Wiseman&Zachos (Eds) - Landscape Archaeology in Southern Epirus, Greece

LANDSCAPE ARCHAEOLOGY IN SOUTHERN EPIRUS, GREECE I

HESPERIA SUPPLEMENTS

1* S. Dow, Prytaneis: A Study of the Inscriptions Honoring the Athenian Councillors (1937)

2* R. S. Young, Late Geometric Graves and a Seventh-Century Well in theAgora (1939) 3* G. P. Stevens, The Setting of the Periclean Parthenon (1940) 4* H. A. Thompson, The Tholos of Athens and Its Predecessors (1940) 5* W. B. Dinsmoor, Observations on the Hephaisteion (1941) 6* J. H. Oliver, The Sacred Gerusia (1941) 7* G. R. Davidson and D. B. Thompson, Small Objectsfrom the Pnyx: I (1943) 8* Commemorative Studies in Honor of Theodore Leslie Shear (1949) 9* J. V. A. Fine, Horoi: Studies in Mortgage, Real Security, and Land Tenure in Ancient

Athens (1951) 10* L. Talcott, B. Philippaki, G. R. Edwards, and V. R. Grace, Small Objectsfrom the

Pnyx: II (1956) 11* J. R. McCredie, FortifedMilitary Camps inAttica (1966) 12* D. J. Geagan, The Athenian Constitution after Sulla (1967) 13 J. H. Oliver, MarcusAurelius:Aspects of Civic and Cultural Policy in the East (1970) 14 J. S. Traill, The Political Organization of Attica (1975) 15 S. V. Tracy, The Lettering of an Athenian Mason (1975) 16 M. K. Langdon, A Sanctuary of Zeus on Mount Hymettos (1976) 17 T. L. Shear Jr., Kallias of Sphettos and the Revolt of Athens in 268 B.C. (1978) 18* L. V. Watrous, Lasithi:A History of Settlement on a Highland Plain in Crete (1982) 19 Studies in Attic Epigraphy, History, and Topography Presented to Eugene Vanderpool

(1982) 20 Studies in Athenian Architecture, Sculpture, and Topography Presented to Homer

A. Thompson (1982) 21 J. E. Coleman, Excavations at Pylos in Elis (1986) 22 E. J. Walters, Attic Grave Reliefs That Represent Women in the Dress oflsis (1988) 23 C. Grandjouan, Hellenistic Relief Moldsfrom the Athenian Agora (1989) 24 J. S. Soles, The Prepalatial Cemeteries at Mochlos and Gournia and the House Tombs

of BronzeAge Crete (1992) 25 S. I. Rotroff and J. H. Oakley, Debris from a Public Dining Place in the Athenian

Agora (1992) 26 I. S. Mark, The Sanctuary of Athena Nike in Athens:Architectural Stages and Chro-

nology (1993) 27 N. A. Winter, ed., Proceedings of the International Conference on GreekArchitectural

Terracottas of the Classical and Hellenistic Periods, December 12-15, 1991 (1994) 28 D. A. Amyx and P. Lawrence, Studies in Archaic Corinthian Vase Painting (1996) 29 R. S. Stroud, TheAthenian Grain-Tax Law of 374/3 B.C. (1998) 30 J. W. Shaw, A. Van de Moortel, P. M. Day, and V. Kilikoglou, A LMIA Ceramic

Kiln in South-Central Crete. Function and Pottery Production (2001) 31 J. Papadopoulos, Ceramicus Redivivus: The Early Iron Age Potters' Field in theArea

of the ClassicalAthenian Agora (2003) * Out ofprint

Hesperia Supplement 32

LANDSCAPE ARCHAEOLOGY IN SOUTHERN EPIRUS, GREECE I

EDITED BY

JAMES WISEMAN AND KONSTANTINOS ZACHOS

The American School of Classical Studies at Athens 2003

Copyright ? 2003 The American School of Classical Studies at Athens

All rights reserved.

