cunningham, 1871). This very closely matches the measured perim-

eter, which would have required a 45 m mean spacing between 140 bastions. Khan (1990) describes the bastions as semicircular and spaced at 33 m intervals with remnants found at a height of 3.5 m. The remains of the bastions are now indistinguishable from the irregular mounds that characterize the level of the city. The several Arabic accounts that describe the founding of the city describe it as an island surrounded by a branch of the Mihran (the Indus). It was noted for its verdure and cleanliness, although some accounts complain of abundant fleas (Elliot, 1867). In the nineteenth century, at the time of the annexation of Sindh to India, the region was arid, with water found only in wells. Twentieth-century irrigation fed by the nearby Jamaro canal has now ponded parts of the abandoned river, which is distinctly concave where it abuts the eastern walls of the city. Elsewhere the ancient path of the river is obscured by agriculture. The description of the city as occupying an island is consistent with its current elevation 5–10 m above the shore of the present water in the river. The straightness of the SW wall and its continuation to a contiguous ruined settlement to the SE are suggestive of the existence of an artificial moat to the SW rather than a river meander. The path of the Jamaro canal was excavated with a 5° inflection to avoid crossing the Mansurah ruins (Fig. 1). Dominating the ruins, there is a single masonry tower known as the Thul (Fig. 4) that afforded subsurface access to a well (Cousens, 1929). The bricks of the town consist of thin fired clay bricks cemented with mud, but in the Thul, these bricks were cemented above ground with a gypsum-based cement. The foundations of the Thul used much larger bricks. Few intact bricks remain in the town, presumably because they have been scavenged for construction elsewhere. Pottery shards and glazed

2 km

Shahdadpur Mansurah Fig. 3 Fig. 2 Fig. 4

Mansurah Berani Tando Alam

10 km Figure 1. Google map view of Mansurah and its suburbs. A canal passes SW and a drain runs NE of the ruins, which show as light gray. An oxbow lake has now occupied the formerly abandoned river to the east of the city. Arrows to the right show photo angles of Figures 2–4.

Strong shaking in Mansurah, a region of low perceived seismic risk in Sindh, Pakistan

31

earthenware fragments are abundant. Coins found among the ruins in the past 150 yr are mostly copper, with some silver coins. A single gold coin has been reported, the location of which is now unknown (Khan, 1990). Even the soils and dust of the ruins have been removed by local farmers who have found their composition desirable as a fertilizer (Cousens, 1929). EVIDENCE FOR AND AGAINST AN EARTHQUAKE According to Bellasis (1857a, 1857b) and Sykes (1857), Mansurah was destroyed by an earthquake sometime after 975 A.D. Their evidence comes from the large number of skeletons discovered during excavations in doorways and room corners, and from the dates of coins scattered throughout the ruins: Figure 2. Close-up of pottery and brick shards with limestone fragments typical of the Mansurah ruins. The muzzle to butt length of the gun is ~80 cm.

Figure 3. A view from the eastern side of the Mansurah ruins near the perimeter wall facing north showing the abandoned river course, now occupied by a lake.

The human bones were chiefly found in doorways, as if the people had been attempting to escape, and others in the corner of rooms. Many of the skeletons were in sufficiently perfect state to show the position the body had assumed; some were upright, some recumbent with their faces down, and some crouched in a sitting posture. One in particular I remember finding in a doorway; the man had evidently been rushing out of his house, when a mass of brickwork had in its fall crushed him to the ground, and there his bones were lying extended full length and the face downwards. (Bellasis, 1857a, p. 417)

The coins provide the latest date for the layer of widespread collapse of structures in the city. Bellasis in his excavation found thousands of badly corroded specimens that were passed (in a 14 kg bag) to experts for identification (Thomas, 1858). The reports on these focused on the earliest coins indicative of minting in Mansurah (e.g., A.D. 750), and dwelled little on the details of coins of younger vintage, and hence the date of the inferred earthquake is poorly bracketed. The latest coins suggest internment after A.D. 975, but the date could be earlier according to Haig (1874, p. 287), who noted that Mansurah was in ruins “when

Figure 4. The Thul at Mansurah in 1897 (left from Cousens, 1929) and in 2008 (right). The integrity of this 12-m-high structure, which is bonded with a tough mortar, suggests that in the past millennium, the maximum shaking intensity in the region cannot have much exceeded Mercalli intensity X. The faced reentrant corner remains approximately vertical. Cousens excavated through the foundation of the Thul to virgin soils at a depth of 5 m below ground level. The 3-m-thick walls enclosed a 2.2 m central spiral staircase leading to a well within.

32

Bilham and Lodi

Biladuri wrote his Futuh-as-Sindh—perhaps about 870–880 A.D.” It is not certain on what authority Haig formed this conclusion because Biladuri died ca. 892 A.D. never having visited Sindh. Biladuri may have been confused by the identity of Mansurah and Brahmanabad mentioned above. Moreover, Elliot’s (1867, p. 122) translation of the Arab conquest indicates that ruination was a consequence of the battle for the city ca. 664 A.D.: “Then Muhammad, son of Kasim, went to old Brahmanabad, two parasangs from Mansurah, which town did not then exist, its site being a forest. The remnant of the army of Dahir rallied at Brahmanabad and resistance being made, Muhammad was obliged to resort to force, when 8, or as some say 26 thousand men were put to the sword. He left a prefect there. The place is now in ruins.” Returning to the ca. 980 A.D. damage following the establishment of the Arab capital, Cunningham (1871) also favored the destruction of the city by an earthquake. He suggested that the sack of the city by an aggressor was unlikely because of the absence of reports of charred timber, pillage being invariably preceded or followed by arson in medieval warfare. He found compelling evidence in the disposition of crushed skeletons, especially a quote by Richardson cited in Bellasis (1857a, p. 423), who described a brick “which entered corner-ways into a skull, and which, when taken out had a portion of the bone adhering to it.” His unstated implication is that a brick, particularly the lightweight Arabic tabular brick, would not have been a weapon of choice for an invading army. Cousens (1929), however, who substantiated the absence of a burn layer, dismissed earthquake damage. He pointed to a slow decline in the city that may have persisted to the thirteenth century before its abandonment. He found the absence of any gold coins or other items of value to support an alternate interpretation— warfare followed by systematic looting. He invoked the scattered skeletons as characteristic of a massacre. Citizens found in doorways, rooms, and streets were killed and interred by invaders intent on hurriedly dismantling a city. Cousens (1929, p. 71) adds, “Walls were thrown down in order to get at the door and window frames and roof timbers; and being brick ornamented with mud, were easily overturned with this rough treatment.” Panhwar (1983b) also attributed the destruction of the city to a punitive army, citing Farrucki’s account of Mahmud of Gazni sacking the city in 1025. It is possible that the city had by then had recovered economically from earthquake damage, sufficiently for survivors to have attracted a punitive attack, but had not yet reconstructed its defensive walls, rendering it an easy target for Mahmud’s army. Although the historical evidence for or against an earthquake remains inconclusive, there is no disagreement on two issues: that around the end of the tenth century, Mansurah underwent a catastrophic change—widespread ruin, followed by a significant decline in the availability of water. The ultimate abandonment of the city is linked, by all, to a change in the course of the river. This chronological sequence of damage followed by a change in hydrology is deduced from archaeological excavations. Cousens’ excavations in 1897 and 1909 revealed three layers of stratig-

raphy at the site: a pre-Muslim layer of occupation, a layer of orderly city structures (Mansurah), and finally a layer of disordered structures with numerous cylindrically lined wells that cut through the two lower levels, suggesting the drying up of the river and the need to tap groundwater supplies. Cousens theorized that following the sack of the city, the survivors reconstructed the city from the ruins of Mansurah and needed to sink wells to access drinking water. Excavations by the Department of Archaeology, Pakistan, have added details to this layered chronology (Farooq, 1986; Khan, 1990), and although the earlier confusion between ancient Brahmanabad and Medieval Mansurah remains unsettled, the excavations agree on the impoverishment of the most recent structures, and the apparent dependence of the later citizens on well water. The village that now occupies the westernmost part of the site also uses well water. Previous historians have considered the timing of the destruction of the city and the drying up of the river to be a coincidence. Certainly, had the events occurred in reverse order, we should agree with them. However, the historical record of large earthquakes contains many examples of rivers for which courses have changed following a nearby earthquake. A well-known case is the permanent shift in the geometry of Mississippi following the 1811–1812 New Madrid earthquakes (Johnston and Schweig, 1996). Other examples are described by Schumm (2005). Assuming that the river near Mansurah was confined within banks and levees not more than 3 m above the normal level of the river, an earthquake with uplift exceeding 3 m would be required to divert the course of the river. This would require a substantial earthquake (e.g., M >7) and the development of a surface uplift feature for which there is no evidence. We note, however, that strong shaking from a deep earthquake (associated with minimal or zero local uplift) could also trigger avulsion. To have affected the flow near Mansurah, avulsion of the river must have occurred upstream. During the 1811 New Madrid (Schumm et al., 2000) and 1897 Shillong earthquakes (Oldham, 1899), the most striking earth movements near rivers involved the collapse of river banks. In the New Madrid earthquake, the banks of the river plunged into the river, creating waves that further undermined the shaken banks, and briefly raised the bed level, impeding channel flow. Whether or not the natural levees that confined the river north of Mansurah would open to permit flooding of the hinterland would depend to some degree on their width, but fissuring and slumping would no doubt have weakened their ability to confine the river in a flood. Overbank flooding may not have been immediate, but in the absence of a reliable levee, extreme flood levels would no longer be a requirement. Flooding could therefore have occurred during the first heavy monsoon. Once the river breached through the banks of the river upstream, avulsion and the abandonment of the former channel would have been almost inevitable. We thus find the account of city collapse followed by a change in the course of the river, typical of earthquakes elsewhere, to represent a neglected consideration relevant to the earthquakerelated destruction of Mansurah. The 1668 Samawani earthquake

Strong shaking in Mansurah, a region of low perceived seismic risk in Sindh, Pakistan (Ambraseys, 2004), seven centuries later but in the same Mughal province of Nasirpur, appears also to have initiated avulsion of the Indus (Bilham et al., 2007). River shift was not immediate. In the half century following the 1668 earthquake, the river slowly shifted its course westward, eventually finding a path to the west of Hyderabad. Cities were hastily founded and abandoned in a half-century-long attempt to keep up with its evolving path, eventually leading to the establishment of Hyderabad in 1768 (Haig, 1894). THE DOOR KNOCKERS OF MANSURAH The collapsed remains of three large structures were exhumed in the post-1966 excavations: a grand mosque (50 m × 80 m with 2-m-thick walls), and two nearby civic buildings with slightly smaller scales (Farooq, 1986; Khan, 1990). These ancillary buildings are interpreted to represent administrative assembly buildings or schools having roofs supported by 1.8-m-square brick pillars. Deep beneath the prodigious quantities of rubble from one of these enormous buildings, four ∼50 kg bronze door knockers were found (Fig. 5). Their improbable survival owes everything to the depth of their burial. The size and quality of these decorative knockers—50 cm wide, 80 cm high, and 10 cm deep—made them a significant feature of this civic structure (Khan, 1990), and it is doubtful that a pillaging army, had they seen them, would have failed to regard them as a valuable souvenir. The knockers were fastened by 2-cm-diameter bolts or nails to what may have been the largest wooden doors in the city; the removal of this material elsewhere, Cousens (1929) argued, was the reason for the collapse of the structures in which they were embedded. We consider it doubtful that an invading army would have the resources or inclination needed to demolish these three large civic structures. To demolish them without first removing these attractive souvenirs appears to us even less likely. The survival of these remarkable fittings is presumably because they were buried by the collapse of the large building of which they were part. The pile of rubble was presumably too daunting to be removed by survivors or subsequent scavengers. It is difficult to escape the conclusion that the door knockers of

33

Mansurah were buried by the collapse of the entrance of a civic structure in an earthquake. DISCUSSION From the foregoing arguments, we present evidence that favors the destruction of tenth-century Mansurah by an earthquake of sufficient severity to tumble the 2-m-thick walls of the huge central Mosque, and other equally substantial civic structures. Much of the city collapsed with these larger structures, but survivors and scavengers were able to remove most of what was of value from the smaller rubble piles of private dwellings and shops of the commercial districts. The Thul, a tower assembled from numerous courses of bricks bound together with a strong cement fared better in the earthquake and has survived ten centuries of weathering. It has also survived determined efforts to dismantle it for its bricks. From these observations, we conclude that the brick-andmud structures of the city were destroyed by at least MSK intensity VII, and probably intensity VIII shaking. The conclusion is similar to that formed by Quittmeyer et al. (1979), who indicated a solitary intensity VIII on their Figure 2. The survival of the Thul (Fig. 4) provides an upper limit to accelerations since its construction ca. 800 A.D. We estimate that it could probably survive intensity X but that higher intensities would have disrupted the structure. The foundations of the tower itself have not been exposed to soil liquefaction, because, despite the fact that its foundations extended to sediments near the medieval water table, its few remaining finished walls appear approximately vertical. The archaeological excavations reveal no obvious examples of liquefaction or warping of the drainage courses within the city. No excavations have been undertaken in the lower ground adjoining the city where liquefaction would have been more likely. Subsequent avulsion of the river is also consistent with intensity VIII shaking. A single breach in a levee upstream caused by bank collapse would be sufficient to facilitate a shift in the river in flood. The intensity of shaking required to do this can be inferred from bank integrity at the time of the Mw = 8.1 1897

Figure 5. The bronze door knockers of Mansurah (Khan, 1990). The diameter of each 1-cm-thick circular plate is 56 cm, with raised relief of 17 cm and an average weight of 53 kg. The Sufic characters engraved on the outer annulus are from the Qu’ran and include the name of the Habbari ruler Abdullah.

50 cm

34

Bilham and Lodi

Great Assam earthquake. Bank collapse was noted by LaTouche in areas assigned MSK intensity VII or greater on the Brahmaputra and Surma River systems (Bilham, 2008). Thus, one would not need to invoke vertical deformation of the form recorded in the 1819 Allah Bund earthquake to restrict or divert flow of the river, nor would avulsion need to have occurred immediately after the earthquake. It may have occurred in the first severe monsoon floods a year or more later. The ensuing shift in the river would have made the site less attractive to survivors, both from the resulting restricted availability of water, but presumably also because the city had lost much of the river trade that supported its former prosperity.

Mach 1931

29

4.6

R.

4.4 4.8 4.4

us

Sibi 4.6

5.9

5.24.5

Ind

3.9 4.7

The causal faults of the ca. 980 Mansurah and Samawani 1668 earthquakes are currently unknown, and few clues to their present-day activity are available from recent microseismicity in the region (Fig. 6). The sediments of the Indus are known to be faulted at depth where these have been subjected to seismic prospecting, but none is known to cut the surface (Nakata et al., 1991). Several lineations have been mapped by Kazmi (1979) near Sibi, and to the SE of the river by Kazmi and Rana (1983) from Landsat imagery. Figure 7 shows an enhanced view of digital elevations in the Indus floodplain surrounding Mansurah, illuminated at an angle to minimize the artifacts of seams in the SRTM image. Undulations with wavelengths of 10–30 km, and

4.2

4.9

Muree 1852

4.8

4.6 4.5 4.8

Sukkur

4.4

H J-K

100 km

igh

3.9 4.3

5.44.9

Fig. 7

4.4 4.7

4.9

Jaisalmer 5.5

4.7

Nawabshah

5.2

4.9 4.5

Mansurah 980

4.4

Samawani 1668

Thar Desert

Hyderabad Karachi

Umarkot

4.8 4.9

Bhanbore

h ug Thatta s Tro Badin du r In e Low 5 5.1

4.7 4.7

4.6

4.4

Allah Bund 1819 4.1

24°N

Lakhpat 67°E

Bhuj 2001 71

Figure 6. Basement faulting between 25.7°N and 27.8°N is inferred from aeromagnetic contours of Zaigham and Mallick (1999). The two parallel lines between 68.3°E and 70.3°E indicate the approximate location of an inferred ancient rift system, and the shaded area delineates an area of shallow subsurface bedrock identified as the Jacobabad-Khairpur (J-K) High (Kazmi and Rana, 1982) that outcrops near Sukkur. Tick marks indicate down. Numbers indicate location and magnitude of instrumental epicenters (omitting the Bhuj 2001 aftershocks) from Villasenor and Engdahl (2007). Less accurate locations for M>3 events in central Sindh from the International Seismological Centre catalog are shown as open circles. Abandoned river channels following the 980 A.D. and 1668 A.D. earthquakes are shown as heavy dashed lines. Place names are in small italics; significant earthquakes are named with dates (Ambraseys and Bilham, 2003).

Strong shaking in Mansurah, a region of low perceived seismic risk in Sindh, Pakistan

35

27 +18 m

50 km

75

12

Jamaro Canal

m

14

/k

m

m

16

57

Figure 7. Shuttle Radar Tomography Mission digital elevation image of the Indus floodplain with smoothed 2 m contours referred to an arbitrary datum to the SE. Illumination is from N68E at an elevation angle of 5° (arrow). A dune field occupies the NE quadrant of the map. The 8 m contour indicates the approximate axis of a reduction in the mean down-valley slope from 75 mm/km in the 100 km north of Mansurah, to 57 mm/yr in the 100 km SE of Mansurah. Circles outline the pre-1668 bed of the Indus, which appears to follow a gentle dome in the topography. Numerous other abandoned paths are evident.

10

m /k

m

8

m

6

Mansurah 4 2 0

l Hyderabad

25°N 68

Indus

pre-1668 Indus

several lineations, are evident in the data, but the most prominent features of note are a subtle reduction of down-valley slope from 75 mm/km NW of Mansurah to 57 mm/km to the south, and gentle dome following the pre-1668 course of the Indus. It is tempting to conclude that the dome-like ridge is causal to the 1668 avulsion, since it is subparallel to the fold belt west of Hyderabad; however, it is possibly a sedimentary feature. The seismic significance of the down-valley change in gradient is unclear; however, the avulsion of the river in the tenth century and seventeenth centuries near this location is suggestive that block tilting may be active. The sense of slope change is, however, opposite to the sense of basement uplift described by Zaigham and Mallick (1999), suggesting that if the morphology is related to tectonics, a reversal of the geological sense of slip is now occurring. The principal seismicity in the surrounding region follows the fold belts along the Kirthar fold-and-thrust belt to the west, and the Sulaiman Range to the north (Fig. 6). Significant 6 < Mw < 7.6 earthquakes have occurred historically north of 28°N (Quittmeyer et al., 1979; Ambraseys and Bilham, 2003), and south of 25°N. Relatively minor seismicity occurs to the east, most notably a shallow Mw = 5.5 dextral earthquake that produced a surface rupture near Jaisalmer in 1992. The region near Mansurah shows little recent microseismicity.

70°E

With the exception of the uplift accompanying the Allah Bund earthquake, geological mapping reveals no recent faults that have disturbed the surface of the Indus sediments. However, Zaigham and Mallick (1999) using aeromagnetic methods identified offset structural features at 5–9 km depth beneath the cover of thin-skinned tectonics in the west that they projected eastward into the Indus basin. These trends suggest that Samawani and Mansurah both may be located above a buried horst 5 km beneath the sedimentary cover. The block on which they are apparently located is elevated 2 km above the contiguous structural block to the north. Strike-slip geological offsets are also observed on these structures, and Zaigham and Mallick (1999) argued that recent seismicity west of Nawabshah indicates that they are presently active. They proposed further that seismicity throughout the region is attributable to reactivation of faults related to an ancient allochogen along the western margin of India. The significance of a buried, poorly defined rift system in the Indus basin on assessing seismic hazards in the region is also discussed by Sawar (2004); however, the lack of clarity of structural features (Kazmi, 1979; Kazmi and Rana, 1982) beneath the Indus provides considerable room for speculation. Spatial variations in river sinuosity support the notion of present-day vertical tectonics in Sindh (Schumm et al., 2000;

36

Bilham and Lodi

Schumm, 2005). Jorgensen et al. (1993) demonstrated that the geometry of the Indus, and its evolving path, responds to geomorphic segmentation of the valley apparently related to tectonic forcing. Basin and domal structures recognized in geological investigations appear to be mimicked by present-day vertical displacements inferred from lateral variations in river sinuosity. The structural trends mapped by Zaigham and Mallick (1999) are orthogonal to the thin-skinned thrusts of the Kirthar range but may be understood in the context of the shear stresses of India’s convergence with Asia, which would tend to activate block rotation in the region. In order to account for the 1819 Allah Bund and 2001 Bhuj Mw ≥7.6 earthquakes Stein et al. (2002) proposed that the Sindh region has been fragmented and is converging with India as a result of this collision. If this is occurring, it must be doing so at rates of less than 1.5 mm/yr, the current rate of global positioning system (GPS) motion we have measured between Karachi and the Indian plate. Seismic Hazards in Sindh The occurrence of two damaging earthquakes NE of Hyderabad in the past millennium suggests that the structures beneath the Indus sediments must be considered active. The Mansurah earthquake resulted in intensity VIII, but it is uncertain how large were the intensities in the Samawani 1668 earthquake. The absence of damage reports from other towns suggests that both earthquakes may have been relatively modest (Mw 10 km in a transtensional relay. This geometry is responsible for the pull-apart basin of Lagunillas that contains the Urao lake (Alvarado, 2008). Fault rupture data were obtained from two paleoseismic sites, the Quinanoque trench on the Boc-b fault segment west of the Urao lake (Figs. 7 and 8) and Pantaleta trench on Boc-a fault segment (Alvarado, 2008; Audemard, 2009, personal commun.). These studies found that the Boc-b and not the Boc-a fault segment ruptured in the 1674 earthquake. The ruins of San Antonio de Mucuñó lie less than 20 km southeast of the Lagunillas pull-apart basin and the active trace of the Boconó fault (section Boc-b; Fig. 1) that ruptured in the 1674 earthquake. Consequently, it is likely that the 1674 earthquake sequence triggered widespread landsliding both on the distant northern flank of the Mérida Andes (Palme and Altez, 2002) and the southwest-facing slope of the Chaquentá flat-top hill where the first village of Mucuñó is situated. DISCUSSION The 1674 seismic events represent a significant earthquake sequence in the seismic history of Venezuela because of both their physical characteristics (multiple premonitory events, hundreds of aftershocks, and more than one main shock) and their macroseismic effects, many of which were associated mainly with

San Antonio de Mucuñó, Mérida Andes, Venezuela: Relocation following the 1674 earthquake

45

Figure 7. Relative location of the Pantaleta and Quinanoque trenches across the northern and southern Boconó fault traces, respectively, in the region of Lagunillas pull-apart basin (Fig. 1; after Alvarado, 2008).

Geomorphology: FT—fault trench; PR—pressure ridge; SR—shutter ridge; SP—sag pond; LV—lineal valley; OD—offset drainage; TH—saddle; BE—bench; Stratigraphy: Q1—Pleistocene; Q2—Middle Pleistocene; Q4—Holocene; 500 m Graphic scale 0 N2-Q—Pilocene-Pleistocene.

—fault scarp 1 Km

Figure 8. Log of the west wall of the Quinanoque trench, where all 14C ages are reported. The youngest recorded deformation is associated with the 1674 earthquake (among organic samples 31 and 32). Details on the chronologic difficulties shown by dated horizons are provided in Alvarado (2008) and Audemard (2008).

coseismic geomorphic events (large and widespread mass wasting in this case). These two aspects should be related in some way. For example, a series of foreshocks of significant magnitude might make the ground prone to large slope failure during the main earthquakes. These factors thus notably complicate the analyses of seismically induced features and damage in the assessment of magnitude of historical earthquakes (e.g., Palme and Altez, 2002). Paleoseismic investigations have provided essential information for the determination of the main earthquake sources along the Boconó fault. In addition, these studies have shed light on the possible rupture mechanism by which two different contiguous seismogenic segments of the Boconó fault (Boc-b and Boc-c) broke together but not necessarily simultaneously (Alvarado, 2008; Audemard et al., 2008). From both geomorphologic interpretation of aerial photographs and fieldwork at the Mucuño town site, it is clear that the first village was deeply affected by a rotational slide that crosscuts the urban grid. Furthermore, the slide has likely been reactivated and grown stepwise as deduced from the historical accounts. The mass movement first destroyed the borders of the village and then the inner parts of it. The landslide triggering mechanism was probably not a continuous or permanent source, but rather an intermittent trigger such as heavy rains or earthquake ground motion. It is plausible that the construction

of the town and human alteration of the general landscape and neighboring slopes helped to destabilize the land and produce conditions favoring mass movements (Laffaille et al., 2002). This issue might be of major significance when evaluating the potential triggering agents. Taking into account that the earthquakes of 1673–1674 have been correlated to rupture of Boc-b and Boc-c segments of the Boconó fault based on four paleoseismic trenches (two in the Lagunillas pull-apart basin and the two in Apartaderos pull-apart basin; Audemard et al., 1999, 2008; Alvarado, 2008; Audemard, 2008; Audemard, 2009, personal commun.), we postulate that the landslide that caused the abandonment of the first location of the village of Mucuñó was due to seismic shaking in an earthquake that occurred sometime between 1620 (when the village was first founded) and 1692 (when the relocation actually occurred). More precisely, historical documents indicate that damage to the village was triggered in an event prior to the 1691 shock that led to the final request for relocation. During this time span (1620–1591), the most likely earthquake in the Venezuelan catalogue of historical seismicity that would be close to the Mucuñó site and strong enough to trigger mass wasting is the earthquakes sequence of 1673–1674. The mass wasting of 1673–1674 was likely favored by rainy conditions along the northern flank of the Mérida Andes. In

46

Laffaille et al.

contrast, the Lagunillas region is characterized by semiarid to arid conditions year-round due to orographic effects that alter the wind patterns and cloud formation. The slide-dammed lakes that formed along the northern flank were suddenly drained, implying enough running water to fill them and later breach the earth dams. These catastrophic failures produced very large debris flows that were responsible for substantial losses in agricultural products within a short time after the earthquakes. Nevertheless, the heavy rains in Chaquentá area alone cannot be completely ruled out as a triggering mechanism for the slide. However, historical accounts reported the formation of debris flows (“bolcanes”) around the town at the time of the opening of the wide cracks, such as the one shown in Figures 4C and 4D. The distance between the first town of Mucuñó and the Boc-b rupture segment of the Boconó fault during the 1673–1674 earthquake sequence is ~20 km. It is much closer than the distance to the larger and wetter mass movements in the northern flank. The smaller size of the landslide at the Mucuñó site compared to those on the northern flank of the Mérida Andes might be better attributed to the dry conditions prevailing in the Las Acequias valley where the ruins of Mucuñó are located rather than to its proximity to the epicenter. Assuming that the epicentral region of one of the main events of the 1673–1674 earthquake sequence was located near the middle of the Boc-b segment, the wave propagation direction (white arrows in Fig. 1) might have impacted almost perpendicular to the SW-facing hillside of the Chaquentá mesa and Mucuñó site. The foliation dip direction and the steepness of the slope, among other factors, probably facilitated downslope movement. Additional geologic, photo interpretation, and detailed fieldwork on the Boc-b fault segment, as well as an exhaustive search of new and relevant historical data, would help to further resolve the triggering and timing of the mass-wasting events in relationship to earthquake faulting. This study combined detailed geomorphologic studies of mass-wasting processes and paleoseismic interpretation of the timing of faulting together with the evaluation and interpretation of historical accounts, not only in terms of past seismic activity and the associated phenomena, but also in terms of social sciences (history, economy, anthropology, sociology, theology, urbanism, among others). The use of this type multidisciplinary approach has allowed us to better interpret the historical data in relationship to earthquake rupture models and ultimately to provide better earthquake source parameters for seismic hazard analyses. ACKNOWLEDGMENTS Our gratitude goes to the Venezuelan Foundation for Seismological Research (FUNVISIS) and to the Foundation for Prevention of the Seismic Risk of the State Merida (FUNDAPRIS) for their unconditional support to this investigation. This research is a contribution to project FONACIT 2001002492 and 2002000478. This article is a contribution to the United Nations Educational, Scientific and Cultural Organization (UNESCO)–funded Inter-

national Geoscience Programme (IGCP) 567, “Earthquake Archaeology: Archaeoseismology along the Alpine-Himalayan Seismic Zone.” REFERENCES CITED Acosta, L., 1982, San Antonio de Mucuño. Formación de un Pueblo Indígena de Encomienda y de Doctrina en el Valle de Acequias, 1558–1620 [tesis de grado]: Mérida, Venezuela, Escuela de Historia, Universidad de Los Andes, 160 p. Alvarado, M., 2008, Caracterización Neotectónica de la Cuenca La González, Estado Mérida, Venezuela [M.Sc. thesis]: Mérida, Universidad Central de Venezuela, 220 p. Audemard, F.A., 2005, Paleoseismology in Venezuela: Objectives, methods, applications, limitations and perspectives: Tectonophysics, v. 408, p. 29–61. Audemard M., F.A., Pantosti, D., Machette, M., Costa, C., Okumura, K., Cowan, H., Diederix, H., and Ferrer, C., 1999, Trench investigation along the Merida section of the Boconó fault (central Venezuelan Andes), Venezuela, in Pavlides, S., Pantosti, D., and Peizhen, Z., eds., Earthquakes, Paleoseismology and Active Tectonics: Selected Papers to the 29th General Assembly of the Association of Seismology and Physics of the Earth’s Interior (IASPEI), Thessaloniki, Greece, August 1997: Tectonophysics, v. 308, p. 1–21, doi: 10.1016/S0040-1951(99)00085-2. Audemard, F.A., Machette, M., Cox, J., Dart, R., and Haller, K., 2000, Map and Database of Quaternary Faults in Venezuela and Its Offshore Regions: U.S. Geological Survey Open-File Report 00-0018, including map at scale 1:2,000,000 and 78 p. text. Audemard M., F.A., Ollarves, R., Betchtold, M., Díaz, G., Beck, C., Carrillo, E., Pantosti, D., and Diederix, H., 2008, Trench investigation on the main strand of the Boconó fault in its central section, at Mesa del Caballo, Mérida Andes, Venezuela: Tectonophysics, v. 459, p. 38–53, doi: 10.1016/j.tecto.2007.08.020. Audemard, F.E., and Audemard, F.A., 2002, Structure of the Mérida Andes, Venezuela: Relations with the South America–Caribbean geodynamic interaction: Tectonophysics, v. 345, no. 1–4, p. 299–327, doi: 10.1016/ S0040-1951(01)00218-9. Bakun, W., and Wentworth, C., 1997, Estimating earthquake location and magnitude from seismic intensity data: Bulletin of the Seismological Society of America, v. 87, p. 1502–1521. Centeno, M., 1940, Estudios Sismológicos: Caracas, Venezuela, Litografía El Comercio, 365 p. Clarac, J., 1990, Etnohistoria de San Antonio de Mucuñó: Mérida, Venezuela, Universidad de Los Andes, Boletín Antropológico 20, 120 p. Grases, J., Altez, R., and Lugo, M., 1999, Catálogo de Sismos Sentidos o Destructores. Venezuela 1530–1998: Caracas, Academia de Ciencias Físicas, Matemáticas y Naturales, Facultad de Ingeniería, Universidad Central de Venezuela, Editorial Innovación Tecnológica, 654 p. Laffaille, J., Ferrer, C., and Rengifo, M., 2002, San Antonio de Mucuñó: Evidencias históricas de la actividad antrópica como detonante de amenazas naturales: Caracas, Venezuela, Memorias del las III Jornadas de Sismología Histórica, FUNVISIS, Serie Técnica 1-2002, p. 219–221. Nadal, A., and Villafañe, M., 2004, Mudanza del pueblo de San Antonio de Mucuño para otro sitio más apropiado en tierras de la Encomienda del Capitán Alonso de Toro Holguín, en el Valle de Acequias, 1692. Procesos Históricos: Revista de Historia, Arte y Ciencias Sociales, no. 5: Mérida, Venezuela, Universidad de Los Andes, 125 p. Palme, C., and Altez, R., 2002, Los terremotos de 1673 y 1674 en los Andes Venezolanos: Interciencia, v. 27, no. 5, p. 220–226. Palme, C., Morandi, M., and Choy, J., 2005, Re-evaluación de las intensidades de los grandes sismos históricos de la región de la cordillera de Mérida utilizando el método de Bakun and Wentworth: Revista Geográfica Venezolana, Número Especial, v. 2005, p. 233–253. Rod, E., 1956, Earthquakes of Venezuela related to strike slip faults?: American Association of Petroleum Geologists Bulletin, v. 40, p. 2509–2512.

MANUSCRIPT ACCEPTED BY THE SOCIETY 19 MAY 2010

Printed in the USA

The Geological Society of America Special Paper 471 2010

New interpretations of the social and material impacts of the 1812 earthquake in Caracas, Venezuela Rogelio Altez* School of Anthropology, Universidad Central de Venezuela, Caracas 1040, Venezuela, and Venezuelan Society of Geosciences History, Caracas 1070, Venezuela

ABSTRACT This work sheds light on one of the most important earthquakes in Venezuelan history. At 16:07 on Holy Thursday, 26 March 1812, Caracas and the surrounding province of Venezuela suffered a very destructive earthquake. The earthquake occurred at a time of great political, economic, and social upheaval, with the beginning of the republican revolution and the Spanish royalist military response. Within this historical context of conflict, documentary information may be biased and subjective. This chapter is a methodological and epistemological analysis of the 1812 earthquake damage from letters and manuscripts and an interpretation of the social impact of the earthquake within ideological, subjective, and political context. The widespread destruction of the city of Caracas was heterogeneous in its distribution. Damage was determined largely by the differences in the construction style and quality and by the maintenance status of the building. Based on analyses of three funeral books from the era, the number of earthquake victims in Caracas in 1812 may have been close to 2000. This value is lower than regional estimates of the death toll.

INTRODUCTION

Although it is generally accepted that the Caribbean plate moves eastward with respect to South America, this plate boundary is not a simple dextral type (Soulas, 1986; Beltrán, 1994; Audemard and Audemard, 2002). Instead, it is a broad active deformation zone resulting from a long-lasting oblique-collision process (Audemard, 1993, 1998). Nevertheless, a large portion of this right-lateral motion seems to take place along the dextral Boconó–San Sebastian–El Pilar fault system (Schubert, 1984; Soulas, 1986; Audemard and Singer, 1996). Seismicity of the fault zones aligned along the southern boundary of the Caribbean plate is known from historical earthquakes (Centeno Graü, 1940; Grases, 1990; Grases et al., 1999). Moreover, the seismicity of northern Venezuela (over the Boconó–San Sebastian–El Pilar fault system) suggests that margin deformation does not occur along a single fault zone (Audemard and Singer, 1996). Other

This study is a qualitative analysis of the documentary evidence of the 26 March 1812 earthquake in Venezuela and the associated seismic damage in the city of Caracas. The earthquake ruptured the San Sebastian system fault, one of the many structures along the transform plate boundary that separates South America from the Caribbean plate (Fig. 1). According to Grases and Rodriguez (2001), the magnitude of the 1812 earthquake is estimated to have been between M 6.9 and 7.2. This event is a good example of the expected seismic hazard from the interconnection between southern boundary of the Caribbean plate and the other system faults related to the South American plate. *[email protected]

Altez, R., 2010, New interpretations of the social and material impacts of the 1812 earthquake in Caracas, Venezuela, in Sintubin, M., Stewart, I.S., Niemi, T.M., and Altunel, E., eds., Ancient Earthquakes: Geological Society of America Special Paper 471, p. 47–58, doi: 10.1130/2010.2471(05). For permission to copy, contact [email protected]. © 2010 The Geological Society of America. All rights reserved.

47

48

Altez

Figure 1. Diagram of plate boundaries of Venezuela (after Audemard et al., 2000).

small earthquakes can be localized on minor faults that everywhere cross the principal systems (Audemard et al., 2005). The fundamental aim of this investigation is to estimate the earthquake intensity, earthquake fatalities, and potential microzonation of damage (e.g., Altez and Laffaille, 2006). This historiographic approach to assessing damage in the 1812 earthquake (see Altez, 1998, 2005a, 2006) contrasts with previous studies that have focused solely on assigning earthquake intensities, drawing isoseismals, or exclusively on the style of structural damage (as an example of that, see FUNVISIS, 1997). As stated by Mocquet (2005, p. 130): “The assessment of macroseismic intensities from historical reports requires one to combine information on vulnerability and damages. Similar earthquakes occurring at different epochs can produce different effects, depending on the demography and local life conditions at the time of occurrence.” Critical analyses of historical information pertaining to earthquakes, i.e., the field of historical seismology, utilize methodological approaches and knowledge across the disciplines of seismology, history, anthropology, and earthquake-resistance engineering (Guidoboni and Ferrari, 2000). Three fundamental variables should be understood and analyzed in order to sufficiently reconstruct damage from a seismic event. The most important of these variables is the historical context in which the seismic phenomena took place and the information and direct testimonies concerning the impacts and

effects that are recorded. Understanding this historical context leads us to comprehend the information that was elaborated on and discussed at that moment in history. This has been called a “critical joint” (Olson and Grawonski, 2003) or “disaster juncture” (Altez et al., 2005). Emerging from complex circumstances at a time of military conflict, the 1812 earthquake is recorded with different perspectives from the initial experience and later interpretations of the event. A second variable in understanding this historic seismic event lies in the material context of the built environment. Observations of the constructive typology and construction conditions at the time of the earthquake are important parameters for assessing the intensity of damage (Altez, 2005a; Yamazaki et al., 2005). Damage in the earthquake is quantified on the European Macroseismic Intensity Scale (EMS, 1998), ranging from I through XII. In assessing the specific construction damage, the “Classification of damage to masonry buildings” (also from EMS, 1998), which ranges from I to V, is particular useful. Lessons from the social impacts of this disaster also are learned from the extent of material destruction and the subsequent consequences felt in the population. Finally, the number of earthquake fatalities is an important variable that is often open for interpretation. The number of deaths in the earthquake provides a good estimation of the severity of the event. To calculate the number of deaths in the earthquake, I extrapolate data from multiple sources. The results of these analyses of the historical

The social and material impacts of the 1812 earthquake in Caracas, Venezuela text for seismic intensities and the number of fatalities seem to be far from classic studies on the 1812 Caracas earthquake and provide a new interpretation on this catastrophic phenomenon in the history of Venezuela. HISTORICAL CONTEXT OF VENEZUELA IN 1811–1812 The political events in South America in the early nineteenth century would transform American society forever. Venezuela, as well as most of the Latin American continent, was on a tortuous road from 300 yr of colonial rule toward independence. In July 1811, Venezuela declared independence. This caused an acceleration of the breakup of the institutional and administrative bonds with the Spanish crown, which was then in crisis due to the Napoleonic invasions. In March 1812, before the earthquake, a small royal army started out from the city of Coro, 500 km to the west of Caracas, heading toward the main city with the objective of recapturing the rebellious counties (for a general approach to the historical context, see Lynch, 1985, or Guerra, 1992). The earthquake of 1812 occurred during the Holy Week between Palm Sunday and Easter, with all of its Christian celebrations, magnificence, and ritual ceremonies. The 26 March 1812 was Holy or Maundy Thursday. The religious mandates for the day were rigorously fulfilled, even more in those days, when faith still had an indisputable nature. Most of the population of the region was concentrated in the city of Caracas (Fig. 2), the province capital (territory) of the General Captainship (jurisdic-

49

tion) and of the Archdiocese (major ecclesiastical authority) of Venezuela. Caracas was also the administrative, institutional, and political center. Thus, Holy Thursday included the presence in the capital city of the most prominent civil, military, and ecclesiastical authorities. In the middle of this religious ceremony, and the threat from the royal invasion and political tension, a dreadful earthquake struck the city at 4:07 in the afternoon. Because the earthquake occurred at this turning point in history between colonial rule and independence, the extent of the damage in the event has become an interpretive problem complicated by heroic and nationalistic reconstructions of history. The fall of the patriots’ ambitions, the solid discourse of the Christian faith, the royal manipulation, the denial of the independence movement, and the exaggeration of the context narrators are all factors that have colluded in the overestimation of the importance and magnitude of the tragedy. The critical conditions imposed by the war, including destruction, abandonment, and migration, contributed to the hopeless loss of information. Files were destroyed in the collapse of buildings, lootings occurred on both sides, paper was used for fuel in bonfires that protected against the cold night in the outdoors, and any type of paper was used to ram the gunpowder for combat; all this combined to produce the information loss that is sorely missed today (for a detailed description on the matter, see Altez, 2006). The 1812 Venezuelan earthquake is categorized by all historians and investigators as one of the largest in the country’s history. Information written about this earthquake was motivated by many reasons. This has potentially contaminated the context

Figure 2. Caracas and its faults (detail from Audemard et al., 2000).

50

Altez

of the point of view of the chronicler, suggesting bias in opposing or supporting one side of the nationalism movement and the war of independence. MATERIAL CONTEXT: THE CITY OF CARACAS IN 1812 From an urban point of view, in 1812 Caracas was not a major city compared to other capitals of the Spanish colonial world. More impressive economic and architectural development and higher population densities were found in cities such as Mexico, Lima, and Bogotá. However, Caracas had shown signs of institutional growth since the late eighteenth century.

The appointment of the General Captainship in 1777, the Royal Audience in 1786, the Royal Consulate in 1793, and the Archdiocese in 1804, granted to the capital of the Province of Venezuela considerable status. By 1810, Caracas was the center of political power and economic growth. However, these aspects had not yet decisively impacted the city’s architecture. Urban development in Caracas at the time of the 1812 earthquake lagged behind its other American contemporaries (Fig. 3). Basically, in early nineteenth-century Caracas, the public buildings were old and ill-maintained (Table 1). Most documentation from Caracas and the region, written by the priests narrating the churches’ condition, denounced the dramatic situation of their structures. They describe cracks, roof problems, and infestations

Figure 3. Map of Caracas near 1810, by Mendoza Solar (1910).

The social and material impacts of the 1812 earthquake in Caracas, Venezuela of ants and termites. On 25 August 1812, the Archbishop of Caracas, Narciso Coll y Prat, when speaking about conditions of the cathedral prior to the earthquake, stated that the church was “old, incompetent, disproportionate…” (Coll y Prat, 1812). In 1812, the only structures in the city of considerable height were the churches, based on the rationale that no construction could show more height than “the house of God.” The only tall structures in Caracas in 1812 were the towers of the churches (Grases et al., 1999). Thus, the remaining constructions were of lower and smaller size, as seen on the perspective view of Caracas in Figure 4. Most houses were of typical colonial architecture, with a single story and a central patio. Building materials varied between adobe, bahareque, and masonry. The testimonies of travelers and visitors at the beginning of the nineteenth century attest to the fact that Caracas hardly had any two-storied houses (e.g., Ker Porter,

51

1997; Duane, 1968). In this sense, when speaking of “buildings,” one should emphasize both the small size and scale. Constructions in the city can be classified into three groups: churches (high buildings with towers), public administration buildings (generally two-storied and spacious houses), and living quarters or housing (low structures built in three different styles). The role of construction type was a determining factor in whether the building collapsed in the earthquake. As deduced from the historic documents, the materials used for the housings and buildings construction were not able to withstand the seismic shaking. According to Pedro Cunill Graü (1987), the housings of the poorest members of society were made of bahareque (stick interwoven with canes and mud) and were roofed with straw or palms. Constructions in the middle sector of the society (merchants and artisans) had adobe, walls reinforced by thin trunks to hold roofs of straw or tiles. The houses of the richest levels of

TABLE 1. AGE OF SOME OF THE CONSTRUCTION OF MAJOR STRUCTURES IN CARACAS AT THE TIME OF THE 26 MARCH 1812, VENEZUELAN EARTHQUAKE Construction Approximate year of construction Cathedral Church Built as parish church toward the second half of the sixteenth century. Remodeled in 1636 as a cathedral and as the seat of the Diocese of Venezuela. Rebuilt after the 1641 earthquake. Altagracia Church 1656 La Pastora Church 1745 San Pablo Church 1580 La Merced Church 1638 San Mauricio Church 1570. Rebuilt in 1667, after the damages received in the 1641 earthquake. San Jacinto Church Late sixteenth century San Lázaro Hospital Construction initiated in 1759. La Candelaria Church 1708 Santa Rosalía Church Built as church in 1732, before it was a chapel. La Concepción Convent 1619 Carmelitas Nuns Convent 1739 Seminar School 1675. Recently repaired, with construction works that lasted from 1809 through 1811.

Figure 4. Detail of the painting titled Nuestra Señora de Caracas, painted in 1766 (artist unknown) and currently located in the Consejo Municipal of Caracas. Note the height of constructions of the time. Church towers and some few two-storied buildings are the only tall buildings and are concentrated around downtown.

52

Altez

the community were built with adobe and stone walls with brick facades, wooden barred windows, and tiled roofs. The houses of the upper classes generally had very elevated roofs that made them more fragile during the earthquake (Ker Porter, 1997). Civic and religious constructions were mostly built of wooden frames with stone facades brick arches, tiled roofs, and whitewashed walls. According to Alejandro Ibarra (1862), the quality of the period constructions was highly responsible for the earthquake damages: “The limited depth of the bases or lack thereof, materials of diverse density and shapes and even of inappropriate nature, which were indiscriminately used to make the main walls of even the larger buildings: their lack of thickness, the use of poor quality mortar and other mistakes related to the art of building, were causes that undoubtedly contributed to increase the devastation of the 1812 earthquake” (p. 2). The conditions pointed out by Ibarra (1862) are of great value, since he was a pioneer in the investigation of the effects of the 1812 Venezuelan earthquake closer to the date of the earthquake and prior to the massive transformations of the city in the twentieth and twenty-first centuries. EARTHQUAKE DAMAGE The earthquake effects of 26 March 1812 were devastating in Caracas. As all documentation points out, Caracas was in ruins (Delpeche, 1813; Díaz, 1829; Palacio Fajardo, 1817). Descriptions near the date and detached from all political or romantic ideas are very eloquent. In the first days after the disaster, the Cabildo (the city council) meetings reassured the citizens, calling on them to make efforts to better the conditions of the city. The city council ordered cleanup of the debris in the streets, thus allowing access to Caracas for food deliveries from outside of the city (Actas del Cabildo, 1972). These first orders exemplified the critical necessity of provisioning the city. The roads drew a perimeter around the city, leading to the Capuchinos square, where provisions would be received. The types of structural damage in the earthquake were very diverse and largely determined by the type of construction, which in turn was based on social class. According to Cunill Graü (1987), the underprivileged classes formed suburbs in the periphery of the city, while the most notable and powerful citizens resided in blocks near the city center. The same author states that the housing of the poorest suffered less damage in the earthquake because they were built of bahareque (more earthquake-resistant material in his opinion). It can be inferred that there would be fewer victims in these houses due to light cane roofs that would be less fatal than the collapse of a tile roof. However, testimony of the city council and the orders of Spanish government given in August (Monteverde, 1812; Actas de Cabildo, 1972) indicate that debris was abundant throughout the city. The house construction types of the middle and upper classes were vulnerable, with very heavy and high roofs of tiles propped up by weak thin trunks. It is assumed that these houses caused the largest quantity of deaths and destruction. In the burial

certificates that could be found (Book of Burials, La Candelaria Parish, number 7, 1806–1817; Book of Burials, Adults, San Pablo Parish, 1808–1812; Book of Burials, Chacao Parish, number 2, 1797–1821), most of the deceased registered by the priests in charge died in their houses. There was a high death toll of maids, children, and slaves because families were at the celebrations in the cathedral at the time of the earthquake. The damage in the more important buildings seems to have been related to the quality of their construction and materials. Ibarra (1862, p. 2–3) stated:

An example of all this is provided in the temple of Altagracia whose walls especially the South one and surely that of the North, also built as just said, came to earth and with them all the temple; at the same time its beautiful front that is of good construction resisted the shakes of the tremor and the shudder that should cause the fall of its arched roofs and its very solid and heavy construction. In the same way in the convent of San Jacinto whose interior new factory, was built with bad materials, came to the floor, conserving the old external one that had resisted the earthquake of 1641; and the one that was made immediately after, heavy and rough, but of solid construction, being noticed that the walls built with raw adobes made of pure mud and straw, resisted perfectly, when the poor built masonries with mezclote [some kind of varied and mixed materials] were quartered, disjointed and destroyed they didn’t fall, already decomposed the mezclote and loose fitting bodies that entered in the formation.