Out-of-print Hesperia supplements may be purchased from Swets & Zeitlinger Backsets Department P.O. Box 810 2160 SZ Lisse The Netherlands

E-mail: [email protected]

Cover illustration: The eroded landscape of Kokkinopilos above the Louros River gorge

Library of Congress Cataloging-in-Publication Data

Landscape archaeology in southern Epirus, Greece I / edited by James Wiseman and Konstantinos Zachos.

p. cm.-(Hesperia Supplement; 32) Includes bibliographical references (p.). ISBN 0-87661-532-9 (alk. paper) 1. Preveza (Greece)-Antiquities. 2. Excavations (Archaeology)-Greece-

Preveza. 3. Landscape archaeology-Greece-Preveza. 4. Arta (Greece: Nome)- Antiquities. 5. Excavations (Archaeology)-Greece-Arta (Nome) 6. Landscape archaeology-Greece-Arta (Nome) I. Wiseman, James. II. Zachos, Konstantinos L. III. Hesperia (Princeton, NJ.). Supplement; 32.

DF9oI.P72L36 2003 938'.2-dc2I 2002044060

CONTENTS

List of Illustrations vii List of Tables xii Preface and Acknowledgments xv

Chapter 1 THE NIKOPOLIS PROJECT: CONCEPT, AIMS, AND

ORGANIZATION

by James Wiseman and Konstantinos Zachos 1

Chapter2 THE ARCHAEOLOGICAL SURVEY: SAMPLING

STRATEGIES AND FIELD METHODS

by Thomas F. Tartaron 23

Chapter 3 THE EARLY STONE AGE OF THE NOMOS OF

PREVEZA: LANDSCAPE AND SETTLEMENT

by Curtis N. Runnels and Tjeerd H. van Andel 47

Chapter 4 EARLY UPPER PALAEOLITHIC SPILAION: AN

ARTIFACT-RICH SURFACE SITE

by Curtis N. Runnels, Evangelia Karimali, and Brenda Cullen 135

Chapter 5 THE COASTAL EVOLUTION OF THE AMBRACIAN

EMBAYMENT AND ITS RELATIONSHIP TO

ARCHAEOLOGICAL SETTINGS

by Zhichun Jing and George (Rip) Rapp 157

VI CONTENTS

Chapter 6 THE LOWER ACHERON RIVER VALLEY: ANCIENT ACCOUNTS AND THE CHANGING LANDSCAPE

by Mark R. Besonen, George (Rip) Rapp, and Zhichun Jing 199

Chapter 7 SUMMARY OBSERVATIONS

by James Wiseman and Konstantinos Zachos 265

References 269 Index 283

ILLUSTRATIONS

Illustrations are by members of the project except where noted.