Another document (Larrain, 1958) estimated that of the 5000 existing houses in Caracas before the earthquake, only 2000 were left standing. Larrain was referring to the housings, and not to the religious buildings or public administration houses. Little mention is given to civic buildings, since the government had been displaced by the revolution. However, it calls attention to the omission of this information. It may be inferred that they suffered less damage. This type of construction (two-story, roomy, stone, adobe houses with lime walls and tile roofs) resisted destruction in the earthquake. Beyond the materials used in construction, it seems that the quality of the building was decisive in the survival of the structure. Houses built with greater care and better resources, such as the spacious public administration and those of the most prominent citizens in Caracas, seem to have suffered less damage, while those that were built with fewer resources and were old or poorly maintained in their structure (bahareque with cane, heavy roofs propped with fragile and weak tied supports) were less lucky. It can be inferred that the collapse of the roofs was responsible for most of the damage and fatalities. IMPACT ON THE SOCIETY The earthquake of 26 March 1812 occurred in an area unprepared to face earthquakes, and thus produced impressive losses. Alex Scott, commissioned from the United States to observe closely the new revolutionary government, calculated the losses

The social and material impacts of the 1812 earthquake in Caracas, Venezuela in Caracas and La Guaira (near the port 20 km to the north of Caracas) at about four million dollars (Scott, 1812). War across the whole territory had bankrupted the administration by 1812. Also, public funds had collapsed. The country’s largest population by percentage, mestizos (natives and slaves), was structurally poor. Losses in the state treasury must then be envisioned with respect to a population with little wealth. The loss of the churches also meant sacrifice in funds and labor, since the majority was not able to assume the responsibility of reconstruction. Economically, the earthquake affected the religious population, who hardly had the means to go to mass. In a structurally poor society, churches were in a poor state of repair and underfunded, since most of the parishioners were mestizo and not white criollos (creoles) and wealthy people. Descriptions of the ruins repeat in all the documentation, giving account of the earthquake damages and sufferings of the reconstruction. The population was devastated by the lack of resources, besides being pressured by ongoing war. In Caracas, the earthquake damage blocked roads and access to provisions. Later, when the disaster reached its maximum expression, the Consulate of Caracas, from which the Intendant Dionisio Franco (1813) had requested a loan of 100,000 pesos, responded as follows: “The same powerful causes that have influenced from the year 1810 until the present in these Provinces, reducing them to the impossibility of covering the expenses that circumstances demand, has caused the extraordinary backwardness that experiences the trade and the agriculture, particularly in this Province, that above all not only suffering losses being consequence of the political events occurred in that epoch, but also those of the great earthquake of last March 26 that had just ruined the harvesters class” (Consulate answer to Intendant, 29 January 1813). The same text later argued that among other reasons, the consulate by itself was not able to ordinarily congregate, because there was not “...in the City enough number of individuals... whom to summon for it by effect of the same things and in the emigration that there was towards the country following the earthquake in where they still remain in terms that Government’s Meetings have sometimes stopped to be made by lack of vowels….” The merchants were also paralyzed due to the blockade suffered by the region. Even so, it was frankly recognized that, although “they are the only ones who can be counted on in similar difficulties,” the same ones were in a “general poverty.” “New contributions and loans cannot be counted on due to the general poverty of these inhabitants, caused by the revolution, earthquakes and also the recovery of the legitimate Government…” (Dionisio Franco, manuscript dated 13 February 1813). To this crisis picture in the public and private funds of the administration, the characteristic poverty of the society should be added. In this way, all suffered the catastrophe, although not equally. Church restorations represented the same tragedy in Caracas as in all the towns of the region. Undoubtedly, this was due to the lack of resources of all kinds. By 1816, efforts were made to plead the situation of Caracas churches to the city council. How-

53

ever, the political passions would continue slowing things. When the city council requested the report about “the resulting destruction caused in the two revolutions that have afflicted them and by the horrendous general earthquake of March twenty six 1812,” the secretary answered: “…almost all Churches are ruined more or less and many totally. (…) that the [Metropolitan] Church for repairs, like the ruins of their building are repairing, to restore some necessary pieces of jewellery and to make the costs of the precise one for the divine cult, is using extraordinary wills, or taking borrowed money, or receiving it to census…” (Guzmán, 26 April 1816). The earthquake’s damage repairs were delayed by the economic and societal problems. The recruitment and maintenance of a labor force were drained by migration. The repair of churches, room houses, and public administration was also delayed many years because of the deep economic crisis. Similarly, the negative consequences in the population distribution in the urban core caused the city housing to be abandoned and the development of new suburbs. This seriously affected location and image of the capital: “…to which is added that thinking that most of the inhabitants not to live more in the City again, with the object of living in the uninhabited places, unroofed their houses, pulled out their doors and windows with their wood, and these pieces to build shacks where to be…” (Méndez, 18 April 1816). With the destruction caused by the 1812 earthquake, certainly the horizons of the cities, towns, and villages were changed forever. That destruction also contributed to the transformation of the colonial society. To lift a nation from that debris was not, in spite of the nationalist historiography, a heroic gesture, but an unavoidable situation accompanied by hopeless circumstances. EARTHQUAKE FATALITIES At the time of the earthquake, the city of Caracas was the most populated in the whole Province of Venezuela. According to Cunill Graü (1987), in 1812, Caracas had 50,000 inhabitants. A more precise figure is given by Díaz (1817) of 31,813 people. If we compare the population figures with the number of houses (5000 noted by Larrain, 1958), then the average number of inhabitants per home was about six people. The earthquake occurred on Holy Thursday in the middle of the afternoon when most people of the city were in the Main Square in front of the Cathedral celebrating the services. Therefore, houses were not fully occupied at the time of the earthquake. It is quite likely that had the houses been fully occupied, the number of deaths would have been higher. The number of deaths for the 1812 Caracas earthquake has been at times confused with the death totals caused by the war, the mandatory migrations, and the famine. To understand the disastrous consequences of these last aspects, Cunill Graü (1987) conducted a detailed analysis in this respect. As noted earlier, that confusion is clearly linked to the critical historical context and to the meaning of that same context subsequently for the country’s

54

Altez

later history. At a time in which such a profound transformation of society was starting to take place with the birth of a new nation, information was managed or exaggerated for political gain (Guidoboni and Ferrari, 2000; Rodríguez and Audemard, 2003). Later, when those years of crisis were transformed by historical discourse into a revolutionary story, nobody dared to question the heroic conditions of the transformation. The tragedy of the earthquake helped manifest and heighten the revolutionary ideal of heroic triumph over adversity. The nation’s genesis mythology needed its heroes to sufficiently overcome her adversaries. This is why the war, the earthquake, and all the circumstances of that moment became the most critical moments in Venezuelan history: They are the heroic genesis of the nation. However, this does not make the history of the period an unbiased record of the truth. The death figures for the 1812 earthquake vary widely, as seen in Table 2. These figures were obtained from documents from contemporary narrators and from nineteenth- and twentiethcentury investigators. They are certainly the well-known death numbers related to the earthquake in Venezuelan historiography. Between the 20,000 of Forrest (1812) and the 1000 of Roscio (1812), the difference is very significant. Of all of the estimations, only Coll y Prat (Archbishop of Caracas in 1812) declares that his figures come from an order to the priests of his diocese. Indeed, that order was extended to all his parishes on 6 April 1812 (Coll y Prat, manuscript of 31-03-1812), but the replies from his whole jurisdiction have not been found in their entirety. Fatality figures by Coll y Prat (1816) include all the parishes and towns to which the request was made, and not exclusively from the city of Caracas. Furthermore, Díaz (1817) stated that the number of inhabitants in the province indicated that 13,000 persons “died with the earthquake.” This value accounts for the entire population of the provinces in the whole jurisdiction of Venezuela. The 1000 deaths in Caracas pointed out by Roscio (1812) are derived from a seemingly very conservative number. In the San Carlos military quarter of the city, ~500 individuals are said to have perished (Delpeche, 1813). Díaz (1829) wrote an eyewitness account of 40 people dying in the collapse of one convent of the Order of Predicadores in the Square of San Jacinto. One way to derive an independent estimation of the earthquake fatalities is to consult the funeral records. However, it was only possible to find three funeral books that correspond to March 1812 (Table 3). Only two of the funeral books are from the urban area of Caracas. The third funeral book is from the Chacao parish, a small town on the outskirts of the city. The funeral book data show a total of 137 deaths from three different parishes. Extrapolation of this fatality rate to the rest of the city is probably not appropriate because it may not represent earthquake damage and building collapse in other parishes of Caracas. In total, Caracas had as many as 19 parishes inside the urban perimeter, and thus calculations of the death numbers with data coming from only two of the parishes are highly uncertain. It is significant that the death numbers registered by the priests of these parishes take into account the total of the jurisdiction, including, obviously, the housing and not only the churches.

TABLE 2. NUMBER OF DEATHS FROM THE 1812 EARTHQUAKE IN CARACAS No. of death s Source 15,000–20,000 Forrest (1812) 16,000 Ker Porter (1825–1842) 14,000 Centeno Graü (1940) 10,000–12,000 Coll y Prat (1960, originally from 1818) 10,000 Rojas (1879) 10,000 Heredia (1895) 9000–10,000 Delpeche (1813) 8000 Irvine (1818) 7000 Parra Pérez (1939) 6000–7000 Urquinaona y Pardo (1820) More than 6000 Ascanio (1813) 3000 Semple (1812) 1000 Roscio (1812)

TABLE 3. NUMBER OF DEATHS FOR THE 1812 VENEZUELAN EARTHQUAKE DERIVED FROM FUNERAL BOOKS OF THE CARACAS REGION No. of deaths Parish Source 83 La Candelaria Book of Burials, 1806–1817 38 San Pablo Book of Burials, 1808–1812 16 Chacao Book of Burials, 1797–1821

Consequently, it is a registration that embraces several (perhaps 8–10) blocks around a church. In a radius of some 10 blocks (the probable maximum for the case of La Candelaria), there were 83 deceased. With this value, it is difficult to reconcile a total Caracas death toll of 20,000 (Forrest, 1812) or 12,000 (Coll y Prat, 1818) or 6000 (Parra Pérez, 1939). Even if the La Candelaria parish death toll is increased to 100 fatalities and applied to each parish and the 500 deaths from the San Carlos Military Headquarter are added, the total number of deaths for the area of 19 parishes is 2400 people. The conservative figure of Roscio (1812) and the approximate value of 3000 by Semple (1812) are close to this calculation. The larger 100 deaths per parish is approximately a 20% increase over the funeral records from the La Candelaria parish. The 38 deaths from the San Pablo parish funeral book are more than 50% less than that of the La Candelaria parish. The total number of deaths in the earthquake across Caracas would be an even lower number if we assumed 38 deaths from each of 19 parishes. Utilizing an average fatality rate from the funeral books of 60 deaths per parish yields a death toll estimation of 1640. Together, an average of these calculations suggests that ~2000 people died in Caracas in the earthquake (Altez, 2005a, 2006). INTENSITIES The widespread destruction of Caracas city was heterogeneous in its distribution. Damage was determined largely by the differences in the construction quality and by the maintenance status of the buildings. A spatial relationship between damage and social class therefore does not exist (Cunill Graü, 1987). The condition of the building at the time of the earthquake explains why there were churches with more or less damage. Some buildings

The social and material impacts of the 1812 earthquake in Caracas, Venezuela were virtually unharmed, and others totally collapsed. There were houses that resisted shaking and others that acted as the burial ground of the people. Estimation of the intensity of the earthquake is more difficult if the extent of damage is related to construction age and quality. The heterogeneous destruction of the buildings and a lower death count found in one historical record for Caracas can now be better appreciated. Table 4 summarizes the reported structural damage in Caracas. Values range from 33% to 90% of the buildings destroyed. These values appear as broad estimations of the areal percentage of damage. Modern assignment of Modified Mercalli scale (MMS) intensity values for Caracas are based on the historical damage records and include MMS X (FUNVISIS, 1997), MMS IX (Fiedler, 1972; Grases, 1990; Altez, 2005b), and MMS VIII–IX (Altez, 2005a). Thorough investigation of the historical sources suggests a lowering of the intensity values (Altez, 2005a). In all these observations, the earthquake intensity values for Caracas are high. However, the Modified Mercalli scale allows only a general perspective of the macroseismic effects. According to the map presented in this work (Fig. 5), it is possible to assign intensities to each damage site reported in the city using the EMS scale. This map shows that the damage distribution varies and can lead to a microzonation of intensity values. The largest variety of building responses to the earthquake is around the Main Square. In the north area, the damage level is V in the EMS scale. For the rest of the city, EMS values of IV are based on the analyses of the quantity and level of destroyed houses. Toward the north, the city felt a tremor that produced general damages with an EMS intensity of V. Most of the city of Caracas felt EMS intensity IV values (Altez, 2005a). CONCLUSIONS From the analyses of the historical data, the following conclusions are drawn: 1. The intensity values for the earthquake of 26 March 1812 in the city of Caracas have been overestimated by previous investigators. The reason for the higher intensity values is the lack of a critical evaluation of the historical context of the primary sources and the initial conditions of the buildings at the time of the earthquake.

55

2. The disaster was not the action of only one negative force (i.e., the earthquake). The earthquake occurred during a time of war and at a critical transition point in Venezuelan political, economic, and societal history. The war, the paradigmatic change represented by the pass from colonial society to forced modern institutions, and the social and political ambition of the criollos produced a vulnerable historical context. 3. The damage distribution from the earthquake of 1812 in Caracas is heterogeneous and determined by the quality and state of construction. This explains why the damage does not directly correlate with the constructions of the social classes or with the size of the construction. 4. The number of earthquake victims in Caracas in 1812 may be closer to 2000, based on analyses of three funeral books from the era. This value is lower than previous estimates about the deaths cipher. 5. The number of fatalities caused by the earthquake in the city of Caracas has been generally overestimated and historically confused with the deaths caused by other factors that were occurring concurrently, including war, migration, and famine. 6. A greater number of deaths is attributed to collapse of houses and not of religious and civic buildings. Most inhabitants were at the cathedral in celebration of Holy Thursday. Those inhabitants who remained in domestic dwellings were more likely to be killed because the roofs of houses were high, heavy, and had fragile support, making them very susceptible to collapse. 7. A seismic microzoning analysis of the city shows damages distributed among intensity values III, IV, and V on the EMS scale, with 46% of buildings and 60% of houses with damage V, while 46% of the buildings show damage IV. 8. The north area of the city of Caracas experienced higher seismic intensities (EMS V) compared to the rest of the city (EMS IV). This conclusion was also reached using similar methods by Altez (2005a), Yamazaki et al. (2005), and Schmitz et al. (2005, 2008). ACKNOWLEDGMENTS Thanks are due to Franck Audemard, André Singer, Jaime Laffaille, and Franco Urbani for their advice and lessons in seismology and geology. Special thanks go to Tina Niemi for her effort in the corrections to and observations about the manuscript. This article is a contribution to the United Nations Educational,

TABLE 4. DAMAGE ESTIMATIONS FROM THE 1812 EARTHQUAKE IN THE CITY OF CARACAS Damages Source 9/10 of the city destroyed Delpeche, mentioned in Centeno Graü (1940), 15 May 1813 3000 destroyed houses Larrain (1958) Almost all the churches and 2/3 of the houses Urquinaona y Pardo (1820) 50% of the city ruined Méndez (1957, originally 1816) 7 churches completely ruined and the rest can be repaired Linares (1816) 1/3 of the houses completely ruined 1/3 of buildings fallen Coll y Prat (1960) 50% of houses fallen Díaz (1829) 8/10 of the city destroyed Ibarra (1862 )

Figure 5. Detailed damages of Caracas. The dark circles show the most destructive effects (V in “Classification of damage to masonry buildings” from EMS scale), and the light circles show the minor damages (III–IV).

The social and material impacts of the 1812 earthquake in Caracas, Venezuela Scientific and Cultural Organization–funded International Geoscience Programme IGCP 567, “Earthquake Archaeology: Archaeoseismology along the Alpine-Himalayan Seismic Zone.” REFERENCES CITED Historical Primary and Secondary Sources Actas del Cabildo de Caracas, 1972, Volume II, 1812–1814: Caracas, Venezuela, Concejo Municipal del Distrito Federal, Tipografía Vargas, 412 p. Ascanio, 1813, Antonio Ascanio, Autobiografía, may be from 1813: National Academy of History Archive, Francisco Javier Yanes Section, volume 28. Book of Burials, Adults, San Pablo Parish, 1808–1812, Book IX, 1808–1812: Caracas, Venezuela, Archdiocese Archive of Caracas, Episcopal Section. Book of Burials, Chacao Parish, No. 2, 1797–1821, Book 2, 1797–1821: Caracas, Venezuela, Archdiocese Archive of Caracas, Episcopal Section. Book of Burials, La Candelaria Parish, No. 7, 1806–1817, Libro de Entierros de la Parroquia La Candelaria: Book 7, p. 1806–1817. Coll y Prat, manuscript from March 31, 1812, Narciso Coll y Prat, Circular a Todas las Parroquias y Pueblos del Arzobispado de Caracas: Caracas, Venezuela, Archdiocese Archive of Caracas, Parishes Section. Coll y Prat, 1957, April 18, 1816: Narciso Coll y Prat, in Los Desastres del Terremoto de 1812: Crónica de Caracas, v. 32, p. 535–536. Coll y Prat, N., 1960, Exposición al rey, 1818, in Memoriales de la Independencia de Venezuela: Caracas, Venezuela, Biblioteca de la Academia Nacional de la Historia, p. 85–386. Consulate answer to Intendant, January 29, 1813, Archivo General de Indias (General Indias Archive): Caracas, Venezuela, File 824. Delpeche, L., 1813, Relación del último terremoto de Caracas: Journal de Paris, 15 May 1813, p. 1. Díaz, J.D., 1817, A los autores y agentes del 19 de abril: Gaceta de Caracas (journal from Caracas), May 21, p. 1027–1034. Díaz, J.D., 1829, Recuerdos sobre la Rebelión de Caracas: Madrid, Imprenta de León Amarita, 408 p. Dionisio Franco, manuscript dated in February 13, 1813, Archivo General de Indias (General Indias Archive): Caracas, Venezuela, File 824. Duane, W., 1968, Viaje a la Gran Colombia en los años 1822–1823: De Caracas y La Guaira a Cartagena, por la Cordillera hasta Bogotá, y de aquí en adelante por el Río Magdalena: Caracas, Venezuela, Instituto Nacional de Hipódromos, 372 p. Forrest, 1812, Forrest (captain) to admiral Stirling, Curazao, March 30, 1812, in Parra Pérez, C., 1939, Historia de la Primera República en Venezuela, Caracas, Volume II: Caracas, Venezuela, Tipografía Americana, p. 211–212. Guerra, F.-X., 1992, Modernidad e Independencia. Ensayos Sobre las Revoluciones Hispánicas: México DF, México, Fondo de Cultura Económica, 407 p. Guzmán, April 26, 1816, Juan Joseph Guzmán, in Los Desastres del Terremoto de 1812: Crónica de Caracas, 1957, v. 32, p. 537. Heredia, J.F., 1895, Memorias Sobre las Revoluciones de Venezuela: Paris, Librería de Garnier Hermanos, 304 p. Ibarra, A., 1862, Temblores y Terremotos: El Independiente (journal from Caracas), April 1862, no. 587, p. 3–4. Irvine, 1818, John Baptiste Irvine, cited by Robertson, W.S., 1918, Francisco de Miranda y la Revolución de la América Española: Bogotá, Colombia, Biblioteca de Historia Nacional, p. 335. Ker Porter, Sir R., 1997, Diario de un Diplomático Británico en Venezuela, 1825–1842: Caracas, Venezuela, Fundación Empresas Polar, Caracas, 1039 p. Larrain, J.B., 1958, Representación ante el Muy Ilustre Ayuntamiento, 15 de febrero de 1813: Boletín de la Academia Nacional de la Historia, v. 162, p. 122–127. Linares, 1816, Pablo Linares, May 18, 1816, in Los Desastres del Terremoto de 1812: Crónica de Caracas, 1957, v. 32, p. 546–548. Lynch, J., 1985, Las revoluciones hispanoamericanas, 1808–1826: Barcelona, Spain, Editorial Ariel, 435 p. Méndez, 1957, April 18, 1816: Silvestre José Méndez, in Los Desastres del Terremoto de 1812: Crónica de Caracas, 1957, v. 32, p. 550–551. Monteverde, manuscript from August 21, 1812, Domingo de Monteverde, Ordenes a los Habitantes de la Provincia de Caracas: General National

57

Archive, Gobernación y Capitanía General section, volume CCXX, Doc. 214, folio 310. Palacio Fajardo, M., 1817, An account of the earthquake of Caracas: The Quarterly Journal of Science, London, v. 2, p. 400–402. Parra Pérez, C., 1939, Historia de la Primera República en Venezuela, Volume II: Caracas, Venezuela, Tipografía Americana, 520 p. Rojas, A., 1879, La catástrofe de 1812: La Opinión Nacional (journal from Caracas), July 12, p. 2. Roscio, 1812, Juan Germán Roscio to Luis López Méndez, Caracas, April 9, 1812: John Boulton Found Archive, Microfilmed Documents, rol c-13. Scott to Monroe, manuscript, Caracas, November 16, 1812, Alexander Scott to James Monroe, Baltimore: Colección de Documentos Diplomáticos, Biblioteca Nacional, Venezuela. Semple, 1812, John Semple a Mathew Semple, Tócome, April 3, 1812, in Tres Testigos Europeos de la Primera República, 1974: Caracas, Ediciones de la Presidencia de la República, p. 86–89. Urquinaona y Pardo, P., 1820, Relación documentada del origen y progresos del trastorno de las provincias de Venezuela hasta la exoneración del Capitán General Don Domingo de Monteverde hecha en el mes de diciembre de 1813 por la guarnición de la plaza de Puerto Cabello: Madrid, Imprenta Nueva, 322 p.

Specialized Literature Altez, R., 1998, Cronometrización extemporánea: Los sismos del 26 de marzo de 1812 en Caracas y Mérida: Revista Geográfica Venezolana, v. 39, no. 1–2, p. 297–325. Altez, R., 2005a, El terremoto de 1812 en la ciudad de Caracas: Un intento de microzonificación histórica: Revista Geográfica Venezolana, Special Issue, p. 171–198. Altez, R., 2005b, Los sismos del 26 de Marzo de 1812 en Venezuela: Nuevos aportes y evidencias sobre estos eventos: Boletín Técnico IMME (Instituto de Materiales y Modelos Estructurales de la Universidad Central de Venezuela), v. 43, no. 2, p. 11–34. Altez, R., 2006, El desastre de 1812 en Venezuela. Sismos, vulnerabilidades y una patria no tan boba: Caracas, Venezuela, Universidad Católica Andrés Bello-Fundación Empresas Polar, 522 p. Altez, R., and Laffaille, J., 2006, La microzonificación sismo-histórica como complemento fundamental de la evaluación de la amenaza sísmica: Revista de la Facultad de Ingeniería, v. 21, no. 4, p. 117–127. Altez, R., Parra, I., and Urdaneta, A., 2005, Contexto y vulnerabilidad de San Antonio de Gibraltar en el siglo XVII. Una coyuntura desastrosa: Boletín de la Academia Nacional de la Historia, v. 352, p. 181–209. Audemard, F.A., 1993, Néotectonique, Sismotectonique et Aléa Sismique du Nord-Ouest du Vénézuéla (Système de failles d’Oca-Ancón) [Ph.D. thesis]: Montpellier, Université Montpellier II, France, 369 p. + appendix. Audemard, F.A., 1998, Evolution géodynamique de la façade Nord Sud-Américaine: Nouveaux apports de l’histoire géologique du Bassin de Falcón, Vénézuéla, in Transactions of the 3rd Geological Conference of the Geological Society of Trinidad and Tobago and the XIV Caribbean Geological Conference, Trinidad, 1995, Volume 2: San Fernando, Trinidad and Tobago, Geological Society of Trinidad and Tobago, p. 327–340. Audemard, F.A., and Singer, A., 1996, Active fault recognition in northwestern Venezuela and its seismogenic characterization: Neotectonic and paleoseismic approach: Geofísica Internacional, v. 35, no. 3, p. 245–255. Audemard, F.A., Machette, M., Cox, J., Hart, R., and Haller, K., 2000, Map and Database of Quaternary Faults in Venezuela and Its Offshore Regions: U.S. Geological Survey Open-File Report 00-18, 79 p. + map. Audemard, F.A., Romero, G., Rendón, H., and Cano, V., 2005, Quaternary fault kinematics and stress tensors along the southern Caribbean from microtectonic data and focal mechanism solutions: Earth-Science Reviews, v. 69, no. 3–4, p. 181–233, doi: 10.1016/j.earscirev.2004.08.001. Audemard, F.E., and Audemard, F.A., 2002, Structure of the Mérida Andes, Venezuela: Relations with the South America–Caribbean geodynamic interaction: Tectonophysics, v. 345, p. 299–327, doi: 10.1016/S0040 -1951(01)00218-9. Beltrán, C., 1994, Trazas activas y síntesis neotectónica de Venezuela a escala 1:2.000.000, in VII Congreso Venezolano de Geofísica: Caracas, Venezuela, Sociedad Venezolana de Ingenieros Geofísicos, p. 541–547. Centeno Graü, M., 1940, Estudios Sismológicos: Caracas, Academia de Ciencias Físicas, Matemáticas y Naturales, 365 p.

58

Altez

Cunill Graü, P., 1987, Geografía del Poblamiento Venezolano en el Siglo XIX: Caracas, Venezuela, Ediciones de la Presidencia de la República, 3 volumes. European Seismological Commission, Subcommission on Engineering Seismology, Working Group Macroseismic Scales, European Macroseismic Scale, 1998, EMS-98: Luxembourg, European Seismological Commission: http://www.gfz-potsdam.de/portal/gfz/Struktur/Departments/ Department+2/sec26/projects/04_seismic_vulnerability_scales_risk/ EMS-98 (accessed 21 February 2010). Fiedler, G., 1972, La liberación de energía sísmica en Venezuela, volúmenes sísmicos y mapas de isosistas, in Memorias del IV Congreso Geológico Venezolano, Volume IV: Caracas, Venezuela, p. 2441–2462. FUNVISIS (Fundación Venezolana de Investigaciones Sismológicas), 1997, Estudio neotectónico y geología de fallas activas en el Piedemonte surandino de los Andes venezolanos, Proyecto INTEVEP 95-061: Caracas, FUNVISIS, 155 p. Grases, J., 1990, Terremotos Destructores del Caribe 1502–1990: Montevideo, UNESCO-RELACIS, 132 p. Grases, J., and Rodriguez, J.A., 2001, Estimaciones de magnitud de sismos venezolanos a partir de mapas de isosistas, in Memorias del 2nd Seminario Iberoamericano de Ingeniería Sísmica: Madrid, Spain, 12 p., digital version on CD. Grases, J., Altez, R., and Lugo, M., 1999, Catálogo de sismos sentidos y destructores, Venezuela, 1530–1998: Caracas, Venezuela, Academia de Ciencias Físicas, Matemáticas y Naturales, Facultad de Ingeniería, Universidad Central de Venezuela, Editorial Innovación Tecnológica, 654 p. Guidoboni, E., and Ferrari, G., 2000, Historical variables of seismic effects: Economics levels, demographic scales and buildings techniques: Annali di Geofisica, v. 43, no. 4, p. 687–705. Mendoza Solar, E., 1910, Plano de la Ciudad de Santiago de León de Caracas en el año 1810 (Map of Caracas near 1810): Caracas, Venezuela, Litografía del Comercio, scale 1:1000, 1 sheet. Mocquet, A., 2005, Geological and architectural context of historical earthquakes in eastern Venezuela: Journal of Earthquake Engineering, v. 9, no. 1, p. 129–146, doi: 10.1142/S136324690500175X.

Olson, R.S., and Grawonski, V.T., 2003, Disasters as critical junctures? Managua, Nicaragua, 1972, and Mexico City, 1985: International Journal of Mass Emergencies and Disasters, v. 21, no. 1, p. 5–36. Rodríguez, J.A., and Audemard, F.A., 2003, Sobrestimaciones y limitaciones en los estudios de sismicidad histórica con base en casos venezolanos: Revista Geográfica Venezolana: Universidad de Los Andes, v. 44, no. 1, p. 47–75. Schmitz, M., Hernández, J., Audemard, F., Malavé, G., and Andrade, L., 2005, Proyecto de Microzonificación Sísmica en las ciudades Caracas y Barquisimeto: Serie Técnica Fundación Venezolana de Investigaciones Sismológicas No. 1, p. 260–263. Schmitz, M., Hernández, J.J., Morales, C., Molina, D., Valleé, M., Domínguez, J., Delavaud, E., Singer, A., González, M., Leal, V., and el Grupo de Trabajo del Proyecto de Microzonificación Sísmica de Caracas, 2008, Resultados principales del Proyecto de Microzonificación Sísmica en Caracas, in Conferencia 50 Aniversario de la Sociedad Venezolana de Geotecnia (SVDG), 6 al 9 de noviembre 2008: Caracas, Venezuela, Sociedad Venezolana de Geotecnia, 11 p. Schubert, C., 1984, Basin formation along Boconó–Morón–El Pilar fault system, Venezuela: Journal of Geophysical Research, v. 89, p. 5711–5718, doi: 10.1029/JB089iB07p05711. Soulas, J.-P., 1986, Neotectónica y tectónica activa en Venezuela y regiones vecinas, in Memorias del VI Congreso Geológico Venezolano, Tomo 10: Caracas, Venezuela, Sociedad Venezolana de Geólogos, p. 6639–6656. Yamazaki, Y., Audemard, F.A., Altez, R., Hernández, J., Orihuela, N., Safina, S., Schmitz, M., Tanaka, I., Kagawa, H., and Jica Study Team—Earthquake Disaster Group, 2005, Estimation of the seismic intensity in Caracas during the 1812 earthquake using seismic microzonation methodology: Revista Geográfica Venezolana, Special Issue, p. 199–216.

MANUSCRIPT ACCEPTED BY THE SOCIETY 19 MAY 2010

Printed in the USA

The Geological Society of America Special Paper 471 2010

The impact of the 1157 and 1170 Syrian earthquakes on Crusader– Muslim politics and military affairs Kate Raphael Institute of Earth and Sciences, The Hebrew University, Jerusalem 91904, Israel

ABSTRACT This paper examines the development of a crisis over a critical military-security issue raised by the severe earthquakes that destroyed defensive structures throughout Nur al-Din’s Sultanate of Syria, the Crusader Principality of Antioch, and the County of Tripoli. The earthquakes that struck Syria in 1157 and 1170 are well documented by contemporary historians. The accounts of destruction concentrate on the collapse of many fortresses and town walls. This circumstance strongly influenced regional politics and military affairs. While the first earthquake led to an increase in tension and a rise in violence between the Crusader Kingdom of Jerusalem and the Muslim Sultanate in Syria, the destruction wrought by the 1170 earthquake forced the two sides to accept a formal peace treaty. The two case studies presented here examine the impact of earthquake destruction on decision makers in the complex international arena of medieval Syria.

role in the way each case developed and the decisions made by the regional rulers. The earthquakes in medieval Syria have been studied by several scholars, who have used the historical sources to enable them to understand the seismic dynamics and assess earthquake hazards (Ambraseys and Jackson, 1998; Amiran, 1952; Amiran et al., 1994; Ben-Menahem, 1979, 1991; Guidoboni et al., 2004a, 2004b; Sbeinati et al., 2005). The impact of earthquakes on regional affairs has seldom been examined (Tucker, 1981, 1999; Little, 1999).

INTRODUCTION Most of the political and military conflicts between the Crusader kingdom and principalities and the Muslim Sultanate of Syria during the second half of the twelfth century were related to religious tensions and territorial disputes. In several cases, however, crisis was not triggered by extreme acts or shifts in the religious or political views of regional leaders, but rather by severe natural disasters, such as long periods of drought, crop failure, and earthquakes. The aim of this paper is to examine the impact of the 1157 and 1170 earthquakes that struck Syria on the political and military affairs of the region. Did such events cause an increase or decrease in the level of animosity and violence? What forces influenced decision makers? Who was responsible for the reconstruction of private and public property? The two earthquakes struck at almost the same sites; each set off a chain of political and military reactions. However, each crisis developed in a very different manner. The force of the earthquake, the scale of damage, and its geographical distribution played an important

SHORT HISTORICAL BACKGROUND The area under discussion is what was known in contemporary Arabic sources as Bilad al-Sham (Greater Syria), corresponding to the modern states of Syria, Lebanon, southeast Turkey, northern Jordan, and northern Israel. The narrow strip along the coast was ruled by the Crusader Principality of Antioch and the County of Tripoli. The Kingdom of Jerusalem ruled the

Raphael, K., 2010, The impact of the 1157 and 1170 Syrian earthquakes on Crusader–Muslim politics and military affairs, in Sintubin, M., Stewart, I.S., Niemi, T.M., and Altunel, E., eds., Ancient Earthquakes: Geological Society of America Special Paper 471, p. 59–66, doi: 10.1130/2010.2471(06). For permission to copy, contact [email protected]. © 2010 The Geological Society of America. All rights reserved.

59

60

Raphael

territory that partly correlates with the modern state of Israel and the Palestinian Authority; it also controlled some of the lands south of the Dead Sea in the region of the fortresses of Karak, today southern Jordan. The Muslim territory included the cities of central Syria from Aleppo in the north to Damascus. During the 1140s, the individual Muslim principalities in the region were slowly being united under the house of Zengi. Lands that had been conquered by the Crusaders were gradually being recovered. In 1144, ‘Imad al-Din Zengi, ruler of Mosul (d. 1146), captured the Crusader County of Edessa (modern Ruha, southeast Turkey) and established his rule over Aleppo. A decade later his son, Nur al-Din (1118–1174) had seized Damascus and made it his capital (Prawer, 1984; Runciman 1994; Riley-Smith, 1990). During the 1160s, the interests of the political entities in the region shifted far south, and the struggle between the Crusader Kingdom and the Syrian Sultanate of Nur al-Din was waged in Egypt. Egypt’s declining Fatimid dynasty (a Shiʿite dynasty) and its wealth from international trade and agriculture made it an attractive prize. Amalric (1136–1174), ruler of the Crusader Kingdom of Jerusalem, fought against Nur al-Din’s armies for the control of Cairo and the Nile Delta for almost a decade. The Crusader armies carried out five successive attacks; on one occasion they allied themselves with the Byzantine emperor in order to receive the assistance of his fleet. The Crusader assaults failed, and Amalric eventually had no choice but to acknowledge defeat. Salah al-Din (Saladin), who was one of the leading officers in this campaign, became the administrator of the newly conquered territory in Egypt. Officially, he remained a dependent of Nur al-Din. By 1170, the geopolitical balance in the region had changed considerably. Nur al-Din’s rule over Syria was consolidated, and Egypt was administered under his suzerainty. The earthquake of 1170 thus struck after a decade of intensive fighting in which large Muslim and Christian armies had constantly been on the march back and forth from Syria or the Kingdom of Jerusalem to the Nile Delta and Cairo. Lengthy sieges were conducted, and a number of open-field battles took place (Prawer, 1984; Runciman, 1994). CONTEMPORARY WRITTEN EVIDENCE ON EARTHQUAKE DAMAGE OF 1157 AND 1170 Earthquakes are habitually described by contemporary eyewitnesses in a dramatic tone of voice. Christian and Muslim chroniclers often use the same phrase “a great and terrible earthquake, far more violent than any other within the memory of men now living” (William of Tyre, v. 2, p. 370–371). This is by no means the language of exaggeration, but some caution is necessary, and more than one contemporary source is required in order to verify each case. The destruction of urban defenses and the repairs that followed are described in detail for both 1157 and 1170. Although the number of casualties was high and the damage to private property and public buildings was considerable, the rulers were more concerned with the state of their fortifications and defenses.

In 1157, town walls, towers, citadels, and fortresses were damaged throughout Nur al-Din’s territories. There is sufficient evidence to show that in the neighboring Crusader Principality of Antioch and the County of Tripoli, the scale of damage was significantly lower. The only repairs carried out by the Crusaders due to earthquake damage were to the large fortress of Hisn al-Akrad (Crac des Chevaliers) held by the Order of the Hospitallers. The Grand Master of the Order, Raymond of Le Puy, received a generous donation from Wladislas II, King of Bohemia; this financed the reconstruction works (Elisséeff, 1986). Strangely enough, the main Crusader source for this period, William of Tyre, does not mention the 1157 earthquake at all. Ibn al-Jawzi (1126–1200), the author of al-Muntazam fi Tarikh al-Umam wal-Muluk (The Order of History, the Nations and the Kings), belonged to the intellectual elite of Baghdad and was one of the most influential people in this circle (Laoust, 1986). Although he himself lived in Baghdad, he was well informed and gives a detailed report of the cities that were hit by this earthquake (Fig. 1; Table 1). He is clearly more concerned with the suffering of the local population than with the damage caused to public and domestic building.

…thirteen cities were destroyed in this earthquake, eight in the land of Islam and five in the land of the infidels. In the land of the Muslims: Aleppo, Hama, Shayzar, Kafar-Tab, Aphamia, Hims, al-Maʿarra, and Tell-Harān; and in the land of the Franks: Hisn al-Akrad, ʿArqa, Lattakia, Tripoli and Antioch. At Aleppo one hundred people were killed; as for Hama few survived and in Shyzar only a woman and a slave survived; all the rest perished. At Kafar-Tab only one person survived and at Aphamia the citadel sank. At Homs many scholars died and at Maʿarra a number of people were killed. Tell-Harān was divided in two, exposing the interior of tombs and many houses. At Hisn al-Akrad and ʿArqa all was destroyed and at Lattakia all that remained was one man and a spring and in it there was a hole with mud in its center [in which] stood a statue. Much of Tripoli was ruined and only parts of Antioch remained. (Ibn al-Jawzi, Muntazam, 1992, v. 18, p. 119)

The following is an extract from an account written by the Muslim chronicler Ibn al-Athīr (1160–1233), author of the al-Kāmil fi’l-ta’rīkh (The Complete Work of History). Ibn al-Athīr spent most of his life in Mosul. In later years, he moved to Aleppo. In contrast to Ibn al-Jawzi, he pays greater attention to the urban defenses. In Rajab this year (9 August–7 September 1157) there were many strong earthquakes in Syria, which destroyed much of the country and which caused the death of more people than could be counted. In one moment Hama, Shayzar, Kafar-Tāb, al-Maʿarra, Homs, H ⋅ is⋅ n al-Akrād, ʿArqa, Lattakia, Tripoli and Antioch were ruined. All Syria suffered damage in most of its parts, even if the damage was not total. City walls and citadels were demolished. Nūr al-Dīn Muh⋅ mūd dealt with this in an exemplary manner. He feared for the land since the city walls had been destroyed. He assembled the troops and camped on the frontiers of his land, carrying raids on Frankish territory, while working on the walls in the rest of his lands. He kept this up until he had completed

Impact of the 1157 and 1170 Syrian earthquakes on Crusader–Muslim politics and military affairs

61

Figure 1. Towns and fortresses struck by the 1157 earthquake according to Ibn al-Jawzi’s account.

all the city walls. The great number of people who were killed is sufficiently indicated by the fact that a teacher who was in his town, namely Hama, left the Koran school for some matter of business that occurred, when the earthquake came and destroyed the town. The school collapsed on all the children. The teacher said, “Not a single person came to enquire after any child of his.” (Ibn al-Athīr, 2007, v. 2, p. 87)

TABLE 1. DISTRIBUTION OF DAMAGE FROM THE 1157 EARTHQUAKE, ACCORDING TO IBN AL-JAWZI Muslim Sultanate Crusader Principality Crusader County of Nur al-Din of Antioch of Tripoli Aleppo Antioch ૚iଙn al-Akrād Hama Lattakia Tripoli Shayzar

ȾArqa

Aphamia

According to Ibn al-Athīr, while repairing the fortifications of the central Syrian cities, Nur al-Din organized raiding contingents and ordered them to attack the neighboring Crusader territories. He was not aiming at long-term political or territorial achievements, but rather it seems he feared that Baldwin III (r. 1152–1163), ruler of the Kingdom of Jerusalem, might take advantage of the poor state of the defenses and launch an attack in order to destabilize his rule and conquer part of his newly acquired lands. These fears were not unfounded, for in fact the Crusader principalities had not suffered severe damage, and the

Kafar-Tab al-MaȾarra Hims Tall-Harān

Kingdom of Jerusalem lay outside the zone of the earthquake. Baldwin did not lose much time; a large army was assembled and joined by the forces of the Prince of Antioch and the Count of Tripoli. The Count of Flanders, who was visiting the Holy Land, joined the campaign with his own men. This army entered

62

Raphael

the Sultanate and attacked the Muslim fortress of Qalʿat Yahmur (Chastel Rouge, roughly 20 km northeast of Tripoli), but to no avail; the garrison held its ground (William of Tyre, 1943, v. 2, p. 265). Ibn Qalanisi (d. 1160), a contemporary Muslim chronicler who lived in Damascus, says Nur al-Din was quick to act and recruited a Muslim army to meet them.

Mention has already been made of the departure of al-Malik al-ʿAdil Nur al-Din from Damascus with his troops towards the cities of Syria, on receipt of news that the factions of the Franks (God forsake them) were assembling together and proceeding against them, being emboldened to attack them by reason of the continuous earthquakes and shocks which afflicted them and of the destruction wrought amongst the castles, citadels and dwellings in their districts and marches. [Nur al-Din therefore took measures] to protect and defend them and to bring solace to those of the men of Hims, Kafr Tab, Hamah and elsewhere who had escaped with their lives, whereupon there assembled to join him a great host and vast numbers of men from the fortresses and provincial cities and from the Turkmens. He encamped with them in exceeding force opposite the army of the Franks in the neighborhood of Antioch and encompassed them so that not one horseman of theirs could set out to make a raid. (Ibn Qalanisi, 1932, p. 340–341; my emphasis)

Soon after, the Muslim army dispersed due to the illness of Nur al-Din. Many thought he was on his deathbed and left the field since there was no strong commander able to replace him. Once the siege around the Crusader army was broken, it resumed its march and launched an attack on the fortress of Shayzar. The fortress was saved thanks to a dispute that broke out among the Christian commanders, and a strong Isma’ili force that managed to defeat the Franks (Ibn Qalanisi, 1932, p. 342; William of Tyre, 1943, v. 2, p. 226–268). The Crusader forces returned to their own lands empty-handed. 1170, THE EARTHQUAKE THAT FORCED PEACE BETWEEN THE CRUSADER PRINCIPALITIES AND NUR AL-DIN, THE SULTAN OF SYRIA The evidence for the 1170 earthquake presented next is based on two contemporary historians. William, Archbishop of Tyre (ca. 1130–1185), whose work “The Deeds beyond the Sea” is considered one of the most reliable sources for the study of the Crusader Kingdom of Jerusalem, was probably born in Jerusalem; he spoke the local languages and was familiar with the local cultures. He received his higher education in Europe. On his return, he became close to the royal court in Jerusalem and served as a tutor to Baldwin IV, who later became king. From 1175 until his death, he was both Archbishop of Tyre and Chancellor of the Crusader Kingdom. The second source is Ibn al-Athīr, mentioned previously. In writing about the 1170 earthquake, both Ibn al-Athīr and William of Tyre emphasize the great damage to fortifications. They both describe the deep concern of each ruler for the defense of his territories.

In contrast to the 1157 earthquake, which William ignored, suggesting that the Crusader principalities were hardly damaged, here he describes the destruction in detail.

Strongly fortified cities dating from very early times were completely demolished….The largest cities of our provinces and those of Syria and Phoenicia as well, cities famous throughout the ages for their noble antiquity were prostrated. In Coelesyria, Antioch, the metropolis of several provinces and once the head of many kingdoms, was utterly overwhelmed and its entire population destroyed. The massive walls and the immensely strong towers along their circuit fell in ruins. Churches and buildings of every kind were thrown down with such violence that even now, although much labor and expense have been devoted to their restoration, they are only partially repaired. Among other places destroyed in that same province were Gabala and Laodicea, famous cities on the coast. Of the cities further inland which were still held by the enemy there were destroyed Beroea, also known as Aleppo, Shayzar, Hama, Hims and others. The number of fortresses wrecked was beyond counting. …the great and populous city of Tripoli was suddenly shaken by a violent earthquake, and scarcely a person within the walls escaped. The entire city was reduced to a heap of stones and became a burial place and common sepulcher of the citizens who perished with it. At Tyre, the most famous city of the province, the earth movement was so violent that several massive towers were overthrown. There was, however, no loss of life here. (William of Tyre, 1943, v. 2, p. 370–371)

William ends his account with a sigh of relief, reassuring his readers that his Muslim neighbors were facing similar troubles. He mentions the damage to the main towns of Syria, Aleppo, Shayzar, Hama, and Hims, and he remarks that many other towns lay in ruins (Fig. 2; Table 2). Ibn al-Athīr’s description conveys the fear and urgency that filled the Syrian ruler as he surveyed the poor state of his fortifications. Nur al-Din quickly organized garrisons to safeguard the towns where the defenses had been destroyed. In some towns, the horrendous scale of destruction and the aftershocks drove the citizens away; certain sites were completely abandoned. The sultan then set out and personally supervised the construction of some of the main citadels in the towns along the frontier with the Crusaders.

When Nūr al-Dīn received the news, he went to Baalbek to repair the damage to its wall and citadel. When, however, the news from the rest of the towns came to him, news of the destruction of their walls and citadels and their abandonment by the inhabitants he placed men in Baalbek to repair, protect and guard it and went to Homs, where he did the same, and then to Hama and then to Baʿrin. He was extremely wary of the danger for the towns from the Franks. Then he came to Aleppo, where he saw effects of the earthquake greater than elsewhere, for it had destroyed it utterly and the survivors were totally terror stricken. They were unable to shelter in their houses for fear of aftershocks. They remained in the open. Nūr al-Dīn personally took part in the repair work and so continued until he had rebuilt its walls and mosques. (Ibn al-Athīr, 2007, v. 2, p . 185–186)

Impact of the 1157 and 1170 Syrian earthquakes on Crusader–Muslim politics and military affairs

63

Figure 2. Towns and fortresses struck by the 1170 earthquake according to William of Tyre’s account.

TABLE 2. THE DISTRIBUTION OF DAMAGE FROM THE 1170 EARTHQUAKE, ACCORDING TO WILLIAM OF TYRE AND IBN AL-ATHĪR Muslim Sultanate Crusader Principality Crusader County of Nur al-Din of Antioch of Tripoli Hama Antioch Tripoli Shayzar

Lattakia

Baalbek

Gabala

It is evident that each side was aware of the fact that the neighbor’s fortifications were in a state of ruin. Nevertheless, one is left with a strong impression that each ruler suspected that his enemy might take advantage of the situation and launch a surprise attack. This notion is clearly conveyed by Ibn al-Athīr.

Tyre

Hims Aleppo Beroea (BaȾrin)

Ibn al-Jawzi’s report on the earthquake is brief and his account only mentions Aleppo.

Half of Aleppo collapsed and it was said that eighty thousand people were killed. (Ibn al-Jawzi, 1992, v. 18, p. 188; my translation)

As for the Frankish territory, the earthquake tremors also had the same effect there. They were kept busy repairing their towns, fearful of Nūr al-Dīn for them. Each side was occupied with repair work for fear of the other. (Ibn al-Athīr, 2007, v. 2, p. 186)

Yet, in order to ensure that no side would make a move against the foe and set out to raid or launch a full-scale attack, a treaty of some definition was necessary. Decision makers in times of crisis are subject to both external and internal pressures that may force them to make peace

64

Raphael

(Randle, 1973). Nur al-Din is described as committed to the security of his lands and their inhabitants. William of Tyre presents the treaty as a decision that was made out of sheer fear: fear of Nur al-Din and fear of God.

Both in our territories and in those of the enemy were found halfruined fortresses, open on every side and freely exposed to the violence and the wiles of the foe. But since each man feared that the wrath of the Stern Judge might descend upon him individually, none dared molest his fellow man. Each was engrossed in his own troubles and weighed down by the burden of his own affairs; hence none thought of injuring his neighbor. Peace brought about by the desire of all, ensued, albeit for a short interval, and a truce was arranged through fear of divine wrath. Each, while momentarily expecting the outpouring of righteous anger from heaven in punishment for his sins, refrained from acts of hostility and curbed his own evil impulses. (William of Tyre, 1943, v. 2, p. 371; my emphasis)

The Crusaders’ fear of Nur al-Din indicates a shift in the regional balance of power. Nur al-Din’s recent conquest of Egypt had made clear his determination and his military abilities. William does not speak of any disagreements or difficulties in concluding the truce. There was no exchange of prisoners of war; neither side was required to pay a tribute, no specific conditions were made, and no lands were demanded or exchanged. William’s Latin terminology is somewhat elusive. “Pax hominum studio procurata, et foedus compositum, divinorum iudiciorum timore conscriptum.” Pax may mean: a pact to end or avert hostilities; a pact granted by God; or a settlement; or simply peace. Foedus is a formal agreement between states or peoples. Conscriptum may refer here to a charter. The frequent mentioning of divine wrath and anger may well signify that William of Tyre was referring to a pact granted by God. Thus, on this reading, it is possible that there was no proper legal written document. William’s words and phrasing seem to indicate that this was a gentleman’s agreement, or a quiet mutual understanding, as suggested by Prawer (1984). Ibn al-Athīr’s report, on the other hand, leaves no room for speculation concerning the nature of this truce, and whether there was a formal agreement rather than a loose understanding. According to Ibn al-Athīr, a formal truce was indeed drawn up between Nur al-Din’s Sultanate, the Principality of Antioch, and the County of Tripoli. This is made clear in a detailed episode that took place not long after the truce was concluded. The term used by Ibn al-Athīr is hudna, meaning peace, truce, or armistice (Ibn al-Athīr, 1966, v. 11, p. 373–374). In the autumn of 1171, a Frankish force from Tripoli and Antioch seized two Muslim merchant ships. Nur al-Din was furious and accused the Franks of violating the truce, demanding that the merchandise be returned.

…Between them [the Franks] and Nur al-Din there was a truce which they treacherously broke. Nur al-Din sent to them about the matter

and about their restoring the merchants’ property they had taken. (Ibn al-Athīr, 2007, v. 2, p. 200)

The Franks ignored the sultan’s demand. Nur al-Din did not hesitate to act. He sent a force to raid the cities of Tripoli and Antioch, and a number of smaller fortresses in the neighborhood were sacked. Frankish territory was set ablaze, plundered, and a number of people were killed. The Muslim force returned with a large amount of booty. Following this destructive raid, the Franks reviewed their situation and decided to renew the truce.

The Franks made contact with him [Nur al-Din] and offered to restore what they had taken from the two ships and to renew the truce. This was accepted…. (Ibn al-Athīr, 2007, v. 2, p. 200)

REPAIRING THE DAMAGE The twelfth century witnessed an exceptionally high number of earthquakes that struck central Syria and the coast. The magnitude of the major 1157 earthquake has been estimated at 7–7.8 (Amiran et al., 1994; Ben-Menahem, 1979; Guidoboni et al., 2004b). Four strong shocks preceded the 1157 late summer earthquake; two were felt in April and two in July. The latter caused some damage in Shayzar, Hama, Kafr Tab, Aphamia, and the area of Aleppo (Ibn Qalanisi, 1932, p. 328–329). These smaller events are well documented by Ibn Qalanisi. The 1170 earthquake has been graded 7–8 (Amiran et al., 1994; BenMenahem, 1979; Guidoboni et al., 2004a). The reconstruction of large-scale fortifications was a long and expensive process. The frequent earthquakes, the tremors that preceded them, and the aftershocks that followed undermined many of these defenses and rendered them unstable and dangerous. The Latin sources provide little evidence of Frankish rulers intervening in the reconstruction work. The only information we have concerns the post-1157 repairs to the large fortress of Crac des Chevaliers. The funds for the reconstruction came from Europe, and the Grand Master of the Order of the Hospitallers oversaw the work. By the late twelfth century, the military orders were well established, almost entirely independent and often better financed than most of the Crusader kings, princes, and counts. On the Muslim side, matters seem better organized. Concerning the financing, it appears that Nur al-Din carried out a sound economic policy carefully adjusted to the financial difficulties of the local inhabitants. An interesting observation is made by Professor Lev in an article titled “The Social and Economic Policies of Nur al-Din.” Lev suggests that disrupted economic activity after the 1157 earthquake caused Nur al-Din to reduce the taxes throughout the sultanate (Lev, 2004). These steps no doubt helped to encourage and revive trade and commercial activity in local markets, enabling the country to recover at a faster pace.

Impact of the 1157 and 1170 Syrian earthquakes on Crusader–Muslim politics and military affairs Nur al-Din was actively involved in the survey and repair of his fortifications. His political power in Syria was stronger and more centralized than that of the Crusader rulers in the County of Tripoli and the Principality of Antioch, where presumably each town and fortress saw to its own defenses. In the eulogy of the Sultan, Ibn al-Athīr dedicates a long passage to Nur al-Din’s public works. Not surprisingly, he opens this chapter with the following passage:

As for the public works, he built the walls of the cities and castles of all Syria, for example Damascus, Homs, Hama, Aleppo, Shayzar, Baalbek and others. (Ibn al-Athīr, 2007, v. 2, p. 223)

The rebuilding of the defenses was seen as the Sultan’s responsibility. It seems that thanks to his sound economic policy, his treasury was well balanced, and the funds for the reconstruction work came from the sultanate’s coffers. AN “EMERGENCY PEACE” In 1170, Nur al-Din acted with care and caution, avoiding a military conflict. Why was the military and political response to the earthquake of 1170 different from that of 1157? What drove the two sides to accept a truce? The answer lies in the historical background and perhaps in a reassessment of the strength of the 1170 earthquake, which seems to have been considerably stronger and more destructive than that of 1157. The damage spread throughout the entire region, affecting both sides equally. Each side no doubt moved to overcome its own internal conflicts, social tensions, and disputes. A few months prior to the 1157 earthquake, the Crusader and Muslim forces were engaged in an ongoing struggle over the town and region of Banyas (at the southern foot of Mount Hermon), on the Muslim-Crusader frontier. The town itself was held by Humphrey of Toron. During the winter, the region was settled by nomadic tribes who reared large herds of horses on the rich pasture. They paid rent to the King of Jerusalem for the land they were using. In the winter of 1157, a Crusader force together with Baldwin III (the king did not initiate this act, but joined the raid) massed a large-scale attack. The nomadic population was slain, and a huge amount of booty was collected, including numerous horses. William of Tyre describes the results of this raid:

…the amount of booty taken in this raid was never equaled in our land. Yet this deed brought no glorious or laudable renown to our people, for they had violated a treaty of peace. (William of Tyre, 1943, v. 2, p. 256)

Nur-al-Din was determined to conquer Banyas. He besieged the town twice during the late spring and early summer of 1157. Both sieges failed owing to the arrival of Crusader reinforce-

65

ments. After the first siege, Baldwin and a large Crusader force were surprised by a Muslim ambush near Banyas. The king barely escaped, and a large number of his knights were taken hostage and paraded through the streets of Damascus. This period of raids, ambushes, and sieges around the locality of Banyas was interrupted by the earthquake that struck the region in the late summer. The destruction on the Muslim side was considerably greater than that in the Crusader territory. This had important military and political implications. The destruction of castles and town fortifications led to a state of emergency within the Muslim sultanate. Its influence on political and military affairs was striking: It changed the balance of power in the region, which for a short time tilted in favor of the Crusader Kingdom. The Crusaders seized the opportunity to attack their chief enemy, and they made a decisive move toward Syria. The level of the conflict was immediately upgraded. Nur al-Din was aware that the Crusader rulers had no interest or reason to negotiate a truce. He recruited an army that prevented the Crusaders from taking two of his fortresses, although it seems that the disputes among the Crusader leadership had an equally significant part in the failure of their campaign. As noted already, the earthquake of 1170 came after a decade of fighting in Egypt. Both sides were engaged in expensive and distant campaigns that involved large armies. Both seem to have exhausted their resources. The destruction of fortifications on both sides by the earthquake brought the conflict to a standstill. It forced Nur al-Din and Amalric to keep their forces at home and refrain from violent military acts. Under certain circumstances, severe natural disasters may change or tilt the balance of power and alter the course of events. In order to do so, they must be of extreme strength and leave behind them mass destruction, or threaten the lives of large populations (in the case of droughts and famines). The 1170 truce was an emergency policy, necessary to allow each side to recover, not from the damage of war, but from an earthquake that brought down the fortifications throughout the region. However, the disaster was not strong enough to change military or political ideology or concepts. William of Tyre clearly states this was to be a short-term agreement. He knew that once the fortifications were in order, their storerooms stocked, and garrisons reestablished and armed, the truce would not hold. CONCLUSIONS The aim of this paper was to examine the development of political and military affairs between the Crusader states and the Muslim Sultanate after the severe damage caused by the earthquakes of 1157 and 1170. While the first earthquake led to an increase in tension and a rise in violence, the destruction wrought by the 1170 earthquake forced the two sides to accept a formal peace treaty. Although there are a number of cases that clearly show that severe environmental disasters will force rival rulers to sign short-term peace treaties, there is no pattern or rule to the behavior and the way rulers make decisions in times of crisis.

66

Raphael

Each case must be carefully studied in the light of the wider international affairs in the region. The king’s or sultan’s personality, military strength, wealth, and experience play an important role in their ability to cope with large-scale disasters. While controlling earthquakes and eliminating the death and destruction that followed was impossible, maintaining peace or ending aggressions in times of a severe crisis were, under certain circumstances, relatively feasible policies. Some leaders ignored the damage, failed to read the political map, and missed the opportunity to reduce the violence; others were wiser, more experienced, and negotiated for a short-term peace treaty. The main benefactors were the people, for it was no doubt easier to attend to the damage and reach full recovery within a shorter period when the region was peaceful. ACKNOWLEDGMENTS The research for this paper was funded by the Galilee Project, affiliated with the Hebrew University of Jerusalem, Israel, and the University of York, England. I would like to thank Michal Kidron from the Cartographic Laboratory, Department of Geography, The Hebrew University of Jerusalem, for preparing the figures. This article is a contribution to the United Nations Educational, Scientific and Cultural Organization–funded International Geoscience Programme IGCP 567, “Earthquake Archaeology: Archaeoseismology along the Alpine-Himalayan Seismic Zone.” REFERENCES CITED Ambraseys, N.N., and Jackson, J.A., 1998, Faulting associated with historical and recent earthquakes in the Eastern Mediterranean region: Geophysical Journal International, v. 133, p. 390–406, doi: 10.1046/j.1365 -246X.1998.00508.x. Ambraseys, N.N., Melville, C.P., and Adams, R.D., 1994, The Seismicity of Egypt, Arabia, and the Red Sea: A Historical Review: Cambridge, UK, Cambridge University Press. Amiran, D.H.K., 1952, A revised earthquake catalogue of Palestine: Israel Exploration Journal, v. 2, p. 48–62. Amiran, D.H.K., Arieh, E., and Turcotte, T., 1994, Earthquakes in Israel and adjacent areas: Macroseismic observations since 100 B.C.E.: Israel Exploration Journal, v. 44, p. 260–305.

Ben-Menahem, A., 1979, Earthquake Catalogue for the Middle East (92 B.C.– 1980 A.D.): Bollettino di Geofisica Teorica ed Applicata, v. 21, no. 84, p. 245–310. Ben-Menahem, A., 1991, Four thousand years of seismicity along the Dead Sea rift: Journal of Geophysical Research, v. 96, p. 20,195–20,216, doi: 10.1029/91JB01936. Elisséeff, N., 1986, Hisn al-Akrad, in Encyclopedia of Islam, Volume 3 (2nd ed.): Leiden, the Netherlands, Brill, p. 503–506. Guidoboni, E., Bernardini, F., and Comastri, A., 2004a, The 1138–1139 and 1156–1159 destructive seismic crises in Syria, south-eastern Turkey and northern Lebanon: Journal of Seismology, v. 8, p. 105–127, doi: 10.1023/B:JOSE.0000009502.58351.06. Guidoboni, E., Bernardini, F., Comastri, A., and Boschi, E., 2004b, The large earthquake on 29 June 1170 (Syria, Lebanon, and central southern Turkey): Journal of Geophysical Research, v. 109, p. B07304, doi: 10.1029/2003JB002523. Ibn al-Athīr, I., 2007, The Chronicle of Ibn al-Athīr for the Crusading Period from Al-Kāmil fi’l-Ta’rīkh, Volume 2 (translated by D.S. Richards): Aldershot, Ashgate, 401 p. Ibn al-Athīr, I., and ʿIzz al-Dīn ʿAlī, 1966, Al-Kāmil fī’l-Ta’rīkh, Volume 11 (C.J. Tornberg ed.): Beirut, Dar Beirut, 585 p. Ibn al-Jawzi, 1992, Muntazam, Volume 18: Beirut, Dar Beirut. Laoust, H., “Ibn al-Djawzi,” 1986, Encyclopaedia of Islam, Volume 3 (2nd ed.): Leiden, the Netherlands, Brill, p. 751–752. Lev, Y., 2004, The social and economic policies of Nur al-Din (1146–1174): The Sultan of Syria: Der Islam, v. 81, p. 218–242, doi: 10.1515/islm .2004.81.2.218. Little, D.P., 1999, Data on earthquakes recorded by Mamluk historians, in Zachariadou, E., ed., Natural Disasters in the Ottoman Empire: Rethomnon, Crete University Press, p. 137–151. Prawer, J.A., 1984, History of the Latin Kingdom of Jerusalem, the Crusades and the First Kingdom: Jerusalem, Bialik Institute. Qalanisi, I., 1932, The Damascus Chronicle of the Crusades (translated by H.A.R. Gibb): London, Luzac and Co., 368 p. Randle, R.F., 1973, The Origins of Peace; A Study of Peacemaking and Structure of Peace Settlement: New York, The Free Press, 307 p. Riley-Smith, J., 1990, The Crusades: A Short History: London, Athlone Press, 302 p. Runciman, S.A., 1994, History of the Crusades, Volume 2: London, The Folio Society, 428 p. Sbeinati, M.R., Darawcheh, R., and Mouty, M., 2005, Catalog of historical earthquakes in and around Syria: Annali di Geofisica, v. 48, p. 347–435. Tucker, W., 1981, Natural disasters and the peasantry in Mamluk Egypt: Journal of Economic and Social History of the Orient, v. 24, pt. 2, p. 215–224, doi: 10.2307/3631995. Tucker, W., 1999, Environmental hazards, natural disasters, economic loss, and mortality in Mamluk Syria: Mamluk Study Review, v. 3, p. 109–128. William Archbishop of Tyre, 1943, A History of Deeds Done beyond the Sea (translated by E.A. Babcock and A.C. Krey), 2 vols.: New York, Columbia University Press. MANUSCRIPT ACCEPTED BY THE SOCIETY 19 MAY 2010

Printed in the USA

The Geological Society of America Special Paper 471 2010

Western Crete: From Captain Spratt to modern archaeoseismology Manolis I. Stefanakis* Classical Archaeology and Numismatics, Department of Mediterranean Studies, University of the Aegean, 1 Democratias Ave, GR-851 00 Rhodes

ABSTRACT The earliest use of seismological observation to identify and date archaeological sites in western Crete was attempted by Captain T.A.B. Spratt in the late nineteenth century. Since then, the development of the subdiscipline of archaeoseismology has offered a great deal to our understanding of western Crete, especially regarding major sites such as Phalasarna and Kissamos. This paper is a review and summary of archaeoseismology in western Crete, presenting the archaeoseismological and excavation evidence from Phalasarna and Kissamos. It also presents evidence from other archaeological sites in western Crete and expresses the potential the region has for future archaeoseismological research.

ley describes them without having been sensible of their purpose. I was instantly impressed, for several reasons, that here was the ancient or artificial port, although full 200 yards from the sea and nearly 20 feet above it. My first idea was, that the ancients had the means of hauling their vessels into it as a dry dock; but at last the coast elevation was remembered, and on measuring the sea marks at its upper level here, I found that the bed of this ancient port is now 3 or 4 feet below that level; so that I had only to imagine the coast again let down 22 feet 6 inches, the amount it has been elevated here and at Grabusa, when the sea would immediately flow into the ancient port, and float any small craft within it. Geologically the recognition of this ancient port has another interest; it establishes the recent origin of this remarkable up heaving of the western end of Crete, which, however, is not surprising, as elsewhere ancient harbours have been lifted into the air, rocks have become islets, and maritime cities or buildings placed many yards from the shore. These facts will enable me to reconcile in some instances the ancient geography with the modern, and thus to verify points otherwise very difficult. For example, Suia is noticed in the Stadiasmus as a town with a good port (πόλις εστί και λιμένα καλόν έχει), and as following next to Poekilassos, its position is easily recognized. There are so few of the ports of Crete so described in the Stadiasmus, that I naturally looked for a well-sheltered harbour. Pashley says nothing about it, and to look at the locality, few would hope to find a port. A straight and steep shingle beach, off which there is no anchorage, stretches across the mouth of the valley of Suia, and beyond the points of the hills on either side. These points, however, were sea-cliffs, formerly rising out of the beach, to about the height

PHALASARNA 1851: ARCHAEOSEISMOLOGY AT BIRTH?

I made an interesting discovery in the western part of the island, viz., that it has been subject to a series of elevations, amounting to the maximum of 24 feet 6 inches, which occurs near Poekilassos and Suia. In the middle of the island, at Messara, the Fair Havens, and Megalo Kastro, there is none. The eastern end of the island has dipped a little. The up heaving is towards the western end. I had observed it to be about 7 feet in Suda Bay many years ago; but supposed it to be of a time prior to history, although there was a freshness in the markings which might have induced me to suspect they were of a more recent date. When at Kissamo, I observed that the ancient mole was remarkably high out of the water, and the port almost choked by sand. But the latter is so common an occurrence that it did not open my eyes, although the height of the naked unhewn rocks which formed the mole ought to have done so. On going to Phalasarna I looked for its ancient port, mentioned by Scylax, and in the Stadiasmus as the Emporium; but I could find no artificial work in the sea. There is, however, a long ledge of rocks, or rather an islet which lies off it, helping to form a natural but not an artificial harbour. This satisfied me in part, till, on examining the ruins, I saw in the plain a square place, enclosed by walls and towers, more massive and solid than those of the city. Pash*[email protected]

Stefanakis, M.I., 2010, Western Crete: From Captain Spratt to modern archaeoseismology, in Sintubin, M., Stewart, I.S., Niemi, T.M., and Altunel, E., eds., Ancient Earthquakes: Geological Society of America Special Paper 471, p. 67–79, doi: 10.1130/2010.2471(07). For permission to copy, contact [email protected]. © 2010 The Geological Society of America. All rights reserved.

67

68

Stefanakis

of 23 feet; and on them the old sea level is shown distinctly by the appearance of the rock, as well as by a line of cylindrical holes, the cells of boring sea-shells, in some of which the shells still remain. Pashley speaks of the town and ruins of Suia as lying on the E. side of the torrent or valley, but takes no notice of the western side, where a little plain within a long ridge of ruined buildings, and nearly 300 yards long and 60 or 70 broad, runs parallel to the shore. This was undoubtedly the tongue of land which sheltered the port lying behind it. The position of the port itself is indicated by a hollow or flat depression of the plain, which depression would even now be overflowed by the sea, if the island was again let down to its old level. Hence it seems evident that this great elevation of the coast must be looked upon as subsequent to the existence of these ancient cities, and subsequent, therefore, to the decline of the Roman Empire. —Spratt and Leake (1854 )

Captain T.A.B. Spratt, who surveyed Crete in the early 1850s (Spratt, 1865; Psillakis, 2007), sent the above report to the Royal Geographic Society of London in 1854. He can therefore be credited with the earliest correct “reading” of the geomorphology of western Crete, having observed the coastal uplift west from Souda Bay, westward and all the way south, to modern Loutro (Figs. 1 and 2). Captain Spratt was also the first scholar to locate the ancient harbor of Phalasarna (Fig. 3), where the geomorphology was radically changed in late antiquity. He attributed this change to the uplift/ natural catastrophe he had observed affecting the west coast of Crete, assuming that this event occurred after Phalasarna’s blossoming, in Hellenistic times and connecting it—not correctly but still rationally—with the decline of the Roman empire (Spratt, 1865, chapter XIX). Thus, Captain Spratt of the Royal British Navy, archaeologist, historian, geologist, paleontologist, and naturalist (Richards, 1888), was also the first archaeoseismologist in Crete. If Spratt had been familiar with the ancient chronicler Ammianus Marcelinus (ca. A.D. 330–ca. 392), he probably would have gotten the date of the uplift correct too. Ammianus was witness to the one of the stronger earthquakes in the history of Mediterranean, which was accompanied by the most violent tsunami (Ammianus Marcelinus Res Gestae 26.10, lines 15–19), a natural catastrophe recorded by many ancient sources (Stiros, 2001, 2009; Kelly, 2004; Stiros and Drakos, 2006). The emergence of archaeoseismology, the scientific discipline that studies past earthquakes in the archaeological record, in the twentieth century, came to bridge the gap between instrumental and historical seismology, on the one hand, and paleoseismology and earthquake geology, on the other hand and marked the beginning of a fertile cooperation between different sciences, with archaeology, tectonics, sedimentology, paleontology being among the most significant. Through the production of quantitative parameters, necessary to fully describe a past earthquake, a multidisciplinary approach, and in situ analysis of the evidence provided under different contexts, archaeoseismology offers solutions to a great number of problems, such as the positioning of the destruction level of past earthquake activity, the analysis of the deformations applied to

static structures, and the analysis of the depositional characteristics of any collapsed constructions. In addition, archaeoseismology is able to construct maps of past seismic activity concerning regions under surveillance and stretches the importance of the history of the relationship between humans and the environment. In the case of historical times (after the seventh–sixth century B.C.), seismic activity may be often traced more easily through the study of historical sources (Stiros and Jones, 1996; Stiros, 1996a, 2001, p. 547–549; Caputo, 2004; Bottari, 2005; Galadini et al., 2006; Marco, 2008). Western Crete offers an excellent field for archaeoseismological study, and it is the aim of this paper to review the results of this interdisciplinary approach in western Crete, with particular reference to Phalasarna and Kissamos. RECENT ARCHAEOSEISMOLOGICAL RESEARCH IN WESTERN CRETE Phalasarna Being one of the most important sites of western Crete, Phalasarna (Fig. 4) has offered—to date—very satisfactory results in terms of excavation stratigraphy and geomorphology, thanks to which a fruitful interdisciplinary scientific collaboration has produced, not only secure dating, but also a good reconstruction of the uplifted and silted harbor. Ammianus’s report in A.D. 365 and Spratt’s observations in 1851 were verified in the years soon after 1986, when Elpida Hadjidaki began excavations at Phalasarna (Fig. 5). Trenches in Spratt’s “square place, enclosed by walls and towers,” which now lies almost 100 m away from the sea front, provided very useful information about the depth of the harbor entrance and lagoon and its political and geologic history. Evidence for blockage of the entrance channel was found during the excavation of the channel trench (Figs. 5 and 6) and has been related to the Roman invasion of Crete, in 68 B.C. under the leadership of Quintus Cecilius Metellus Creticus. The harbor was vital to Phalasarna’s well-being, and its destruction eventually led to the city’s abandonment (Hadjidaki, 1988, 1990, 1992, 2001; Frost, 1989, 1997; Frost and Hadjidaki, 1990; Pirazzoli et al., 1992). A second trench sunk in the middle of the harbor basin (Figs. 5 and 7A) illuminated the gradual silting-up of the harbor after it was blocked. Successive layers of mud, marine shells (Fig. 7B), sand, and building ruins provide an extremely secure relative chronology for the site over the centuries and its geologic history (Hadjidaki, 1988; Pirazzoli et al., 1992). At this point, geology and environmental archaeology combined with traditional archaeology for a fuller and more secure interpretation of the finds and corrected Spratt’s inspiring original reconstruction of the harbor basin (Fig. 8). Seismic and geological research conducted by Pirazzoli and this team in the early 1990s helped to reconstruct the shape of the lagoon and the process of harbor elevation and silting in relation to human intervention and

Adriatic Sea Epidamnus

Aegean Sea Sicily Ionian Islands Peloponnese Cyprus Crete Sabartha Oea

Figure 1. Map of the Eastern Mediterranean clarifying all geographically relevant places used in the paper and situating Crete in context.

EASTERN MEDITERRANEAN SEA Lepcis Magna Alexandria

North Africa

Figure 2. Detailed map of western Crete with all geographically relevant places cited in the paper.

70

Stefanakis

Figure 3. Map of Phalasarna from Spratt (1865).

natural (seismic and tsunamigenic) events, over a span of eight centuries (fourth century B.C. to fourth century A.D.). According to Pirazzoli (Fig. 9), the silting of the harbor started gradually, shortly after the blocking of its entry, as the formation of a stratum lying 20 cm above today’s sea level consisting of deeper-water species Cerastoderma glaucum has been dated by radiocarbon method between 41 B.C. and A.D. 145 (Fig. 9B). Above this stratum, at +5.70 to +5.90 m above sea level, there lies a distinctive layer of terrestrial sediments, debris, and rounded blocks of stone, washed from the surrounding area of the city, showing the arrival of a tsunami (Fig. 9C). Some 20 cm higher, on top of the tsunami sediments, a layer containing Hydrobia acuta, dated between 54 B.C. and A.D. 137, renders identification of the origin of the stratum beneath from the A.D. 66 tsunami possible (Pirazzoli et al., 1992; DomineyHowes et al., 1998; Stiros and Papageorgiou, 2001; DomineyHowes, 2002). From +6.0 m up and for ~15 cm, marine sediments continue, denoting that seawater was still coming into the harbor (Fig. 9D). Prior to A.D. 169, however, the harbor entrance became completely blocked from the sea and was only occasionally breached by storm surges (Pirazzoli et al., 1992). A second tsunami has been blamed for the uppermost layer of angular limestone blocks and rubble within a silty clay layer in the stratigraphic section seen between +6.4 and +6.7 m (Fig. 9E) (Pirazzoli et al., 1992), although is effects were relatively limited, since no marine stuff is observed (Dawson, 1996; DomineyHowes et al., 1998; Pirazzoli, 1999; Stiros and Papageorgiou,

2001; Price et al., 2002), and it does not seem to have entered very far into the town. Seismic research and radiocarbon analysis date the incident to 1530 (±40) radiocarbon years, which calibrates to ca. A.D. 365 (Pirazzoli et al., 1992). This tsunami was therefore related to the seismic event responsible for raising the coast. Although the A.D. 365 tsunami was stronger than the A.D. 66 one, it does not seem to have affected Phalasarnas’ harbor as much. If the same seismotectonic movement that generated the tsunami had also uplifted western Crete, then perhaps the coast at Phalasarna had already been uplifted by 6.6 m (Fig. 10) when the tsunami hit (Pirazzoli et al., 1992; Pirazzoli, 1999). This may also account for the absence of evidence of A.D. 365 tsunami deposits along the western and southwest coast of Crete (Scheffers and Scheffers, 2007). To sum up, archaeological, seismic, and geological studies at Phalasarna, assisted by late antique chronicles have not only identified the seismic event of A.D. 365 on the ground, but they have shown that by the time of the great earthquake and the land uplift, the city’s infamous “closed” military port (Hadjidaki, 1990, 2001; Stefanakis, 2006b) had already been blocked, abandoned, and partly silted up. The A.D. 66 tsunami wave completed the destruction begun by the Romans more than a century earlier (68 B.C.). This wave swept anything in its path on land into the harbor basin during its retreat, contributing greatly to the siltingup of the harbor. Next, in A.D. 365, the harbor of Phalasarna was transformed into a piece of dry land due to an earthquake that raised the coast almost 9 m above sea level (Pirazzoli et al., 1992; Kelletat, 1998; Stiros and Papageorgiou, 2001; Price et al., 2002; see also Hadjidaki, 1988, 1990, 2001; Frost, 1997; Zouros et al, 2002; Stefanakis, 2006a, 2006b). Its associated tsunami deposited the last—but not much—settlement remains, along with other terrestrial sediments and rubble in the harbor basin. The A.D. 365 uplift of western Crete coast offers a unique opportunity for archaeological research in Phalasarna, since the partly artificial harbor channels, basin, and installations have become part of the land, revealing much useful information on harbor architecture and use and harbor defensive constructions (Sanders, 1982; Gondicas, 1988; Hadjidaki and Stefanakis, 2003; Stefanakis, 2006a). As for the architecture, however, exposure on the land definitely made it more accessible and easier to excavate by landlubbers. In fact, the uplift exposed the site to all kinds of land-based destruction, including being robbed out for building stone and burned up in limekilns. Kissamos At a distance of 5 km east of Phalasarna, in the bay of modern Kissamos, lies Kissamos, the port of ancient Polyrrhenia, which developed after Phalasarna’s destruction and flourished during the first centuries of the Roman conquest of Crete. Archaeological and seismic research in the area indicated that it had been heavily damaged by the A.D. 365 earthquake. Captain Spratt had originally observed an uplift of 5.5 m at the

Western Crete: From Captain Spratt to modern archaeoseismology

71

Figure 4. The site of Phalasarna. Reconstruction is from Hadjidaki and Stefanakis (2003) (courtesy of Kretiko Panorama). 1—military port; 2—fort; 3–4—temples; 5—fish tank; 6—industrial area; 7—south tower; 8—northwest tower; 9—northeast tower; 10—quarries.

harbor (Spratt, 1865, chapter XVIII), a short distance west of the city, at the site “Mavros Molos” (Pologiorgi, 1985; Gondicas, 1988; Stiros and Papageorgiou, 2001; Markoulaki et al., 2004). Indeed, the Kissamos coast was also uplifted (coastal uplift estimated to 6.5 m above sea level), and the harbor is now located quite a few meters inland (Fig. 11) (Davaras, 1967a; Flemming and Pirazzoli, 1981; Pirazzoli, 1999; Stiros and Papageorgiou, 2001; Papadimitriou and Karakostas, 2008). At the city of Kissamos itself, archaeological data from more than 50 rescue excavations over the last decades by the 25th Ephoreate of Prehistoric and Classical Antiquities have contributed to our knowledge about a destructive earthquake that abruptly ended a prosperous period of the city. Archaeological data include, among others, destruction layers, demolished houses, human corpses trapped under the ruins, as well as signs of destructive fire at some point shortly after A.D. 355–361 (Pologiorgi, 1985; Stiros, 2001; Stiros and Papageorgiou, 2001; Vlazaki-Andreadaki, 2002, 2004; Markoulaki, 2002, 2006). Geologic research indicates an earthquake of a minimum

seismic intensity of XI (MM scale) from an epicenter less than 100 km away. All this, considered together with the numerous copper coins found in the destruction layer and dating up to A.D. 355–361 (reign of the roman Emperor Constant II), makes the A.D. 365 earthquake the most probable cause of the catastrophe, which led to “nearly total physical destruction of the community” (Stiros and Papageorgiou, 2001, p. 387) since the victims failed to receive a proper burial (Stiros and Papageorgiou, 2001; Papadimitriou and Karakostas, 2008). Kissamos followed the fate of Phalasarna, with a coastal elevation of 5.5 m, which resulted in the uplift and exposure of its harbor installations. Unlike Phalasarna, Kissamos did not receive any tsunami impacts, for the town was safely built within the homonymous secure closed bay. Although it has not been verified for Phalasarna, Kissamos was seized to ground by the intense earthquake and never managed to recover. For her, as for the whole of western Crete, the A.D. 365 event signified the end of pagan era and the beginning of Christian era (Stiros and Papageorgiou, 2001; Markoulaki, 2006).

72

Stefanakis

Figure 5. Plan of the harbor area after Frost and Hadjidaki (1990). Arrows indicate the spots of two important test trenches (Figs. 6 and 7).

SEISMIC EVENT OF A.D. 365 AND WESTERN CRETE The powerful tectonic earthquake that took place during the night of 21 July A.D. 365 off the southwest coast of Crete, a result of the subduction of the African under the Aegean tectonic plate (Fig. 12), raised the west coast of the island 6–9 m above sea level (Thommeret et al., 1981; Jacques and Bousquet, 1984a, 1984b; Papazachos and Papazachos, 1989; Pirazzoli et al., 1992; Stiros, 1996b, 2009; Spyropoulos, 1997; Stiros and Papageorgiou, 2001; Stiros and Drakos, 2006) (Fig. 13). According to seismic stud-

ies on Crete and Antikythera, the earthquake took place as the result of 20 m slip on a fault ~100 km long, and its epicenter was between the SW edge of Crete and the Hellenic trench (Pirazzoli et al., 1992; Kelletat, 1998; Stiros and Papageorgiou, 2001; Shaw et al., 2008). Evaluations of the event estimate the intensity of the earthquake at M >8.5. Its focal depth was between 40 and 70 km and caused a tsunami of unknown magnitude, but of an intensity reaching 5 (in Ambrasey’s 1962 scale) and a surfacewave magnitude reaching 8 (Papazachos and Papazachos, 1989; Papazachos and Dimitriou, 1991; Papadopoulos, 2001; Stiros

Western Crete: From Captain Spratt to modern archaeoseismology

Figure 6. Drawing of the channel trench, where eight huge stone blocks (A–Θ) were revealed during the excavation in 1987 (from Hadjidaki, 1988).

A

73

the southern Peloponnese, west and south Crete, and Alexandria in the Nile Delta. Destruction probably from the same event is recorded in North Africa at Leptis Magna, Oea, and Sabartha (Papazachos and Dimitriou, 1991; Pirazzoli et al., 1992; Guidoboni et al., 1994; Kelletat, 1998; Stiros, 2001, 2009; Stiros and Papageorgiou, 2001; Price et al., 2002; Dominey-Howes, 2002; Kelly, 2004; Stiros and Drakos, 2006; Papadimitriou and Karakostas, 2008). The absolute date of the uplift, however, is still controversial (Stiros and Drakos, 2006). A reassessment of radiocarbon dates on material from the solution notches on the Sphakia shoreline, based on the Bayesian approach and new calibration data, produced a date between A.D. 480 and 550 for the land uplift, suggesting that the major uplift has not been detected in archaeological record (Price et al., 2002). A more recent study however, by Shaw et al. (2008), reassessing radiocarbon dates based on material from present sea level and the uplifted paleoshoreline, came to reinforce the earlier date of A.D. 365 for western Crete’s uplift and tsunami. According to the study and the new radiocarbon dates, the A.D. 365 event had a magnitude of Mw 8.3–8.5, and the paleoshoreline “was lifted close to its present position within a few decades of the AD 365 earthquake” (Shaw et al., 2008, p. 273). ARCHAEOSEISMOLOGY AND WESTERN CRETE: POTENTIAL RESEARCH

B

Figure 7. The harbor trench. (A) Stratigraphy of the two tsunami deposits (courtesy of Kretiko Panorama). (B) A.D. 66 tsunami layer, detail (courtesy of Kretiko Panorama).

and Papageorgiou, 2001; Stiros and Drakos, 2006; Shaw et al., 2008). It was the biggest tsunami reported in and near ancient Greece, claiming thousands of lives and causing widespread devastation in various parts of eastern Mediterranean (Fig. 1), including Sicily, the Ionian Islands, Epidamnus in the Adriatic,

Even if controversy over the dating of the seismic event persists, it still remains that western Crete offers a unique opportunity for archaeoseismological study. The coastal uplift in late antiquity is a fact, while tsunami impact and seismic destruction have been identified at a number of archaeological sites (Fig. 14) (Stiros et al., 2004). Costal uplifts attributed to the A.D. 365 earthquake (Shaw et al., 2008) have been observed at the south coast: Tarrha, modern Hagia Roumeli (Weinberg, 1960; Price et al., 2002; The Sphakia Survey, 2010), has been uplifted by 6 m (Price et al., 2002), Poikilasion (Price et al., 2002; The Sphakia Survey, 2010) by 7 m (Pirazzoli, 1999; Price et al., 2002), and Phoenix, modern Loutro (Price et al., 2002; The Sphakia Survey, 2010), by 3.5–4 m (Price et al., 2002). All three sites have already been thoroughly surveyed (Moody et al., 1998; The Sphakia Survey, 2010) and studied from the seismic and geological point of view, although the original studies did not attribute the coastal uplift to the A.D. 365 earthquake but to a later event between A.D. 405 and A.D. 615 (Price et al., 2002). Uplift, known since Spratt’s time and verified by later scientific observation, can be also seen at the coastal cities: Kydonia, uplift of 2 m (Stiros and Papageorgiou, 2001); Inachorion, uplift of 8 m at the nearby sites Ormos Stomiou and Mavros Bay (Scheffers and Scheffers, 2007); Kalamyde, modern Palaiochora, port of ancient Kandanos, uplift of 7–8 m (Scheffers and Scheffers, 2007), Lissos, uplift of 7 m (Pirazzoli, 1999); and Syia, modern Sougia (Pirazzoli, 1999), uplift of 6.6 m.

Figure 8. Suggested section plan of the harbor basin in relation to ancient and modern sea level from Spratt (1865).

A

B

C

D

E

F

Figure 9. Model to explain the stratigraphy in the Phalasarna harbor basin in relation to tectonic movement and gradual harbor silting from late classical period to modern era, based on simplified stratigraphy of the harbor trench (after Pirazzoli et al., 1992). The A–F labels represent stages of MSL (mean sea level) in relation to the harbor basin, from the late fourth century B.C. to present. F—freshwater deposits; T—tsunami deposits; C—confined marine deposits; M—marine deposits.

Western Crete: From Captain Spratt to modern archaeoseismology

Figure 10. Traces of ancient sea line southeast of the harbor entrance at Phalasarna coast (courtesy E. Hadjidaki).

Remains of uplifted ancient jetty

Figure 11. Mavros Molos, ancient harbor of Kissamos. Uplifted jetty is made of unhewn blocks (from Davaras, 1967a).

75

Tsunami impact evidence (deposits) has been observed on the whole southwestern shore of Crete, from Palaiochora (ancient Kalamyde) to Gramvousa Island, north of Phalasarna. Among other sites, stratigraphic evidence of tsunami impacts is observed in the area of ancient Kalamyde, Inachorion, and Phalasarna (Scheffers and Scheffers, 2007), while the area of ancient Viennos is also worth testing. It is very interesting that although geomorphological studies in western Crete have produced enough proof of tsunami impacts from the late Holocene to the late antiquity, they have produced no absolute data or any evidence, so far, of fourth century and the A.D. 365 tsunami deposits (DomineyHowes, 2002; Stiros and Drakos, 2006; Scheffers and Scheffers, 2007; Stiros, 2009). More evidence of the destructive earthquakes of A.D. 66 and A.D. 365 probably awaits discovery at other inland archaeological sites in western Crete. A prime candidate is the big city of Polyrrhenia, in the mountains east of Phalasarna and south of Kissamos (Theophaneides, 1948; Davaras, 1967b; Sanders, 1982; Gondicas, 1988; Markoulaki, 1992), which was possibly damaged by the A.D. 66 earthquake and eventually relocated to Kissamos (Pologiorgi, 1985; Stiros and Papageorgiou, 2001). The prosperous city of Aptera (Drerup, 1951; Sanders, 1982; Niniou-Kindeli, 2006), overlooking Souda Bay, where Spratt (1865, chapter XI) observed an elevation of ~2 m, may also bear traces of this earthquake. A destroyed Roman villa of the first century B.C.–A.D. first century, in sector VI, for example, is described by the excavator as giving “…the impression as having collapsed after a strong earthquake” (Niniou-Kindeli, 1999a, p. 169; see also Niniou-Kindeli, 1994–1996, 1999b, 2003).

Figure 12. Crete situated in the Mediterranean tectonic context (after http://commons.wikimedia.org/wiki/File:Tectonic _map_Mediterranean_EN.svg).

76

Stefanakis

Figure 13. Western Crete elevation (m) shown by uplift curves (Papadimitriou and Karakostas, 2008).

Figure 14. Sites of archaeological interest of western and central Crete affected by the land elevation or by the A.D. 365 earthquake in general. 1—Aptera; 2—Kydonia; 3—Kissamos; 4—Polyrrhenia; 5—Phalasarna; 6—Inachorion; 7—Viennos; 8—Kandanos; 9—Kalamyde; 10—Hyrtakina; 11—Lissos; 12—Elyros; 13—Syia; 14—Poekilassion; 15—Tarrha; 16—Phoenix; 17—Diktynnaion; 18—Eleutherna; 19—Knossos; 20—Gortyna. (Map with uplift curves is after Pirazzoli et al., 1992.)

Many inland cities in southwestern Crete such as Kantanos (Sanders, 1982; Gondicas, 1988; Stefanakis, 2000), Hyrtakina (Sanders, 1982), and Elyros (Sanders, 1982) must have also been damaged by the destructive earthquakes. Solution notches and seismic damage should also be looked for near the ancient sites of the Diktynnaion Temple at Spatha promontory (Welter and Jantzen, 1951; Gondicas, 1988; Markoulaki, 2000; Markoulaki and Martinez, 2000–2001), Inachorion

(Gondicas, 1988) and Viennos (Gondicas, 1988) on the west coast, and Kalamyde (Hood, 1967; Gondicas, 1988) on the south. The list of the aforementioned sites of western Crete is, however, selective, and traces of the seismic event of A.D. 365 may exist in many more archaeological sites of western Crete (Hood, 1967; Sanders, 1982; Gondicas, 1988; Faure, 1989; Andreadaki-Vlazaki, 1997; Faraklas et al., 1998; The Sphakia Survey, 2010).

Western Crete: From Captain Spratt to modern archaeoseismology At the same time, research should also extend to central Crete, since Eleutherna, in the western foothills of Mount Ida, Knossos on the north coast, and Gortyna in south-central Crete (Fig. 14) also have evidence for a destructive earthquake in the second half of the fourth century A.D. (Stiros, 2001, 2009; see also Themelis, 1988; Sidiropoulos, 2004; Guidoboni et al., 1994), which could be associated with the A.D. 365 event. So many earthquake-affected sites on Crete in the fourth century A.D. give credence to Athanasius of Alexandria’s claim that in A.D. 365, more than 100 Cretan cities were destroyed by an earthquake of unprecedented magnitude, followed by a tsunami (Migne, 1857, v. 25, p. ccx). Evidence exists, but recovering it is not simply a matter of observation. Systematic archaeological survey and excavation are needed to reveal stratigraphic sections for chronological, geologic, and seismic research. For many sites, however, such research may be too late. For example, in the early 1850s, Captain Spratt produced a map of ancient Syia showing an elevated harbor (coast uplifted by 6.70 m) to the west of the ancient settlement (Fig. 15A). Today, however, the modern village of Sougia has gradually built up over the harbor basin (Fig. 15B), obscuring the archaeological record. It is important to explore and record these sites before more are buried by modern development. ACKNOWLEDGMENTS The author wishes to thank Elpida Hadjidaki, director of the excavations at Phalasarna, for her kind permission to present data from the excavations at the site, plans, and photographs, as well as Giorgos Patroudakis, publisher of Kretiko Panorama, for his permission to reproduce photographs. Thanks are also addressed to Manuel Sintubin for the invitation and encouragement to participate in the International Geo-

A

77

science Programme (IGCP) 567 on Earthquake Archaeology and to Stathis Stiros and Jennifer Moody for saving the manuscript from many mistakes. This article is a contribution to the United Nations Educational, Scientific and Cultural Organization– funded IGCP 567, “Earthquake Archaeology: Archaeoseismology along the Alpine-Himalayan Seismic Zone.” REFERENCES CITED Andreadaki-Vlazaki, M., 1997, The County of Chania through Its Monuments: Athens: Archaeological Receipts Fund, 75 p. Bottari, C., 2005, Ancient constructions as markers of tectonics deformation and strong seismic motions: Pure and Applied Geophysics, v. 162, p. 761– 765, doi: 10.1007/s00024-004-2639-6. Caputo, R., 2004, Historical seismology, archaeoseismology and palaeoseismology: Three distinct approaches to a natural phenomenon, in Archaeoseismology at the Beginning of the 21st Century (Atlas Conferences): Rome. Davaras, K., 1967a, Kastelli Kissamou: Archaeologiko Deltion, v. 22, part B2, Chronicles, p. 498–499 [in Greek]. Davaras, K., 1967b, Polyrrhenia: Archaeologiko Deltion, v. 22, part B2, Chronicles, p. 499 [in Greek]. Dawson, A.G., 1996, The geological significance of tsunamis: Zeitschrift für Geomorphologie N.F., supplement, v. 102, p. 199–210. Dominey-Howes, D.T.M., 2002, Documentary and geological records of tsunamis in the Aegean Sea region of Greece and their potential value to risk assessment and disaster management: Natural Hazards, v. 25, p. 195–224, doi: 10.1023/A:1014808804611. Dominey-Howes, D.T.M., Dawson, A.G., and Smith, D.E., 1998, Late Holocene coastal tectonics at Phalasarna, western Crete; a sedimentary study, in Stewart, I.A., and Vita-Finzi, C., eds., Coastal Tectonics: Geological Society of London Special Publication 146, p. 343–352. Drerup, H., 1951, Paläokastro-Aptara. Bericht über eine Untersuchung und Vermessung des Stadtgebietes, in Matz, F., ed., Forschungen auf Kreta, 1942: Berlin, W. de Gruyter, p. 89–98. Faraklas, N., Kataki, E., Kossyva, A., Xifaras, N., Panagiotopoulos, E., Tassoulas, G., Tsatsaki, N., and Chatzipanagioti, M., 1998, The Territories of the Ancient Cities of Crete: Rithymna 6: Rethymno, University of Crete, 242 p. [in Greek]. Faure, P., 1989, Cités antiques de Crète de l’oust: Cretan Studies, v. 1, p. 81–96. Flemming, N., and Pirazzoli, P., 1981, Archéologie des côtes de la Crète, in Ports et Villes Engloutis, Dossiers d’Archéologie, v. 50, p. 66–81.

B

Figure 15. (A) Map of Syia from Spratt (1865). (B) Aerial view of modern Sougia.