1.1. Map of Epirus and adjacent regions 2

1.2. Map of survey zone with selected toponyms 3

1.3. Multispectral image (SPOT) of the northern part of the survey zone 14

1.4. Multispectral image (SPOT) of the southern part of the survey zone 14

1.5. The eroded landscape of Kokkinopilos 16

1.6. Aerial view of the fortified town site at Kastro Rogon 18

1.7. Aerial view of the water channel and aqueduct bridges across the Louros River 19

2.1. Map of southwestern Epirus 29

2.2. Archaeological survey tract form 36

2.3. Examples of spatial relationships between tracts and site/scatters 41

2.4. General view of the site at Grammeno (SS92-6) 44

3.1. Map of Epirus and surrounding areas 49

3.2. Tectonics of northwestern Greece and the Ionian Sea 55

3.3. Possibly active (Late Quaternary) tectonic features of western Epirus 56

3.4. Present tectonic activity in western Epirus as indicated by fresh striae on fault planes 56

3.5. Simplified bedrock map of western Epirus 57

3.6. Formation of a doline (sinkhole) 58

3.7. Diagram of the genesis of loutses and poljes on a karstic peneplain 59

3.8. Poljes and loutses in western Epirus 60

ILLUSTRATIONS

3.9. View ofValtos Kalodiki 63

3.10. The eponymous loutsa on the raised peneplain south of the lower Acheron valley 63

3.11. The polje of Cheimadio 63

3.12. Red sediments and paleosols 64-65

3.13. Terra rossa redeposited in fan complex 66

3.14. Typical grain-size frequency diagrams of terra rossa redeposited in poljes and loutses 67

3.15. The raised polje of Kokkinopilos 71

3.16. Badland erosion at Kokkinopilos 72

3.17. Cross section through the incised polje deposits of Kokkinopilos 73

3.18. Morphi polje outcrop with paleosols forming hard, protruding benches 74

3.19. Composite profile of Ayia loutsa 74

3.20. Stratified lower section of the Ayia loutsa looking west; detail of Mousterian artifacts in situ 75

3.21. The Adriatic Sea during the last glacial maximum 76

3.22. Global sea-level variations for the past 140,000 years 77

3.23. The emerged coastal plain off Epirus at six key moments 79

3.24. Two sea-level rise curves for the deglaciation interval of late OIS 2 79

3.25. Locations of raised paleoshore deposits of the last interglacial in coastal Epirus 81

3.26. Cumulative grain-size distributions of coastal sediments of the last interglacial and early Holocene 81

3.27. The raised Tyrrhenian beach at Tsarlambas 82

3.28. Climate and vegetation changes during the last two glacial- interglacial cycles 84

3.29. Maturity stages and approximate ages of the Mediterranean paleosol chronosequence 87

3.30. Relationship between paleosol maturity, terra rossa deposition rate, and Palaeolithic stone tool age in poljes and loutses 94