78

Stefanakis

Frost, F.J., 1989, The last days of Phalasarna: Ancient History Bulletin, v. 3, p. 15–17. Frost, F.J., 1997, Tectonics and history at Phalasarna, in Swiny, S., Hohlefelder, R.L., and Swiny, H.W., eds., Res Maritimae: Cyprus and the Eastern Mediterranean from Prehistory to Late Antiquity: Atlanta, American School of Oriental Research, p. 107–115. Frost, F.J., and Hadjidaki, E., 1990, Excavations at the Harbour of Phalasarna in Crete: Hesperia, v. 59, p. 513–527, doi: 10.2307/148300. Galadini, F., Hinzen, K.-G., and Stiros, S.C., eds., 2006, Archaeoseismology: Methodological issues and procedure: Archaeoseismology at the Beginning of the 21st Century (Atlas Conferences): Journal of Seismology, v. 10, no. 4, p. 395–414, doi: 10.1007/s10950-006-9027-x. Gondicas, D.G., 1988, Recherches sur la Crète Occidentale: Amsterdam, Adolf M. Hakkert, 365 p. Guidoboni, E., Comastri, A., and Traina, G., 1994, Catalogue of Ancient Earthquakes in the Mediterranean up to the 10th Century: Rome, Istituto Nazionale di Geofisica, 504 p. Hadjidaki, E., 1988, Preliminary report of excavation at the harbour of Phalasarna in west Crete: American Journal of Archaeology, v. 92, p. 463–479, doi: 10.2307/505244. Hadjidaki, E., 1990, Excavations at the classical and Hellenistic harbor at Phalasarna, west Crete, Greece: Acts of the 6th International Cretological Congress, v. A1, Archaeological Section: Chania, Philologikos Syllogos “O Chryssostomos,” p. 355–361. Hadjidaki, E., 1992, Phalasarna: Archeologikon Deltion, Xρονικά, v. 42 (1987), part B2, Chronicles, p. 566–567 [in Greek]. Hadjidaki, E., 2001, The Roman destruction of Phalasarna, in Higham, N., ed., Archaeology of the Roman Empire; a Tribute to the Life and Works of Professor Barri Jones: British Archaeological Reports International Series, v. 940, p. 155–166. Hadjidaki, E., and Stefanakis, M.I., 2003, The secrets of Phalasarna: Kretiko Panorama, v. 2, p. 10–135. Hood, M.S.F., 1967, Some ancient sites in southwest Crete: Annual of the British School of Athens, v. 62, p. 47–56. Jacques, F., and Bousquet, B., 1984a, Le raz de marée du 21 juillet 365—Du cataclysme local à la catastrophe cosmique: Mélanges de l’École Française de Rome, v. 96, p. 423–461. Jacques, F., and Bousquet, B., 1984b, Le cataclysme du 21 juillet 365: Phénomène régional ou catastrophe cosmique?, in Helly, B., and Pollino, A., eds., Tremblements de Terre. Histoire et Archéologie: IVèmes Rencontres Internationales d’Archéologie et d’Histoire d’Antibes: Valbonne, Editiones ADPCA, p. 183–193. Kelletat, D., 1998, Geologische Belege katastrophaler Erdkrustenbewegungen 365 AD im Raum von Kreta, in Olshausen, E., and Sonnabend, H., eds., Stuttgarter Kolloquium zur historischen Geographie des Altertums 6, 1996 Naturkatastrophen in der antiken Welt: Geographica Historica 10: Stuttgart, Frank Steiver Verlag, p. 156–161. Kelly, G., 2004, Ammianus and the Great Tsunami: Journal of Roman Studies, v. 94, p. 141–165, doi: 10.2307/4135013. Marco, S., 2008, Recognition of earthquake-related damage in archaeological sites: Examples from the Dead Sea fault zone: Tectonophysics, v. 453, p. 148–156, doi: 10.1016/j.tecto.2007.04.011. Markoulaki, St., 1992, Polyrrhenia: Archaeologiko Deltion, Chronicles, v. 42 (1987), part B2, Chronicles, 563 p. [in Greek]. Markoulaki, St., 2000, Stele Telephou: Actes of the 8th International Cretological Congress, v. A2, p. 239–257 [in Greek]. Markoulaki, St., 2002, Kentro Hygias: Kretike Estia, v. 9, p. 270–271. Markoulaki, St., 2006, Kissamos, in Andreadaki-Vlazaki, M., and NiniouKindeli, V., eds., Ancient Sites and Monuments. The Chania Prefecture: Chania, KE Ephoreate of Prehistoric and Classical Antiquities, p. 22–23 [in Greek]. Markoulaki, St., and Martinez, A.F., 2000–2001, Psefisma proxenias apo tin Kissamo: Kretike Estia, v. 8, p. 147–158. Markoulaki, St., Christodoulakos, G., and Fragkonikolaki, C., 2004, I archaia Kissamos kai I poleodomike tes organose: Creta Romana e Protobyzantina II: Padova, Bottega d’Erasmo A. Ausilio, p. 355–374 [in Greek]. Migne, J.P., 1857, Patrologiae cursus completus: Omnium SS. patrum, doctorum scriptorumque ecclesiasticorum; sive latinorum, sive graecorum, in Diotis, J., ed.: vols. 25–28 (Athanasius of Alexandria), Helleniki Patrologia (Patrologia Graeca): Athens, Centre for Patrological Editions.

Moody, J., Nixon, L., Price, S., and Rackham, O., 1998, Surveying poleis and larger sites in Sphakia, in Cavanagh, W.G., and Curtis, M., eds., PostMinoan Crete: Proceedings of the Colloquium organised by the British School at Athens and the Institute of Archaeology, University of London, November 1995: British School at Athens Studies Series 2, p. 87–95. Niniou-Kindeli, V., 1994–1996, Aptera: Kretike Estia, v. 5, p. 210–212 [in Greek]. Niniou-Kindeli, V., 1999a, Aptera (Aptara): Kretike Estia, v. 7, p. 167–175. Niniou-Kindeli, V., 1999b, Aptera: Archaeologiko Deltion, v. 49 (1994), part B2, Chronicles, p. 721. Niniou-Kindeli, V., 2003, Aptera: Archaeologiko Deltion, v. 52 (1997), part B3, Chronicles, p. 1017–1019. Niniou-Kindeli, V., 2006, Aptera, in Andreadaki-Vlazaki, M., and NiniouKindeli, V., eds., Ancient Sites and Monuments. The Chania Perfecture: Chania, KE Ephoreate of Prehistoric and Classical Antiquities, p. 8–9 [in Greek]. Papadimitriou, E., and Karakostas, V.G., 2008, Rupture model of the great A.D. 365 Crete earthquake in the southwestern part of the Hellenic Arc: Acta Geophysica, v. 56, no. 2, p. 293–312. Papadopoulos, G.A., 2001, Tsunamis in the East Mediterranean: A catalogue for the area of Greece and adjacent seas, in Proceedings of the Intergovernmental Oceanographic Commission/International Union of Geodesy and Geophysics International Workshop Tsunami Risk Assessment Beyond 2000 Theory, Practice and Plans: In Memory of Professor S.L. Soloviev, Moscow, 14–16 June 2000: Moscow, p. 34–43. Papazachos, B., and Dimitriou, P.P., 1991, Tsunamis in and near Greece and their relation to the earthquake focal mechanisms: Natural Hazards, v. 4, p. 161–170, doi: 10.1007/BF00162785. Papazachos, B., and Papazachos, K., 1989, The Earthquakes in Greece: Thessaloniki [in Greek], 356 p. Pirazzoli, P.A., 1999, Les ports antiques soulevés de la Méditeranée orientale, in Rosselló, V.M., ed., Geoarqueologia I Quartenari Litoral, Memorial Maria Pilar Fumanal: Valencia, Valencia University, p. 391–401. Pirazzoli, P.A., Ausseil-Badie, J., Giresse, P., Hadjidaki, E., and Arnold, M., 1992, Historical environmental changes at Phalasarna Harbour, west Crete: Geoarchaeology, v. 7, no. 4, p. 371–392, doi: 10.1002/gea.3340070406. Pologiorgi, M., 1985, Kissamos; the topography of an ancient polis: Archaeologika Analekta ex Athenon, v. XVIII, p. 65–79 [in Greek]. Price, S., Higham, T., Nixon, L., and Moody, J., 2002, Relative sea-level changes in Crete: Reassessment of radiocarbon dates from Sphakia and west Crete: Annual of the British School at Athens, v. 97, p. 171–200. Psillakis, M., and Psillakis, N., 2007, T.B.A. Spratt. Taxidia kai Erevnes stin Kriti tou 1850, v. 2: Herklion, Karmanor Publications, 408 p. [in Greek]. Richards, G.H., 1888, Obituary: Vice-Admiral Thomas A.B. Spratt, C.B., F.R.S.: Proceedings of the Royal Geographical Society and Monthly Record of Geography, New Monthly Series, v. 10, no. 4 (Apr.), p. 242–244. Sanders, I.F., 1982, Roman Crete. An Archaeological Survey and Gazetteer of Late Hellenistic, Roman and Early Byzantine Crete: Warminster, Aris and Phillips, 185 p. Scheffers, A., and Scheffers, S., 2007, Tsunami deposits on the coastline of west Crete (Greece): Earth and Planetary Science Letters, v. 259, p. 613– 624, doi: 10.1016/j.epsl.2007.05.041. Shaw, B., Ambraseys, N.N., England, P.C., Floyd, M.A., Gorman, G.J., Higham, T.F.G., Jackson, J.A., Nocquet, J.-M., Pain, C.C., and Piggott, M.D., 2008, Eastern Mediterranean tectonics and tsunami hazard inferred from the AD 365 earthquake: Nature Geoscience, v. 1, p. 268–276 (published online: 9 March 2008; doi: 10.1038/ngeo151). Sidiropoulos, K., 2004, Numismatic history of Roman and Protobyzantine Crete (67 BC–AD 827). Testimonia et Desiderata, in Livadiotti, M., and Simiakaki, I., eds., Creta Romana e Protobyzantina: Padova, Bottega d’Erasmo A. Ausilio. The Sphakia Survey, 2010, The Sphakia Survey, Internet edition: http:/sphakia .classics.ox.ac.uk (accessed June 2010). Spratt, C.T.A.B., 1865, Travels and Researches in Crete, Volume II: London, John van Voorst MDCCCLXV(=1865), 327 p. Spratt, C.T.A.B., and Leake, C., 1854, Extract of a Letter from Captain Spratt, R.N., on Crete: Journal of the Royal Geographical Society of London, v. 24, p. 238–239, doi: 10.2307/3698110. Spyropoulos, P.I., 1997, The Chronicle of Earthquakes in Greece: Athens, Dodone, 453 p. [in Greek].

Western Crete: From Captain Spratt to modern archaeoseismology Stefanakis, M.I., 2000, Polyrrhenia, Oreioi and Kandanos. A relationship of the second half of the third century BC: Actes of the 8th International Cretological Congress, Herakleion, v. A3, p. 249–261 [in Greek]. Stefanakis, M.I., 2006a, Natural catastrophes in the Greek and Roman world: Curse or blessing? Four cases of earthquake-generated tsunamis: Mediterranean Archaeology and Archaeometry Journal, v. 6.1, p. 61–88. Stefanakis, M.I., 2006b, Phalasarna: Un port antique, un espace d’échanges en Méditerranée, in Clément, F., Tolan, J., and Wilgaux, J., eds., Espaces d’Échanges en Méditerranée. Antiquité et Moyen-Age: Lyon, PU Rennes, p. 41–75. Stiros, S., 1996a, Identification of earthquakes from archaeological data: Methodology, criteria and limitations, in Stiros, S., and Jones, R., eds., Archaeoseismology: British School at Athens, Fitch Laboratory Occasional Paper 7, p. 129–152. Stiros, S., 1996b, Late Holocene relative sea level changes in SW Crete: Evidence of an unusual earthquake cycle: Annali di Geofisica, v. XXXIX, no. 3, p. 677–687. Stiros, S., 2001, The AD 365 Crete earthquake and possible seismic clustering during the 4–6th centuries AD in the Eastern Mediterranean: A review of historical and archaeological data: Journal of Structural Geology, v. 23, p. 545–562, doi: 10.1016/S0191-8141(00)00118-8. Stiros, S.C., 2009, The 8.5+ magnitude, AD 365 earthquake in Crete: Coastal uplift, topography changes, archaeological and historical signature: Quaternary International, doi: 10.1016/j.quaint.2009.05.005. Stiros, S., and Drakos, A., 2006, A fault-model for the tsunami-associated, magnitude ≥8.5 Eastern Mediterranean, AD 365 earthquake: Zeitschrift für Geomorphologie, supplement, v. 146, p. 125–137. Stiros, S., and Jones, R.E., eds., 1996, Archaeoseismology: Oxford, UK, Institute of Geology and Mineral Exploration/British School at Athens, 268 p.

79

Stiros, S.C., and Papageorgiou, S., 2001, Seismicity of western Crete and the destruction of the town of Kissamos at AD 365: Archaeological evidence: Journal of Seismology, v. 5, p. 381–397, doi: 10.1023/A:1011475610236. Stiros, S., Papageorgiou, S., and Markoulaki, S., 2004, The destruction of Cretan towns in AD 365, in Livadioti, M., and Simiakaki, I., eds., Creta Romana e Protobyzantina: Atti del Congresso Internazionale (Iraklion, 23–30 Settembre 2000): Padova, Bottega d’Erasmo A. Ausilio, p. 193– 223 [in Greek]. Themelis, P., 1988, Eleutherna: Kritiki Estia, v. 2, p. 298–302. Theophaneides, V., 1948, Excavational research and finds from western Crete. The province of Kissamos. (B) Excavations: Archaeologiki Ephimeris 1942–44, part B, Chronicles, p. 17–31 [in Greek]. Thommeret, Y., Laborel, J., Montaggioni, L., and Pirazzoli, P., 1981, Late Holocene shoreline changes and seismotectonic displacements in western Crete (Greece): Zeitschrift für Geomorphologie, supplement, v. 40, p. 127–149. Vlazaki-Andreadaki, M., 2002, Kissamos: Kretike Estia, v. 9, p. 266–271. Vlazaki-Andreadaki, M., 2004, Kissamos: Archaeologiko Deltion, v. 53 (1988), part B3, Chronicles, p. 864–868. Weinberg, G.D., 1960, Excavations at Tarrha 1959: Hesperia, v. 29, p. 90–108, doi: 10.2307/147333. Welter, G., and Jantzen, U., 1951, Das Diktynnaion, in Matz, F., ed., Forschungen auf Kreta, 1942: Berlin, W. de Gruyter, p. 106–117. Zouros, N., Velitzelos, E., Mountrakis, D.M., and Soulakelis, N., eds., 2002, Atlas of the Geological Monuments of the Aegean: Athens, Ministry of the Aegean, 350 p. [in Greek]. MANUSCRIPT ACCEPTED BY THE SOCIETY 19 MAY 2010

Printed in the USA

The Geological Society of America Special Paper 471 2010

Earthquake archaeology in Japan: An overview Gina L. Barnes School of Oriental and African Studies (SOAS), University of London, Thornhaugh Street, Russell Square, London WC1H 0XG, UK

ABSTRACT Earthquake archaeology developed in Japan simultaneously with that in the Mediterranean in the mid-1980s. By 1996, evidence of earthquake occurrence had been documented at 378 sites throughout the archipelago. The main features identified include various results of liquefaction, faults, landslips, and surface cracking. This evidence differs greatly from the standard Mediterranean focus on building damage, and the reasons for the very different natures of archaeoseismology in these world regions are explained herein. This article recounts the development of this new subfield, inspired by the interest of geomorphologist Sangawa Akira and taken to its most recent advances in identifying soft-sediment deformation structures by geoarchaeologist Matsuda Jun-ichirō. The evidence of earthquake activity at archaeological sites can be matched with earthquakes caused by either active fault movement or subduction. The historical record of earthquake occurrence, already documented back to 599 C.E., is extended into the prehistorical record through relative dating of artifacts and features on archaeological sites. Both the identification and the dating of the archaeological evidence of earthquakes can be challenged, though the “territorial approach” gives the data a significance that is not achieved through analysis of single sites.

THE NEW SUBDISCIPLINE

Sangawa to develop the idea that even human-made constructions in Japan can give evidence of past earthquake activity. This may sound odd to the Mediterranean archaeologists, who deal with earthquake damage to buildings and site destruction layers as a matter of course (cf. Galadini et al., 2006), but the novelty of this idea in Japan is illustrative of the difference between earthquake archaeology in Japan and elsewhere—a difference discussed in detail herein. Sangawa’s ideas were introduced to the archaeological community in 1987 through the Kodaigaku Kenkyūkai (Paleology Research Group) at Dōshisha University in Kyoto (Sangawa, 1988). Primarily through this publication, archaeologists became aware that past earthquake activity can be seen in excavations and that many unexplained features in some sites were, in fact, the unrecognized effects of earthquakes. Such discoveries are

The 1995 Kobe earthquake (officially called the HanshinAwaji Dai-shinsai in Japanese) brought to the fore a developing subdiscipline of Japanese archaeology focused on identifying the effects of earthquakes in archaeological sites. Named jishin kōkogaku (“earthquake archaeology”) in the mid-1980s, the field has been developed in Japan primarily through the initiative of one geomorphologist with the Japan Geological Survey, Sangawa Akira, who participated in the compilation of the national survey of active faults and the rendering of these onto 1:50,000 scale maps (RGAFJ, 1992). A chance encounter in his student days with the fifth-century Kondayama mounded tomb (see periodization in Table 1), which shows evidence of dike slippage along a fault line (Fig. 1), led

Barnes, G.L., 2010, Earthquake archaeology in Japan: An overview, in Sintubin, M., Stewart, I.S., Niemi, T.M., and Altunel, E., eds., Ancient Earthquakes: Geological Society of America Special Paper 471, p. 81–96, doi: 10.1130/2010.2471(08). For permission to copy, contact [email protected]. © 2010 The Geological Society of America. All rights reserved.

81

82

Barnes TABLE 1. CHRONOLOGY OF JAPANESE HISTORY Dates Period Character Ca. 15,000–800 B.C.E. Jōmon Hunting, gathering, fishing, horticulture 800 B.C.E.–C.E. 250 Yayoi Agriculture, bronze, iron 250–710 Kofun Mounded tomb culture, state formation 710–794 Nara Early court culture, archaic state 794–1185 Heian Late court culture, state decentralization 1185–1603 Medieval Warrior-dominated culture, rise of samurai 1603–1868 Edo Tokugawa family rule, feudal administration 1868–present Modern Westernization, industrialization, militarism

Figure 1. The Kondayama Tomb (K) situated in a late fourth- to fifthcentury cluster of keyhole-shaped mounded tombs of early Japanese ruling elites, located on a Pleistocene terrace in southeastern Osaka. Some tombs are moated with additional greenbelts or dikes, which exhibited displacement along the Konda fault. (Courtesy of A. Sangawa, redrawn by Durham Archaeological Services.)

continuously being published in the journal Kodaigaku Kenkyū, and Sangawa has expanded upon his original views in 1995, 2001, and 2007, in addition to many individual articles—most recently a personalized excursion through the progress of his earthquake archaeology research (Sangawa, 2009). As Sangawa was formulating his approach to archaeological data in Japan, the field of earthquake archaeology was developing overseas, particularly in the Mediterranean (Rapp, 1986, p. 365). Archaeological sites with building damage and destruction layers were well known, but it was not until the late 1980s that a systematic approach to the data began to be constructed. In 1991, an international conference in Athens “brought together around a hundred specialists from several countries ... to exchange ideas and discuss the problems of identification and study of ancient earthquakes from the complementary standpoints of their social, cultural, historical and physical effect” (Stiros and Jones, 1996, p. 1). Many of these conference papers were published in the volume entitled Archaeoseismology (Stiros and Jones, 1996). In Japan, it was the 1995 Kobe earthquake that galvanized archaeological action on a multifaceted front. Young archaeologists in the Kinai region around Osaka, Nara, and Kyoto joined together to form the “Disaster Concerned Archaeologists’ Network” (Maibun Kankei Kyūen Renraku Kaigi; see, e.g., DCAN, 1996) to assist with communications related to cultural properties damaged in the earthquake. A newsletter edited by Okamura Katsuyuki conveyed important findings, and one of the network’s outstanding contributions to the scholarship of earthquake archaeology was the joint sponsorship of a symposium together with the “Buried Cultural Properties Research Group” (Maizō Bunkazai Kenkyūkai) on evidence of earthquakes in the archaeological record. Over 150 scholars from around Japan gathered near Kobe just 19 mo. after the earthquake to present evidence on 378 sites throughout Japan assessed as having sustained earthquake damage. Their published report runs to 826 pages and contains a master list of the sites covered (DCAN, 1996). Their purpose in this

Earthquake archaeology in Japan compilation, however, was more than academic: they saw that if the cycle of earthquakes could be established through time in different regions, they could help contribute to the future of society. The Athens conference discussed and defined the terms seismic archaeology, archaeological seismicity, earthquake geology, and paleoseismology. It seems that archaeoseismology has been generally adopted as the name of the field since it interfaces with archaeology. Sangawa preferred the term seismic archaeology, but for the purposes of this article I have chosen to translate jishin kōkogaku directly as earthquake archaeology (as Sangawa also does now) and will use this term only with reference to Japan to avoid confusion. In Japan, earthquake archaeology is considered a branch of archaeology, which itself is considered a branch of history. Its sister discipline, historical seismology (Ishibashi, 2004), has concentrated on the documentary evidence for earthquakes in Japan, beginning with the earliest chronicles of 712 C.E. However, the editors of the most recent compilation of papers in the special issue of the Journal of Seismology (Galadini et al., 2006) treat archaeoseismology as a developing branch of seismology, emphasizing the refinement of quantitative seismological analyses. The Stiros and Jones volume (1996) stressed that this new subdiscipline, whatever it is called, needs full interdisciplinary status to accomplish the goals of both neotectonic geology and archaeology in terms of identification and documentation of past seismic events, reconstruction of seismic cycles, assessment of earthquake damage on cultural properties, and methods of prevention and conservation in earthquake-prone areas. Caputo and Helly (2008) have contrasted several of these new subdisciplines, but not always to the best informative level, as we shall see in the following discussion. Earthquake archaeology in Japan accomplishes only a few of Stiros and Jones’ objectives. It is a rather narrowly defined field, confined to what is discovered at archaeological sites rather than including damage to cultural properties in general. However, given the reduced rate of excavation in today’s economic retrenchment, publicly employed archaeologists (more than 7000 at the peak time in the late 1980s) are being asked to address the built environment as well, assessing extant cultural properties. Still, there is a split between what is seen as earthquake damage to standing buildings and earthquake activity in the archaeological record. For reasons detailed later herein, building damage is not a major concern in Japanese earthquake archaeology, unlike archaeoseismology in other parts of the world (Table 2). The types of earthquake evidence at archaeological sites are totally different in the Mediterranean than in Japan: in the former area, buildings and layers of destroyed cultural remains (destruction layers) are of concern, while in the latter, sediment deformation is the focus. Around 75% of the sites in the DCAN volume (1996) bore evidence of liquefaction, while fissures and faults were reported at far lower levels of incidence. This chapter will explain why buildings per se are not a prominent component of Japanese earthquake archaeology and how sediment deformation, in its more geological sense, is used instead to glean information about past earthquakes.

83

BUILDINGS Damage to Traditional and Monumental Architecture Buildings rarely survive in Japan’s archaeological record, so Japanese archaeology is, perforce, mainly post-hole archaeology. Prehistoric buildings were mostly pit houses constructed of organic materials such as wood, thatch, reed mats, and wattleand-daub; they have not survived except for their house pits and post-hole impressions, or sometimes as architectural fragments in water-logged strata (Fig. 2). From the sixth to eighth centuries, Chinese-style palatial architecture was introduced, first through Buddhist temple architecture, and then through state administrative architecture (Fig. 3). The buildings themselves were similar to the indigenous architecture, using load-bearing wooden posts and wattle-and-daub walls; however, the roofs were laid with ceramic tiles, and the posts were set on foundation stones, while the whole structure was underlain by a pounded-earth foundation platform. Bricks, whether sun-baked or fired, were unknown until the nineteenth century, though some earlier Chinese-style buildings had ceramic tile floors. Western-style brick buildings were very

TABLE 2. PREDOMINANT TYPES OF EARTHQUAKE DAMAGE AT ARCHAEOLOGICAL SITES IN THE MEDITERRANEAN COMPARED WITH JAPAN Mediterranean (Galadini et al., 2006) 1. Displacement 2. Building damage 3. Building deformation 4. Destruction layers Japan (Sangawa, 1995) 1. Fault slippage (dansō) 2. Landslides (jisuberi) 3. Cracks (kiretsu), fissures (jiwari) 4. Sand eruptions (funsa)

Figure 2. A raised storehouse of Yayoi period agriculturalists, ca. first– second century C.E., reproduced at the Toro site, Shizuoka Prefecture. The superstructure is based on rare wood fragments preserved in water-logged strata. (Author’s photo.) Total height of storehouse: 4.3 m; floor: 1.45 m above ground; floorspace: 4 × 2.5 m.

84

Barnes

Figure 3. Chinese-style architecture with foundation platforms, foundation stones, loadbearing pillars, wattle-and-daub walls, heavy roof bracketing, and ceramic roof tiles (after Kidder, 1972, Figs. 45, 52, 56 therein; author’s photo). Top: Scale drawing of the Main Hall at Yakushiji Temple, late seventh century. Lower left: Carved foundation stones and bases of pagoda central pillars and Yakushiji pagoda, built in 698 C.E. Lower right: Bracketing system of Main Hall at Tōshōdaiji Temple, postdating 759 C.E.; note uncarved foundation stones and pillar bases.

vulnerable to earthquake damage, as discovered in the 1891 Nobi earthquake (Clancey, 2006). Stone was not a common building material until the Medieval period (1185–1603) when castles began to be constructed; before that, however, tombs of the Kofun period often had burial chambers of dry-wall stone construction and coffins carved of tuff. Chinese-style buildings also used stone as facing material for the foundation platforms, for eaves-drip drainage facilities around the structure, and for the foundation stones themselves. Tombs can be considered as monumental architecture, and Caputo and Helly (2008) include them in their “buildings” category while bundling “buildings” into their “artifacts” category. Nevertheless, these authors misrepresent the Japanese archaeological record (as well as Japan’s seismic instrumentation record), indicating only a 500 yr depth back to the sixteenth century in their Figure 3 for artifacts (buildings?) in Japan. Mounded tombs (artifacts or buildings in Caputo and Helly’s terms) have been built in Japan over several centuries, dating back to the first century B.C.E., but mostly between 250 and 710 C.E. They are

particularly subject to landslips, fissures, and fault displacement, which can disrupt both their internal stone chamber structures and their external surfaces (Fig. 4). These deformations are particularly difficult to date except as occurring after tomb construction (a terminus post quem date); since the tombs are rarely reused, there are usually no later cultural materials to sandwich the seismic event. Many building foundations and facilities such as house pits, pounded-earth platforms, and stone-lined drainage canals will also show fault displacement, fissures, and crack damage (Fig. 5), but the buildings themselves have often collapsed and/or were burned in the fires that accompanied the earthquakes; rain and rot then dispersed the remains. Survival of Traditional Architecture Rapp (1986, p. 368) noted that “well built wooden structures will flex under the stress of strong earth vibrations” but that “in most structures the locations where components are joined

Earthquake archaeology in Japan

85

Figure 4. Small landslides on the Ōyama Tomb, estimated five occurrences (A–Dʹ) (courtesy of A. Sangawa, modified by Durham Archaeological Services).

Figure 5. An Early Kofun–period house pit cut diagonally by a sand dike, assigned to the earthquake of 701 C.E. Various pits and circles of burned earth are illustrated on the house floor. (After DCAN, 1996, p. 759.)

together are the weakest points.” Despite Rapp’s assessment, there are no better built wooden buildings that those in the Chinese style, and it is exactly their joinery that makes them flexible. Two aspects of this style of architecture (Fig. 3) allowed buildings to survive earthquakes if they avoided fire: their mortice and tenon joinery construction and their foundation-stone

setting. The former allowed the building to flex and shake with the tremors, then come to rest in its original state. The latter, load-bearing pillars set on stones without any other fixtures, allowed the pillars to move separately from the foundation platform, comprising a sophisticated and ingenious ancient baseisolation system.

86

Barnes

Though both China and Japan can be considered “seismic cultures” (Stiros, 1995, p. 726), there is as yet no evidence that these building techniques, dating to the late second millennium B.C.E., were active responses to earthquake damage. They might, however, be viewed as the result of “natural selection,” in that buildings of such structure survived more often than others. However, we do know that the Chinese were sensitive to earthquakes: the first seismograph was invented in China by Zhang Heng, a Han court mathematician who operated it between 132 and 139 C.E. (Pajak, 2005). His bronze seismograph has recently been re-created from fragments buried in his tomb (Xinhua News Agency, 13 June 2005). Japan has the oldest extant wooden buildings in the world, dating back to the seventh century, and most are Chinese-style architecture. Due to the ravages of fire, many original buildings have been replaced through time, but even these replacements are often several hundred years old. In 1964, Mutō Kiyoshi, professor of structural engineering at Tokyo University, ostensibly modeled his innovative flexible steel-frame lattice on the Kan’eiji Temple pagoda, one of the few buildings in Tokyo to survive the 1923 Kantō earthquake intact (NHK, 2002). The structure of a pagoda, with a central pillar set into a foundation stone and the wooden multistoried structure arranged around it with bracketing (Fig. 3), enabled the building to shake independently around the central pillar. Mutō’s flexible steel-frame lattice was first used in the Kasumigaseki Building, the first skyscraper in Japan, completed in 1968. Thus, Japan has a history of wooden architecture, and those wooden buildings that have survived through the ages are generally shrine or Chinese-style temple buildings still in use, which were usually repaired relatively quickly if they sustained earthquake damage. Most other wooden buildings, whether collapsed by earthquakes or not, succumbed to rot or fire and have disappeared from the archaeological record, unless fragments are preserved by water-logging. Thus, earthquake archaeology will recover data from building foundation remains but not superstructure. Japanese archaeologists do not generally work on standing buildings, which are assessed within the Cultural Heritage sector, particularly in conjunction with designations as National Treasures or Historic Sites. The UNESCO World Heritage Site program was inspired by Japan’s preservation efforts, dating back to the late 1800s, in designating important cultural properties for protection and maintenance. Unlike World Heritage Sites in many other parts of the world, however, Japan’s historic sites are still living buildings, not ruins. EARTHQUAKE TYPES AND ARCHAEOLOGICAL CORRELATIONS No country is more likely to be subjected to earthquakes than Japan. The archipelago sits on a conjunction of at least four plates in a subduction zone (Fig. 6). Subduction earthquakes are caused by the descent of the Pacific plate in the northeast and the Philippine plate in the southwest; there is also an incipient sub-

duction zone developing along the edge of the Japan Sea in the northwest as the backarc basin begins to close. Most subduction earthquakes to date are recorded for the Pacific seaboard, while Japan Sea–side earthquakes have so far been assigned to landbased active fault activity. Active faults, defined as rock fractures caused by pressure with vertical and/or lateral movement of rock bodies against each other, occur only in the upper 20 km of brittle crust. In Japan, these are created by the archipelago itself being squeezed between the continental Amur plate in the west (part of the Eurasian plate) and the oceanic plates to the east. The continental plate is moving eastward at up to 1 cm/yr, through Himalayan collision escape tectonics, while the oceanic plates are moving westward at 2–5 cm/yr (Taira, 2001, p. 112, 114). The stress loading on Japan, caught in between, is much higher than for subduction earthquakes, so that active fault earthquakes are both more infrequent and much stronger than ones originating in subduction. Active faults are said not to produce earthquakes of anything less than magnitude 6.5, notated here as M 6.5. Between 1975 and 1979, the Active Fault Survey Group (RGAFJ, 1992) documented thousands of faults across the archipelago and on surrounding continental shelves as known through surface and submarine features such as horizontal displacements, fault scarps, etc. These faults are ranked into three “Certainty” groups with reference to the likelihood of their having been active in the present Quaternary period: I > 90% probability; II > ~50% probability; and III lower probability of activity. Some 80% of Japanese earthquake epicenters of ≥M 6.5 in the last century have occurred on or within 5 km of a mapped active fault (RGAFJ, 1992, p. 37), and maximum magnitudes have been predicted for active fault earthquakes likely to occur in specific fault zones within Japan (RGAFJ, 1992, Fig. 5.1 therein). Earthquakes from both subduction and active fault causes are noted in many historical sources but are, of course, undifferentiated according to type. The earliest believable historical mention of an earthquake in Japan, attributed to 599 C.E., occurs in the Nihon Shoki chronicle, which was compiled in 720 C.E. The oldest actual listing of earthquakes dates to 900 C.E., documenting 700 earthquakes before 887 C.E. (Ishibashi, 2004, p. 340, 344; see also Usami, 1988). The current collations of “historical earthquakes”—defined by Ishibashi as those occurring before seismological instrumentation was developed in Japan—run to 25 volumes and are cited in Ishibashi (2004). The first modern seismographs were developed in Japan between 1880 and 1883 through the collaboration of four men (J.A. Ewing, T. Gray, S. Sekiya, and J. Milne), and from 1884, Gray-Milne seismographs began to collect data in Japan and Britain (Utsu, 2003). Thus, instrumentation records in Japan run back ~125 yr and are not limited, as implied by Caputo and Helly (2008), to the modern seismological network covering Japan today. This modern network gives immediate information on current earthquakes, and within 4 or 5 min of occurrence, the location of the epicenter, depth, and magnitude are posted on the website of the Japan Meteorological Agency (www.jma.go.jp/en/quake/).

Earthquake archaeology in Japan

87

Figure 6. The Japanese archipelago at the junction of two continental plates, the Amur (Eurasian) and Okhotsk (North American) plates, and two oceanic plates, the Pacific and Philippine plates (redrawn by Durham Archaeological Services as Barnes, 2008, Fig. 2 therein). I-STL—Itoigawa-Shizuoka tectonic line; MTL—median tectonic line.

One of the more interesting aspects of earthquake archaeology is the ability of scientists today to collate these historical and instrumentation records of earthquakes and differentiate their causes, as described in the next two sections. Subduction Earthquake Damage Figure 7, compiled by Sangawa (2001), correlates earthquake evidence sets recovered from 33 archaeological excavations with major subduction earthquakes as recorded in the historical documents after 684 C.E. The map in Figure 7 shows the archaeological sites with earthquake damage; the offshore troughs (Nankai to the left, Sagami to the right) are indicated by dotted lines. The troughs are divided into regions (A–F) of earthquake activity. The chart below this map shows the dates of the earthquakes on horizontal lines that indicate their regions of occurrence; the short vertical lines indicate the site (numbers keyed to the map) and date range of its earthquake evidence. Site damage occurs in clusters of sites located in the Tōkai (1–18), Nankai (19–30), and Kantō regions (31–33). Earthquakes in the

Nankai and Tōkai regions are generated in Nankai Trough rupture zones A–B and C–E, respectively, by Philippine plate subduction under southwestern Japan. Kantō earthquakes are generated northeast of the Sagami Trough (F) by Pacific plate subduction. The last great Kantō earthquake was in 1923, and the next big one is apparently being delayed by the wedging of a broken piece of Pacific plate in the subduction zone between the continental plate and oceanic plates (Toda et al., 2008; Miller, 2008). Two patterns are apparent in the horizontal date lines of Figure 7, indicating the regional extent of historically documented earthquakes. First, earthquakes occur simultaneously or nearly so in rupture zones A–E, as would be expected by the subduction of a single plate; however, not all zones are always attested in the documentary or archaeological records. Second, the time depths for the different zones are quite different, with far fewer earthquakes recorded in regions D–F before 1498. This has to do with the location of the ancient capitals where literacy rates were higher: areas of old literacy documented more incidents. The capitals were located in the Kinai (around sites 11–14, 15, 17) until the late twelfth century, and then in Kantō thereafter.

88

Barnes The archaeological dates indicated by vertical date ranges were obtained from stratigraphic and artifactual associations with damage assigned to earthquakes. Several sets of relative dates predate historical records at sites 8, 9, 13, 17, 18, and 26–28. These sites were assigned dates according to their cultural contents, and at least at site 8, it appears that earthquakes struck twice. The problems of relative dating of earthquakes, as well as of assigning observed damage to earthquake causes, are discussed more fully later herein. Active Fault Earthquake Damage The 1596 Fushimi earthquake in the Kinai region (Fig. 8) was copiously documented in the historical record, and several active faults are proven (through trench excavations on the fault lines) to have caused the widespread damage (Sangawa, 2001, p. 109). Such excavations have been carried out in the last few decades as part of another new discipline, earthquake geology (cf. Caputo and Pavlides, 2008). Geological excavations took place in the Arima-Takatsuki tectonic zone and included the Nodao and Higashiura faults, both on Awaji Island. Furthermore, 31 archaeological excavations in the Kinai region have revealed damage that may be correlated with the 1596 event (Fig. 8), several of which are detailed in Table 3. Identifying and Dating Earthquake Damage

Figure 7. Subduction earthquakes of southwestern Japan from the historical and archaeological records (after Sangawa, 2001, Figs. 64 and 65 therein; modified by Durham Archaeological Services). See text for explanation.

As seen in Table 3, the relative dating by archaeological artifacts is rather loose, and many assumptions are involved in choosing to assign cause of the damage to the Fushimi 1596 active fault earthquake. The subduction earthquake nearest in time, the 1605 Keichō earthquake (M 7.9) (Fig. 7, zones A–D), is the best competitor, but Sangawa states that its intensity inland would have been too weak to cause the observed liquefaction damage (seen at sites 10, 25, 26, 31). Liquefaction is acknowledged to occur at not less than M 5 and to become common at M 5.5–6 (Obermeier, 1996, p. 331); the land-based Fushimi earthquake has been assessed at M 7.5 (Kanaori and Kawakami, 1996), and the hypocenter(s) of the various faults was located much closer to the area of damage than was the Keichō earthquake. Thus, the late Medieval deformation in archaeological sites is assigned to the nearby active fault activity causing the Fushimi earthquake rather than to distant subduction activity causing the Keichō earthquake. For reference, the Kobe earthquake of 1995, caused by the Nojima fault in line with the Arima-Takatsuki tectonic line, measured M 7.3 and caused extensive liquefaction in the reclaimed land areas near Kobe as well as widespread damage in the Kinki region of Osaka, Nara, and Kyoto. A significant difference in recurrence time characterizes the two types of earthquakes: subduction (interplate) earthquakes recur within 100 yr cycles, while active fault (intraplate, plate boundary–related) earthquakes recur in periodicities of more than 1000 yr (RGAFJ, 1992, p. 41). Paleoseismologists

Earthquake archaeology in Japan

89

Figure 8. Active faults (FS) in the Kinai region and archaeological sites exhibiting damage from the 1596 Fushimi earthquake (after Sangawa, 2001, p. 88, redrawn by Durham Archaeological Services). R—Rokkō, A—Arima-Takatsuki, U—Uemachi, I—Ikoma, N—Nara, MTL—median tectonic line.

TABLE 3. DAMAGE ASSIGNED TO THE FUSHIMI EARTHQUAKE OF 1596, DATED WITH REFERENCE TO THE MEDIEVAL PERIOD (1185–1603) AND EDO PERIOD (1603–1868) 8. Imashiro-zuka tomb: A landslide of the early sixth-century mound face into the inner moat, covered with sediment dated by radiocarbon. 10. Tamakushi site: Sand liquefaction dike cut Medieval stratum, then was overlaid by Edo stratum. 13. Osaka Castle, outer moat: A landslide archaeologically dated between 1583 and 1598. 21. Ashiya ruined temple: A late seventh-century temple through which ran a 1-m-wide fissure, which collected sand and dirt clods from a rainy period, then was filled with roof tiles made only in the late sixteenth century. Contemporaneous documents describe how a typhoon hit 5 days after the 1596 earthquake, filling the fissure with sand and roof tiles. The earthquake is written to have destroyed the temple, and it was abandoned at that time. 25. Nishi-motomezuka tomb: Sand in moat liquefied to intrude Medieval stratum above, then was overlaid by Edo stratum. 26. Hyōgo-no-tsu site: Sand boils existed just underneath a fire layer dated to 1596. 31. Shimonaizen site: A sand dike cutting through a stratum containing Medieval artifacts, then covered with an Edo stratum. Note: Site numbers are keyed to Figure 8; descriptions were extracted from Sangawa (2001, p. 110–115).

interested in the elucidation of earthquake cyclicity are the main consumers of earthquake archaeology data. Based on current information, a very large combined Tōkai-Nankai earthquake is now expected before the mid-twenty-first century; it promises to affect all A–E rupture zones with great damage (Sangawa, 2001, p. 83–84). In filling out the cyclicity chart to refine expectations of future earthquakes, the main problems in Japan (as elsewhere) relate to identifying damage or deformation at archaeological sites that can be unequivocally assigned to an earthquake of a specific type and not some other cause, and to date them sufficiently.

EARTHQUAKE EVIDENCE IN SEDIMENTS Liquefaction Features Soil liquefaction was first recognized in Japan in the 1948 Fukui earthquake; studies of it increased after the 1964 Niigata earthquake (Yasuda, 2005, p. 152), when apartment buildings fell over sideways in softened sediments with relatively little damage. Liquefaction occurs only in sediments (e.g., silt, sand, gravel) saturated with water and usually requires a relatively impermeable layer (e.g., of clay) somewhere above them, which prevents

90

Barnes

the water from dispersing as pressure builds (Obermeier, 1996, p. 334–335). Ground shaking increases the pressure of the water, which fills the “pore” spaces between the sediment grains, and in extreme cases, puts the sediment particles into suspension. Ishihara and Cubrinovski (2005, p. 10) stated that “the motion in the direction of maximum intensity is the key component of the ground motion that effectively controls the development of pore water pressures and consequent ground deformation.” Sand eruptions are the usual result of subsurface liquefaction, as sediments (not just sand but silt and gravels) are pushed out to the surface with the pressurized groundwater as sand dikes, blows, or boils. Usually, the ground surface becomes depressed and cracked as water is expelled from the lower layers (Obermeier, 1996, p. 335). Obermeier also described the phenomenon of lateral spreading as an adjunct to liquefaction, happening on level ground of less than 5% incline. When an underlying stratum liquefies, the capping stratum fissures or breaks into segments, spreading laterally, especially when the cap is faced on one side by an abrupt change of slope like a terrace scarp or river bank (Obermeier, 1996, p. 335–337). Although Obermeier states that the liquefied layer itself does not undergo a change in volume (Obermeier, 1996, p. 333–334), Ishihara et al. (1997, p. 23), based on data from the 1995 Kobe earthquake lateral spreading, postulated that dilation of the overlying ground mass must have occurred, and that “it appears highly likely that the liquefied soil was sucked into the masses instead of being ejected to the surface.” The year 1985 was the year that liquefaction features were first recognized at Japanese archaeological sites (Takahama et al., 2000, p. 157), but as far as is known, lateral spreading, with its surface fissures indicative of subsurface liquefaction, has not yet been identified in the archaeological record and has not been methodologically addressed, though its existence is well attested in modern earthquakes (Ishihara et al., 1997; Kuribayashi and

Tatsuoka, 1975; Nagase et al., 2006). Galadini et al. (2006, p. 400) include liquefaction and lateral spreading as “off-site paleoseismological” effects that may damage buildings. In Japan, liquefaction dikes characteristically cut through cultural layers, or through features such as pits and ditches; the cut-and-fill relations help date the earthquake occurrence rather than contributing greatly to an understanding of destruction to a built environment. Numerous forms that liquefaction features can take at archaeological sites are illustrated by Sangawa (Fig. 9). One of the most useful aspects of earthquake archaeology is instructing archaeologists how to recognize sand dikes from their widening at the bottom and connection to the liquefied sand layer underneath, so that these features are not mistaken for canals or ditches. Another diagnostic comes from grain-size analysis: sand dike sediments fine upward, distinguishing them from fissures propagating from above, which fill in from above and fine downward. Takahama et al. (2000) have also elucidated the “draw-in” phenomenon of liquefaction, where at the end of a sand eruption, surface materials are sucked back into the top of the eruption path (Fig. 10). Often, the surface layer contains cultural materials that can be used to relative-date the eruption. Again, careful excavation will reveal the connection with the disturbed sand layer below, so that the feature—if circular—is not mistaken for a post hole. Shaking of saturated sediments can have quite opposite effects in the sedimentary record. The most extraordinary case of liquefaction in the Japanese archaeological record is at the Izumida site in Fukui Prefecture, dated to the middle of Late Yayoi (ca. 100 C.E.) (Tomiyama, 2009). Here, a large vein of cobbles up to 20 cm in diameter breached the surface; Yayoi people were apparently so surprised at this that they embedded a 35-cm-tall standing stone at the head of the eruption (Fig. 11). Conversely, at the Nishi-Sanso/Yakumo-Higashi site in Osaka, mild liquefaction features have been recovered in section (Fig. 12). Dish and pillar structures are identifiable in the upper sand stratum; dike

A F E

D C

B

G

I H

Figure 9. Liquefaction and faulted structures revealed in archaeological sites (modified from Sangawa, 2001, p. 52). I–IV—strata; black areas—cultural features; A—sand boil later than strata II and III but earlier than stratum I; B—sand dike cutting through earlier sand boil D into preexisting cultural feature; C—sand eruption cutting through existing stratum III and cultural feature but truncated by surface razing before stratum II was laid; D—sand boil cutting through existing stratum III before stratum II was laid, cut by later sand dike B; E—deflected sand eruption; F—fizzled sand eruption; G—sand dike cutting through existing strata II and III but cut by later cultural feature; H, I—normal listric fault slips; X—pillar structures, Y—dish structures; Z—disturbed area.

Earthquake archaeology in Japan

91

Figure 10. Liquefaction draw-in of cultural materials from overlying cap (after Takahama et al., 2000, their Fig. 9). (Top) Natural stratigraphy before earthquake. (Middle) Sand and gravel erupts to surface. (Bottom) Erupted material partially drawn back into eruption vent.

intrusions strike upward, some obviously abortive, while two small efforts at sill formation are visible. In sum, sand eruptions and lateral spreading are two immediate effects of liquefaction that can be incontrovertibly assigned an earthquake cause. Obermeier (1996, p. 334–337) has implied that sand dikes, sills, and dish and pillar structures are known results of liquefaction and that liquefaction itself is a result of earthquake activity; no other earth motions have the shaking severity necessary to raise pore pressure to explosive levels. Accepting this, we may assume that liquefaction features at an archaeological site, if properly identified, are evidence of past earthquake events. The advantage of finding these features at sites rather than in the great geological outdoors is the association of cultural materials that may aid in dating the events. Soft-Sediment Deformation Structures A more difficult arena in determining earthquake effects is soft-sediment deformation. Fine-grained sedimentation is characteristic of low-energy depositional environments such as ocean floors and lake bottoms. The muddy deposits that collect there can be disturbed and distorted either during or after sedimentation

Figure 11. Liquefaction eruption of ~20-cm-diameter cobbles and cultural isolation as sacred site, Late Yayoi Izumida site, Fukui Prefecture (courtesy of A. Sangawa). Cobble layer can be seen in lower strata in trench exposure.

92

Barnes

A C

B

by several processes—not only by seismic shaking but by “loading during rapid sedimentation, localized artesian conditions, or slumping” (Obermeier, 1996, p. 332). The deformation structures formed under these pressures are patterned and have been given evocative names, including ball-and-pillow, flame, water-escape structures, neptunian dikes, load casts, slumps, and convolutions. These comprise plastic deformations of the soft sediments, which then lithify over time; they can be seen either in rock formations going back to Paleoproterozoic times or in modern unlithified sediments. The challenge is to prove which ones were caused by earthquakes rather than by some other disturbance. As Guidoboni emphasized (1995, p. 10–11), structural damage and landscape changes cannot be assumed to result from earthquakes but must be proven so. Obermeier (1996, p. 332) noted that the geological literature is “replete” with references to “seismites”—just such patterned disturbances as are assumed to have been caused by earthquake shaking. The last decade has seen a concerted effort to assess critically the formation processes of soft-sediment deformation structures (Ettensohn et al., 2002; Shiki et al., 2000), and the term “seismite” has somewhat fallen out of favor as an initial descriptor. Even after sufficient investigation, a suspected seismite might be prefaced with different degrees of inclusion in that category as certain, probable, likely, or possible (Greb et al., 2002). Several authors offer criteria by which to determine an earthquake cause for deformed sediments (e.g., Rossetti and Góes, 2000; Moretti, 2000; Greb et al., 2002; Greb and Dever, 2002; Davies et al., 2004; Bowman et al., 2004; Mazumder et al., 2006). Of these seven sources cited, six specify two important criteria: exclusion of normal processes, and correlative horizons (i.e., lateral continuity) over wide areas. Four speak of the importance of regional abundance of data, so that earthquake attribution does not rely on a single site or single observation. Five specify the obvious criteria of being located in a tectonically active zone, having other tectonic evidence nearby, and occurring in conjunction with deposition (synsedimentary evidence). Three emphasize that the deformed strata should be discrete, occurring sandwiched

C

A

Figure 12. Liquefaction structures at the Nishi-Sanso/Yakumo-Higashi site, Osaka (after Sangawa, 1999, Fig. 8 therein). Eruptions of sand from stratum II into stratum I; A—disturbed area, B—pillar structures, C—dish structures.

between other undeformed strata or strata of different origins. Two note that such deformed strata should recur, possibly rhythmically or cyclically. Two emphasize the importance of deformation immediately postdepositional while still unconsolidated. Finally, individual sources cite further interesting criteria: the necessity to define a core zone of activity; the presence of complex upward deformation; the presence of long periods of quiescence; and the similarity of deformed patterns to modern cases or experimental results. Thus, we cannot immediately discuss the presence of seismites in Japanese archaeological sites without first proving their seismic origins; instead, we must first talk about soft-sediment deformation structures. The criteria for judgment should be the same as proposed already for other areas outside Japan, and it is notable that the initial forays in this direction within Japan very much take the regional or “territorial” approach common to most of the aforementioned authors, as well as Rapp (1986), Obermeier (1996), and Galadini et al. (2006). Damage at any one site or location may have been caused by local conditions, but an earthquake usually has broader geographical repercussions. This approach simply relies on recovering similar evidence of damage from a wide area (several sites’ worth) at a certain date, so that through repetition of data, the conclusion that the damage was caused by an earthquake and not a more localized situation is reinforced. A limitation of this approach, however, is chronological uncertainty that each instance of damage was caused by the same earthquake (Galadini et al., 2006, p. 407, 408): There is always the potential of collapse of different events into one, resulting in the exaggeration of event size (Galadini et al., 2006, p. 408). Matsuda Jun-ichirō’s research in the eastern Osaka Basin, with its attention to regional occurrences of soft-sediment deformation structures in the late Holocene archaeological record, is an important new direction in Japanese earthquake archaeology. A trained geomorphologist and sedimentologist working as a geoarchaeologist, Matsuda has focused on disturbance layers in the stratigraphic succession of alluvial and cultural deposits

Earthquake archaeology in Japan from postglacial Jōmon to recent times. His 2000 publication synthesizes the sequences from six archaeological sites clustered in an area 7 km E-W by 4 km N-S in the southern Osaka Plains (cf. Fig. 8, just east of site 13). The deformation layers are demonstrated to occur in a tectonically active zone at multiple, separated site intervals, and are territorially correlative (Fig. 13). These data appear to fulfill the territorial approach criterion for assessing seismogenic status; their repetitive nature at intervals of several centuries, occurring between undisturbed layers of parallel laminated strata, also argues for occasional large disturbances separated by periods of quiescence rather than routine storm or slumping conditions. For these reasons, the deformations identified by Matsuda may be considered to be seismites. At the Kitoragawa site (Fig. 14), the evidence of softsediment deformation structures in layers relatively dated by cultural materials can be correlated with six or seven historically known earthquakes, all assessed by Usami at magnitudes between M 7 and M 8.4 (Matsuda, 2000, Fig. 11 therein). This seems to go against Obermeier’s statement that “soft-sediment deformations often form at such low levels of seismic shaking

93

that the shaking poses no hazard” (1996, p. 332). Several of the named earthquakes were Nankai subduction earthquakes, which have already been shown to have affected archaeological sites in the eastern Inland Sea region in and around Osaka (Fig. 7). The unaddressed question is, what was the level of intensity of ground shaking at these various sites to create greater (liquefaction) or lesser (soft-sediment deformation) disturbance features? The features investigated by Matsuda were all generated in muddy strata submerged under standing water at the time of deformation: in deltaic back marshes, small shallow lakes, or paddy fields. They appeared in the plans and sections of archaeological trenches; Matsuda cut out blocks and returned them to the laboratory for further analysis, including X-ray radiographs of 1-cm-thick slices (Fig. 15). He identified several kinds of features in stratigraphic succession, going upward: downward fissures and microfaults at the bottom, load structures then plumose patterns, and homogenization at the top. The extensive use of radiographs in addition to sketch drawings of the structures answers Obermeier’s lament concerning liquefaction that “the deformational effects…have rarely been illustrated and discussed

Figure 13. Correlation of soft-sediment deformation zones at different archaeological sites on the Osaka Plains (modified from Matsuda, 2000, Fig. 9 therein).

Figure 14. Soft-sediment deformation zones (DZ) and anthropogenic zones (EZ) at Kitoragawa site (modified from Matsuda, 2000, Fig. 10 therein).

Figure 15. Radiographs of soft-sediment deformation structures (photograph courtesy of J.-i. Matsuda; caption based on Matsuda, 2000). (Left) Deformation zone characterized by, from the top down, liquidized deformation unit with laminae of a weak load structure from a younger deformation, a hydroplastic deformation unit with plumose pattern of light and white intermixed and depressions at its base, and a lower, white hydroplastic deformation unit. Kitajima site. (Middle) Two deformation zones. The top light-colored layer is a young deformation zone. The older zone begins with the dark liquidized deformation unit having an undulated base in response to downward pressure. This is underlain by a hydroplastic deformation unit with twisted plumose pattern; the lower area of microfaults (white vertical fracture lines) indicates a brittle deformation unit. Kitoragawa site. Lower half of this image is a normal photograph taken in daylight. (Right) Two deformation zones with load structure. The darker liquidized homogenized band through the center marks the top of the older deformation zone; the lowermost peaty mud bed is deformed into blocky shapes. The light-colored layers of the younger deformation zone are separated horizontally into a hydroplastic deformation unit with plumose patterns above and a hydroplastic deformation unit below having downward lobes that involve the older homogenized layer. Kitoragawa site. Scale bar = 10 cm.

Earthquake archaeology in Japan in vertical section, the view most useful for paleoliquefaction studies” (1996, p. 332). Several earthquakes were identified through these deformation patterns that are not attested in the historical record: one in early Heian, one in Late Kofun, and at least one in the first half of Yayoi. Prehistoric subduction earthquakes are also attested at other archaeological sites, as shown in Figure 7, though the evidence is quite different. It remains to correlate the soft-sediment deformation structures with other dated sites to construct a potential prehistoric series of earthquakes. Soft-sediment deformation studies have broadened the earthquake data-gathering capacity of archaeological excavation, but few specialists are available to do them. Japanese geoarchaeologists have unknowingly responded to Rapp’s original goal of paying more attention to the sediments in archaeoseismic excavations (Rapp, 1986, p. 370), and this methodological aspect of Japanese earthquake archaeology is well worth exporting, since many other countries are also subject to post-hole archaeology without substantial building remains. CONCLUSIONS Earthquake archaeology in Japan arose simultaneously in the mid-1980s with archaeoseismology in the Mediterranean, but it has taken off in quite different directions. Having few building structures in which to assess earthquake damage, the actual sediments at archaeological sites and their tectonic disruption have come under examination. Geomorphology rather than paleoseismology or historical seismology has been the main disciplinary inspiration; thus, it is the identification of morphological features that becomes the foremost task. Archaeologists are being encouraged to recognize features such as fault displacement, fissures, and liquefaction eruptions, while a few geoarchaeological specialists are engaged in sediment analysis for soft-sediment deformation patterns. Much of the earthquake damage recovered in archaeological excavations in Japan is geological damage to soils, sediments, and strata. The main consumers of earthquake archaeology in Japan have been paleoseismologists interested in establishing recurrence rates. Since such rates are different for subduction and active fault earthquakes, the challenge has been to divide the archaeological evidence to support one or the other series. This relies both on accurate dating and on the definition of broad areas of coseismic damage. Although relative dating by association with cultural artifacts or layers is not as precise as documentary or absolute dating, the Japanese ceramic chronology is the most refined in the world, with generational (20 yr) spans of earthenware types and 5 yr spans of stoneware types in the prehistoric and early historic periods in the best-case examples. This is better than establishing “seismic periods” (Caputo and Pavlides, 2008, p. 3) or spans of a “few decades” (Galadini et al., 2006, p. 407). That relative dates are imprecise may irritate instrumentalists used to dealing with precise measurements, but any time human activities enter the research picture, we must deal with

95

probabilities rather than certainties. Rather than ignore or discard the datings provided by archaeoseismologists, these should be taken as supporting data for working hypotheses about past earthquake occurrence. Earthquake archaeology provides a finer framework for discovering and dating past earthquakes than do trenching activities undertaken by earthquake geologists, and so the additional data should be welcomed. Moreover, extension of research interests beyond earthquake damage to former civilizations (Caputo and Helly, 2008) by including worldwide Paleolithic and Neolithic sites should enormously expand the data available. As Noller and Lightfoot (1997) did for prehistoric sites in America, Japanese archaeologists have demonstrated that any earthquake evidence that occurs in conjunction with cultural materials can be relatively dated, regardless of period or level of social development and whether or not those “artifacts” suffered “damage.” A more thorough treatment of approaches to archaeoseismology worldwide, however, would need to take account of Japanese-language sources and even the increasing body of English-language materials on the Japanese situation, including the depth of historical information about earthquakes in Japanese texts (cf. Ishibashi, 2004) and the instrumentation records that go back to the invention and deployment of the seismograph in Japan and England in the 1880s (Clancey, 2006; Utsu, 2003). Earthquake archaeology in Japan has demonstrated that sites of all ages are repositories of earthquake damage, once archaeologists learn to read the traces. Also, damage need not be restricted to buildings but can affect the actual sediments at archaeological sites. The inclusion of the full range of Japanese historic and archaeological earthquake data cannot but enhance the emerging discipline of archaeoseismology, as would information from other tectonically active regions of the world. ACKNOWLEDGMENTS I am very grateful to Sangawa Akira, Okamura Katsuyuki, and Matsuda Jun-ichirō, who inspired my interest in earthquake archaeology through their writings, who generously discussed with me the finer points of their new discipline, and who have given permission for reuse of their published material. Iain Stewart, Manuel Sintubin, and Tina Niemi facilitated making these Japanese data available in English through inclusion in their International Geoscience Program IGCP 567 project, including presentation of the data at the Seismological Society of America meetings in Santa Fe, 9 April 2008, and the Geoarchaeology meetings in Sheffield, 17 April 2009. Thanks also go to Bruce Batten for his constructive refereeing. My research on this topic has been supported by Durham University, the Arts and Humanities Research Council (AHRC) of England, the School of Oriental and African Studies (SOAS) at the University of London, and the International Research Center for Japanese Studies (Nichibunken) in Kyoto, Japan. Several illustrations were prepared by Durham Archaeological Services (DAS), Durham University.

96

Barnes

REFERENCES CITED Barnes, G.L., 2008, The making of the Japan Sea and the Japanese mountains: Understanding Japan’s volcanism in structural context: Japan Review, v. 20, p. 3–52. Bowman, D., Korjenkov, A., and Porat, N., 2004, Late-Pleistocene seismites from Lake Issyk-Kul, the Tien Shan range, Kyrghyzstan: Sedimentary Geology, v. 163, p. 211–228, doi: 10.1016/S0037-0738(03)00194-5. Caputo, R., and Helly, B., 2008, The use of distinct disciplines to investigate past earthquakes: Tectonophysics, v. 453, p. 7–19, doi: 10.1016/j.tecto .2007.05.007. Caputo, R., and Pavlides, S.B., 2008, Earthquake geology: Methods and applications: Tectonophysics, v. 453, p. 1–6, doi: 10.1016/j.tecto.2008.01.007. Clancey, G., 2006, Earthquake Nation: The Cultural Politics of Japanese Seismicity, 1868–1930: Berkeley, University of California Press, 331 p. Davies, N.S., Turner, P., and Sansom, I.J., 2004, Soft-sediment deformation structures in the Late Silurian Stubdal Formation: The result of seismic triggering: Norwegian Journal of Geology, v. 85, p. 233–243. DCAN (Disaster Concerned Archaeologists’ Network), 1996, Hakkutsu sareta jishin konseki (Excavated evidence of earthquakes): n.p., Osaka, Maibun Kankei Kyūen Renraku Kaigi, 826 p. Ettensohn, F.R., Rast, N., and Brett, C.E., eds., 2002, Ancient Seismites: Geological Society of America Special Paper 359, 190 p. Galadini, F., Hinzen, K.-G., and Stiros, S., 2006, Archaeoseismology: Methodological issues and procedure: Journal of Seismology, v. 10, special issue, p. 395–414, doi: 10.1007/s10950-006-9027-x. Greb, S.F., and Dever, G.R., Jr., 2002, Critical evaluation of possible seismites: Examples from the Carboniferous of the Appalachian basin, in Ettensohn, F.R., Rast, N., and Brett, C.E., eds., Ancient Seismites: Geological Society of America Special Paper 359, p. 109–125. Greb, S.F., Ettisohn, F., and Obermeier, F., 2002, Developing a classification scheme for seismites: Geological Society of America Abstracts with Programs, v. 34, no. 2, p. A-102; also available at http://gsa.confex.com/ gsa/2002NC/finalprogram/abstract_32750.htm (accessed February 2008). Guidoboni, E., 1995, Archaeology and historical seismology: The need for collaboration in the Mediterranean area, in Stiros, S., and Jones, R.E., eds., Archaeoseismology: Athens, British School in Athens, p. 7–13. Ishibashi, K., 2004, Status of historical seismology in Japan: Annals of Geophysics, v. 47, no. 2/3, p. 339–368. Ishihara, K., and Cubrinovski, M., 2005, Characteristics of ground motion in liquefied deposits during earthquakes: Journal of Earthquake Engineering, v. 9, Special Issue 1, p. 1–15, doi: 10.1142/S1363246905002304. Ishihara, K., Yoshida, K., and Kato, M., 1997, Characteristics of lateral spreading in liquefied deposits during the 1995 Hanshin-Awaji earthquake: Journal of Earthquake Engineering, v. 1, no. 1, p. 23–55, doi: 10.1142/ S1363246997000039. Kanaori, Y., and Kawakami, S., 1996, The 1995 7.2 magnitude Kobe earthquake and the Arima-Takatsuki tectonic line: Implications of the seismic risk for central Japan: Engineering Geology, v. 43, no. 2–3, p. 135–150, doi: 10.1016/0013-7952(96)00056-7. Kidder, J.E., Jr., 1972, Early Buddhist Japan: London, Thames & Hudson, 212 p. Kuribayashi, E., and Tatsuoka, F., 1975, Brief review of liquefaction during earthquakes in Japan: Soil and Foundation, v. 15, no. 4, p. 81–92. Matsuda, J.-i., 2000, Seismic deformation structures of the post-2300 a BP muddy sediments in Kawachi lowland plain, Osaka, Japan: Sedimentary Geology, v. 135, p. 99–116, doi: 10.1016/S0037-0738(00)00066-X. Mazumder, R., van Loon, A.J., and Arima, M., 2006, Soft-sediment deformation structures in the Earth’s oldest seismites: Sedimentary Geology, v. 186, p. 19–26, doi: 10.1016/j.sedgeo.2005.12.002. Miller, M., 2008, Breaking the slag: Nature Geoscience, v. 1, p. 730–731, doi: 10.1038/ngeo341. Moretti, M., 2000, The interpretation of soft-sediment deformation structures as seismites, in European Geophysical Society, 25th general assembly (10297006) 01/01/2000 abstracts: www.copernicus.org/EGS/egsga/nice00/ programme/abstracts/aac6959.pdf (accessed February 2008). Nagase, H., Zen, K., Hirooka, A., Yasufuku, N., Kasama, K., Kobayashi, T., Maeda, Y., Uno, K., Hashimura, K., and Chen Guangqi, 2006, Zoning for liquefaction and damage to port and harbor facilities and others during the 2005 Fukuoka-ken Seiho-Oki earthquake: Soil and Foundation, v. 46, no. 6, p. 805–816.

NHK (Japan Broadcasting Company), 2002, Kasumigaseki-Biru—Chōkōsō e no hatenaki tatakai: Jishin rettō Nihon no kakumei gijutsu (Kasumigaseki Building—Endlessly fighting towards skyscrapers: Revolutionary technologies for the Japanese earthquake-prone archipelago): Tokyo, NHK Sofutowea (DVD) [in Japanese]. Noller, J.S., and Lightfoot, K.G., 1997, An archaeoseismic approach and method for the study of active strike–slip faults: Geoarchaeology, v. 12, p. 117–135, doi: 10.1002/(SICI)1520-6548(199703)12:23.0.CO;2-7. Obermeier, F., 1996, Using liquefaction-induced features for paleoseismic analysis, in McCalpin, J.C., ed., Paleoseismology: San Diego, Academic Press, p. 331–396. Pajak, J., 2005, Signal processing in the “Zhang Heng Seismograph” for remote sensing of impending earthquakes, in Gupta, G.S., Mukhopadhyay, S.C., and Messom, C.H., eds., Proceedings of the 1st International Conference on Sensing Technology, 21–23 November 2005, Palmerston North, New Zealand: www-ist.massey.ac.nz/conferences/icst05/proceedings/ ICST2005-Papers/ICST_112.pdf, p. 669–673 (accessed September 2008). Rapp, G., Jr., 1986, Assessing archaeological evidence for seismic catastrophes: Geoarchaeology, v. 1, p. 365–379, doi: 10.1002/gea.3340010403. RGAFJ (Research Group for Active Faults in Japan), 1992, Maps of Active Faults in Japan with an Explanatory Text: Tokyo, University of Tokyo Press, 73 p., 3 maps [English condensed version of 1991 full publication in Japanese]. Rossetti, D.F., and Góes, A.M., 2000, Deciphering the sedimentological imprint of paleoseismic events: An example from the Aptian Codó Formation, northern Brazil: Sedimentary Geology, v. 135, p. 137–156, doi: 10.1016/ S0037-0738(00)00068-3. Sangawa, A., 1988, Kōkogaku no kenkyū taishō ni mitomerareru jishin no konseki: Kodaigaku Kenkyū, v. 116, p. 1–16 [in Japanese]. Sangawa, A., 1995, Kōkogaku no shiryō kara kojishin o saguru (Searching for ancient earthquakes in archaeological data), in Ōta, Y., and Shimazaki, K., eds., Kojishin o Saguru: Tokyo, Kokon Shoin, p. 107–124 [in Japanese]. Sangawa, A., 1999, Palaeoliquefaction features at archaeological sites in Japan: Chigaku Zasshi, v. 108, no. 4, p. 391–398. Sangawa, A., 2001, Jishin: Namazu no Katsudōshi (Earthquakes: A History of Catfish Activities): Tokyo, Daikōsha, 173 p. Sangawa, A., 2007, Jishin no Nihonshi (Earthquakes in Japanese History): Tokyo, Chūō Kōron Shinsha, 268 p. Sangawa, A., 2009, A study of paleoearthquakes at archeological sites: A new interdisciplinary area between paleoseismology and archeology: Synthesiology, English edition, v. 2, p. 84–94. Shiki, T., Cita, M.B., and Gorsline, D.S., 2000, Sedimentary features of seismites, seismo-turbidites and tsunamiites—An introduction: Sedimentary Geology, v. 135, no. 1–4, p. vii–ix, doi: 10.1016/S0037-0738(00)00058-0. Stiros, S.C., 1995, Archaeological evidence of antiseismic constructions in antiquity: Annali di Geofisica, v. 35, p. 725–736. Stiros, S., and Jones, R.E., 1996, Archaeoseismology: Athens, British School at Athens, Fitch Laboratory Occasional Paper 7, 268 p. Taira, A., 2001, Tectonic evolution of the Japanese island arc system: Annual Review of Earth and Planetary Sciences, v. 29, p. 109–134, doi: 10.1146/ annurev.earth.29.1.109. Takahama, N., Ōtsuka, T., and Brahmantyo, B., 2000, A new phenomenon in ancient liquefaction, the draw-in process, its final stage: Sedimentary Geology, v. 135, p. 157–165, doi: 10.1016/S0037-0738(00)00069-5. Toda, S., Stein, R.S., Kirby, S.H., and Bozkurt, S.B., 2008, A slab fragment wedged under Tokyo and its tectonic and seismic implications: Nature Geoscience, v. 1, p. 771–776, doi: 10.1038/ngeo318. Tomiyama, M., 2009, Hayashi-Fujishima Iseki Izumida-chiku. Fukui-ken Maizou Bunkazi Chosa Houkoko, v. 106, p. 87, 89, + photos. Usami, T., 1988, A study of historical earthquakes in Japan, in Lee, W.H.K., and Shimazaki, K., eds., Historical Seismograms and Earthquakes of the World: San Diego, Academic Press, p. 276–288. Utsu, T., 2003, Historical development of seismology in Japan: Swiss National Centennial Report to the International Association of Seismology and Physics of the Earth’s Interior, Ch. 79.33 Japan, Part 2: www.iris.edu/ seismo/info/historical/Utsu2003.pdf, 19 p (last accessed August 2008). Yasuda, S., 2005, Survey of recent remediation techniques in Japan, and future applications: Journal of Earthquake Engineering, v. 9, Special Issue 1, p. 151–186, doi: 10.1142/S1363246905002213. MANUSCRIPT ACCEPTED BY THE SOCIETY 19 MAY 2010 Printed in the USA

The Geological Society of America Special Paper 471 2010

Historical earthquake catalogues and archaeological data: Achieving synthesis without circular reasoning John D. Rucker Tina M. Niemi Department of Geosciences, University of Missouri–Kansas City, Kansas City, Missouri 64110, USA

ABSTRACT The field of archaeoseismology has been plagued by a persistent problem. The problem has been the integration of several lines of evidence to produce a holistic conclusion without entering into a situation of circular reasoning, wherein the sources are used to build on each other without foundation. The four main sources of evidence are historical texts, epigraphy, archaeology, and geology. Any seismic event may appear in any or all of them, but only the most extreme events in fortuitous locations would be expected to appear in all four. This paper uses some aspects of the interpretation of the 551 C.E. earthquake in the Levant to illustrate how this circular reasoning can develop, and how it tends to corrupt the different lines of evidence. We conclude with a suggested new approach, making the database of regional seismic events both more specific and more complete. toriography such as bias, both cultural and geographic, accident of preservation, and availability, confusion, amalgamation, and conflation in the sources. Since 1994, the three primary earthquake catalogs in English for the Levant, Amiran et al. (1994), Ambraseys et al. (1994), and Guidoboni (1994), in addition to Russell’s 1980 and 1985 papers, have been widely utilized. These earthquake catalogues have been viewed as authoritative within the community of users. However, a troubling tendency toward circular reasoning has developed. It is fair to note that two recent articles have begun the process of presenting new comprehensive catalogs and separating the sources of evidence. Particularly notable in doing so are Sbeinati et al. (2005) and Guidoboni and Comastri (2005). Ambraseys (2009) also turned a more critical eye toward separating the evidence, especially in sorting out conflating historical sources. Within the subfield of archaeoseismology, a holistic approach including all of the possible sources of information is especially important, since major seismic events are by their nature regional, with sometimes far-flung and occasionally subtle effects. In this article, we attempt to show how these historical

INTRODUCTION Because the ultimate goal of all historical and archaeological research is to achieve a better understanding of ancient events and cultural lifestyles, it is best to utilize all available sources of information. An archaeological study that ignores historical sources is doomed to at best, much additional work and at worst, failure. Historians who ignore archaeological sources will certainly miss the best independent corroboration of the ancient written sources with which they are working. Both disciplines are wise to consider newly available data from unexpected and far-flung sources as they become available (the revolution that was radiocarbon dating, for example). Thus, working in the Near East, we find ourselves being informed by sources of data as disparate as core samples from the Greenland ice sheets and historical sources from all over Eurasia, as well as our own direct excavations and historical sources from the region itself. Typically, modern earthquake catalogues are constructed by compiling all possible textual references to historic earthquakes. This makes them subject to many of the common concerns of his-

Rucker, J.D., and Niemi, T.M., 2010, Historical earthquake catalogues and archaeological data: Achieving synthesis without circular reasoning, in Sintubin, M., Stewart, I.S., Niemi, T.M., and Altunel, E., eds., Ancient Earthquakes: Geological Society of America Special Paper 471, p. 97–106, doi: 10.1130/2010.2471(09). For permission to copy, contact [email protected]. © 2010 The Geological Society of America. All rights reserved.

97

98

Rucker and Niemi

records, inscriptions, archaeological excavations, and geological data can and should mutually reinforce each other, but are susceptible to misuse, resulting in a situation of runaway circular reasoning, which taints the sources of data. The problem of circular reasoning in archaeoseismology is illustrated in Figure 1. Simply put, once this circular cycle begins, both lines of evidence become corrupted. SOURCES OF EVIDENCE The four main sources of evidence are historical texts, epigraphy, archaeology, and geology. For any given earthquake, evidence from any or all of these sources may be available. Each source, however, is only capable of providing certain kinds of information, and may be completely silent on other aspects of the seismic event. It is thus extremely important that scholars keep firmly in mind what a piece of evidence is not revealing—not just what it is revealing. In many ways, historical text is the most straightforward of the four sources of earthquake data. A respected ancient historian who says “City X was destroyed by an earthquake at 9 p.m. on Monday the 19th of September” is hard to dispute. The historical record can have the advantages of specificity, and clarity of message, but it may also have the familiar historigraphical disadvantages of secondary evidence, exaggeration, and deliberate or accidental misinformation (e.g., Guidoboni and Ebel, 2009). Some problems specific to historical seismology are: There is considerable bias toward regions of denser population and earthquakes of greater magnitude. That is, earthquakes of greater magnitude and/or occurring in areas of denser populations are much more likely to enter the historic record. Also, there is a tendency toward amalgamation of earthquakes for which the occurrences were closely spaced in time. This is due both to the limitations of ancient knowledge and the vagaries of preservation and copying amongst historic texts. When we consider the widely spaced geographic locations of ancient sources, it is not surprising that

Archaeologist finds a destruction layer at Site X and interprets it as evidence for an earthquake.

Historical seismologist adds Site X to the catalogue of cities damaged by that specific earthquake.

Archaeologist uses earthquake catalogue to assign a date to the destruction layer.

Figure 1. Circular reasoning model.

moderate earthquakes with smaller felt areas, though possibly quite severe in their local effects, might escape the notice of a distant chronicler entirely. Earthquake catalogues are collections of dates and reports of the effects of earthquakes as recorded from written records. Most catalogues are thought to be complete for major M >7 earthquakes but may be silent on less severe or less widespread earthquakes. This is, of course, a summary of an extremely complicated type of evidence that must not be taken at face value without consideration of its various problems. Guidoboni and Ebel (2009) dealt with the problems of historical seismology at a length and detail not possible in this paper. Also, Karcz (2004) contains case studies helpful in sorting out the historical sources for several specific ancient earthquakes in the Levant. Another important source for historical seismology is epigraphy. This is the study of surviving inscriptions, usually on structures, statuary, or ostraca, but also including ancient text on any artifact. These are often dedicatory in nature, but include more mundane inscriptions and even graffiti. Since inscriptions are also a written record, they are often lumped together as a subcategory of the historic data. For the purposes of historical seismology, there is an important reason to separate epigraphy from historical sources. Epigraphy is usually a written primary source, without the problems of later copying or interpolation. Thus, it is a slightly but significantly different line of evidence from the rest of the historic record. By separating it, we maintain an independent line of evidence. However, the main disadvantage of inscriptions is that they are generally short, and may or may not do more than hint at the events or issues in question, often enigmatically, as we discuss later with the Areopolis inscription. Epigraphy relies on history to flesh out the details, and archaeological and geological stratigraphy to help constrain both the date and context of the evidence. Obviously, an earthquake happens when a fault ruptures, so investigations of the stratigraphy and geomorphology of seismogenic faults in the region are critical to understanding the history of earthquakes in an area. The main advantage of this line of evidence is that it can provide an absolute record of slip events on that fault. The main disadvantage is that sediments of appropriate age to constrain relevant seismic events may not be present. In addition, the chronological resolution may be too coarse to define a specific historic earthquake. Even though geological study is the baseline of seismology, in many ways, archaeological sites offer better research opportunities, because anthropogenic features may provide better information on both dating and seismic effects than is typically present in a natural geologic setting. Paleoseismic research based predominantly on geologic data also relies on historical accounts and archaeological data to interpret the record of fault movement and secondary seismic effect. Archaeological evidence has its own set of strengths and weaknesses. The first advantage to the archaeological line of evidence is that there are many more archaeological sites than there are surviving ancient texts, and thus is a wider potential distribution of data is available. The second main advantage of

Historical earthquake catalogues and archaeological data archaeological data for historical seismology is that it is free of bias—you are not looking at the evidence through the lens of an ancient historian, nor even through that of an ancient stone carver. However, the evidence is difficult to interpret and can be obscure. Any bias ultimately introduced comes from the modern scholars who interpret (or misinterpret) it. Finally and most importantly, archaeological artifacts can provide a fine resolution of time within the stratigraphic section. The main disadvantage of archaeology as a line of evidence for archaeoseismology is that the stratigraphic and structural damage data are often open to varying interpretations. Furthermore, archaeological information is only available from those sites that have been studied or excavated. Thus, it is at best incomplete, and possibly contains a sampling bias. One area in which a sampling bias already exists in the archaeological record is a definite bias toward study and excavation of larger, more obtrusive sites. This sampling bias may impact any given question in historical seismology. An important example of the ways in which these lines of evidence have been used in the past is the study of the 551 C.E. earthquake in the Levant. This example is important both in that the earthquake in question was a major seismic event, and in that we argue that the way the evidence has been used typifies the problem with circular reasoning we seek to elucidate. THE LEVANT EARTHQUAKE OF 9 JULY 551 Historical Text The main historical sources for the 9 July 551 C.E. earthquake as described by Guidoboni (1994) are Antoninus of Piacenza, Malalas, John of Ephesus, Theophanes, Agathias Scholasticus, and the accounts of the life of St. Symeon the Stylite the Younger. Contemporary or primary sources (those alive at the time of the event) include Malalas, Agathias, John of Ephesus, and Antoninus Placentinus. Later chroniclers such as Theophanes, who wrote in the eighth–ninth century C.E., and others are considered noncontemporary or secondary sources. Secondary sources recount, copy, and often embellish what they have heard or read, thus diminishing the reliability of the information. However, it can also be said that the similarity of prose between primary sources often suggests that these text have also been copied. Several lines of historical evidence in these texts indicate that the earthquake damage was centered on the coast of Phoenicia, modern Lebanon. First, Antoninus of Piacenza describes massive destruction of the cities between Tripoli (Tripolis) and Beirut (Berytus). Second, the account in St. Symeon indicates minor damage north of Beirut and “the area to the south from Tyre to Jerusalem was also preserved” (Guidoboni, 1994, p. 335). Third, the text by Agathias indicates that the law school of Beirut was transferred after the earthquake to Sidon, corroborating evidence that the intensity of seismic shaking was weaker toward the south, and structures there remained intact and functioning. Finally, the occurrence of a seismic sea wave (tsunami) in

99

551 C.E. suggests slip on an offshore fault in addition to the described coastal landslide (Elias et al., 2007). This position is partially shared by Salamon et al. (2007), although they also allow the possibility of a submarine landslide as the result of an onshore rupture. Guidoboni (1994) and a number of previous earthquake catalogers place the epicenter of the 551 C.E. earthquake in coastal Lebanon (Fig. 2A). Two sources of confusion arise pertaining to the epicenter of the 551 C.E. earthquake. First, descriptions found in the texts of Malalas and Theophanes list such widely separated areas as Syria, Arabia, Mesopotamia, Palestine, and Phoenicia as areas affected by the earthquake. The second source of confusion is that several different earthquakes occurred in Turkey, the Levant, and Egypt during the period between 551 C.E. and 561 C.E. (Russell, 1985; Guidoboni, 1994). This has led to several different interpretations of the historical texts. Ambraseys et al. (1994), reading the same historical sources, places the epicenter of the 551 earthquake in the Jordan Valley based on damage in Alexandria, Egypt (Fig. 2B). These authors state “Modern writers place the epicentral region of this event offshore from Lebanon. This is due to the bias of information from the more populous coastal region” (p. 24). Ben-Menahem in his 1979 earthquake catalogue defines a Lebanese coastal epicentral zone, but in a later publication (1991), he lists a 7 July 551 C.E. earthquake in the Gulf of Corinth. Amiran et al. (1994) in their earthquake catalog for Israel and adjacent areas for the 551 C.E. earthquake include a Latin quote (apparently from Theophanes) that translates as “a great and terrible earthquake occurred throughout the regions of Palestine, Arabia, Mesopotamia, Syria, and Phoenica” and also lists several cities damaged in the earthquake beyond the Lebanese coast, including Jerusalem, Jerash, Mt. Nebo, Areopolis, el-Lejjun, and Petra. They state that the “el-Lejjun fortress east of Kerak destroyed. Petra destroyed and never rebuilt.” The inclusion of the earthquake damage at cities in present-day Jordan (Jerash, Mt. Nebo, Areopolis [Rabbath], el-Lejjun, and Petra) is based on archaeological data presented in Russell (1985). In the conclusion of her discussion of the 551 C.E. Beirut earthquake, Guidoboni (1994, p. 336) writes: “It is therefore very likely that the surrounding regions (Arabia, Mesopotamia, Palestine and Syria) mentioned by Malalas and Theophanes, were either subject to secondary effects or to after-shocks with different epicenters.” This leaves open the possibility that historical accounts, although focusing specifically on the major damage in 551 C.E. along the important commercial centers between Beirut and Tripoli, allude to other potential seismic damage farther afield. Were there other local source moderate earthquakes within this decade of 551–561 C.E. centered in Palestine? Guidoboni (1994) describes an account by the chronicler Agathias who personally experienced an earthquake in Alexandria on 15 October 554 C.E. Agathias relates that earthquakes were uncommon in this region, and the population did not build to withstand strong earthquake ground motion. No damage is described for

100

Rucker and Niemi

Figure 2. (A) Felt area of 551 C.E. Levant earthquake (after Guidoboni, 1994). (B) Epicenter of 551 C.E. Levant earthquake (after Ambraseys et al., 1994).

Geological Data

this seismic tremor in Egypt. Could the shaking in Alexandria in 554 C.E. have been caused by an earthquake with an epicentral region located somewhere along the Dead Sea transform in Byzantine Palestine? One must ask, why has a mid-sixth-century layer of destruction been described for sites from northern to southern Jordan? Did the 9 July 551 C.E. earthquake actually cause major damage away from the Lebanese coast, contrary to the historical accounts? Was there a different local earthquake source that was largely contemporaneous to 551 C.E. that has been conflated with it? What exactly does the archaeological data really tell us about seismic activity in Jordan and the Levant in the mid-sixth century C.E.?

Darawcheh et al. (2000) recognized the discrepancy between the two divergent epicentral locations for the 551 C.E. earthquake—one offshore of the Lebanese coast (Guidoboni, 1994), and the other in the Jordan Valley (Ambraseys et al., 1994). A reevaluation of the evidence from primary and secondary Byzantine sources and an assessment of seismic parameters of the 551 C.E. Lebanese earthquake thus seemed warranted. From the available macroseismic data, Darawcheh et al. (2000) designated the Roum strike-slip fault and its potential offshore extension as the likely seismic source and calculated a surface-wave earthquake magnitude of Ms = 7.1–7.3 for the 551 C.E. earthquake. These authors also suggested that assigning earthquake damage to archaeological sites in western Jordan “may be an interpretative error.” New offshore seafloor mapping by Elias et al. (2007) identified the source rupture of the 551 C.E. earthquake as an active low-angle thrust fault located along the Lebanese coast (Fig. 3). This east-dipping Mount Lebanon thrust fault is exposed along the coast south of Beirut, where a beach terrace has been elevated by ~80 cm. Radiocarbon analyses of shells on the uplifted beach terrace date to the sixth century C.E. (Morhange et al., 2006). Comparison of the 551 C.E. uplift and older marine terraces on Mount Lebanon suggests that the repeat time of this type of earthquake may range between 1500 and 1750 yr. Elias et al. (2007) suggested a moment magnitude (Mw) of 7.5 for the 551 C.E. seismic event. Archaeological Data It is very clear that Kenneth Russell’s 1985 summary article entitled “The Earthquake Chronology of Palestine and Northwest Arabia from the Second through the Mid-Eighth Century A.D.”

Historical earthquake catalogues and archaeological data

A B

Figure 3. (A) Epicenter of the 551 C.E. earthquake along the offshore Mount Lebanon thrust fault. Abbreviations: Ar.—Arqa; Ba.—Batroun; By—Byblos; Ch.—Chekka; Sa—Sarafand. AT—Aakkar thrust; TT—Tripoli thrust; R-AF—Rankine-Aabdeh fault; RF—Roum fault; RaF—Rachaya fault; SaF—Saida fault. Image from Elias et al. (2007). (B) Example of the elevated beach terrace near Tripoli (from Elias et al., 2007).

has had a profound effect on both the entries in the regional earthquake catalogues and on defining the stratigraphy at archaeological sites in this region. This may largely be due to the fact that the publication was in a journal widely read by archaeologists. Two commonly used earthquake catalogues at the time prior to Russell (1985) were Amiran (1950–1951, 1952) and Ben-Menahem (1979).

101

Four archaeological sites in Jordan were cited by Russell (1985) supporting the interpretation that the 551 C.E. earthquake damaged the region in present-day Jordan (Mt. Nebo, Jerash, el-Lejjun, and Petra). In Petra, Russell was part of the University of Utah’s excavation team at the Temple of the Winged Lions, where they purportedly found evidence for a mid-sixth-century collapse and possible earthquake. There are very few published data on this site available for us to understand the chronological basis of this interpretation. Russell’s (1985) evidence of earthquake damage at Jerash is based on Crowfoot (1938), Mount Nebo from Saller and Bagatti (1949), and el-Lejjun from Parker (1982). We are not in a position to evaluate the archaeological evidence at Mt. Nebo or Jerash. However, recent published excavation reports from Petra and el-Lejjun do allow us to evaluate the evidence at those two sites. Russell in 1985 (p. 45) wrote “Petra, the capital of Palestina tertia, was never rebuilt after the 551 C.E. earthquake, and by the end of the sixth century C.E., its ruins had become a quarry for liming and smelting operations.” However, recent excavations at the Petra Church archaeological site refute these conclusions (Fiema, 2001a, 2001b). Scrolls found in the Petra Church provide an unprecedented record of Late Byzantine Petra (Fiema, 2002). The church was destroyed in a fire at the end of the sixth or the beginning of the seventh century C.E. The fire carbonized papyrus scrolls that were being stored in the church. These scrolls, known as the Petra Papyri, are a collection of documents predominantly relating to taxes and property ownership, dating from 537 C.E. to at least 13 April 593 C.E., and may postdate this range by several years. Therefore, the last recorded date of the Petra papyri scrolls may extend to 597 C.E. “Neither the effects of the earthquake of 551 C.E. nor the mid-sixth century C.E. plague can be discerned from the texts” of the scrolls (Fiema, 2002, p. 4). After the fire and into the seventh century C.E., the church ceased to function as an ecclesiastical structure, building materials were salvaged for reuse, and the shell of the structure was converted to a domestic complex. Fiema (2001a, 2001b) noted evidence for two earthquakes in the later phases of the Petra Church— one in the seventh century C.E. and one in the medieval to Ottoman period—at which time, no columns remained standing. As recounted already, excavations in the 1990s at the Petra Church and textual evidence from the newly translated Petra Papyri have convincingly demonstrated that the city of Petra was not apparently appreciably affected by the 551 C.E. earthquake. Unfortunately, some excavators still designate a 551 C.E. earthquake in the stratigraphic sequence at Petra. The site of el-Lejjun, a fourth-century Roman legionary fort located east of the Dead Sea, excavated by Parker (2006) over five seasons from 1980 to 1988, is in some ways an example of the way in which the correlation of the several lines of evidence should work (Fig. 4). Parker (2006) found evidence of a collapse horizon that contained coins, the most recent of which dated to 540–541 C.E. Admittedly, this does support an interpretation that this collapse horizon may be dated to the historically recorded earthquake of 551 C.E. It is possible that this is entirely correct,

102

Rucker and Niemi

Jerusalem

although unlikely based on the majority of the other evidence. It is extremely important to note, however, that Parker’s evidence in no way precludes a slightly more recent date for the destruction layer at el-Lejjun. Based on Parker’s excavations at el-Lejjun, Amiran et al. (1994) lists el-Lejjun as destroyed in 551 C.E. Thus, this piece of only archaeological evidence, which is only partially corroborated, enters the historical record as fact. What should be a fruitful cross-pollination (and might still be in this specific case) has brought into question the integrity of both lines of evidence.

Madaba

Hebron

Diban

Dead Sea

Epigraphic Data and/or Earthquakes Not in the Catalogue?

Areopolis En Boqeq

Kerak

Betthorus (el-Lejjun)

Zoara (Safi)

Tamar

31°00’

Praesidium (Fifa) Toloha (Telah)

Mampsis

rabah

Hatzeva

Wadi 'A

Feinan

Petra Bir Madkhur

Udruh

Areopolis, modern-day Ar-Rabbath (Fig. 4), located east of the Dead Sea and west of el-Lejjun, is listed as one of the cities damaged in the earthquake of 9 July 551 C.E. in the earthquake catalogue of Amiran et al. (1994). Zayadine (1971) published the translation of a dedicatory inscription found at Areopolis that states “Restored in 492 (597–598 C.E.) after the earthquake” (Fig. 5). This block was found out of context, reused in a structure, and interpreted as referring to a previously unattested earthquake (Zayadine, 1971). Russell (1985) suggested that there was a long (46 yr) time lag between the 551 C.E. earthquake and the reconstruction of Areopolis. The long recovery time was postulated to be due to “a depressed economic environment” (Russell, 1985, p. 50). This suggestion, that the general economic depression of the region caused a 46 yr delay in reconstruction after the earthquake, is unlikely for several reasons. First, earthquake reconstruction typically takes place soon after the event or not at all. For example, the Byzantine Emperor Justinian provided funds for the immediate rebuilding of the cities of

Ma'an

Gharandal 30°00’

Humeima

Gulf of Aqaba

0

N

Aila ('Aqaba)

Via

No va

Tra

ian

a

Yotvata

30 km

35°30’

Figure 4. Map of southern Jordan showing the sites and roads during the Roman period. (Modified from Talbert [2000], p. 71 and 76).

Figure 5. A dedicatory inscription found in secondary context at the ancient site of Areopolis, east of the Dead Sea, which reads: “Restored in 492 [597–598 C.E.] after the earthquake” (from Zayadine, 1971).

Historical earthquake catalogues and archaeological data the Lebanese coast after 551 C.E. (Russell, 1985, p. 45). When Pompeii, was entombed by ash from the eruption of Mount Vesuvius in 79 C.E., damage from the devastating earthquake of 62 C.E. had mostly been repaired. San Francisco was rebuilt and able to host the 1915 Panama Pacific International Exposition after the1906 earthquake and fire. As we see in the aftermath of the 2010 earthquakes in Haiti, Chile, and Mexico, societies both modern and ancient quickly repair earthquake-damaged areas (Sintubin et al., 2008). Second, if it were to occur at such a chronological distance as 46 yr after the event, it is unlikely that a dedicatory inscription would mention the earthquake. Average life expectancy would suggest that there would be relatively few people still around who remembered the event. Is it possible that the earthquake that destroyed Areopolis was a later event that is not chronicled by historical text? Since the Petra Papyri do not date beyond 593–597 C.E., could an earthquake around 597 C.E. have also caused damage in Petra? Do archaeoseismic and paleoseismic methodologies have the ability to discern between a 551 C.E. and 597 C.E. earthquake? It is interesting to note that the earthquake catalogue of Ambraseys (2009, p. 216– 217) now includes an earthquake entry for a seismic event at “IX shaking in the past millennium. This conclusion is, however, more tenuous than the conclusions one can draw about shaking during nineteenthand twentieth-century earthquakes. Estimates of shaking intensity at a single site provide little indication of earthquake magnitude; instrumental recordings of recent earthquakes reveal that very high (PGA [peak ground acceleration] >1 g) accelerations can be generated by relatively moderate (M 6.5–7) earthquakes, while surprisingly low accelerations are sometimes recorded in the near field of very large events. Our findings are thus clearly insufficient to draw conclusions about the magnitudes of earthquakes that have shaken Kashmir Valley in the past millennium, nor do our results provide upper limits to the shaking experienced in historical times in nearby Srinagar, where thick sediments in the Jhelum River valley and around lakes are likely to amplify shaking significantly. Careful analysis of other ancient monuments in the valley, in particular dating of damage, may usefully supplement the sparse historical record. We note in closing that the Pandrethan Temple serves as both an encouraging and a cautionary case study: encouraging to the extent that the structure does provide useful clues that help elucidate the earthquake history of the region; cautionary to the extent that, if not for the fortuitous existence of repeat historical photographs, one could easily be led to the same obvious but mistaken conclusion implied by R.D. Oldham’s 1887 photograph, that damage to the temple evident in 1887 was caused by the 1885 earthquake.

CONCLUSIONS

REFERENCES CITED

The survival of the small masonry tenth-century Shiva Temple at Pandrethan, near Srinagar, through more than a dozen damaging earthquakes suggests that shaking greater than intensity IX has not occurred in the past 200 yr, and possibly the past 1000 yr. The case is based on “calibration” earthquakes in 1885 and 2005 that shook the temple with intensities of VII or less. From photographs in 1868 and 1887 that show the rapid growth of a small tree that grew through cracks in its roof, we deduce that the temple had been damaged by an earlier earthquake, probably in 1828, and/or between 1778 and 1885, in which local intensities must have exceeded VIII. The absence of damage to the ornate ceiling of the Pandrethan temple and most of the temple blocks suggests that accelerations in the past millennia have been insufficient to destroy the

Ambraseys, N.N., and Douglas, J., 2004, Magnitude calibration of North Indian earthquakes: Geophysical Journal International, v. 158, p. 1–42, doi: 10 .1111/j.1365-246X.2004.02323. Ambraseys, N.N., and Jackson, D., 2003, A note on early earthquakes in India and southern Tibet: Current Science, v. 84, p. 570–582. Bernier, F., 1891, Travels in the Mogul Empire, A.D. 1656–1668 (translated by I. Brock and revised by A. Constable): London, Constable & Co., 497 p. Bernier, R.M., 1997, Himalayan Architecture: Madison, Fairleigh Dickinson University Press, 196 p. Bashir, A., Bhat, M.I., and Bali, B.S., 2009, Historical record of earthquakes in the Kashmir Valley: Journal of Himalayan Geology, v. 30, no. 1, p. 75–84. Bilham, R., 2008, Tom LaTouche and the Great Assam Earthquake of 12 June 1897; letters from the epicenter: Electronic supplement: Seismic Research Letters, v. 79, no. 3, p. 426–437. (Transcriptions of LaTouche’s letters 1882–1913 illustrated by archival photographs from Calcutta and London: http://www.seismosoc.org/publications/SRL/SRL_79/srl_79-3 _hs.html [accessed 3 August 2010].) Burke, J., 1868, “Temple of Meruvarddhanaswami at Pandrethan near Srinagar,” British Library: Shelfmark: Photo 981/1(40).