3.31. Palaeolithic site/scatters in the Thesprotiko valley 99

3.32. Palaeolithic and Mesolithic site/scatters in the Acheron valley 100

3.33. View of a stone cluster at Alonaki 101

VIII

ILLUSTRATIONS

3.34. Early Palaeolithic artifacts from Alonaki 102

3.35. Early Palaeolithic choppers from Alonaki 102

3.36. Early Palaeolithic core-choppers from Alonaki 103

3.37. Early Palaeolithic core from Alonaki 103

3.38. Early Palaeolithic biface (handaxe) from Ormos Odysseos 104

3.39. Interglacial sand dune (SS92-25) at Ormos Odysseos 104

3.40. Ormos Odysseos, biface findspot (W94-20) 104

3.41. Early Palaeolithic biface or bifacial core from Ayios Thomas 105

3.42. The Palaeolithic site of Ayia and its setting 109

3.43. Middle Palaeolithic (Mousterian) artifacts from Ayia 110

3.44. Middle Palaeolithic (Mousterian) artifacts from Ayia 110

3.45. Palaeolithic findspots in the vicinity of Kastrosykia 111

3.46. Anavatis site/scatter 94-13, looking northeast 111

3.47. View of Rodaki (SS92-15) 112

3.48. Middle Palaeolithic artifacts from Rodaki 112

3.49. Early Upper Palaeolithic end scrapers from Spilaion 115

3.50. Late Upper Palaeolithic backed blades 116

3.51. Palaeolithic and Mesolithic site/scatters in the Preveza area 118

3.52. Mesolithic artifacts from Tsouknida and Ammoudia 120

3.53. Mesolithic trapeze from Ammoudia 120

3.54. View of Ammoudia, looking northwest, with stone feature visible at left 121

3.55. Mesolithic artifacts from Loutsa 122

3.56. Typical Preveza Mesolithic findspot (SS94-23), looking southwest 123

3.57. Typical Mesolithic artifact scatter near Preveza (SS94-22) 123

3.58. Mesolithic artifacts from the Preveza area 124

4.1. Map showing the location of Spilaion at the mouth of the Acheron River 136

4.2. Map of Spilaion showing topographic contours 139

4.3. View of Spilaion, looking southwest 139

4.4. View of the rugged karst surface on the southeast slope of Spilaion at the time of collection 140

4.5. Sample grid on the southeast slope of Spilaion during collection 141

IX

ILLUSTRATIONS

4.6. Lithic artifacts from Spilaion 145





4.11. End scrapers from Spilaion 146

4.12. Spatial distribution of lithic debitage and retouched tools at Spilaion 151

4.13. Spatial distribution of individual categories of retouched tools at Spilaion 152

5.1. Geology and geomorphology of the Ambracian embayment and its vicinity 158

5.2. Locations of geologic cores and cross sections 159

5.3. Map of the Nikopolis isthmus showing the location of

geologic cores and cross sections 163

5.4. Map of Ormos Vathy showing the location of geologic cores and cross section 163

5.5. Stratigraphic cross section D-D', parallel to the axis of the

Nikopolis isthmus 165

5.6. Stratigraphic cross section E-E', parallel to the axis of the

Nikopolis isthmus 166

5.7. Stratigraphic cross section A-A', perpendicular to the axis of the Nikopolis isthmus 170

5.8. Stratigraphic cross section B-B', perpendicular to the axis of the Nikopolis isthmus 171

5.9. Stratigraphic cross section C-C', perpendicular to the axis of the Nikopolis isthmus 172

5.10. Paleogeographic reconstruction of the eastern side of the Nikopolis isthmus showing the shorelines at different periods 173

5.11. Stratigraphic cross section along the west arm of Ormos Vathy 175

5.12. Paleogeographic reconstructions of Ormos Vathy indicating shoreline changes from the Neolithic through modern periods 176

5.13. Stratigraphic cross section near the Grammeno plain 178

5.14. Map of Kastro Rogon and vicinity showing the location of geologic cores and cross sections 180

5.15. Stratigraphic cross section C-C' at Kastro Rogon 181

x

ILLUSTRATIONS

5.16. Stratigraphic cross section B-B' near Kastro Rogon 183

5.17. Stratigraphic cross section A-A' near Kastro Rogon 185

5.18. Stratigraphic cross section north of the Ambracian Gulf showing sedimentary sequences and environments across the entire coastal plain-lagoon-barrier system 187

5.19. Paleogeographic reconstructions of Kastro Rogon and vicinity showing the changing coastlines and environments from 7000/6500 B.P. through 1000/500 B.P. 190-191

5.20. Changes in relative sea level as indicated by the radiocarbon-dated peat samples from swamp deposits north of the Ambracian Gulf 193

5.21. Paleogeographic reconstructions of the Ambracian embayment showing the shoreline changes from 7000/6500 B.P. through 1000/500 B.P. 196-197