ACKNOWLEDGMENTS We thank the British Library for permission to publish Burke’s 1859 photo, and the director general of the Geological Survey of India for permission to reproduce R.D. Oldham’s photographs of Pandrethan. The investigation was funded by the U.S. National Science Foundation.

Historical earthquakes in Srinagar, Kashmir Cole, H.H., 1869, Illustrations of Ancient Buildings in Kashmir, India Museum: London, India Museum, W. H. Allen and Co., publishers to the India office, 31 p. Cunningham, A., 1848, An essay on the Arian Order of Architecture, as exhibited in the Temples of Kashmir: Journal of the Asiatic Society of Bengal, September 1848, p. 241–327. Doughty, M., 1902, Afoot through Kashmir Valleys: London, Sands and Co., 276 p. Fergusson, J., 1867, History of Indian and Eastern Architecture: London, Murray, v. 3, 756 p. Hough, S.E., Bilham, R., and Bhat, I., 2009, Kashmir Valley megaearthquakes: American Scientist, v. 97, no. 1, p. 42–49, doi: 10.1511/2009.76.1. Hügel, C.A., 1845, Travels in Kashmir and Panjab: Containing a Particular Account of the Government and Character of the Sikhs (translated by T.B. Jervis, East India Company): London, J. Petheram, v. 1845, 423 p. Hunter, W.W., 1881, Gazetteer of India, v. VIII, 284 p. Iyengar, R.N., and Sharma, D., 1996, Some earthquakes of Kashmir from historical sources: Current Science, v. 71, no. 4, p. 300–331. Iyengar, R.N., and Sharma, D., 1998, Earthquake History of India in Medieval Times: Roorkee, Central Building Research Institute, 124 p. Iyengar, R.N., Sharma, D., and Siddiqui, J.M., 1999, Earthquake history of India in medieval times: Indian Journal of History of Science, v. 34, no. 3, p. 181–237. Jones, E.J., 1885, Report on the Kashmir earthquake of 30 May 1885: Records of the Geological Survey of India, v. 18, no. 4, p. 221–227. Kak, R.C., 1933, Ancient Monuments of Kashmir: Delhi, India Society, 172 p., 1971 reprint. Knight, W.H., 1863, Diary of a Pedestrian in Cashmere and Thibet: London, Bentley, 385 p. Lawrence, W.R., 1895, The Valley of Kashmir: Gulshan, 478 p., 2007 reprint.

117

Martin, S., and Szeliga, W., 2010, A catalog of felt intensity data for 589 earthquakes in India, 1636–2008: Bulletin of the Seismological Society of America, v. 100, no. 2, p. 562–569, doi: 10.1785/0120080328. Moorcroft, W., and Trebeck, G., 1841, Travels in the Himalayan Provinces of Hindustan and the Panjab, in Ladakh and Kashmir, in Peshawar, Kabul, Kunduz, and Bokhara from 1819 to 1825, Volume 2: London, J. Murray, 565 p. Oldham, R.D., 1887, Photographic Archives #257(288): Temple at Pendrethan showing stones displaced by earthquakes, location: near Srinagar; and #258(289): Temple at Pandrethan, location: Srinagar, Kashmir: Calcutta, Geological Survey of India. Oldham, T., 1833, A catalogue of Indian earthquakes from the earliest times to the end of AD 1869 (edited by R.D. Oldham): Memoirs of the Geological Survey of India, v. 19, no. 1, p. 163–215. Prinsep, J., 1858, Essays on Indian Antiquities, Historic, Numismatic, and Palæographic, of the Late James Prinsep: To Which Are Added His Useful Tables, Illustrative of Indian History, Chronology, Modern Coinages, Weights, Measures, Etc., Volume 2: London, J. Murray, 336 p. Szeliga, W., Martin, S., Hough, S., and Bilham, R., 2010, Intensity, magnitude, location and attenuation in India for felt earthquakes since 1762: Bulletin of the Seismological Society of America, v. 100, no. 2, p. 570–584, doi: 10.1785/0120080329. Temple, R., 1887, Diary of a Journey into Jammun and Kashmir between 8th June and 8th July 1859, in Temple, R.C., ed., Journals Kept in Hyderabad, Kashmir, Sikkim and Nepal, v. 2: London, W. H. Allen, 303 p. Vigne, G.T., 1844, Travels in Kashmir, Ladak and Iskardo, the Countries Adjoining the Mountain Course of the Indus, and the Himalaya, North of Panjab, with Map, Volume 1 (2nd ed.): London, H. Colburn, 406 p. MANUSCRIPT ACCEPTED BY THE SOCIETY 19 MAY 2010

Printed in the USA

The Geological Society of America Special Paper 471 2010

Earthquakes and civilizations of the Indus Valley: A challenge for archaeoseismology Robert L. Kovach* Department of Geophysics, Stanford University, Stanford, California 94305, USA Kelly Grijalva Department of Earth and Planetary Science, University of California–Berkeley, Berkeley, California 94720, USA Amos Nur Department of Geophysics, Stanford University, Stanford, California 94305, USA

ABSTRACT Civilizations have existed in the proximity of the Indus River Valley regions of modern Pakistan and India from at least 3000 B.C. onward. Geographically, the region encompasses a swath of the Makran coast, the alluvial plain and delta of the Indus River, and the Runn of Kachchh. The regional tectonic setting is controlled by the collision of the Indian and Eurasian plates and the subduction of the Arabian plate beneath the Eurasian plate. Earthquakes have undoubtedly struck many ancient sites, but finding their footprint in a riparian environment represents a challenge for archaeoseismology. However, some insight into seismoarchaeological indicators can be gleaned from examining the earthquake effects produced by historical infrequent large-magnitude events that have occurred in the region. Studies of these earthquakes emphasize the importance of repeated reconstructions, direct faulting, river damming from seismic uplift, and coastal elevation change as indicators of past earthquakes. Examples of past earthquake effects are presented for Banbhore in the Indus Delta, Brahmanabad, and the Harappan sites of Kalibangan and Dholavira. Future hermeneutic investigations in the region need to incorporate a seismological/ tectonic perspective and not rely solely on serendipity. INTRODUCTION For many thousands of years, civilizations have occupied regions of the southwestern Indian subcontinent stretching from the Arabian Sea to the foothills of the Himalayas. The tectonic environment argues that earthquake occurrences are a continuing integral facet of the region and must have occurred in the historical past. Earthquakes often leave their mark in the myths, leg*[email protected]

ends, and written accounts of ancient peoples, the stratigraphy of their historical sites, and the structural integrity of their constructions. Such information contributes to a better understanding of the irregularities in the spatial-temporal patterns of earthquake occurrences and whether earthquakes have contributed to the abandonment of ancient occupational sites. The objective of this contribution is to present some examples of seismic damage for a few archaeological sites in the southwestern Indian subcontinent, in particular, the Indus River Valley region and the Runn of Kachchh. Much work is yet to be

Kovach, R.L., Grijalva, K., and Nur, A., 2010, Earthquakes and civilizations of the Indus Valley: A challenge for archaeoseismology, in Sintubin, M., Stewart, I.S., Niemi, T.M., and Altunel, E., eds., Ancient Earthquakes: Geological Society of America Special Paper 471, p. 119–127, doi: 10.1130/2010.2471(11). For permission to copy, contact [email protected]. © 2010 The Geological Society of America. All rights reserved.

119

120

Kovach et al.

done in this region in deciphering the effects of ancient earthquakes from the archaeological record. TECTONIC FRAMEWORK AND SEISMICITY OF THE REGION A generalized map of the region under discussion is shown in Figure 1. A more detailed tectonic map can be found in Nakata et al. (1991). The east-west Makran coast of southern Pakistan forms a section of the convergent boundary of the EurasianArabian plate. It is bounded to the east by a transform fault, the Murray Ridge, which defines the western oceanic edge of the Indian plate. On land, the Chaman fault system, including the Ornach Nal fault zone in the south, forms the eastern boundary of the Makran region, accommodating the motion between the Eurasian and Indian plate. The convergence rate between the Arabian and Eurasian plate is estimated from modern global positioning system (GPS) studies to be ~22–30 mm/yr along the Makran coast (Apel et al., 2006; Sella et al., 2002; Vernant et al., 2004). Such a high collision rate implies that a large amount of potential earthquake slip should be accumulating each century to be periodically released in large earthquakes. However, the historic seismic record of the Makran coast does not indicate many great events, arguing that perhaps much of the subducting plate motion takes place aseismically or that great earthquakes are separated by periods, longo intervallo, that exceed the seismic record. Long-term, average rates of uplift have been determined for various segments of the Makran coast by Page et al. (1979) and Vita Finzi (1980). Radiometric dating on marine shells found in various depositional horizons in marine terraces indicates a Holo-

60˚ 61˚ 62˚ 63˚ 64˚ 65˚ 66˚ 67˚ 68˚ 69˚ 70˚ 71˚ 72˚ 73˚ 74˚ 75˚E 31˚N ASIAN PLATE 30˚ 30˚ Helmand block lock INDIAN PLATE Pakistan 29˚ 29˚ 0.8 cm/yr er iv R 28˚ 28˚ s 3.5 cm/yr du In 27˚ 27˚

cene uplift rate of 0.1–0.3 cm/yr for portions of the Makran coast. The inferred recurrence time for an M ~8 event is 660–1000 yr, suggesting that 4–6 similar sized events have occurred within the past 4000 yr. The Chaman fault system, which defines the western edge of the Indian plate, has a slip rate of movement of 20–40 mm/yr based on geological offsets (Lawrence et al., 1992) and a relative plate velocity of 26–34 mm/yr (Apel et al., 2006). The rates of movement point to a recurring potential of significant amounts of seismic slip accumulation over time scales of several hundred years. The pattern of current and historical seismicity roughly mirrors the plate boundary framework of the region (Fig. 2). The Makran region of southern Pakistan and southeastern Iran is characterized as a region with a relatively low level of seismic activity and infrequent great earthquakes. Along the Makran subduction zone, most of the seismic activity takes place at focal depths less than 24 km and is confined within an ~26° northward-dipping zone beneath the coast (Engdahl et al., 2005). The spatial distribution of epicenters forms a relatively narrow east-west band. Along the coast, the seismic activity is accented by two large earthquakes that took place in the 1940s. The MW = 8.1 event of 27 November 1945 produced a tsunami, significant coastal uplift, and the eruption of several offshore mud volcanoes (Byrne et al., 1992). A second event (M = 6.9) occurred approximately in the same area as the 1945 event on 5 August 1947. Indications of earlier seismic activity in the remote Makran coastal area

E N

CFZ

31˚

26˚

Makran

2001

India Arabian Sea

23˚ 22˚ 21˚

~2.8 cm/yr

1819

26˚

25˚ 24˚

1945

y rra Mu

ge Rid

~2.9 cm/yr Kachchh

km 0 100 200

20˚

25˚ 24˚ 23˚ 22˚ 21˚ 20˚

60˚ 61˚ 62˚ 63˚ 64˚ 65˚ 66˚ 67˚ 68˚ 69˚ 70˚ 71˚ 72˚ 73˚ 74˚ 75˚

Figure 1. Regional tectonic setting of the Indus River Valley and its environs (modified from Regard et al., 2005; Vernant et al., 2004). CFZ indicates the Chaman fault zone.

Figure 2. Seismicity of the Indus Valley region from 1905 to 2005. The circles of varying size are for events of magnitude 5 or greater. The stars represent the assumed locations for pre-instrumental events producing seismic intensities of VIII or greater. Locations of the 1945 M = 8.1 Makran coast, the 1819 Runn of Kachchh, and the 2001 M = 7.6 Bhuj earthquakes are labeled. Data taken from Quittmeyer and Jacob, 1979; Endahl et al., 2005; International Seismological Center, 2006; and National Earthquake Information Center, 2006.

Earthquakes and civilizations of the Indus Valley are sparse, so that any inferences about its long-term history are poor and incomplete. Historical damaging events are reported to have occurred in 1483, 1765, and 1851 (Ambraseys and Melville, 1982; Bilham et al., 2007; Oldham and Oldham, 1883). The highest level of current seismic activity is found along the western boundary of the Indian plate, extending northeasterly from the Arabian coast up to latitude 29°N along the Chaman fault zone. The recent seismic activity of the Chaman fault zone primarily consists of moderate-sized earthquakes with maximum magnitudes less than 5. Seismic activity increases northward from the coast, and at latitude of 29°N, a zone of high seismic activity is reached, characterized by several damaging earthquakes possessing magnitudes of 7 or greater. Epicenters in this region align well with the mapped Chaman fault system but also occur to the east toward the Indus River. Several earthquakes that took place from 1931 to 1935 caused significant fatalities (West, 1934, 1936). Damaging earthquakes in the Indus River floodplain and Kachchh regions, distant from the known plate boundaries, are infrequent. These intraplate earthquakes, however, such as the 1819 event in the northern Runn (Rann) of Kachchh, have produced a significant number of fatalities and also altered the geomorphic landscape of the region. It is believed that in the historical past, the Runn of Kachchh was at a lower elevation and linked to the Arabian Sea. Gupta (1975) has demonstrated that as late as 2000 yr ago, portions of Runn had a water depth of 4 m and thus was inundated throughout the year. Today, the Runn is not perennially under water, and so its elevation has been altered in recent times. What cannot be unambiguously resolved, however, is what percentage of the present configuration was produced by deltaic sedimentation and what percentage by tectonic uplift (Sivewright, 1907)? Burnes (1835, p. 570) described an oral tradition of the local natives arguing that the Runn was formerly opened to the sea and was altered by an earthquake. Bearing in mind that many cultures do not want to separate mythology from legendary tradition, the marvelous tale given next appears to contain umbilical elements of plausibility: “… a Hindu saint, named Dhuramanat’ha, a Jogi [a holy and hospitable man] underwent penance, by standing on his head … for a period of 12 years. At that time he resumed his proper position, and God became visible to him, when a convulsion of nature took place, and the hill on which he stood split in two, the sea that lay northward of him [which is the present Runn] dried up, and the ships which than navigated it were wrecked and its harbours destroyed, with other miraculous and wonderful events.” What is important, in the context of seismoarchaeology, is that geological forces, particularly earthquakes, have greatly altered the geography of the region over the past 4000–5000 yr or so. As a consequence, this may have accelerated the demise of many ancient settlements by altering the water supply, modifying trade routes, producing the need for continual rebuilding, and ultimately forcing migration.

121

SEISMOARCHAEOLOGICAL INDICATORS Earthquakes can cause elevation changes and alter the drainage pattern of life-giving rivers. Other direct effects of earthquakes that might be present at ancient sites in the Indus Valley and its environs under archaeological investigation are often more difficult to document because individuals with seismic expertise have been rarely involved during the actual excavations themselves. The archaeological approach at many ancient sites mainly concentrates on documenting stratifications of the ground unless the investigator has a curiosity as to the seismic history of the site or is intent on finding evidence for a catastrophic event of the past. Nevertheless, earthquakes have left their imprint on many archaeological sites in the Indus Valley region. Unlike volcanic eruptions and floods that can be recognized in excavations from an examination of soil composition and stratigraphy, the traces of past earthquakes in buildings and other structures are usually more difficult to interpret. Aging of materials and burial and architectural innovations over time have to be considered. However, the effects of ancient seismic destruction can often be ascertained from a careful reading and syntheses of archaeological records and written accounts when available. Even if the evidence for earthquake damage is only suggestive, it can be bolstered by plausibility and consistency arguments, and the onus probandi is then placed upon those who wish to prove one point or another. We here point to the statement by Ambraseys (2005), who asserts that earthquakes are often looked to as an easy solution for explaining civilization and cultural gaps. With these prefatory comments being said, Table 1 enumerates a few key diagnostics that can often be gleaned from the archaeological record and that are appropriate for ancient locations in the Indus Valley region. In the following sections, we will discuss a few examples for which evidence of ancient earthquakes has been found.

TABLE 1. SOME CRITERIA FOR IDENTIFYING EARTHQUAKE OCCURRENCES 1. Direct evidence of faulting observed in excavations. 2. Skeletons of people found in nonburial positions, particularly if found buried beneath the debris of fallen structures. 3. Abrupt geomorphological changes, such as changes in river drainage associated with subsequent abandonment or reconstructions. 4. Fallen walls with no evidence of rebuilding. 5. Evidence of destruction and hasty reconstruction using debris from earlier constructions. 6. Reconstruction of walls with what might be called “antiseismic” additions, such as buttressing. 7. Well-dated building destructions correlating with historical accounts of earthquakes. Earthquakes mentioned in myths and legends. 8. Numismatic evidence such as the leaving behind of coins and valuables due to a hasty abandonment.

122

Kovach et al.

Brahmanabad An interesting set of archaeological ruins, discovered in 1854, is known as Brahmanabad-Mansura in Sind. The ruins are situated in an open sandy plain upon the bank of an ancient bed of the Indus River, ~200 km northeast of Karachi, Pakistan (Fig. 3). A broken brick tower was the only recognizable standing feature seen in 1894 (Fig. 4). The deserted site had the circuitous shape of a boot with its sole facing northwest and its leg pointing toward the southeast. It was not a large site, as the distance around the periphery was only 9.2 km. The excavations of Bellasis (1856a, 1856b) led him to believe that the site had been devastated by an ancient earthquake:

66˚

67˚

68˚

69˚

70˚E 26˚N

26˚ Brahmanabad Umarkot

25˚

25˚

Karachi

BANBHORE

712 coastline

1833 coastline

24˚

24˚ Lakhpat

Bhuj

We had not dug two feet before we came to quantities of bones … skeletons were so numerous … The human bones were found in doorways, as if the people had been attempting to escape, and others in the corners of the rooms. Many of the skeletons were in a sufficiently perfect state to show the position the body had assumed; some were upright, some recumbent, with their faces down, and some crouched in a sitting posture. One in particular, I remember, finding in a doorway: the man had evidently been rushing out of his house, when a mass of brick-work had, in its fall, crushed him to the ground, and there his bones were lying extended full length…. (p. 417)

Three arguments were given in favor of destruction by an earthquake. The observed destruction was far too complete to have been solely the work of time; an invading army would have not produced such complete destruction, yet leave behind coins and other valuables. Finally, if the city had only been deserted in an orderly manner, the inhabitants would most likely have carried off their valuables with them. Brahmanabad was not visited again by archaeologists until 1897. At this time, it was observed that much of the surficial soil cover at the site had been destroyed by local cultivators. Subsequent excavations carried out by Cousens (1906, 1912) led him to suggest that the site was also the location of Mansura, the first Arab capital in Sind, and that Mansura had been built upon the ruins of ancient Brahmanabad. Of particular relevance to the earthquake question was the observation that the earlier excavations of Bellasis were not deep enough to have reached the Brahmanabad horizon. It was in the uppermost layer, that of Mansura, or a subsequent Mansura rebuilt after the earthquake disaster, that bones, ash, broken pottery and quantities of charcoal were found that had led Bellasis to his original supposition. It now remains to place bounds on the date of the postulated earthquake disaster. Brahmanabad was conquered by Muhammad Qasim in A.D. 712, and Mansura is reputed to have been built by his son Amru. Local coins found in the upper layers belong to subsequent Arab governors of Mansura, ca. A.D. 750 (Sykes, 1857), and Arab chroniclers describe its existence until at least A.D. 1020 (Bellasis, 1856a). The terminous post quem for the earthquake that leveled the walls of Mansura, overthrew

km

23˚ 0 66˚

100 67˚

23˚ 200 68˚

69˚

70˚

Figure 3. Location of archaeological sites of Brahmanabad and Banbhore. Shoreline and the inland sea at the time of the Arab conquest of Sind in A.D. 712 are shown by thin, solid black lines (Sivewright, 1907). The routes of the Indus River and its branches to the sea in 1833 are also shown (Burnes, 1833).

Figure 4. View from the west of the ruined tower at Brahmanabad (Cousens, 1912).

Earthquakes and civilizations of the Indus Valley its housing, and crushed many of its inhabitants, therefore, is the early eleventh century. On the other hand, if it is not accepted that Mansura was reconstructed on the site of Brahmanabad, but instead was located some 8 km to the northeast, then a slightly earlier date for the terminal earthquake at Brahmanabad is suggested (Sykes, 1857; Bilham et al., 2007). Banbhore Banbhore, an archaeological site 64 km east of Karachi (Fig. 3), is located on a flat prominence on the right bank of Gharo Creek, a creek that formed the main course of the Indus prior to A.D. 1250. Gharo Creek, now filled with silt and sand, marks the northern edge of the alluvial fan of the Indus Delta. The key structure at Banbhore was its central mosque. Its earliest constructed wall was built with dressed blocks of sandy limestone. The feature of particular interest is that the mosque was subjected to three successive reconstructions with increasing deterioration in the quality of work. Based on ceramics and renovations, it is inferred that there were four principal phases of Muslim occupation under which construction and repairs took place (Ashfaque, 1969). The earliest construction, in which the foundation was laid, took place in the Umayyid (Ummeide) Period from A.D. 715 to 750. This was followed by the Abbasis (Abbasides) Period from A.D. 750 to 892, during which the mosque was damaged by an earthquake, and major repairs were carried out. The southern wall of the mosque suffered the greatest damage and was rebuilt from undressed stones picked from the rubble. In addition, buttressing was placed against the weakened eastern boundary walls. Written documents state that the ancient town of Banbhore was destroyed during the time of the military exploits of Shekh Abu Turab during the caliphate of Harun Al Rashid (A.D. 786–808) to reestablish Islam in Sind, placing the time of the earthquake sometime between A.D. 787 and 790 (Fredunbeg, 1902). The third period of Muslim occupation, called the Late Abbasid Period, covered the time interval from A.D. 892 through most of the twelfth century A.D. Earlier repairing of the boundary walls did not prove to be lasting, since there is archaeological

123

evidence for an extensive second repair using carved blocks from the original structure and rubble set in mortar. A dated inscription (294 A.H. or A.D. 906) has been found commemorating the completion of the extensive repairs to the mosque (Fig. 5). It has been speculated that this rebuilding may have been needed as a result of an earthquake in 280 A.H. (A.D. 893–894). In Tornberg’s (1865) edition of Ibn-el-Athiri, al-Kaml fi’l-Ta’rikh, volume VII, p. 465, it is stated:

And in the Shawwal of this year (280 A.H.) there was an eclipse of the moon and the people of Dabil and the whole world woke in darkness. And the darkness lasted for a long time and when it was 4 o’clock in the afternoon there blew a black wind, which continued to a third part of the night. And when the third part of the night had come, there was an earthquake, and the town was destroyed, so that only about 100 houses remained, and after this the earth shook five times more and the total number of people that were found killed below the ruins amounted to 150,000.

The description given in A’s Suyuti’s History of the Caliphs (Jarrett, 1881, p. 387–388) strongly suggests the same historical source: “In the same year (280 A.H.) came advices from Daybul (Daibul) that the moon had been eclipsed in the month of Shawwal, and that darkness had spread over the country till the afternoon when a black storm began to blow which continued for a third of the night, followed by a mighty earthquake which destroyed the whole city, and the number of those taken out from the ruins was one hundred and fifty thousand.” Mention of a disastrous earthquake on this date is given by Hoff (1840) quoting the authority of the thirteenth-century Syrian writer Bar Hebraeus (Budge, 1932), who explicitly mentions a location in outer India. However, the earthquake described in these chronicles is well-documented to have taken place in Dvin, Armenia, on 27 December 893. Dabil is the Arabic name for Dvin, leading to misconceptions that this particular earthquake may have occurred in a similarly named place in the Indus Delta (Guidoboni and Traina, 1995; Ambraseys, 2004). The location of Debal in modern-day Pakistan has not been positively identified,

Figure 5. Floral Kufic inscription found at Banbhore commemorating rebuilding after earthquake(?) in 280 A.H. (A.D. 893–894). “In the name of Allah…This is what Amir Muhammad ibn has ordered about its erection in ‘Dhu’l Qadah?’ in the year 294” (Ghafur, 1966).

124

Kovach et al.

but some scholars have argued that a location near Karachi and possibly Banbhore itself is not unlikely. Regardless of its location, however, the archaeological evidence for repeated damage and reconstruction at Banbhore lends credence to its past vulnerability to earthquakes. The final period of the settlement of Banbhore covers the time interval, described as a period of decadence and extinction, from the late twelfth to the mid-thirteenth century A.D. During this third period of reconstruction, arches were repaired, and entrances were made narrower and strengthened by the stacking of stones and bricks on either side. Khan (1960) stated that the settlement of Banbhore came to a sudden end following a violent disturbance. A large number of human skeletons were found lying in a disorderly manner, and bricks and stones were found fallen from their original structures. Kalibangan Kalibangan was an Early and Mature Harappan site that was situated on the left bank of the Ghaggar River (known in ancient times as the Sarasvati), near its intersection with the Drishadvati River (Fig. 6). These rivers were active and perennial during the time of occupation from 3000 B.C. to 1750 B.C. It is generally accepted that the final abandonment of Kalibangan was the consequence of major shifts in the courses and volume of water associated with the Sarasvati and Drishadvati Rivers. The temporal details of the changing drainage patterns of these ancient rivers, however, are a topic of continuing debate (Possehl, 1999). The Early Harappan settlement at Kalibangan, found beneath a mound labeled KLB-1, was not large. It consisted of a parallelogram-shaped walled enclosure of mud bricks with an entrance on the northwestern side (Fig. 7). Several trenches were

made during the excavation of KLB-1 that reveal definite evidence for one or more ancient earthquakes. The archaeological diggings show clear signs of fault rupture and displacement of horizons (Figs. 8 and 9). Lal et al. (2003, p. 100) stated that “… even the uppermost layers are affected by this earth-movement. All this implies that the violent shaking of the area, which in other words would mean an earthquake, brought about the end of the Early Harappan occupation at Kalibangan.” Offset stratigraphic horizons are present in trench XD2-XE2 in the southwest corner of the excavated mound. To the north, a tilted brick wall produced by some disturbance was exposed near YA17. We can speculate that surface faulting produced both disturbed structural features. The strike of the deduced rupture is N12°E, roughly parallel to the northeasterly trend of the Indian plate boundary in this region. It is possible to estimate the magnitude of this earthquake in 2700 B.C. A fault offset of 30–40 cm is present in the depositional layers in trench XD2-XE2. From the empirical relations of Wells and Coppersmith (1994), a fault offset of this size implies an earthquake magnitude of at least 6.5, and implies that the seismic intensities that struck Kalibangan were in the range of VIII to IX on the modified Mercalli intensity scale. The supposition that the earthquake caused abandonment is strengthened by the presence of an overlying layer of infertile, windblown sand covering the ruins. Kalibangan was subsequently reoccupied after a short hiatus, placing the date of the earthquake sometime between Period I (Early Harappan) and Period II (Mature Harappan), or 2700 B.C. (Lal, 1998). Final abandonment of Kalibangan occurred around 1750 B.C., presumably as a result of the change in river flow of the Sarasvati (Raikes, 1968).

E N

Figure 6. Map showing the location of Kalibangan at the junction of the paleorivers Sarasvati and Drishadvati in 2000 B.C. (generalized from Wilhelmy, 1969).

Figure 7. Site layout of fortified structure at the Mature Harappan site of Kalibangan.

Earthquakes and civilizations of the Indus Valley

125

An item of added interest to the thesis of earthquake damage is that archaeologists have found evidence for nine reconstructions at Kalibangan during the Mature Harappan occupational period. What is unusual about the archaeological site of Kalibangan is that it is not located in an area of current seismic activity. Nevertheless, the archaeological identification of an ancient earthquake(s) that is contemporary with the level of excavation is beyond conjecture. The direct evidence of this ancient earthquake is, therefore, doubly interesting. It demonstrates a cause and effect between an earthquake occurrence and abandonment and allows us to see traces of past seismic activity in an area that one would conclude from a modern-day seismicity map to be an area that is relatively “nonseismic.” Secondly it emphasizes that the seismic effects from damaging earthquakes with very large intervals of time between occurrences can be seen in archaeological excavations. Dholavira

Figure 8. Fractured and offset cultural horizons found in trench XD2-XE2 at Kalibangan. View is looking south. Evidence of ancient earthquake rupture can be seen in three places (Lal, 1998). Height of vertical wall is approximately 2 m.

Dholavira was a well-planned Harappan enclave built on a low plateau on Khadir Island, a true island ca. 4000 B.P. in the Runn of Kachchh (Fig. 10). The site is enclosed by a stone and brick wall, 5 m thick at its base, with rectangular dimensions of 700 × 750 m. Constructed on a slope between two nullahs (seasonal drainage gullies), the city contained a sophisticated watercollection system of giant reservoirs to collect and store seasonal rainwater. Archaeologists have identified seven states of occupation beginning in 2650 B.C. and extending to 1450 B.C. Stages I and II, which lasted ~150 yr, involved constructions of molded mud brick houses with a substantial plastered fortification wall surrounding a fort. Stage III saw a substantial residential area added to the north, together with reservoirs carved into the host rock (Bisht, 1991; Joshi and Bisht, 1994). Somewhere near the end of Stage III (ca. 2200 B.C.), Dholavira was struck by a major earthquake, as evidenced by slip faults

Figure 9. Stratigraphic section of trench shown in Figure 8. Faulted section is highlighted (Lal et al., 2003).

126

Kovach et al.

67˚

68˚

69˚

70˚

71˚

72˚

73˚ 73˚E

26˚

26˚N 26˚ 1030

25˚ 1668

25˚

893 1819

Runn of Kachchh

24˚

Dholavira 1956 2001 Bhuj

Ahmedabad

Kachchh Anjar

23˚

24

Surkotada

23

Lothal 0

Kathiawar Peninsula

22˚ 67˚

68˚

69˚

70˚

71˚

72˚

22 73˚

Figure 10. Map showing the location of the Harappan site of Dholavira in the Runn of Kachchh. Epicenters of the 1819 Allah Bund event and the M = 7.6 Bhuj earthquake of 2001 are shown.

Figure 11. View to the north of partially damaged entrance to fort at Dholavira from 2001 Bhuj earthquake. Structural damage from ancient earthquake in 2200 B.C. is preserved on adjacent tilted wall to the left of 2001 damaged wall. The inset shows a flexured wall in the central part of the fort damaged by an earlier earthquake (Rajendran and Rajendran, 2003). Width of flexured wall is ~1 m.

exposed in vertical sections and the displacement and tilting of walls (Fig. 11). The earthquake apparently produced extensive devastation at the site, because large-scale repairs were undertaken near the end of Stage III (Singh, 1996; Possehl, 2002). Stages IV and V, from 2200 to 1900 B.C., encompasses the Mature Harappan period of architectural flourishing, followed by gradual decline and a short period (~100 yr) of abandonment. The subsequent posturban stages had alternating periods of occupation and abandonment, characterized by an inferior level of construction compared to earlier periods. Natural events such as flooding, drought, or earthquakes are believed to have triggered the short periods of abandonment. Dholavira is located in close proximity to the epicenters of the M = 7.7 Allah Bund event of 1819 (Bilham, 1999) and the M = 7.6 Bhuj event of 2001 (Hough et al., 2002), supporting the contention that Dholavira was subjected to direct earthquake effects during Harappan times.

is that ancient earthquakes with long intervals of time between occurrences may have taken place in regions not indicated by present-day patterns of seismic activity. Traces of damage may also have disappeared during reconstruction works carried out at later times. Challenges remain for future seismoarchaeological investigations in the Indus River valley and delta region. It is hoped that future hermeneutic investigations will incorporate a seismological/tectonic perspective, so that future discoveries are not accidental.

CONCLUDING REMARKS

REFERENCES CITED

The examples presented here demonstrate that earthquakes have produced direct damage at several historical archaeological sites in the Indus Valley region. In addition, ancient earthquakes, similar to the 1819 Runn of Kachchh event, the 1945 Makran coast event, and the 2001 Bhuj event, may also have produced significant ground uplift and deformation affecting the geographic setting of early settlements. We here point to significant changes over historical time of the fluvial system of the Indus River, the consequences of river damming and subsequent flooding, and coastal elevation changes. For this reason, the lack of apparent seismic destruction at other ancient sites in the region should not be necessarily taken as evidence that a specific site was free from earthquake effects in the past. Another factor to consider

Ambraseys, N.N., 2004, Three little known early earthquakes in India: Current Science, v. 86, no. 4, p. 506–508. Ambraseys, N.N., 2005, Archaeology and neo-catastrophism: Seismological Research Letters, v. 76, p. 560–564. Ambraseys, N.N., and Melville, C.P., 1982, A History of Persian Earthquakes: Cambridge, UK, Cambridge University Press, 219 p. Apel, E., Burgmann, R., Bannerjee, P., and Nagarajan, B., 2006, Geodetically constrained Indian plate motion and implications for plate boundary deformation: Eos (Transactions, American Geophysical Union), v. 85, abs. 52T51B-1524. Ashfaque, S.M., 1969, The grand mosque of Banbhore: Pakistan Archaeology, no. 6, p. 182–209. Bellasis, A.F., 1856a, An account of the ancient and ruined city of Brahminabad, in Sind: Journal of the Bombay Branch of the Royal Asiatic Society, v. 5, p. 413–425. Bellasis, A.F., 1856b, Further Observations of the ruined city of Brahminabad, in Sind: Journal of the Bombay Branch of the Royal Asiatic Society, v. 5, p. 467–477.

ACKNOWLEDGMENTS This article is a contribution to the United Nations Educational, Scientific and Cultural Organization–funded International Geoscience Programme IGCP 567, “Earthquake Archaeology: Archaeoseismology along the Alpine-Himalayan Seismic Zone.”

Earthquakes and civilizations of the Indus Valley Bilham, R., 1999, Slip parameters for the Rann of Kachchh, India, 16 June 1819 earthquake, quantified from contemporary accounts, in Stewart, I.S., and Vita-Finzi, C., eds., Coastal Tectonics: Geological Society of London Special Publication 146, p. 295–319. Bilham, R., Lodi, S., Hough, S., Bukhary, S., Khan, A.M., and Rafeeqi, S.F., 2007, Seismic hazard in Karachi, Pakistan: Uncertain past, uncertain future: Seismological Research Letters, v. 78, no. 6, p. 601–613, doi: 10 .1785/gssrl.78.6.601. Bisht, R.S., 1991, Dholavira: A new horizon of the Indus civilization: Puratattva, v. 20, p. 71–82. Budge, E.A.W., translator, 1932, The Chronography of Gregory Abu’l Faraj, Commonly Known as Bar Hebraeus: London, Oxford University Press, 2 volumes, 151 p. Burnes, A., 1833, Substance of a geographical memoir on the Indus: Journal of the Royal Geographical Society of London, v. 3, p. 113–156, doi: 10 .2307/1797602. Burnes, A., 1835, Memoir on the eastern branch of the River Indus, giving an account of the alterations produced on it by an earthquake, also a theory of the formation of the Runn, and some conjectures on the route of Alexander the Great; drawn up in the years 1827–1828: Transactions of the Royal Asiatic Society of Great Britain and Ireland, v. 3, p. 550–588. Byrne, D.E., Sykes, L.R., and Davis, D.M., 1992, Great thrust earthquakes and aseismic slip along the plate boundary of the Makran subduction zone: Journal of Geophysical Research, v. 97, p. 449–478, doi: 10.1029/91JB02165. Cousens, H., 1906, Brahmanabad-Mansura in Sind: Annual Report 1903–04: Calcutta, Archaeological Survey of India, Office of the Superintendent of Government Printing, p. 133–144. Cousens, H., 1912, Excavations at Brahmanabad-Mansura in Sind: Annual Report 1908-09: Calcutta, Archaeological Survey of India, Office of the Superintendent of Government Printing, p. 79–87. Engdahl, E.R., Bergman, E.A., Myers, S.C., Jackson, J.A., and Priestley, K., 2005, Relocation and assessment of seismicity in Iran region: Eos (Transactions, American Geophysical Union), v. 86, Fall meeting supplement, abstract 52T51-C-1352. Fredunbeg, M.K., 1902, The History of Sind, Volume 2: Karachi, Printed at the Commissioner’s Press, 346 p. Ghafur, M.A., 1966, Fourteen Kufic inscriptions of Banbhore: Pakistan Archaeology, no. 3, p. 65–90. Guidoboni, E., and Traina, G., 1995, A new catalogue of earthquakes in the historical Armenia area from antiquity to the 12th century: Annali di Geofisica, v. 38, no. 1, p. 85–147. Gupta, S.K., 1975, Silting of the Rann of Kutch during Holocene: Indian Journal of Earth Sciences, v. 2, p. 163–175. Hoff, K. von, 1840, Geschichte der durch uberlieferung nachgewiesenen naturlichen veranderungen der erdoberflache, Volume 4: Gotha, Justus Perthes, 275 p. Hough, S., Martin, R., Bilham, R., and Atkinson, G., 2002, The 26 January 2001, Bhuj India earthquake: Observed and predicted ground motions: Bulletin of the Seismological Society of America, v. 92, p. 2061–2079, doi: 10.1785/0120010260. International Seismological Centre, 2006, International Seismological Centre Bulletin Thatcham, United Kingdom: http://www.isc.ac.uk/ (accessed 14 July 2010). Jarrett, M., H.S., 1881, History of the Caliphs by Jalalu’ddin A’s Suyuti: Calcutta, Asiatic Society, p. 387–388. Joshi, V.P., and Bisht, R.S., 1994, India and the Indus Civilization: New Delhi, National Museum Institute, p. 22–31. Khan, F.A., 1960, Preliminary Report on Banbhore Excavations: Karachi, Government of Pakistan, Department of Archaeology. Lal, B.B., 1998, India 1947–1997: New Light on the Indus Civilization: New Delhi, Aryan Books International, 135 p. Lal, B.B., Joshi, J.P., Thapar, B.K., and Bala, M., 2003, Excavations at Kalibangan: The Early Harappans (1960–1969): Archaeological Survey of India Memoir 98, 340 p.

127

Lawrence, R.D., Khan, S.H., and Nakata, T., 1992, Chaman fault, PakistanAfghanistan: Annales Tectonicae, v. 6, p. 196–223. Nakata, T., Tsutsumi, H., Khan, S.H., and Lawrence, R.D., 1991, Active Faults of Pakistan, Map Sheets and Inventories: Hiroshima University Research Center for Regional Geography Special Publication 21, 141 p. National Earthquake Information Center, 2006, National Earthquake Information Center: http://neic.usgs.gov/neis/epic/ (accessed 14 July 2010). Oldham, T., and Oldham, R.D., 1883, A catalogue of Indian earthquakes from the earliest time to the end of A.D. 1869: Memoirs of the Geological Survey of India, v. 19, pt. 2, p. 163–213. Page, W.D., Alt, J.N., Cluff, L.S., and Plafker, G., 1979, Evidence for the recurrence of large magnitude earthquakes along the Makran coast of Iran and Pakistan: Tectonophysics, v. 52, p. 533–547, doi: 10.1016/0040-1951 (79)90269-5. Possehl, G.L., 1999, Indus Age, the Beginnings: Philadelphia, University of Pennsylvania Press, 1063 p. Possehl, G.L., 2002, The Indus Civilization: A Contemporary Perspective: Walnut Creek, California, AltaMira Press, 276 p. Quittmeyer, R.C., and Jacob, K.H., 1979, Historical and modern seismicity of Pakistan, Afghanistan, northern India and southern Iran: Bulletin of the Seismological Society of America, v. 69, p. 773–823. Raikes, R.L., 1968, Kalibangan, death from natural causes: Antiquity, v. 42, no. 168, p. 286–291. Rajendran, K., and Rajendran, C.P., 2003, Seismogenesis in the stable continental regions and implications for hazard assessment: Two recent examples from India: Current Science, v. 85, no. 7, p. 896–902. Regard, V., 2005, Cumulative right-lateral fault slip rate across the ZagrosMakran transfer zone: Role of the Minab-Zedan fault system in accommodating Arabia-Eurasia convergence in southeast Iran: Geophysical Journal International, v. 162, no. 1, p. 177–203, doi: 10.1111/j.1365-246X .2005.02558.x. Sella, G.F., Dixon, T.H., and Mao, A., 2002, REVEL: A model for recent plate velocities from space geodesy: Journal of Geophysical Research, v. 107, p. 2081–2111, doi: 10.1029/2000JB000033. Singh, B.P., ed., 1996, Indian Archaeology 1991–92—A Review: New Delhi, Archaeological Survey of India, 105 p. Sivewright, R., 1907, Cutch and the Ran: The Geographical Journal, v. 29, p. 518–539, doi: 10.2307/1776171. Sykes, W.H., 1857, Relics from the buried city of Brahmunabad in Sind: Illustrated London News, v. 30, no. 846, p. 166–167. Tornberg, C.J., 1865, Ibn-el-athiri, chronicon quod perfectissimum inscribitur (12 volumes in Arabic): Leiden, E.J. Brill, v. 7, 465 p. Vernant, P., Nilforoushan, F., Hatzfeld, D., Abbassi, M.R., Vigny, C., Masson, F., Nankali, H., Martinod, J., Ashtiani, A., Bayer, R., Tavakoli, F., and Chéry, J., 2004, Present-day crustal deformation and plate kinematics in the Middle East constrained by GPS measurements in Iran and northern Oman: Geophysical Journal International, v. 157, p. 381–398, doi: 10 .1111/j.1365-246X.2004.02222.x Vita-Finzi, C., 1980, 14C dating of recent crustal movements in the Persian Gulf and Iranian Makran: Radiocarbon, v. 22, no. 3, p. 767–773. Wells, D.L., and Coppersmith, K.J., 1994, New empirical relationships among magnitude, rupture length, rupture width, rupture area, and surface displacement: Bulletin of the Seismological Society of America, v. 84, p. 974–1002. West, W.D., 1934, The Baluchistan Earthquakes of August 25th and 27th, 1931: Geological Survey of India Memoir 67, no. 1, p. 1–82. West, W.D., 1936, Preliminary geological report on the Baluchistan (Quetta) earthquake of May 31st, 1935: Records of the Geological Survey of India, v. 69, p. 203–240. Wilhelmy, H., 1969, Das urstromtal am istrand der Indusebene das Sarasvati- problem: Zeitschrift für Geomorphologie, Supplementband, v. 8, p. 76–93. MANUSCRIPT ACCEPTED BY THE SOCIETY 19 MAY 2010

Printed in the USA

The Geological Society of America Special Paper 471 2010

Comparing semiquantitative logic trees for archaeoseismology and paleoseismology: The Baelo Claudia (southern Spain) case study Christoph Grützner* Klaus Reicherter* Institute of Neotectonics and Natural Hazards, RWTH Aachen University, Lochnerstrasse 4-20, 52056 Aachen, Germany Pablo G. Silva Departamento de Geología, Universidad de Salamanca, Escuela Politécnica Superior de Ávila, Calle Hornos Caleros, 50, 05003 Ávila, Spain

ABSTRACT The Bolonia Bay, close to the Strait of Gibraltar, hosts the Roman ruins of Baelo Claudia. In the first and third century A.D., this ancient town was affected by two earthquakes. Several earthquake-related damages can be found inside the ruins, and the adjacent mountain ranges show features of Quaternary activity. Extensive paleoseismological and archaeoseismological investigations have been conducted at the archaeological site and in its environs. The first 14C dating results from damaged infrastructure are presented in this paper, together with the preliminary results of fault-trenching on one of the closest suspect seismogenic faults near the archaeological site. The observations have been quantified using the two logic trees for paleoseismology and archaeoseismology. Our results show that a mere paleoseismological classification of the geological features leads to a paleoseismic quality factor (PQF) of 0.03, which is low compared to other studies. Taking into account the additional information from archaeoseismological work (archaeoseismological quality factor [AQF] is 0.5), it becomes clear that the Baelo Claudia study site provides an opportunity for detailed earthquake investigations. Therefore, it has a high potential for reliable seismic hazard analyses. A complementary application of both logic trees is recommended in future studies if sufficient data are available. INTRODUCTION When searching for historical and prehistorical earthquakes, many uncertainties due to the natural limits of the archives and methods need to be taken into account. When employing written records of historical events, one has to take into account inaccurate or biased reports, incomplete lists, geographic deviations,

political interest, event doubling, unreliability of witnesses, etc. (Guidoboni and Traina, 1996). Paleoseismological methods are used to evaluate prehistorical events; however, these techniques are necessarily confined to the geological environment that preserves the record and to the resolution available by the methods applied (McCalpin and Nelson, 1996). Archaeoseismology is not exactly in between those two methods, but

*E-mails: [email protected]; [email protected]; [email protected].

Grützner, C., Reicherter, K., and Silva, P.G., 2010, Comparing semiquantitative logic trees for archaeoseismology and paleoseismology: The Baelo Claudia (southern Spain) case study, in Sintubin, M., Stewart, I.S., Niemi, T.M., and Altunel, E., eds., Ancient Earthquakes: Geological Society of America Special Paper 471, p. 129–143, doi: 10.1130/2010.2471(12). For permission to copy, contact [email protected]. © 2010 The Geological Society of America. All rights reserved.

129

130

Grützner et al.

it certainly combines some advantages and disadvantages of both disciplines (Ambraseys et al., 2002; Galadini et al., 2006; Reicherter et al., 2009). For seismic hazard analysis, reliable data on earthquake location, magnitudes, and recurrence intervals are required. The most precise data are provided by instrumental seismology, covering only the past 100 yr, and even less in most places. Historical archives are often far less reliable but reach back to the beginning of written records. Finally, information on earthquakes in a geological time frame can be obtained by paleoseismology, which is often lacking the reliability and accuracy necessary for precise seismic hazard analysis. Here, archaeoseismological research can extend the time span of comparatively accurate information on old events. While the seismotectonic setting of the research area has to be investigated in terms of paleoseismology, information about use, construction, and the response of the archaeological buildings provided by archaeologists, architects, and engineers, geologists, and historians needs to be acquired in order to understand the site’s potential as an archaeoseismological record. With such a vast number of variables and uncertainties, archaeoseismology is often thought to be inaccurate and therefore is questioned as to how reliably the events described might be used in seismic hazard analyses. These contradictions are important and necessary, because they force scientists to review their own work, to find weaknesses in their arguments, and to accept and develop new methods that can resolve doubts. During the last years, several proposals have been implemented in order to establish a stringent and transparent methodology and provide a quantitative analysis of archaeoseismic damages (Karcz and Kafri, 1978; Nikonov, 1988; Guidoboni and Traina, 1996; Stiros, 1996; Galadini et al., 2006; Caputo and Helly, 2008; Sintubin and Stewart, 2008; Reicherter et al., 2009). Sintubin and Stewart (2008) created a logic tree approach on archaeoseismology based on the one proposed for paleoseismology by Atakan et al. (2000). This approach is supposed to “track uncertainties in successive stages of archaeoseismological investigation” (Sintubin and Stewart, 2008, p. 2209) and might therefore constitute an important step on the way to a systematization of the methods employed in researching ancient earthquakes. In this study, we apply the logic trees of Atakan et al. (2000) and Sintubin and Stewart (2008) to a site where both paleoseismological and archaeoseismological investigations have been conducted. We test the logic trees and aim to improve the database necessary for comparative site potential estimations. Furthermore, we show that a joint analysis of both approaches can improve the assessment of a particular site in terms of earthquake investigation. The paleoseismological observations described in the following sections are based on field work conducted by the authors between 2005 and 2009 and particularly on the results of the fault-trenching studies carried out in 2008 in the vicinity of Baelo Claudia. Archaeoseismological data have been published by Silva et al. (2005, 2006, 2009) and were mainly collected by the authors during the same time span.

BAELO CLAUDIA AREA Geology and Tectonics The Roman ruins of Baelo Claudia are situated on Bolonia Bay (Cádiz, south Spain), within the Strait of Gibraltar area (Fig. 1). In the westernmost part of the Gibraltar arc, which was formed by the assemblage of the Betic Cordilleras and the Moroccan Rif, the convergent plate boundary between Eurasia and Africa is less clearly defined than in the Eastern Mediterranean (Stich et al., 2006). Instrumental seismicity records list only 14 events with magnitudes >4.5 since 1929 (IGN) in the working area (latitude 35°N–37°N, longitude 4.5°W–6.5°W). Intensities have not exceeded VII; the strongest event reached M = 5.4 in 1976 and occurred on the Moroccan side of the Strait of Gibraltar. Compared with the entire south Iberian margin, the study area marks a seismic gap in the plate-boundary zone. Historical records mention earthquakes with local intensities greater than MSK VIII in southwestern Andalusia and in the Gulf of Cádiz (compiled by Reicherter, 2001). Silva et al. (2006) described several faults in Bolonia Bay. The SW-NE–trending Cabo de Gracia fault and the La Laja fault have a clearly expressed morphology, but their active phase might date back to the late Pliocene–early Pleistocene. However, there are few hints for more recent movements, since this fault has vertically displaced late Pleistocene marine terraces belonging to the oxygen isotopic substage (OIS) 5c, isotopically dated here ca. 125 ka (Zazo et al., 1999). On the other hand, N-S–trending normal faults in the Gibraltar Strait area account for moderate instrumental seismicity in this zone (Goy et al., 1995; Silva et al., 2006). The Carrizales, La Laja, and San Bartolome range-front faults display examples of Quaternary activity around Bolonia Bay (Silva et al., 2009) in a radius of 5 km from Baelo Claudia (Fig. 2). These faults show evidence for Quaternary paleoseismological events and offset prominent mountain ranges, resulting in step-shaped profiles of the ridges. All these faults also display relevant evidence of related landslides, massive rockfalls, liquefaction, and differential ground subsidence (Silva et al., 2009), which in some cases can be catalogued as secondary or sympathetic earthquake ground effects in the classification developed for the Environmental Seismic Intensity Scale (ESI-2007) developed by Michetti et al. (2007). Neotectonics of the study area were described in detail by Silva et al. (2006) (Figs. 2 and 3). Intensely folded Tertiary flysch deposits of the Aljibe, Algeciras, and Bolonia sandstone formations (Eocene to Aquitanian) dominate the study area. During the Alpine orogeny (Burdigalian to late Tortonian), they were overthrust by turbiditic sediments of Cretaceous to Eocene age (Sanz de Galdeano, 1990; Weijermars, 1991; Silva et al., 2006). The Facinas and Almarchal formations of the flysch complex consist of very plastic clayey and sandy layers, which are involved in the large number of mass movements observed in the study area (Fig. 2). Bolonia Bay is surrounded by three steep mountain ranges of Aquitanian sandstone: the Cabo de Gracia and La Laja mountain

The Baelo Claudia (southern Spain) case study 7°00’ W

131 4°00’

5°00’ W

6°00’ W

Córdoba

37°45’ N

37°45’ N

Seville 37°15’ N

Huelva

36°45’ N

Cádiz

Málaga

SRTM outline

Fig. 2 Bolonia Bay

Tarifa

ait Str

of

r

a ralt b i G

Figure 1. Bolonia Bay, location of the Roman remains of the Baelo Claudia village close to the Gibraltar Strait in southern Spain (SRTM data). Legend: 1—Upper Miocene–Quaternary sedimentary rocks; 2—Subbetic unit; 3—Internal Subbetic; 4—Prebetic; 5—Oligocene–Lower Miocene sedimentary rocks including flysch and olistostromes; 6—Campo de Gibraltar Flysch; 7—Predorsalian, Dorsalian, and Maláguide complexes; 8—Alpujárride complex; 9—Nevado-Filábride complex; 10—Iberian Massif with cover rocks. See Figure 2 for the regional tectonic setting. Figure is modified after Reicherter et al. (2003).

ranges in the west and the San Bartolome mountain range in the east (Fig. 3). All of them are significantly affected by landslides and extraordinary rockfall events (Höbig et al., 2009; Vollmert et al., 2009). The prominent La Laja and Cabo de Gracia mountain fronts, the morphotectonic features of which are discussed later with regard to the Atakan logic tree, form kilometer-scale lineaments (Silva et al., 2009). At the latter, slickensides were observed. Several studies describe the Quaternary tectonic setting (e.g., Zazo et al., 1999; Silva et al., 2006) and paleoseismological and archaeoseismological records at the Roman ruins (Goy et al., 1994; Sillières, 1997; Alonso-Villalobos et al., 2003; Silva et al., 2005, 2006, 2009). The Roman Ruins of Baelo Claudia The Roman ruins of Baelo Claudia are the remains of a small coastal town with a population of 2000 persons in an area of ~0.5 km2, the economic welfare of whom mainly stemmed from

the fishing industry. Its strategic significance is evident through the potential to control parts of the Gibraltar Strait. Prior to the Roman period, no urban settlement was established in the area, albeit tombs in the vicinity give evidence for Neolithic settlements. The small Roman town of Baelo Claudia was founded in the late second century B.C. (described by Strabo in A.D. 17) and was an important strategic and industrial part (tuna fishing, saucemaking, and olive-oil-pressing industries as well as iron-smelting industries) of the Roman Empire. Moreover, it used to be the gateway to Tingis—the modern Tangiers in Morocco. Before the town was founded, a small well-sheltered oppidum was situated in the Sierra La Laja (Silla de Papa). In the last century, Baelo Claudia has been extensively excavated by French archaeologists from the Casa de Velázquez in Madrid (Sillières, 1997). Thanks to these efforts, Baelo Claudia is one of the most complete and best-studied Roman towns in Spain, and its main features are an excellent example of early imperial town planning (Fig. 4). The central parts and buildings have been dated to the early first century A.D., under the reign of Emperor Claudius (A.D. 41–54),

0

5

10 km

La Laja fault El Almarchal

N

Facinas

Zahara

Cabo de La Plata Cabo de Gracia

1 2

CdG fault

3

Bolonia Bay

Carrizales fault

Baelo Claudia

S.Bartolome Mt. front

4 5 Tarifa

At l a n t i c O c e a n

6

Figure 2. Neotectonics and regional geology of Bolonia Bay and its surroundings. Legend: 1—El Almarchal plastic clays; 2—Facinas plastic clays; 3—El Aljibe Flysch nappe (predominantly sandstones), traces of Betic uprighted stratification planes are marked with dashed lines; 4—flysch units activated during the neotectonic period; 5—Pliocene and Quaternary deposits (postcollisional); 6—landslides; CdG—Cabo de Gracia Fault. Figure is modified after Silva et al. (2006). See Figure 1 for location and Figure 3 for a detailed map.

0

1

2

3

5 km

N

Minerva Jupiter Juno

La Laja fault

Cabo de Gracia fault

Forum

Carrizales fault Trenching site T2

ck Ro F G Cd

ck Ro

San Bartolome mountain front

T1

T3

Isis

Roman ruins of Baelo Claudia

Figure 3. Simplified map of Bolonia Bay, showing the location of the three trenches and the outline (city wall) of the archaeological site. A most likely fault-bounded spring is situated at the Cabo de Gracia fault (CdGF) in the trenching area, draining to the north. Geophysical investigations have been carried out at the trenching site and covered almost the entire area outlined by the rectangle in the lower left (GPR and DC-Geoelectrics). Along the La Laja and the San Bartolome, additional measurements have been taken out. See Figure 2 for the legend.

Basilica

Macellum

Decu

man

us M axi

mus

N

Figure 4. Aerial photo of the central or Capitol area of Baelo Claudia, showing the main features of the imperial town. Note the Isis temple, which is off to the side from the geometrical town center with Minerva, Jupiter, and Juno temples, the forum, and the basilica.

The Baelo Claudia (southern Spain) case study during which Baelo became a municipium. Therefore, Claudia was added to the name. Our most recent studies give evidence for at least two strong earthquakes in the first and the third centuries A.D., which destroyed parts of the infrastructure and eventually forced the Romans to gradually abandon the town (Silva et al., 2005, 2009). Excavations of the extramural necropolises, however, showed that at least parts of the town were in use during the fourth to sixth centuries A.D. After the second earthquake in the third century A.D., human activity in the town declined (Silva et al., 2009), and refuse and colluvial deposits began to accumulate in the lower monumental sector of the city, which includes the Theatre, the Decumanus Maximus (main E-W road), the Capitol area, and the Roman Basilica (Collins, 1998). The Basilica and the Roman Market (Macellum) were no longer in use, and the aqueducts supplying water to the city were severely damaged, resulting in water shortage (Sillières, 1997). The findings of the most recent coins point to the reign of Constantine the Great (A.D. 306–337), which is more or less coeval with a major crisis of the Western Roman Empire (Sillières, 1997). Later, Visigothic graves carved around the ancient city wall indicate settling in the city area until the conquest of the Moors in A.D. 711, who built a military base on top of the Roman theater in Baelo Claudia. Many different indicators of archaeoseismic damage in various places within Baelo Claudia have been described and documented, and we were able to date those with archaeological findings (e.g., pottery remains and coins) and 14C dating. These allow a tentative bracketing of the occurrence of repeated strong archaeoseismic damage (intensity ≥IX MSK) at Baelo Claudia to around A.D. 40–60 and A.D. 260–290 (Silva et al., 2009). The Isis Temple Problem The Capitol, devoted to Jupiter, Minerva, and Juno, which consists of three almost identical temples, dominates the forum and central area of the Roman village (Fig. 4). These temples are thought to have been constructed between A.D. 50 and 70—during the first imperial reconstruction period in Baelo Claudia—to represent the official religion (Sillières, 1997). Aside from the central temple area, a temple dedicated to the Egyptian goddess Isis (Figs. 4 and 5) was built in the adjacent area. Especially after the annexation of Egypt in 30 B.C., the Roman Empire adopted the Isis cult and spread it throughout their entire domain. During the reign of Emperor Caligula (A.D. 37–41), the cult was eventually established in the entire Roman Empire. Archaeological studies assumed that the Isis temple of Baelo Claudia was built around the year A.D. 70, with dimensions of 29.85 m × 17.70 m (Fig. 5). The sanctuary was divided into a public cultural area with a cella and private rooms. Currently, the area of the Isis temple is only partly excavated (Fig. 6). As with the other three temples, the Isis temple was built exactly at the toe of a small topographic scarp and displays a set of particular structural deformations that can be attributed to a shallow landsliding event presumably triggered by the second episode of damage recorded in the city during the third century A.D. (Silva et al., 2009).

133

The Isis temple is one of the few buildings of the city in which evidence for two destruction events can be assessed. The temple is partially buried by a colluvial deposit of 2.5–3 m thickness and displays some of the best examples of ground-failure building damage. Beside ceramics, glass shards, shells, and bones, the colluvium also contains slag of iron smelting. A rough stratigraphy based on Roman pottery allows dating the latter collapse event to the third century A.D. The base of the colluvial deposits was dated to be 1955 ± 30 yr B.P., pointing to a first damage event in the first century A.D. (see Table 1 for the dating results). The sample from the top of the section (Fig. 6) yielded an age of 1725 ± 25 yr B.P. This top soil sample was taken from directly underneath a toppled stucco plastered wall. Radiocarbon dating indicates that (1) the construction of the Isis temple occurred earlier than previously assumed by archaeological studies (e.g., Sillières, 1997), and (2) refuse accumulated beside the Capitol area during no more than 200–250 yr. Additionally, another set of soil samples from directly underneath the fallen columns and walls within the Isis temple was collected. Fragments of toppled columns, wall, and pillar collapses are predominantly oriented in a SW direction (Silva et al., 2009), but also parallel to the aforementioned landslide. Dating of soil samples below the column fragments (Fig. 6) yields radiocarbon ages between 2020 ± 25 yr B.P. and 1195 ± 30 yr B.P. These two dates are relatively consistent and related to the destruction of the temple by seismic activity (2020 ± 25 yr B.P. and 1900 ± 30 yr B.P.). The very young age may be interpreted as having originated by later modifications (quarry use) during the Muslim Spanish Period. Taking this into account, we are able to ascribe the event of the destruction of the Isis temple to the first century A.D., relatively soon after its construction, followed by immediate reconstruction, as previously described for other structural elements such as the city wall (Sillières, 1997). However, parts of the temple area were used as a contemporary refuse tip. During the second damage event in the third century A.D., remains of the damaged walls of the temple eventually collapsed on top of the tip. After this episode, the Capitol area was covered by a 1.5–2.0-m-thick post-Roman colluvium (no longer present) and thus was protected from erosion processes, except for the area where the suspected quarry reutilization was carried out during the Muslim period. PALEOSEISMOLOGICAL OBSERVATIONS Application of Atakan’s Logic Tree and UNIPAS v. 3.0 to the Baelo Claudia Site Atakan et al. (2000) created a logic tree approach to quantify the uncertainties related to paleoseismological investigations. At each node, at least two alternatives with their respective uncertainties can be described. The tree takes into account six basic criteria (Atakan et al., 2000, p. 416): “1. tectonic setting and strain-rate; 2. site selection for detailed analysis (site selection criteria); 3. extrapolation of the conclusions drawn from the detailed site analysis to the entire fault; 4. identification of individual

134

Grützner et al.

Sample points

0

10

30

50 m

N

Min2

Min1

Isis1

Isis5

Isis2 Isis4

Isis3

Figure 5. Detailed plan of the Isis temple area, with the mapped archaeoseismic damage and sample points for 14C dating. Small arrows with lines indicate cracks and opening directions; big arrows indicate deformation directions of walls; light gray shaded areas near Isis1 and Isis2 indicate debris of fallen walls, dark gray shaded areas indicate pavement. Min1 and Min2 indicate the position of the Minerva temple sampling points; Isis1–Isis5 indicate the position of the Isis temple sampling points.

paleo-earthquakes (diagnostic criteria); 5. dating of paleo-earthquakes (type of technique); 6. paleo-earthquake size estimates (slip on individual events, correlation between trenches).” These single steps then lead to a final joint probability of the entire estimation, the probability of the preferred end solution, called Pes. For the implementation of a paleoseismological study in seismic hazard analyses, it is necessary to introduce a correction term Cri, which describes the relative importance of the study. In combination with the correction term Cri, a paleoseismic quality factor (PQF) can be calculated as follows: PQF = Pes × Cri. Atakan et al. (2000) introduced the UNIPAS v. 3.0 program, which can be downloaded from the webpage of Bergen University (http://www.geo.uib.no/seismo/software/unipas/ unipas1.html) to facilitate and automate the calculations.

Tectonic Setting and Strain Rate The study area is located at the Eurasia-Africa plate boundary (Fig. 1), where the convergence rate is ~4 mm/yr (DeMets et al., 1990). According to the suggestions of the Atakan’s logic tree, this corresponds to a quality weight factor of 0.8–1.0 (plate boundaries, high strain rates). In the environs of Baelo Claudia, moderate earthquakes (Silva et al., 2009) occur; however, they are not as common as along other plate margins (see Figs. 2 and 3 for the regional tectonic setting). Background knowledge is available and categorized as intermediate. In the field, geomorphological features such as the Cabo de Gracia fault and the La Laja fault are clearly visible from a great distance (Fig. 7). Stepped mountain ridges are interpreted as the morphological expression of the

The Baelo Claudia (southern Spain) case study

135

Isis1: 1725 ± 25 yr B.P.

Isis5: 2020 ± 25 yr B.P. Isis2: 1955 ± 30 yr B.P.

Isis3: 1900 ± 30 yr B.P.

Isis4: 1195 ± 30 yr B.P.

Figure 6. Results of 14C dating and photos of sample locations in the Isis temple area. Here, several toppled columns have a similar orientation. Those columns fallen in the oldest event fell on a clean floor, indicating that the building was not abandoned at that time. The ones giving younger 14C ages fell on colluvium accumulated after the first earthquake.

Name Isis1 Isis2 Isis3 Isis4 Isis5 Min1

Min2

TABLE 1. DATING RESULTS FROM THE TEMPLE AREA Laboratory code Sample Material KIA32686 Isis-1 Soil Sediment, leaching residue Sediment, humid acid KIA32687 Isis-2 Soil Sediment, leaching residue KIA32688 Isis-3 Soil Sediment, leaching residue Sediment, shell KIA32689 Isis-4 Soil Sediment, leaching residue KIA32690 Isis-5 Soil Sediment, leaching residue KIA32691 Min-1 Roots Sediment, leaching residue Sediment, humid acid Charcoal, leaching residue KIA32692 Min-2 Soil Sediment, leaching residue Sediment, humid acid

Carrizales fault. The visibility of paleoseismological features is considered as low to intermediate. Generally, official seismic hazard maps show low to moderate values in Bolonia Bay, while the Sagalassos study site in Turkey, used by Sintubin and Stewart (2008) to check the archaeoseismological logic tree proposal, is situated in a high hazard region. However, an evaluation based on hazard map data leads to circular reasoning as the paleoseismological investigations are not normally included in the mapped hazard estimations. Subse-

Age (radiocarbon) 1955 ± 30 yr B.P. 1885 ± 25 yr B.P. 1725 ± 25 yr B.P. 365 ± 20 yr B.P. 1900 ± 30 yr B.P. 1195 ± 30 yr B.P. 2020 ± 25 yr B.P. 390 ± 25 yr B.P. 1070 ± 40 yr B.P. 1025 ± 25 yr B.P. >1954 A.D. >1954 A.D.

quently, we assume a QWF of 0.67 as derived from the parameters implemented in the UNIPAS v. 3.0 program. Site Selection Ground penetrating radar (GPR) and geoelectric resistivity measurements were conducted along the mountain ranges in order to visualize faults and fault-related features. Most of the GPR data show the accumulation of debris at the base of steep

136

Grützner et al.

W

E

A

Sierra de la Plata

S

N

B

Silla de la Papa La Laja

Figure 7. Geomorphological features associated with neotectonic activity. (A) Georadar and geoelectrical measurements were carried out at the Cabo de Gracia fault prior to trenching; trench T1 is located perpendicular to the linear rock face in the left of the image (see Fig. 3 for location). The Sierra de la Plata cliff is visible in the background. (B) The La Laja fault as seen from San Bartolome. Hanging valleys have formed along the mountain range, shaping triangular facets (black lines). Silla de la Papa in the background is the highest peak in the Bolonia Bay area (430 m).

scarps (La Laja and Sierra de la Plata; Fig. 7) and underlying sandstones and clays subject to differential weathering. Rockfall and colluvial deposits can be identified; however, the occurrence of faulting-related colluvial wedges could not be clarified peremptorily. Only one prospection site allowed crossing the expected fault zone (Fig. 7). Due to the high conductivity of shallow weathered clays, data quality is poor for some locations. Two-dimensional (2-D) geoelectric tomography shows a clear change of apparent resistivity on a cross profile perpendicular to the suspected fault zone. This feature is a result of increased soil moisture due to a nearby spring, which is most likely faultrelated. An offshore seismic survey conducted in 2006 (Meteor Cruise M69/1) concentrated on the proposed southwest extension of the Cabo de Gracia fault (Silva et al., 2006). The data show evidence for Quaternary faulting and allow us to determine the length of the fault. Geomorphological analyses yield clear hints for fault activity, such as large linear lineaments (Fig. 7; see also Silva et al., 2009) with lichen-free ribbons at the base of the mountain fronts, triangular facets, liquefaction structures, and fresh slickensides close to the shore. The Carrizales fault offsets the Cabo de Gracia escarpment at various places and could therefore be responsible for the observed slickensides. Activity during Holocene times is likely and the two faults join at the trenching site. According to Michetti et al. (2005), despite these features, Bolonia Bay cannot be classified as a “classic” seismic landscape with clear primary surface ruptures. However, the zone displays a large number of features related to secondary and sympathetic earthquake ground effects (Silva et al., 2009). The trenching site was chosen at the location where most of the geophysical investigations were conducted and with regard to accessibility. As mentioned already, there is only one location suitable for crossing the suspected fault zone with geophysical measurements and trenching. The observations made correspond

with a QWF of 0.6–0.8 in the logic tree. The lack of shallow seismics and onshore deep reflection studies, the poor quality of some GPR profiles, and the fact that the fault-trenching sites had to be located at a very specific site and close to each other, resulted in a decrease of the QWF to 0.6, as computed by the UNIPAS v. 3.0 algorithm. Data Extrapolation The fault-trenching sites were located at the only site suitable for excavation and within a range of 100 m of each other (Fig. 3). The average depth of the open trenches was 2.2 m, with a total research length of 44 m. The fault is supposed to have a length of no more than 10 km (onshore and offshore). Instrumental seismicity in the area shows shallow and intermediate earthquakes at depths not exceeding 60 km, with a concentration of hypocenters at depths between 10 and 15 km (Silva et al., 2006). With these values, the trench to fault ratio (TFR) calculated is 0.00000015, corresponding with QWF 5.5. Since a combination of primary and secondary evidences was employed, a QWF of 0.5 was chosen.

Sintubin and Stewart Logic Tree

Uncertainties and Application to Seismic Hazard Analysis

The logic tree presented by Sintubin and Stewart (2008) is a modification of the logic tree for paleoseismology (Atakan et al., 2000) and works in a similar way. Accordingly, the first three criteria defined deal with the probability of an earthquake affecting the study site. In the second half of the logic tree, the focus is on the damages observed. The following aspects are to be evaluated: (1) tectonic setting; (2) site environment; (3) site potential; (4) identification of damage; (5) dating of damage; and (6) regional correlation (Fig. 9). The joint probability resulting from the answer to those questions is a value between 0 and 1. Similar to the paleoseismological scheme, a site confidence level (SCL) is introduced. The SCL is divided into seven stages and corresponds to correction term C, which is between 1 and 10. A qualitative evaluation of the excavations and the quality of the excavation reports account for the determination of the SCL. The product of Pes and C then yields the archaeoseismological quality factor AQF, in analogy to the PQF of Atakan et al. (2000).

At the current state of investigations, the Atakan logic tree yields a probability of 0.005 for our solution (Fig. 9). This value is thirty times lower than the one computed on the Bree fault example (Atakan et al., 2000) in Belgium, mainly due to the limited possibility of extrapolating the trench observation to the entire fault, the uncertain earthquake patterns in the trenches, and the lack of dating. Considering that the results of this study will be used to complete the local earthquake catalogue and to give

Tectonic Setting The first criterion is identical to the one in the paleoseismologic tree of Atakan et al. (2000). According to the setting along the active plate boundary of Africa and Eurasia and to the convergence rate of 4 mm/yr, we had to choose a QWF of 0.8–1.0. On the other hand, the plate boundary is diffuse in the study area, the faults in the vicinity are visible but difficult to evaluate, and an M >6 epicenter within a radius of 10 km cannot be proven

The Baelo Claudia (southern Spain) case study

Total number of nodes: 6 Total number of branches: 12 Total number of end solutions: 64

Identification of paleoearthquakes

Extrapolation of data Site selection Tectonic setting 0.67

0.2

0.5

Paleoearthquake Dating size estimation of paleoPes1= 0.005 earthquakes 0.5 6

0.25 5

4

3

0.6

Level of importance 3 (Cr i = 6) PQF= Pes1 x Cri = 0.03

2

1 Atakan et al. (2000) Regional correlation SPF = 0.28

Dating of damage Identification of damage

Site potential Site environment Tectonic setting 0.67

0.6

0.7

0.8

139

0.7

0.4

Pes1= 0.063

6

5

4

3

2

Figure 9. The Atakan et al. (2000) logic tree for paleoseismology consists of 12 branches and six nodes at which certain probabilities must be defined (upper image). The result is the probability of the preferred end solution, Pes. In this study, Pes is 0.005, 30 times lower than the one achieved by Atakan et al. (2000) at the Bree fault example. The paleoseismic quality factor (PQF) is Pes × Cri (Cri is a correction term depending on the level of importance of the study). PQF is 0.03 in our case. UNIPAS v. 3.0 automatically computes the Pes based on the values entered and creates the graph. The logic tree of Sintubin and Stewart (2008) has been applied to the Roman ruins of Baelo Claudia (lower image). The site potential factor (SPF) is 0.28; the resulting overall probability of our preferred end solution Pes1 is 0.063; the archaeological quality factor (AQF) computes to 0.5 (Pes1 × correction term C).

SCL 6 (C = 8) AQF= Pes1 x C = 0.5

1 Sintubin & Stewart (2008)

without any doubts. Background knowledge must be categorized as intermediate, and the tectonic setting points to a value between 0.6 and 0.8. Following the estimations from the previous sections, we assume a QWF of 0.67. Site Environment Sintubin and Stewart (2008) recommend different ways to evaluate the site environment factor. Based on the landscape signature categories introduced by Michetti et al. (2005) (see also Michetti and Hancock, 1997) for active normal faults in the Mediterranean, the QWF can be determined. Another possibility is to use the INQUA (International Union for Quaternary Research) Environmental Seismic Intensity (ESI) scale for categorizing the seismic landscape (Michetti et al., 2007; Reicherter et al., 2009). Bolonia Bay and adjacent areas are situated on an diffuse plate boundary with convergence rates of ~4 mm/yr. Liquefaction structures on Quaternary deposits have been found, as well as a variety of late Pleistocene to historical landslides and rockfalls; faults can be traced over kilometers, and sparse evidence for recent tectonic activity is visible in the field (see also site selection subsection of Paleoseismological Observations). These features correspond to a QWF of 0.6–0.8. In contrast, Bolonia Bay is not a classic seismic landscape with primary surface ruptures, but it displays a large variety of secondary earthquake ground

effects. Environmental earthquake effects recorded in the area do not necessarily point to an M >6 event. Considering the similarity to the Atakan tree, we choose a QWF of 0.6. Site Potential Baelo Claudia was populated by the Romans from the late second century B.C. to the late fourth century A.D. After that, sparse paleo-Christian to Visigothic settlement is documented, and parts of the town’s ruins served as a source for building materials. With the conquest of the Iberian Peninsula, the Moors built a small garrison on the site. Over the centuries, the ruins have repeatedly been used for military purposes (Sillières, 1997; Silva et al., 2009). Anthropogenic disturbances can be observed occasionally due to this history. The upper part of the ruins is situated on a gentle slope consisting of Cretaceous clayey material, which is subject to occasional creeping. Marine terraces are present close to the shore and host the fish factories. Shallow landslides and landsliderelated deformation are documented in some of the upper parts of the ancient town. Ground settlement must be taken into account, but will be small-scaled—if at all existent—due to underground conditions. An extensive GPR survey has been conducted in the ruins to detect shallow landslide structures, to map the event-horizon

140

Grützner et al.

of the seismic events identified, and to estimate the depth of the bedrock (Silva et al., 2009) by the correlation with the different geotechnical drillings performed within the ruins (Silva et al., 2005). Additionally, archaeological remains such as walls, tombs, streets, and houses, as well as damages (fallen boulders, deformed walls), have been imaged. Many buildings of different types are recorded at Baelo Claudia, including brick and masonry houses, a city wall, a cistern, a theater, shops, a market, a forum, a temple area, roads, a necropolis, an entire fish factory district, and many more. Masonry mainly consists of durable Cretaceous thin-bedded and fine-grained sandstones. In contrast, pillars of houses are made of very porous late Tertiary calcarenites and partly Tyrrhenian beach rocks incorporated with mortar. The calcarenites and beach rocks can be easily hewn but are also fragile and vulnerable to erosion, and the ancient quarries are nearby. The third regional construction material is heavy, siliceous sandstone of the Aljibe Formation (Early Tertiary to Mid-Tertiary); however, due to its hardness, it was exclusively used for stairways or raisers. Archaeological investigations have shown that the town was reconstructed after abrupt ruin and depopulation during the Imperial Roman settlement (e.g., walls and houses were reinforced and rebuilt, an artificial terrace was created etc.; details are listed in Silva et al., 2005). After sporadic investigations in the early twentieth century, the extensive systematic excavation began in the 1970s and continues ever since, now conducted by the Autonomic Government of Junta de Andalucía. Since the beginning of the present millennium, there has been a special focus on seismically induced damages in all ongoing archaeological work. Currently, the ruins are of major touristic significance for the area. Hence, the excavations continue in order to improve the touristic appeal, to increase the number of displays and to support further investigations. There are complete yearly reports of all the excavation campaigns carried out since the year 1970, as well as thematic publications dealing with specific buildings of structural elements within the ruins (i.e., forum, basilica, Capitol) and multiple research articles and special publications on the archaeology of the city. Most of this documentation is listed in the work of Sillières (1997) and can be consulted in the Casa de Velázquez at Madrid. The facts mentioned here suggest a QWF of 0.8–1.0, if the evaluation focuses on the excavation history and on the number of the buildings. Taking into account the quality of buildings and the anthropogenic disturbances, the QWF should be on the order of 0.6–0.8. The moderate physiography would lead to values from 0.4 to 0.6, and the evidence for ground instabilities could even produce a QWF 10 km) to the causative fault, as is the case of Baelo Claudia and the Bolonia Bay area (Silva et al., 2006, 2009). Similarly, the archaeoseismological logic tree approach designed by Sintubin and Stewart (2008) seems to have weaknesses with regard to ground conditions and seismic shaking amplification. Secondary earthquake effects are incorporated in the site environment evaluation, but ground instability facilitating secondary coseismic effects also accounts for the quality of archaeological excavations and therefore has an influence on the evaluation of the site potential. In this step, ground instability evidence diminished the site potential factor value, even with a very good excavation record, as is the case for Baelo Claudia. By circular reasoning, however, weak ground conditions may amplify seismic shaking, thereby triggering severe damage. Therefore, in a case like Baelo Claudia, in which site effects may play a vital role as to the observed damage, the occurrence of unstable ground conditions will lower the site potential factor, diminishing the final Pes estimation and resulting in a possible undervaluation of the true archaeoseismological information recorded at this specific site for future seismic hazard estimations. Due to the lack of similar studies and the absence of sufficient estimations of PQFs and AQFs values for different sites, the classification of our results in a logic tree framework is ulti-

mately not possible. Future investigations also have to prove whether the logic trees are suitable for comparing very different settings of study sites. In the probability estimations, tectonic setting and site environment are taken into account, thus providing the basis for a global approach. However, as already mentioned by Sintubin and Stewart (2008), a direct comparison of several studies in a certain region may enhance the reliability of the sites. Eventually, a growing database will also improve the estimates of a specific site, as it facilitates individual classification of observations in terms of the relevant criteria. Subsequently, the numbers we obtained can be interpreted. In any case, a more robust incorporation of secondary earthquake ground effects and their relation to ground geotechnical properties and seismic amplification in the logic tree approaches employed will be required to conduct more realistic assessments of nonfaulted sites, exclusively devastated by ground shaking, which is the case for most of the severely damaged locations during individual earthquakes. ACKNOWLEDGMENTS This study was financially supported by the German Research Foundation (DFG-project Re 1361/9). We thank the Leibniz Institute in Kiel for radiocarbon dating. The authors would like to thank the reviewers for their inspiring and helpful comments and Kay Barbara for linguistic consultation. We are grateful for everyone who was involved in the fieldwork. This chapter is a contribution to the International Geoscience Programme (IGCP) 567, “Earthquake Archaeology: Archaeoseismology along the Alpine-Himalayan Seismic Zone,” funded by the United Nations Educational, Scientific and Cultural Organization (UNESCO) and to the International Union for Quaternary Research Focus Group on Paleoseismology and Active Tectonics. REFERENCES CITED Alonso-Villalobos, F.J., Gracia-Prieto, F.J., Ménanteau, L., Ojeda, R., Benavente, J., and Martínez, J.A., 2003, Paléogeographie de l’anse de Bolonia (Tarifa, Espagne) à l’époque romaine, in Fouache, E., ed., The Mediterranean World Environment and History: Amsterdam, Elsevier S.A.S., p. 407–417. Ambraseys, N.N., Jackson, J.A., and Melville, C.P., 2002, Historical seismicity and tectonics: The case of the Eastern Mediterranean and the Middle East, in Lee, W.H.K., Kanamori, H., Jennings, P.C., and Kisslinger, C., eds., International Handbook of Earthquake and Engineering Seismology: Amsterdam, Academic Press, International Geophysics Series 81A, p. 747–763. Atakan, K., Midzi, V., Moreno Toiran, B., Vanneste, K., Camelbeeck, T., and Meghraoui, M., 2000, Seismic hazard in regions of present-day low seismic activity: Uncertainties in the paleoseismic investigations along the Bree fault scarp (Roer graben, Belgium): Soil Dynamics and Earthquake Engineering, v. 20, p. 415–427, doi: 10.1016/S0267-7261(00)00081-6. Atwater, B.F., 1992, Geologic evidence for earthquakes during the past 2000 years along the Copalis River, southern coastal Washington: Journal of Geophysical Research, v. 97, p. 1901–1919, doi: 10.1029/91JB02346. Bergen University, Institutt for Geovitenskap, 2008, UNIPAS V3.0 Program: http://www.geo.uib.no/seismo/software/unipas/unipas1.html (accessed 14 February 2010).

The Baelo Claudia (southern Spain) case study Caputo, R., and Helly, B., 2008, The use of distinct disciplines to investigate past earthquakes: Tectonophysics, v. 453, p. 7–19, doi: 10.1016/j .tecto.2007.05.007. Castilla, R.A., and Audemard, F.A., 2007, Sand blows as a potential tool for magnitude estimation of pre-instrumental earthquakes: Journal of Seismology, v. 11, p. 473–487, doi: 10.1007/s10950-007-9065-z. Collins, R., 1998, Spain: An Oxford Archaeological Guide: Oxford, UK, Oxford University Press, 328 p. DeMets, C., Gordon, R.G., Argus, D.F., and Stein, S., 1990, Current plate motions: Geophysical Journal International, v. 101, p. 425–478, doi: 10 .1111/j.1365-246X.1990.tb06579.x. Galadini, F., Hinzen, K.G., and Stiros, S., 2006, Archaeoseismology: Methodological issues and procedure: Journal of Seismology, v. 10, no. 4, p. 395–414, doi: 10.1007/s10950-006-9027-x. Goy, J.L., Zazo, C., Mörner, N.A., Hoyos, M., Somoza, L., Lario, J., Bardají, T., Silva, P.G., and Dabrio, J.C., 1994, Pop up–like deformation of a Roman floor and liquefaction structures in SW Spain as possible paleoseismic indicators: Bulletin of the International Union for Quaternary Research (INQUA) Neotectonics Comisión, v. 17, p. 42–44. Goy, J.L., Zazo, C., Silva, P.G., Lario, J., Bardají, T., and Somoza, L., 1995, Evaluación geomorfológica del comportamiento neotectónico del Estrecho de Gibraltar durante el Cuaternario, in Esteras, M., ed., El Enlace Fijo del Estrecho de Gibraltar: Madrid, Sociedad española de estudios para la comunicacion fija a través del Estrecho de Gibraltar S.A (SECEGSA), v. 2, p. 51–69. Guidoboni, E., and Traina, G., 1996, Earthquakes in medieval Sicily: A historic revision (7th–13th century): Annali di Geofisica, v. XXXIX, no. 6, p. 1205–1225. Höbig, N., Braun, A., Grützner, C., Fernández-Steeger, T., and Reicherter, K., 2009, Rock fall hazard mapping and run out simulation: A case study from Bolonia Bay, southern Spain, in Pérez-López, R., Grützner, C., Lario, J., Reicherter, K., and Silva, P.G., eds., Archaeoseismology and Palaeoseismology in the Alpine-Himalayan Collisional Zone: 1st International Union for Quaternary Research (INQUA) and International Geoscience Programme (IGCP) 567 International Workshop on Earthquake Archaeology and Palaeoseismology, 7–13 September 2009, Baelo Claudia (Cádiz, Spain): Madrid, Instituto Geográfico Nacional de España, p. 52–56. Instituto Geográfico Nacional de España: http://www.ign.es/ign/es/IGN/ SisCatalogo.jsp (accessed 14 February 2010). Karcz, I., and Kafri, U., 1978, Evaluation of supposed archaeoseismic damage in Israel: Journal of Archaeological Science, v. 5, p. 237–253, doi: 10.1016/0305-4403(78)90042-0. Koster, B., Vonberg, D., and Reicherter, K., 2009, Tsunamigenic deposits along the southern Gulf of Cádiz (southwestern Spain) caused by tsunami in 1755?, in Pérez-López, R., Grützner, C., Lario, J., Reicherter, K., and Silva, P.G., eds., Archaeoseismology and Palaeoseismology in the AlpineHimalayan Collisional Zone: 1st International Union for Quaternary Research (INQUA) and International Geoscience Programme (IGCP) 567 International Workshop on Earthquake Archaeology and Palaeoseismology, 7–13 September 2009, Baelo Claudia (Cádiz, Spain): Madrid, Instituto Geográfico Nacional de España, p. 73–75. Levret, A., Backe, C., and Cushing, M., 1994, Atlas of macroseimic maps for French earthquakes with their principal characteristics: Natural Hazards, v. 10, p. 19–46, doi: 10.1007/BF00643439. Luque, L., Lario J., Zazo, C., Goy, J.L., Dabrio, C.J., and Silva, P.G., 2001, Tsunami deposits as paleoseismic indicators: Examples from the Spanish coast: Acta Geologica Hispanica, v. 36, no. 3–4, p. 197–211. Martínez Solares, J.M., 2005, Tsunamis en el contexto de la Península Ibérica: Enseñanza de las Ciencias de la Tierra, v. 13, no. 1, p. 52–59. McCalpin, J.P., and Nelson, A.R., 1996, Introduction to paleoseismology, in McCalpin, J.P., ed., Paleoseismology: San Diego, Academic Press, p. 1–32. Michetti, A.M., and Hancock, P.L., 1997, Paleoseismology: Understanding past earthquakes using Quaternary geology: Part 1: Geodynamics, v. 24, no. I-4, p. 3–10, doi: 10.1016/S0264-3707(97)00004-5. Michetti, A.M., Audemard, F.A., and Marco, S., 2005, Future trends in paleoseismology: Integrated study of the seismic landscape as a vital tool in seismic hazard analyses: Tectonophysics, v. 408, p. 3–21, doi: 10.1016/ j.tecto.2005.05.035. Michetti, A.M., Esposito, E., Guerrieri, L., Porfido, S., Serva, L., Tatevossian, R., Vittori, E., Audemard, F., Azuma, T., Clague, J., Comerci, V., Gürpinar, A., McCalpin, J., Mohammadioun, B., Mörner, N.A., Ota, Y., and

143

Roghozin, E., 2007, Intensity Scale ESI 2007: Memorie Descrittive della Carta Geologica D’Italia, v. 74, p. 41. Nikonov, A.A., 1988, Reconstruction of the main parameters of old large earthquakes in Soviet Central Asia using the paleoseismogeological method: Tectonophysics, v. 147, p. 297–312, doi: 10.1016/0040-1951(88)90191-6. Reicherter, K., 2001, Paleoseismologic advances in the Granada basin (Betic Cordilleras, southern Spain): Acta Geologica Hispanica, v. 36, no. 3–4, p. 267–281. Reicherter, K., Jabaloy, A., Galindo-Zaldívar, J., Ruano, P., Becker-Heidmann, P., Morales, J., Reiss, S., and González-Lodeiro, F., 2003, Repeated palaeoseismic activity of the Ventas de Zafarraya fault (S Spain) and its relation with the 1884 Andalusian earthquake: International Journal of Earth Sciences, v. 92, p. 912–922, doi: 10.1007/s00531-003-0366-3. Reicherter, K., Michetti, A.M., and Silva, P.G., 2009, Introduction, in Reicherter, K., Michetti, A.M., and Silva, P.G., eds., Paleoseismology: Historical and Prehistorical Records of Earthquake Ground Effects for Seismic Hazard Assessment: Geological Society of London Special Publication 316, p. 1–10. Sanz de Galdeano, C., 1990, Geologic evolution of the Betic Cordilleras in the Western Mediterranean, Miocene to present: Tectonophysics, v. 172, p. 107–119, doi: 10.1016/0040-1951(90)90062-D. Sillières, P., 1997, Baelo Claudia: Una ciudad Romana de la Bética: Madrid, Junta de Andalucía–Casa de Velázquez, 237 p. Silva, P.G., Borja, F., Zazo, C., Goy, J.L., Bardají, T., Luque, L., Lario, J., and Dabrio, C.J., 2005, Archaeoseismic record at the ancient Roman city of Baelo Claudia (Cádiz, south Spain): Tectonophysics, v. 408, p. 129–146, doi: 10.1016/j.tecto.2005.05.031. Silva, P.G., Goy, J.L., Zazo, C., Bardají, T., Lario, J., Somoza, L., Luque, L., and González-Hernández, F.M., 2006, Neotectonic fault mapping at the Gibraltar Strait tunnel area, Bolonia Bay (south Spain): Engineering Geology, v. 84, p. 31–47, doi: 10.1016/j.enggeo.2005.10.007. Silva, P.G., Reicherter, K., Grützner, C., Bardají, T., Lario, J., Goy, J.L., Zazo, C., and Becker-Heidmann, P., 2009, Surface and subsurface paleoseismic records at the ancient Roman city of Baelo Claudia and the Bolonia Bay area, Cádiz (south Spain), in Reicherter, K., Michetti, A.M., and Silva, P.G., eds., Paleoseismology: Historical and Prehistorical Records of Earthquake Ground Effects for Seismic Hazard Assessment: Geological Society of London Special Publication 316, p. 93–121. Sintubin, M., and Stewart, I.S., 2008, A logical methodology for archaeoseismology: A proof of concept at the archaeological site of Sagalassos, southwest Turkey: Bulletin of the Seismological Society of America, v. 98, p. 2209–2230, doi: 10.1785/0120070178. Stich, D., Serpelloni, E., Mancilla, F.L., and Morales, J., 2006, Kinematics of the Iberia–Maghreb plate contact from seismic moment tensors and GPS observations: Tectonophysics, v. 426, p. 295–317, doi: 10.1016/j .tecto.2006.08.004. Stiros, S., 1996, Identification of earthquakes from archaeological data: Methodology, criteria and limitations, in Stiros, S., and Jones, R.E., eds., Archaeoseismology: British School at Athens, Fitch Laboratory Occasional Paper 7, p. 129–152. Vollmert, A., Reicherter, K., and Grützner, C., 2009,The origin of rockfalls and the formation of hanging valleys along the La Laja range front (Tarifa, S. Spain), in Pérez-López, R., Grützner, C., Lario, J., Reicherter, K., and Silva, P.G., eds., Archaeoseismology and Palaeoseismology in the AlpineHimalayan Collisional Zone: 1st International Union for Quaternary Research (INQUA) and International Geoscience Programme (IGCP) 567 International Workshop on Earthquake Archaeology and Palaeoseismology, 7–13 September 2009, Baelo Claudia (Cádiz, Spain): Madrid, Instituto Geográfico Nacional de España, p. 162–164. Weijermars, R., 1991, Geology and tectonics of the Betic zone, SE Spain: Earth-Science Reviews, v. 31, p. 153–236, doi: 10.1016/0012-8252(91 )90019-C. Wells, D.L., and Coppersmith, J.K., 1994, New empirical relationships among magnitude, rupture length, rupture width, rupture area, and surface displacement: Bulletin of the Seismological Society of America, v. 84, p. 974–1002. Zazo, C., Silva, P.G., Goy, J.L., Hillaire-Marcel, C., Lario, J., Bardají, T., and González, A., 1999, Coastal uplift in continental collision plate boundaries: Data from the Last Interglacial marine terraces of the Gibraltar Strait area (south Spain): Tectonophysics, p. 301, v. 95–119. MANUSCRIPT ACCEPTED BY THE SOCIETY 19 MAY 2010 Printed in the USA

The Geological Society of America Special Paper 471 2010

Long-term effect of seismic activities on archaeological remains: A test study from Zakynthos, Greece Melek Tendürüs* Institute for Geo- and Bioarchaeology, Vrije Universiteit Amsterdam, De Boelelaan 1085, 1081 HV Amsterdam, The Netherlands Gert Jan van Wijngaarden* Amsterdam Archaeological Centre, Universiteit van Amsterdam, Turfdraagsterpad 9, 1012 XT Amsterdam, The Netherlands Henk Kars* Institute for Geo- and Bioarchaeology, Vrije Universiteit Amsterdam, De Boelelaan 1085, 1081 HV Amsterdam, The Netherlands

ABSTRACT During the archaeological and geoarchaeological surveys on the island of Zakynthos, Greece, it has been noted that the distribution and preservation of archaeological remains of Zakynthos present spatially different characteristics. In general, archaeological pottery finds and architectural remains in the eastern part of the island appear to be more fragmented and more widely distributed than in the western part of the island. Due to the high seismicity in the region, the question has come up whether a correlation between seismic activity and distribution and preservation conditions of archaeological remains exists or not. In order to investigate the mentioned relationship, we looked at the cumulative effect of continuing earthquakes for the last hundred years on the island of Zakynthos. We used ground acceleration to quantify the earthquake-induced damage. The predicted cumulative destruction intensity is presented on a map, and it illustrates that we can cautiously attribute the distribution difference of the archaeological remains with different preservation conditions to the seismic activity on the island. It is hoped that this study will initiate new scientific research into the characteristics of the distribution of archaeological remains in seismically active areas. In addition, it is to be expected that this study will contribute to in situ preservation studies relating to the long-term effect of seismic activities on the archaeological record.

*E-mails: [email protected]; [email protected]; [email protected]. Tendürüs, M., van Wijngaarden, G.J., and Kars, H., 2010, Long-term effect of seismic activities on archaeological remains: A test study from Zakynthos, Greece, in Sintubin, M., Stewart, I.S., Niemi, T.M., and Altunel, E., eds., Ancient Earthquakes: Geological Society of America Special Paper 471, p. 145–156, doi: 10.1130/2010.2471(13). For permission to copy, contact [email protected]. © 2010 The Geological Society of America. All rights reserved.

145

146

Tendürüs et al.

INTRODUCTION The Mediterranean area is well known for its archaeological richness and its frequent earthquakes. It is only very recently, however, that archaeologists, historians, seismologists, geologists, and engineers have begun to collaborate systematically in research and heritage management. At the moment, the common concern of these studies is mainly limited to determining, based on the archaeological and geological-geomorphological evidences, whether or not an earthquake of significance was a major cause in the destruction at a particular site, e.g., Stiros et al. (1996), Ellenblum et al. (1998), Galadini and Galli (2001), Marco et al. (2003), Koukouvelas et al. (2005), Caputo et al. (2006), Similox-Tohon et al. (2006), Reinhardt et al. (2006), and Marco (2008). The cause of destruction in an archaeological stratigraphy (e.g., earthquake, land instability, war) can be determined by examining the distortion on structures and the in situ geological data. With developing methodologies for the recognition of earthquake-induced damages, archaeologists can ascertain the reason for massive destruction at their site more reliably. Sometimes, the destruction layer recorded in the destroyed ancient site might be dated, using the archaeological context, or be correlated with a devastating earthquake event in the region, using historical sources of the expected time period mentioning such an event. The inferred earthquake is used by seismologists in the improvement of historical earthquake catalogues for the assessment of seismic hazard; by geologists in understanding of the geodynamic characteristics of the region; and by engineers in the development of new constructional methods to mitigate the seismic risk. However, the effect of seismic activities on archaeological remains does not simply conclude with one devastating event, it is an ongoing process that may continue to destroy the archaeological remains (Papastamatiou and Psycharis, 1993; Psycharis et al., 2000; Cerone et al., 2001). In this paper, the effect of seismic activities on archaeological remains is not considered to be constrained to only one devastating event. The study concentrates on the probable continuing destruction of archaeological record by earthquakes before, during, and after the abandonment of sites. Seismic hazard studies will be used to begin to understand the archaeology of a seismically active region. The area of study, Zakynthos (western Greece), is located in one of the highest seismic activity regions of the world and has been inhabited since the Paleolithic. During the archaeological and geoarchaeological surveys on the island in the period of 2005 to 2008, it was noted that the distribution and preservation of archaeological remains of Zakynthos present spatially different characteristics. In general, archaeological pottery finds and architectural remains in the eastern part of the island appear to be more fragmented and more widely distributed than in the western part of the island. Standing architectural remains also appear to be in a better condition in the west than in the east. It is believed that there may be a correlation between the ongoing seismic activities in the region and the distribution and preser-

vation conditions of archaeological remains in Zakynthos. This paper particularly addresses the development of a method that may be used to investigate spatially such a probable link between the conditions of archaeological remains and the seismicity of the specific region. STUDY AREA Geological and Seismotectonic Background Zakynthos—the southernmost of the Ionian Islands of western Greece—lies in a tectonically complex and active area (Fig. 1; Underhill, 1989; Papazachos and Kiratzi, 1996; Barka and Reilinger, 1997; Hinsbergen et al., 2006; Lagios, et al., 2007). In particular, the Ionian Basin of the African plate subducts beneath the Aegean continental microplate of the Eurasian plate, the Apulian continental crustal part of the African plate collides with the Eurasian plate in the north, and the Cephalonia transform fault zone connects these subduction and collision zones. The seismicity of the area is the highest in Greece, mainly consisting of shallow seismic activities (Papazachos, 1990; Papazachos et al., 1993; Clément et al., 2000). Figure 2 shows the spatial distribution and the magnitude occurrences of the earthquakes that were recorded between 1901 and 2006 (except the last 3 months) with Mw ≥4.5 in the vicinity of Zakynthos, which also form our data set for the study. Within the defined area, there are 1975 events, including the most destructive earthquake of the last century in Greece, which occurred in the Ionian Islands on 12 August 1953 with a surface-wave magnitude of 7.2 (Stiros et al., 1994). Hatzidimitriou et al. (1985) calculated the return period of this and larger magnitudes of earthquakes for the vicinity as 29 yr, based on the data covering the last 81 and 181 yr periods. The important local seismicity of Zakynthos occurs along the Ionian thrust, which also divides the island into two different geological units: Pre-Apulian and Ionian (Fig. 1). The PreApulian (also called Paxos) unit is recognized on the island as the Vrachionas carbonate anticline. It is composed of thick Upper Cretaceous limestones and dominates the western part of the island. Marly limestones of Eocene and Oligocene age are exposed at its eastern slope. Miocene deposits are observed at the central lowlands as sandstones and bluish marls with gypsum intercalations. Toward the east, this range of hills suddenly leaves its place to a plain filled with recent alluvium deposits. The Ionian unit is exposed on the eastern part of the island, namely at the Vasilikos Peninsula. The peninsula mainly consists of Pliocene and Pleistocene marine mudstones and sandstones, and Triassic evaporates and limestones also crop out. Archaeological Investigations Zakynthos is mentioned in various historical sources, indicating its long and intensive habitation. It is mentioned on Linear B tablets from Pylos showing the overseas connection of the island with the mainland in the Mycenaean period (Palaima,

A test study from Zakynthos, Greece

147

Figure 1. Plate boundaries and main tectonic features of Greece (Barka and Reilinger, 1997; Clément et al., 2000) and the geological map of the Zakynthos Island (Perry and Temple, 1980; Underhill, 1989). CTF—Cephalonia transform fault, NAF—North Anatolian fault.

1991). Homer mentions the island as a part of the territory of Odysseus (Iliad II: 634; Odyssey IX: 24). In Classical and Hellenistic periods, the island often was an ally of Athens (Kalligas, 1993). Pliny the Elder mentions the particular fertility of the island (The Natural History 4: 19: 12). During the Venetian period, Zakynthos was known as “the flower of the Levant” due to its beauty and fertility (Ζois, 1955). In spite of its prosperous land and the suitable geographic location with regard to local and Mediterranean maritime traffic, the archaeology of Zakynthos is relatively little known. Sylvia Benton of the British School in Athens was the first archaeologist to systematically describe several archaeological sites on the island (Souyoudzoglou-Haywood, 1999). She excavated a Mycenaean house in Cape Kalogeras on the Vasilikos

Peninsula and a tholos tomb near Alykanas in the 1930s. Unfortunately, her results remained unpublished, and at the time of the 1953 earthquake, all the finds and records were lost. The Greek Archaeological Service carried out excavations in the early 1970s (Mylona, 2006). Their research focused on the hill rising just behind the modern town of Zakynthos and its surrounding area. Today, there is a Venetian castle with extensive British modifications standing on top of the hill facing Peloponnese. The uncovered archaeological artifacts and some standing architecture in the interior of the castle indicate occupation during the Bronze Age and during Archaic, Classical, Hellenistic, and Roman periods. Based on the archaeological evidence and ancient authors’ texts, it can be concluded that the hill used to be an ancient acropolis already fortified by the mid-fifth century B.C.

148

Tendürüs et al. tholoi and settlement remains from the Mycenaean period, and other concentrations of finds, especially pottery, from Archaic, Hellenistic, or Roman times, medieval, and early modern periods have been recorded. Archaeological Remains

Figure 2. The spatial distribution (A) and the magnitude frequency (B) of the earthquakes recorded in the vicinity of Zakynthos between 1901 and 2006 (except the last 3 months). The presented data also form the data set of this study and cover the magnitudes of earthquakes Mw 4.5 and greater.

Currently, systematic archaeological research on the island has been continuing since 2005 as a joint project of the Netherlands Institute in Athens (NIA) and the 35th Ephorate of Prehistoric and Classical Antiquities: the Zakynthos Archaeology Project (ZAP) (Van Wijngaarden et al., 2006, 2007). The project aims to gain more insight into the archaeology of the island, and it combines archaeological surveys with geographic information system (GIS), aerial photography, and geomorphologic studies to better identify the human-nature interaction in the past and also to understand the effect of today’s processes on the archaeological remains (e.g., Rink, 2005; Stoker 2006; Pieters et al., 2007; Horn Lopes, 2008; Storme, 2008; Tendürüs, 2009). In 3 years, the project showed that the island was inhabited in several periods. Lithic tools and flakes from the Paleolithic to Early Bronze Age,

Unlike many other areas in Greece, Zakynthos has very few standing archaeological remains that date back to more than a few centuries ago. The only excavation that was fully published is the Mycenaean cemetery at Kambi in the western mountains of the island (Agalopoulou, 1973). Other notable ancient remains, probably of Roman date, are found built into the little church St. Dhimitrios at Melinado, near Machairado (Fig. 3A; Foss, 1969; Kalligas 1993). Palaiokastro is another significant archaeological site on Zakynthos. It is located on the hills west of Machairado with an imposing view of the plain. The site was considered to be medieval, but the finds in 2007 and 2008 indicated activities also in prehistory and antiquity (Van Wijngaarden et al., 2009, 2010). Apart from the medieval structures, several walls, probably to be dated sometime in the period from Archaic to Hellenistic times, have been recorded on the ridges and the plateaus to the west of the top of the hill (Fig. 3C). Figure 4 presents the archaeological sites known on the island from the literature and from the surveys of the ZAP. Among these sites, there are settlement remains (i.e., architectural remains like foundations and walls), graves, and surface finds (i.e., small finds including mainly pottery sherds and stone tools). In 2006, the survey conducted at the southeastern peninsula (Vasilikos) revealed large quantities of pottery and lithics, mainly lacking of local concentrations. Archaeological ceramics that were recovered, included sherds of considerable quality, but they were generally heavily worn. Standing ancient settlement remains were only observed at Kalogeras where the thick brushes were removed. A view from the discovered various eroding walls is displayed in Figure 3B. The majority of the pottery found in association with these walls dated to the prehistoric period, in particular, to the Middle and Late Bronze Ages (Von Stein, 2009). Elsewhere at the site, Archaic and Classical finds were more abundant. In comparison to the Vasilikos Peninsula, the distribution of archaeological artifacts in the vicinity of Keri, where a pilot survey was carried out in 2005, showed more concentrations of material, probably representing archaeological sites. In a few cases, these sites could be identified with certainty on the basis of the quantity, the quality, and the diversity of the surface material. In general, the archaeological record in the southeast appears to show a higher degree of disturbance and dispersion. Figure 4 also indicates churches and monasteries of Zakynthos, the major architectural remains on the island. More of these monuments are located in or near the town of Zakynthos and in the western high areas. The churches at the town were reconstructed after the catastrophic earthquake of 1953, except the concrete reinforced church of St. Dionysios, built in 1948 (Facaros and Theodorou, 2003). Some of the churches and the

A test study from Zakynthos, Greece

149

A

C B Figure 3. A few example views from the “standing” archaeological remains on Zakynthos. (A) The church of St. Dhimitrios in Melinado containing marble columns and blocks, probably of Roman date. (B) Part of wall remains at Kalogeras, probably from the second millennium B.C. (C) Part of wall remains at the site of Palaiokastro. (D) The monastery of Skopiotissa on Mount Skopos being restored.

D

monasteries in the western part of the island were also badly and partially damaged. The most serious damages mentioned are those on the church of St. Andreas at the northwest and the little church at the east foot of the Vrachionas (Foss, 1969). The monastery of Skopiotissa on Mount Skopos was also destroyed with the earthquake, but it is currently being restored (Fig. 3D).

and concrete supports. Although short-term effects of earthquakes on the archaeological remains receive some attention, their longterm effects have been neglected in archaeological studies. Therefore, there is virtually no directly relevant literature available for our investigation. On the other hand, studies in the fields of seismology, geology, and civil engineering can help us to look into the spatial distribution of the combined effect of past earthquakes.

METHOD Quantification of Earthquake-Induced Damage Human response to the damaging effects of earthquakes on archaeological remains has only recently improved further than trying to provide stability of some standing buildings with steel

Earthquakes have six major effects. Ground motion and faulting are two of them and cause damage directly. Others are fire,

150

Tendürüs et al.

Figure 4. Geographic distribution of the archaeological sites, churches, and monasteries of Zakynthos. A, B, and C represent the intensive archaeological survey areas of the Zakynthos Archaeology Project. Areas A and C were covered in 2005 and 2006. The investigations at area B still carry on.

landslides, cliff collapse, or any mass-wasting movements, liquefaction, and tsunamis (initiated by earthquakes). These are secondary effects that cause damage indirectly (Yeats et al., 1997). In our study, we used the ground motion to quantify the damage due to seismic activities because it is the most common way of determining the level of vulnerability of a region in seismic hazard assessment studies. The level of ground shaking

caused by an earthquake at a site mainly depends on the magnitude of the earthquake, geographical proximity of the site to the seismic source, and local geological characteristics of the media. For instance, ground shaking decreases when the waves propagate outward from the source; soft sediments amplify seismic waves and create more ground shaking than hard rocks (Day, 2002).

A test study from Zakynthos, Greece The level of ground shaking, or damage, is usually measured as peak ground acceleration (PGA), which is often determined by attenuation models (Day, 2002; Bozorgnia and Campbell, 2004). An attenuation model is defined as a mathematical expression developed to estimate the peak ground acceleration at a certain distance from a seismic source, using a given data set of seismological parameters (deterministic method) or using all possible earthquake locations and magnitudes together with their expected probabilities of occurrence (probabilistic method). There are a few attenuation relationships developed specifically for Greece and its neighboring regions. The attenuation model by Theodulidis and Papazachos (1992) provides a reasonable and geographically specific model to apply to PGA seismic evaluations for Greece (Burton et al., 2003) and was selected for the calculations in this study. The model was developed studying 36 shallow earthquakes from Greece with magnitudes Ms 4.5–7.0 and four from Japan and Alaska with magnitudes Ms 7.2–7.5. The attenuation relation is: ln ag = 3.88 + 1.12 Ms − 1.65 ln ( R + 15) + 0.41 S + 0.71 P,

(1)

where ag is the peak horizontal ground acceleration in cm s–2, R is the epicentral distance in km, S is a parameter equal to zero at “alluvium” sites and equal to one at “rock” sites, and P is a parameter equal to zero for mean or 50 percentile values and one for 84 percentile values (taken as zero in our calculations). Figure 5 shows the change of PGA values projected by this attenuation model when the waves are propagating further from the epicenter of an earthquake of magnitude Ms 4.5, 6.0, 7.5, and 9.0 in the same type of geological medium. While the effect of a Ms 4.5 earthquake remains local, an earthquake with a strongermagnitude earthquake, e.g., Ms 7.5, is felt moderately at 60 km distance. Data Collection We put together a collection of the magnitude and location of past earthquakes for the vicinity of Zakynthos, forming

our data set from the recent catalogue published by the Aristotle University of Thessaloniki (Papazachos et al., 2007). The selected seismic events have moment magnitudes of 4.5 and greater and occurred between 1901 and 2006 (except the last 3 mo). The area coverage of the data set is 300 × 300 km2 within the frame bounded by the coordinates of 36.420°N–39.140°N and 19.072°E–22.500°E. The spatial distribution of earthquake events included in our data set and their magnitude occurrences are shown in Figure 2. An earthquake of Richter magnitude ML 4 is felt by almost everybody; it breaks some dishes and windows and can displace unstable objects (Table 1; U.S. Geological Survey, 2010). Since slight damages start appearing during earthquakes of ML 4, equivalent to Mw 4.5 (please refer to the study of Papazachos et al. [1997] for the relationships between the magnitudes in the region), it is set as the threshold value for our data set. Due to their unstable nature, we considered that archaeological remains will be more prone to ground shaking. Although the compiled earthquake catalogues of Greece extend back to 550 B.C.E. (Papazachos and Papazachou, 1997; Papazachos et al., 2000), we chose the earthquake records from 1901 (introduction of seismograph in Greece) onward as our data set because the historical records are lacking a considerable amount of seismic events, most importantly, in the lower-magnitude range (Ambraseys, 1996; Kouskouna and Makropoulos, 2004). Insertion of the earlier records into our data set would promote a few devastating events and omit a very large collection of smaller and unrecorded events, causing a deceptive geographical distribution in our results. Cumulative Destruction Intensities The cumulative damaging effect of past earthquakes, which we call cumulative destruction intensity, was calculated by accumulating the PGA values of each earthquake in our data set for the vicinity of Zakynthos using the attenuation relationship of Theodulidis and Papazachos (Eq. 1). First, the earthquake magnitudes of the data set were converted from moment magnitude to surface-wave magnitude using the following relationship (Papazachos et al., 1997): Mw = MS, 6.0 ≤ MS ≤ 8.0, Mw = 0.56MS + 2.66, 4.2 ≤ MS ≤ 6.0.

Figure 5. The change of peak ground acceleration (PGA) values projected by the attenuation model of Theodulidis and Papazachos (1992).

151

(2)

Second, the geological units of the island were classified as alluvium and rock for the S parameter, a dummy variable taking the value of zero for “alluvium” sites and one for “rock” sites in the attenuation relationship. Figure 6 shows the rock sites of Zakynthos in gray color, including limestones, sandstones, mudstones, and gypsum, and the alluvium sites in white. Finally, the area covered by our data set was divided into 40,000 grid points, and PGA values were calculated on every grid point for each earthquake in the data set (Fig. 7). The computed PGA values for each grid point were accumulated in the process to obtain the total

152

Tendürüs et al. TABLE 1. COMPARISON OF RICHTER MAGNITUDE AND MODIFIED MERCALLI INTENSITY SCALES Richter magnitude

Modified Mercalli intensity

Description

1.0–3.0

I

Not felt except by a very few under especially favorable conditions.

3.0–3.9

II

Felt only by a few persons at rest, especially on upper floors of buildings.

III

Felt quite noticeably by persons indoors, especially on upper floors of buildings. Many people do not recognize it as an earthquake. Standing motor cars may rock slightly. Vibrations similar to the passing of a truck. Duration estimated.

IV

Felt indoors by many, outdoors by few during the day. At night, some awakened. Dishes, windows, doors disturbed; walls make cracking sound. Sensation like heavy truck striking building. Standing motor cars rocked noticeably.

V

Felt by nearly everyone; many awakened. Some dishes, windows broken. Unstable objects overturned. Pendulum clocks may stop.

VI

Felt by all, many frightened. Some heavy furniture moved; a few instances of fallen plaster. Damage slight.

VII

Damage negligible in buildings of good design and construction; slight to moderate in well-built ordinary structures; considerable damage in poorly built or badly designed structures; some chimneys broken.

VIII

Damage slight in specially designed structures; considerable damage in ordinary substantial buildings with partial collapse. Damage great in poorly built structures. Fall of chimneys, factory stacks, columns, monuments, walls. Heavy furniture overturned.

IX

Damage considerable in specially designed structures; well-designed frame structures thrown out of plumb. Damage great in substantial buildings, with partial collapse. Buildings shifted off foundations.

X

Some well-built wooden structures destroyed; most masonry and frame structures destroyed with foundations. Rails bent.

XI

Few, if any (masonry) structures remain standing. Bridges destroyed. Rails bent greatly.

XII

Damage total. Lines of sight and level are distorted. Objects thrown into the air.

4.0–4.9

5.0–5.9

6.0–6.9

7.0 and higher

Earthquake 1 MS(1)

i R(i1, j1, 1) S(i1, j1) R(i2, j2, 1)

S(i2, j2)

R(i3, j3, 1)

j

R(i1, j1, 2) Earthquake 2 MS(2)

R(i2, j2, 2) S(i3, j3)

R(i3, j3, 2)

Figure 7. A simple sketch of the calculation procedure of cumulative destruction intensity values. There are 40,000 grid points in the defined area and 1975 earthquakes in the considered period.

Figure 6. Two basic geological units of Zakynthos: rock sites (black) and alluvium sites (white). A, B, and C represent the intensive archaeological survey areas of the Zakynthos Archaeology Project. Areas A and C were covered in 2005 and 2006. The investigations at area B continue.

A test study from Zakynthos, Greece damage induced by all the earthquakes during the considered time period, using the formula CPGA (i, j ) = ∑ k =1 ag (i, j, k ), N

i = 1… L, j = 1…L,

(3)

where i and j are the horizontal and vertical grid indices, L is the grid resolution, N is the total number of earthquakes in the data set, and ag(i, j, k) is the peak ground acceleration at grid point (i, j) caused by earthquake k. Using Equation 1, this is ultimately

#

N

CPGA i, j  - exp 3.88 1.12 Ms  k < 1.65 ln k 1

}

⎡⎣R (i , j, k ) + 15⎤⎦ + 0.41S (i, j ) + 0.71P ,

(4)

where Ms(k) is the surface magnitude of the kth earthquake, R(i, j, k) is the distance of grid point (i, j) to the epicenter of the kth earthquake, S(i, j) is either zero or one according to whether grid point (i, j) is considered alluvium or rock, and P is the percentile parameter, as described earlier for the attenuation relation given in Equation 1.

153

RESULTS AND DISCUSSION The accumulated PGA values or the cumulative destruction intensities for the last hundred years for the vicinity of Zakynthos are shown in Figure 8. The higher are the cumulative PGA values, the larger is the expected accumulated damage on archaeological remains. However, at this stage of this test study, a comparison of the cumulative destruction intensities with the accumulated damage on the archaeological remains will be possible only relatively. The central part of the island, where recent alluvial sediments are deposited, accommodates the greatest cumulative destruction intensities in the resulting map. Today, this extensive plain is very poor in terms of availability of churches and monasteries. However, since the archaeological surveys from this part are not complete yet, it is difficult to attribute the absence of ancient architectural remains in this area to seismic activities. On the other hand, in the book The Earthquakes of Greece, by Papazachos and Papazachou (1997), one of the most comprehensive accounts concerning strong earthquakes in Greece, the town of Zakynthos and the hill behind it accommodating the Venetian

Figure 8. The cumulative peak ground acceleration (PGA) values in m s–2 for the period of 1901–2006 and Mw ≥ 4.5.

154

Tendürüs et al.

castle are listed most frequently among the heavily damaged areas on the island. After the last catastrophic earthquake, the town was almost entirely rebuilt. It is also known that the castle has undergone many repairs since the Venetian period (Mylona, 2006). Accordingly, Figure 8 indicates that these two sites are subjected to high levels of cumulative destruction due to earthquakes. The cumulative PGA values of the southeastern part of the island (the Vasilikos Peninsula) are found to be comparable to those of the area of the hill behind the town of Zakynthos. The rare preservation of the settlement remains and the relatively less concentrated distribution of the small finds on the peninsula might be explained with these high destruction intensity values in the vicinity. The intense fragmentation of the constructions may result in significant displacement of the pieces by the intervention of humans and nature. Additionally, the noted damages on the church of St. Andreas on the north and the monastery of Skopiotissa on Mount Skopos (listed as number 1 and 16 in Fig. 4) and the archaeological survey results on the west show a good match with the destruction intensities of the resulting map. The church of St. Andreas takes our extra attention because of its contradictory condition with the base rock type underneath. It also agreeably follows the contour lines for the cumulative PGA values on our cumulative destruction intensity map and indicates the value of studying the probable cumulative effect of seismic activities on archaeological remains. A similar situation can be considered for the area south of the Vrachionas mountain range as well. While the archaeological sites have been assigned according to the amount and concentration of the small finds in the vicinity of Keri, the allocated archaeological sites in the central part of the Vrachionas consist of all types of find collections following suitably the picture on the cumulative PGA map. Our cumulated ground acceleration map, once more, points out the significance of the local geological conditions in the ground motion studies and illustrates the vast divergence observed between the destruction levels of alluvium and rock sites. Unfortunately, attenuation models generally use simple binary categories to describe the ground composition, in which local site conditions are classified simply as soil or rock. Campbell and Bozorgnia (2003) showed the importance of a refined geological classification, including soft soil, firm soil, soft or primarily sedimentary rock, and hard rock, in the prediction of ground acceleration. Zakynthos is formed of various geological units such as limestone (also in various ages and characteristics), gypsum, sandstone, and mudstone. The Vasilikos Peninsula mainly consists of Pliocene and Miocene sandstone, in which the shear waves travel slower than in the massive limestone outcrop on the western part of the island, causing more shaking of the ground. Some examples of shear wave velocity for different materials are given in Table 2. Results from a new attenuation relation specific to Zakynthos will certainly be more detailed and more representative. Another important point to mention is that the attenuation model includes only the direct damage caused by ground motion created by earthquakes. However, surface faulting and auxiliary

TABLE 2. P AND S WAVE VELOCITIES OF SOME SELECTED MATERIALS Material

P wave velocity (m/s)

S wave velocity (m/s)

Steel

6100

3500

Concrete

3600

2000

5500–5900

2800–3000

Granite

6400

3200

Sandstone

Basalt

1400–4300

700–2800

Limestone

5900–6100

2800–3000

Sand (unsaturated)

200–1000

80–400

Sand (saturated)

800–2200

320–880

effects such as liquefaction, landslides, rockfalls, tsunamis, fire, looting, etc., can also have considerable roles in the destruction of sites. These types of damages may also be introduced into a cumulative destruction map or be studied separately. Such studies would confine the damage to a local level. CONCLUSIONS This paper aimed to draw attention to the possible correlation between local seismic activity and the different distribution and preservation conditions of archaeological remains. It has shown that the spatial distribution of total destruction intensities of past earthquakes that affected archaeological sites can provide additional insight into the present distribution of remains. However, we recognize that, being a test study, this research will lead to many further questions and discussions on the topic. First, the qualification of the degree of weathering and of the dispersion of archaeological remains has not been done systematically yet. In the near future, we plan to develop methodologies to investigate these issues more systematically. Second, the distribution of archaeological remains in the landscape is subject to many different factors. The influence of seismic activity with regard to the distribution of archaeological material and its relationship to factors such as erosion and sedimentation, and agricultural and cultural factors, still need to be assessed. On the other hand, our study shows that relationships between seismic activities and characteristics of the archaeological record are likely. This result merits similar investigations at other archaeologically rich and seismically active areas in order to validate whether the cumulative destruction intensity maps also show spatial correlation between the ongoing seismic activity and the preservation conditions of the remains. It is hoped that this research may stimulate new studies concerning the interaction of long-term seismic activities and archaeological remains, especially in view of archaeological site preservation. This interaction depends on many factors that can be studied further by employing in situ or experimental methods. For instance, standing or collapsed remains, buried or exposed remains, stone or mud brick remains will respond differently to earthquakes. Results from such studies can potentially be

A test study from Zakynthos, Greece important as an additional decision factor in setting up excavations or making preservation plans for a site. ACKNOWLEDGMENTS We are grateful to the Greek Institute for Geological and Mineralogical Exploration (I.G.M.E.) and the 35th Ephorate of Prehistoric and Classical Antiquities for the possibility to do fieldwork on the island of Zakynthos. We also would like to thank Atılım Güneş Baydin for helping with the implementation of our model and performing the calculations in Mathematica, and Ioannis Koukouvelas for his constructive comments that significantly improved the manuscript. This chapter is a contribution to the International Geoscience Programme (IGCP) 567, “Earthquake Archaeology: Archaeoseismology along the Alpine-Himalayan Seismic Zone,” funded by the United Nations Educational, Scientific and Cultural Organization (UNESCO). REFERENCES CITED Agalopoulou, P.I., 1973, Mycenaean graves at Kambi, Zakynthos: Archaeologikon Deltion 28A, p. 198–214 (in Greek). Ambraseys, N.N., 1996, Material for the investigation of the seismicity of central Greece, in Stiros, S., and Jones, R.E., ed., Archaeoseismology: British School at Athens, Fitch Laboratory Occasional Paper 7, p. 23–36. Barka, A., and Reilinger, R., 1997, Active tectonics of the Eastern Mediterranean region: Deduced from GPS, neotectonic and seismicity data: Annali di Geofisica, v. XL, no. 3, p. 587–610. Bozorgnia, Y., and Campbell, K.W., 2004, Engineering characterization of ground motion, in Bozorgnia, Y., and Bertero, V.V., ed., Earthquake Engineering: From Engineering Seismology to Performance-Based Engineering: Boca Raton, Florida, CRC Press, Chap. 5, p. 1–74. Burton, P.W., Xu, Y., Tselentis, G.A., Sokos, E., and Aspinall, W., 2003, Strong ground acceleration seismic hazard in Greece and neighboring regions: Soil Dynamics and Earthquake Engineering, v. 23, p. 159–181, doi: 10 .1016/S0267-7261(02)00155-0. Campbell, K.W., and Bozorgnia, Y., 2003, Updated near-source ground motion (attenuation) relations for the horizontal and vertical components of peak ground acceleration and acceleration response spectra: Bulletin of the Seismological Society of America, v. 93, no. 1, p. 314–331, doi: 10.1785/0120020029. Caputo, R., Helly, B., Pavlides, S., and Papadopoulos, G., 2006, Archaeoand palaeoseismological investigations in northern Thessaly (Greece): Insights for the seismic potential of the region: Natural Hazards, v. 39, p. 195–212, doi: 10.1007/s11069-006-0023-9. Cerone, M., Croci, G., and Viskovic, A., 2001, The structural behaviour of the Colosseum, International UNESCO-ICOMOS Congress “More than two thousand years in the history of architecture,” Bethlehem, Palestine, 2001: retrieved in 2010 from the UNESCO 2000 archives: http://www.unesco .org/archi2000/bio/crocicoloss.htm. Clément, C., Hirn, A., Charvis, P., Sachpazi, M., and Marnelis, F., 2000, Seismic structure and the active Hellenic subduction in the Ionian islands: Tectonophysics, v. 329, p. 141–156, doi: 10.1016/S0040-1951(00)00193-1. Day, R.W., 2002, Geotechnical Earthquake Engineering Handbook: New York, McGraw-Hill, 700 p. Ellenblum, R., Marco, S., Agnon, A., Rockwell, T.K., and Boas, A., 1998, Crusader castle torn apart by earthquake at dawn, 20 May 1202: Geology, v. 26, p. 303–306, doi: 10.1130/0091-7613(1998)0262.3.CO;2. Facaros, D., and Theodorou, L., 2003, Cadogan Guides: Greece: London, Globe Pequot Press, 880 p. Foss, A., 1969, The Ionian Islands: Zakynthos to Corfu: London, Faber, 272 p. Galadini, F., and Galli, P., 2001, Archaeoseismology in Italy: Case studies and implications on long-term seismicity: Journal of Earthquake Engineering, v. 5, no. 1, p. 35–68, doi: 10.1142/S1363246901000236.

155

Hatzidimitriou, P.M., Papadimitriou, E.E., Mountrakis, D.M., and Papazachos, B.C., 1985, The seismic parameter b of the frequency-magnitude relation and its association with the geological zones in the area of Greece: Tectonophysics, v. 120, p. 141–151, doi: 10.1016/0040-1951(85)90092-7. Hinsbergen, D.J., Meer, D.G., Zachariasse, W.J., and Meulenkamp, J.E., 2006, Deformation of western Greece during Neogene clockwise rotation: International Journal of Earth Sciences, v. 95, no. 3, p. 463–490, doi: 10 .1007/s00531-005-0047-5. Homeros, 1998, Odysseia (translated by A. Erhat and A. Kadir): İstanbul, Can Yayınları, 394 p. Homeros, 1999, İlyada (translated by A. Erhat and A. Kadir): İstanbul, Can Yayınları, 592 p. Horn Lopes, J., 2008, Sun over Zakynthos: Lighting Conditions Investigated [B.Sc. thesis]: Institute for Geo- and Bioarchaeology, Faculty of Earth and Life Sciences, VU University Amsterdam, 36 p. Kalligas, P.G., 1993, Habitation on Ancient Zakynthos, in Settlements of Zakynthos from ancient times to the present day: Athens, Institute for Zakynthian Studies, p. 45–73 (in Greek). Koukouvelas, I.K., Katsonopoulou, D., Soter, S., and Xypolias, P., 2005, Slip rates on the Helike fault, Gulf of Corinth, Greece: New evidence from geoarchaeology: Terra Nova, v. 17, p. 158–164, doi: 10.1111/j.1365 -3121.2005.00603.x. Kouskouna, V., and Makropoulos, K., 2004, Historical earthquake investigations in Greece: Annals of Geophysics, v. 47, p. 723–731. Lagios, E., Sakkas, V., Papadimitriou, P., Parcharidis, I., Damiata, B.N., Chousianitis, K., and Vassilopoulou, S., 2007, Crustal deformation in the Central Ionian Islands (Greece): Results from DGPS and DInSAR analyses (1995–2006): Tectonophysics, v. 444, p. 119–145, doi: 10.1016/j.tecto .2007.08.018. Marco, S., 2008, Recognition of earthquake-related damage in archaeological sites: Examples from the Dead Sea fault zone: Tectonophysics, v. 453, p. 148–156, doi: 10.1016/j.tecto.2007.04.011. Marco, S., Hartal, M., Hazan, N., Lev, L., and Stein, M., 2003, Archaeology, history, and geology of the A.D. 749 earthquake, Dead Sea transform: Geology, v. 31, no. 8, p. 665–668, doi: 10.1130/G19516.1. Mylona, Z.A., 2006, The Castle of Zakynthos (translated by A. Doumas): Athens, Hellenic Ministry of Culture Archaeological Receipts Fund, 56 p. Palaima, T.G., 1991, Maritime matters in the Linear B, in Laffineur, R., and Basch, L., eds., Thalassa: L’Egée Préhistorique et la Mer: Actes de la 3e Rencontre égéenne internationale de l’Université de Liège, Station de recherches sous-marines et océanographiques, Calvi, Corse, 23–25 avril 1990: Annales d’Archéologie égéenne de l’Université de Liège (Aegaeum) 7, p. 273–309. Papastamatiou, D., and Psycharis, I., 1993, Seismic response of classical monuments—A numerical perspective developed at the Temple of Apollo in Bassae, Greece: Terra Nova, v. 5, no. 6, p. 591–601, doi: 10.1111/j.1365 -3121.1993.tb00309.x. Papazachos, B.C., 1990, Seismicity of the Aegean and surrounding area: Tectonophysics, v. 178, p. 287–308, doi: 10.1016/0040-1951(90)90155-2. Papazachos, B.C., and Papazachou, C., 1997, The Earthquakes of Greece: Thessaloniki, P. Ziti and Co., 304 p. Papazachos, B.C., Papaioannou, C.A., Margaris, B.N., and Theodulidis, N.P., 1993, Regionalization of seismic hazard in Greece based on seismic sources: Natural Hazards, v. 8, p. 1–18, doi: 10.1007/BF00596232. Papazachos, B.C., Kiratzi, A.A., and Karacostas, B.G., 1997, Toward a homogeneous moment-magnitude determination for earthquakes in Greece and the surrounding area: Bulletin of the Seismological Society of America, v. 87, no. 2, p. 474–483. Papazachos, B.C., Comninakis, P.E., Karakaisis, G.F., Karakostas, B.G., Papaioannou, C.A., Papazachos, C.B., and Scordilis, E.M., 2000, A Catalogue of Earthquakes in Greece and Surrounding Area for the Period 550 BC–1999: Thessaloniki, Geophysical Laboratory, University of Thessaloniki, 333 p. Papazachos, B.C., Comninakis, P.E., Scordilis, E.M., Karakaisis, G.F., and Papazachos, C.B., 2007, A Catalogue of Earthquakes in the Mediterranean and Surrounding Area for the Period 1901–2006: Department of Geophysics, University of Thessaloniki; retrieved in 2007 from http:// lemnos.geo.auth.gr/the_seisnet/WEBSITE_2005/station_index_en.html. Papazachos, C.B., and Kiratzi, A.A., 1996, A detailed study of the active crustal deformation in the Aegean: Tectonophysics, v. 253, p. 129–153, doi: 10.1016/0040-1951(95)00047-X.

156

Tendürüs et al.

Perry, L.J., and Temple, P.G., 1980, The Geological Map of Greece: Zakinthos Island: Athens, Section of Geology and Geological Mapping, Institute of Geological and Mineral Exploration (I.G.M.E.), scale 1:50,000, 1 sheet. Pieters, N., Stoker, A., Tendurus, M., and Wijngaarden, G.J., 2007, Archeologie in een extreem dynamisch landscap: Het Zakynthos Archeologie project: Tijdschrift voor Mediterrane Archeologie, v. 19, no. 1, p. 26–32. Pliny the Elder, 1855, The Natural History (translated by J. Bostock and H.T. Riley): London, Taylor and Francis, available online at the Perseus Digital Library (http://www.perseus.tufts.edu/hopper/text?doc=Perseus :text:1999.02.0137). Psycharis, I.N., Papastamatiou, D.Y., and Alexandris, A.P., 2000, Parametric investigation of the stability of classical columns under harmonic and earthquake excitations: Earthquake Engineering & Structural Dynamics, v. 29, p. 1093–1109, doi: 10.1002/1096-9845(200008)29:83.0.CO;2-S. Reinhardt, E.G., Goodman, B.N., Boyce, J.I., Lopez, G., van Hengstum, P., Rink, W.J., Mart, Y., and Raban, A., 2006, The tsunami of 13 December A.D. 115 and the destruction of Herod the Great’s harbor at Caesarea Maritime, Israel: Geology, v. 34, no. 12, p. 1061–1064, doi: 10.1130/ G22780A.1. Rink, R., 2005, Geographical Development of Zakynthos during Late Pleistocene–Holocene, Greece [M.Sc. thesis]: Utrecht, Netherlands, Department of Physical Geology, Utrecht University, 34 p. Similox-Tohon, D., Sintubin, M., Muchez, P., Verhaert, G., Vanneste, K., Fernandez, M., Vandycke, S., Vanhaverbeke, H., and Waelkens, M., 2006, The identification of an active fault by a multidisciplinary study at the archaeological site of Sagalassos (SW Turkey): Tectonophysics, v. 420, p. 371–387, doi: 10.1016/j.tecto.2006.03.026. Souyoudzoglou-Haywood, C., 1999, The Ionian Islands in the Bronze Age and Early Iron Age 3000–800 BC: Liverpool, Liverpool University Press, 280 p. Stiros, S., Pirazzoli, P., Laborel, J., and Laborel-Deguen, F., 1994, The 1953 earthquake in Cephalonia (western Hellenic arc): Coastal uplift and halotectonic faulting: Geophysical Journal International, v. 117, p. 834–849, doi: 10.1111/j.1365-246X.1994.tb02474.x. Stiros, S., Pirazzoli, P., Rothaus, R., Papageorgiou, S., Laborel, J., and Arnold, M., 1996, On the date of construction of Lechaion, western harbor of ancient Corinth, Greece: Geoarchaeology, v. 11, no. 3, p. 251–263, doi: 10.1002/(SICI)1520-6548(199605)11:33.0.CO;2-2. Stoker, A., 2006, Remote Sensing in the Zakynthos Archaeology Project. Investigation of the Benefits and Restrictions of Historic Airphotos and Satellite Images in Landscape-Archaeological Research in Greece [M.A. thesis]: Groningen, the Netherlands, Faculty of Arts, University of Groningen, 85 p.

Storme, A.L., 2008, The Landscape at Zakynthos: Detailed Geomorphological Description of the South-Eastern Part of the Vasilikos Peninsula [M.Sc. thesis]: Institute for Geo- and Bioarchaeology, Faculty of Earth and Life Sciences, VU University Amsterdam, 246 p. Tendürüs, M., 2009, Landscape Evolution of the Southern Coastal Plain of Zakynthos, Greece, since the Middle Holocene [M.Sc. thesis]: Institute for Geo- and Bioarchaeology, Faculty of Earth and Life Sciences, VU University Amsterdam, 122 p. Theodulidis, N.P., and Papazachos, B.C., 1992, Dependence of strong ground motion on magnitude-distance, site geology and macroseismic intensity for shallow earthquakes in Greece: I. Peak horizontal acceleration, velocity and displacement: Soil Dynamics and Earthquake Engineering, v. 11, p. 387–402, doi: 10.1016/0267-7261(92)90003-V. Underhill, J.R., 1989, Late Cenozoic deformation of the Hellenide foreland, western Greece: Geological Society of America Bulletin, v. 101, p. 613– 634, doi: 10.1130/0016-7606(1989)1012.3.CO;2. U.S. Geological Survey, 2010, Magnitude / Intensity Comparison in the Severity of an Earthquake: United States Geological Survey’s (USGS) General Interest Publication 1989-288-913, retrieved in 2010 from the USGS Earthquake Hazard Program: http://earthquake.usgs.gov/learn/topics/mag _vs_int.php. Van Wijngaarden, G.J., Arapogianni, X., Rink, R., and Tourloukis, V., 2006, The Zakynthos Survey 2005. Preliminary report of a pilot study, Pharos: Journal of the Netherlands Institute in Athens, v. XIII, p. 59–76. Van Wijngaarden, G.J., Sotiriou, A., Pieters, N., and Tourloukis, V., 2007, The Zakynthos Archaeology Project 2006. Preliminary report of the first season, Pharos: Journal of the Netherlands Institute in Athens, v. XIV, p. 29–46. Van Wijngaarden, G.J., Sotiriou, A., Horn Lopes, J., Gouma, M., Koster, K., Stoker, A., Susan, D., and Tourloukis, V., 2010, The Zakynthos Archaeology Project, 2008. Preliminary report on the 2008 season, Pharos: Journal of the Netherlands Institute in Athens, v. XVI, 61–83. Van Wijngaarden, G., Sotiriou, A., Pieters, N., Abed, K., and Tendürüs, M., 2009, The Zakynthos Archaeology Project. Preliminary report of the 2007 season, Pharos: Journal of the Netherlands Institute in Athens, v. XV, p. 43–57. Von Stein, I., 2009, Just in Time… Assessing Cape Kalogeros before It Is Gone [Master’s thesis]: Amsterdam, Amsterdam Archaeological Centre, University of Amsterdam, 61 p. Yeats, R.S., Sieh, K.E., and Allen, C.R., 1997, The Geology of Earthquakes: New York, Oxford University Press, 568 p. Zois, L., 1955, History of Zakythos: Athens, 439 p. (in Greek). MANUSCRIPT ACCEPTED BY THE SOCIETY 19 MAY 2010

Printed in the USA

The Geological Society of America Special Paper 471 2010

Assessment of seismically induced damage using LIDAR: The ancient city of Pınara (SW Turkey) as a case study Barış Yerli Johan ten Veen* Institute for Geology, Mineralogy, and Geophysics, Ruhr University, Universitätsstrasse 150, D-44801, Bochum, Germany Manuel Sintubin Geodynamics & Geofluids Research Group, Katholieke Universiteit Leuven, Celestijnenlaan 200E, 3001 Leuven, Belgium Volkan Karabacak C. Çağlar Yalçıner Erhan Altunel Engineering Faculty, Department of Geological Engineering, Osman Gazi University, Eskisehir, Turkey

ABSTRACT Seismic-related damages of archaeological structures play an important role in increasing our knowledge about the timing and magnitudes of historical earthquakes. Although quantitative data should form the basis of objective archaeoseismological methods, most studies still do not rely on such methods. Ground-based LIDAR (light detection and ranging) is a promising, rather new, scanning technology that determines spatial position of an object or surface and provides high-resolution threedimensional (3-D) digital data. Using LIDAR, we mapped the damage and overall attitude of a Roman theater in the ancient Lycian city of Pınara (500 B.C.–A.D. 900), located at a faulted margin of the Eşen Basin (SW Turkey). An average 0.81°°NW tilt of the 20 seating rows could be computed from the LIDAR data. Conventional compass readings of these seating rows did not provide the same results because errors involved with this method are generally >2°°. The tilt direction appears perpendicular to the NE-trending basin-margin fault, suggesting that fault-block rotation is the most likely mechanism to have induced the systematic tilt of the theater. The estimated 4 m offset on this normal fault should be seen as a rough estimate of the total displacement and was likely produced by several (more than one) earthquakes with magnitudes of M = 6–6.8. This is consistent with historical records that mention several large earthquakes during the Roman period.

*Current address: TNO Built Environment and Geosciences, P.O. Box 80015, Utrecht 3508 TA, Netherlands. Yerli, B., ten Veen, J., Sintubin, M., Karabacak, V., Yalçıner, C.Ç., and Altunel, E., 2010, Assessment of seismically induced damage using LIDAR: The ancient city of Pınara (SW Turkey) as a case study, in Sintubin, M., Stewart, I.S., Niemi, T.M., and Altunel, E., eds., Ancient Earthquakes: Geological Society of America Special Paper 471, p. 157–170, doi: 10.1130/2010.2471(14). For permission to copy, contact [email protected]. © 2010 The Geological Society of America. All rights reserved.

157

158

Yerli et al.

INTRODUCTION Archaeoseismological research commonly focuses on establishing a link between damages to archaeological building structures (e.g., tilted, toppled blocks, disorientation of structures, fallen columns, etc.) and historical or younger earthquakes. The main challenge lies in separating seismic-related damage from other, natural or human, causes for destruction. Archaeoseismology is faced with the difficulty that constructions are either too intensely destroyed, or that damage is too indistinct to give any detailed information on its nature. Because of poor spatial and temporal resolution, amalgamation or duplication of seismic events is a well-known pitfall (Ambraseys et al., 2004; Guidoboni, 2002). Therefore, the most reliable results of an archaeoseismological investigation are obtained by application of modern geoarchaeological practices (archaeological stratigraphy plus geological-geomorphological data), augmenting a quantitative approach and (if available) historical information (Galadini et al., 2006). Data obtained from archaeoseismological investigation can be either qualitative or quantitative. By qualifying damage to archaeological structures, an earthquake magnitude can be estimated based on the Medvedev-Sponheuer-Karnik (MSK) scale (Medvedev et al., 1964) or the modified Mercalli (MM) scale (Wood and Neumann, 1931). However, the assessment of earthquake magnitude is imprecise and often based on a presumed analogue with earthquake effects on modern man-made structures (e.g., Karcz and Kafri, 1978; Altunel, 1998). Recent engineering seismological models can be used to test the hypothesis that observed building damage is of seismogenic nature by seeking a systematic relation between building response and the seismic source or ground motion (e.g., Hinzen, 2005), and to test the archaeoseismic hypothesis (Galadini et al., 2006). Quantitative data can be grouped as either directional or spatial and can be applied for the purposes of geophysical engineering or kinematic analysis. Directional data include the sense of slip on faulted archaeological relics and can be used (in a similar fashion as in paleoseismology) as supplementary information to discover previously unknown earthquakes (Galli and Galadini, 2001). On the basis of the amount of offset and age of displaced structures, a realistic value of earthquake magnitude and a rough evaluation of the recurrence time can be obtained. In addition, information on faulted relics (fault direction and offset direction) can be treated as fault-kinematic data and as such be used for strain analysis (e.g., Hancock and Altunel, 1997). The parallel direction of fallen columns is often used as an indicator for earthquake damage (Stiros, 1996). Nur and Ron (1996) suggested a relationship between ground motion and fall direction, where the latter is an indicator of the direction of fault-rupture propagation. However, the behavior of columns during shaking is very complex and depends on different physical parameters, such as ground motion, material characteristics, and type of building foundation (Ambraseys, 2006). Numerical modeling studies show that the fall direction of single standing columns

is highly chaotic and can be influenced by small anisotropies in constructional elements or variations of the ground motion (Hinzen, 2009). Up until recently, spatial data of archaeological sites were mostly unavailable, but they are now easier to acquire as a result of the arrival of high-resolution laser detection techniques. Spatial data provide information on the position of archaeological relics such as walls and floors. Such positional data alone is informative, but data treatment can reveal information on tilt, torque, and dislocation of building elements. This type of information is elementary in establishing links between damage and faulting and may find its way in future finite-element simulation and reconstructions as well. Here, we test data obtained by LIDAR (light detection and ranging system) against conventionally obtained data in order to test the hypothesis of faulting-induced tilting of a Roman theater at the archaeological site of Pınara. TECTONIC SETTING Neotectonics of SW Turkey Turkey is characterized by complex neotectonic deformation that is predominantly related to the convergence of Africa and Eurasia. The North and East Anatolian fault systems are the most prominent structures that accommodate the collision of the African plate’s Arabian promontory with Eurasia (e.g., Barka and Kadinsky-Cade, 1988; Westaway, 1994; Armijo et al., 1999; ten Veen et al., 2009). Southwestern Turkey is transected by numerous faults that are thought to connect southward with major fault zones associated with the Hellenic subduction zone (Eyidoğan and Barka, 1996; ten Veen and Kleinspehn, 2002, 2003; ten Veen et al., 2004). The Fethiye-Burdur fault zone (Barka and Reilinger, 1997) is a domain characterized by many subparallel fault segments, roughly aligned in the area between Fethiye and Burdur. Different styles of deformation and a wide variety of fault orientations suggest that this zone has been affected by three different tectonic phases from late Miocene until present (ten Veen, 2004; Alçiçek, 2007; ten Veen et al., 2009). Recent Earthquake Activity in SW Turkey The Fethiye-Burdur fault zone has been attributed one of the highest seismic hazard designations in Turkey, based on the concentration of recent and historical seismicity in the area (see Fig. 1B), comparable to regions situated along the North Anatolian fault zone. Instrumental data show that most earthquakes in southwestern Turkey have a shallow focal depth (