6.1. Area map of Epirus 200

6.2. Area map of the lower Acheron valley 201

6.3. View of concentric accretionary beach ridges surrounding Phanari Bay 202

6.4. Suggested locations of the Acherousian lake in the lower Acheron valley 203

6.5. Satellite image of Epirus 206

6.6. Simplified geology of the lower Acheron valley 207

6.7. Core locations in the lower Acheron valley 210

6.8. Topographic map of the lower Acheron valley bottom 211

6.9. North-south cross section through the Mesopotamon/ Tsouknida valley constriction 218

6.10. East-west cross section through the Mesopotamon/ Tsouknida valley constriction 219

6.11. Northeast-southwest cross section through the valley bottom (area of former marine embayment) 220

6.12. Paleogeographic reconstructions of the lower Acheron valley for 2100 B.C. and the 8th century B.C. 221

6.13. Paleogeographic reconstructions of the lower Acheron valley for 433 B.C. and 1 B.C. 222

6.14. Paleogeographic reconstructions of the lower Acheron valley forA.D. 1100 andA.D. 1500 223

6.15. Paleogeographic reconstruction of the lower Acheron valley for A.D. 1809 and a map of the modern landscape 224

XI

TABLES

1.1. Project Staff and the Years of Their Participation 10-11

1.2. Field School Students and Their Home Institutions 12

2.1. Stratified Sample and Systematic Survey Coverage, Lower Acheron Valley, 1992-1994 31

2.2. Typical Daily Work Assignment, June 28,1994 33

3.1. Dimensions and Elevations of Poljes and Loutses in Western Epirus 61

3.2. Composition of the Fraction >0.064 mm in Redeposited Terra Rossa 67

3.3. Grain-Size Distribution of Redeposited Terra Rossa 68-69

3.4. Mineral Composition of Redeposited Terra Rossa at

Kokkinopilos 70

3.5. Mineral Composition of Redeposited Terra Rossa from

Poljes and Loutses in Western Epirus 71

3.6. Approximate Paleoshoreline Depths and Coastal Plain Widths, 140 kyr B.P. to Present 78

3.7. Mineral Composition of Modern and Last Interglacial Coastal Sands in Western Epirus 83

3.8. Maturity Indicators of the B Horizon of Greek Quaternary Paleosols 87

3.9. Short Descriptions and Maturity Stages of Paleosol Bt Horizons at Key Sites in Coastal Epirus 88

3.10. Thermoluminescence and Infrared Stimulated Luminescence Sediment Dates for Western Epirus 91

3.11. Chronostratigraphic Diagram for Archaeological Sites, Sediments, and Paleosols in the Preveza Region 92

3.12. Early Stone Age Chronology 98

TABLES XIII

4.1. Categories of Flintknapping Debitage 143

4.2. Types of Retouched Tools 144

4.3. Degree of Association between Pairs of Classes of

Flintknapping Debitage 150

5.1. Radiocarbon Dates from the Ambracian Embayment 168

6.1 Radiocarbon Dates from the Acheron River Valley 210

PREFACE AND

ACKNOWLEDGMENTS

As editors of this volume we wish to thank the Hellenic Ministry of Cul- ture for the approval of the permit to conduct archaeological surface investigations in southern Epirus, and to thank as well the directors of the 12th Ephoreia of Prehistoric and Classical Antiquities and the 8th Ephoreia of Byzantine Antiquities, Angelika Douzougli and Frankiska Keph- allonitou, for their positive recommendation to the Central Archaeologi- cal Council and their cooperation for the entire duration of the project. We also want to thank Evangelos Chrysos, then Professor of Byzantine History of the University of Ioannina (now at the University of Athens), for his many different contributions to the success of the project, and Nikolaos Yiannoulis, Mayor of Preveza during our investigations, who helped us in the resolution of a variety of problems that arose in the course of the project. We acknowledge the significant help in geological matters of Panayiotis Paschos, geologist of the Institute of Geology and Mineral- ogy Exploration (Preveza branch) and an expert in the geomorphological investigations of Epirus. During the fieldwork and the subsequent research in the facilities of the Archaeological Museum and the Byzantine Mu- seum of Ioannina, to which the ancient artifacts collected in the surface survey had been brought, the project enjoyed substantial help from the scientific, technical, and security personnel of both ephoreias, to whom we express our warm thanks. The American School of Classical Studies at Athens approved the proposal for American participation in this coopera- tive project, and staff members of the project annually benefited from the superb library and other facilities of the School in Athens. We are grateful to the School, its staff, and its director during those years, the late W. D. E. Coulson. The former comptroller of the School, Joanna Driva, and the School's Administrator, Maria Pilali, were particularly helpful on numerous occasions, and it is a pleasure to acknowledge their congenial advice and cooperation.

The project was sponsored in the United States by Boston University through its Department of Archaeology, the Center for Archaeological Studies, and the Center for Remote Sensing, all of which provided equipment and facilities to the project, and whose faculty, staff, and students have been supportive in many ways. Boston University also provided fi-

PREFACE AND ACKNOWLEDGMENTS

nancial and logistical support through its Office of International Programs, which sponsored an archaeological field school as part of the project in 1992-1994. The American codirector of the project (JW) was director of the field school, and Thomas F. Tartaron and Carol A. Stein were teaching assistants; all senior staff of the project also provided instruction and guidance to the students, whose field and laboratory studies were fully integrated into the project's activities. All staff and field school students are listed in Tables 1.1 and 1.2. Thomas L. Sever, now of NASAs Global Hydrology and Climate Center in Huntsville, Alabama, and Farouk El- Baz, director of Boston University's Center for Remote Sensing, were both supportive and helpful with advice on remote-sensing aspects of the project.

Funding for the Nikopolis Project was provided by grants from the Earth Observing System, NASA in 1991; the National Geographic Soci- ety, 1992; the Institute for Aegean Prehistory, 1993-1995; and contributions throughout the years of the project by a number of private individuals, the Friends of the Nikopolis Project, who are listed below. Special thanks are due to four of the Friends, Martha Sharpe Joukowsky and Arte- mis A. W.Joukowsky, James H. OttawayJr., and Malcolm Hewitt Wiener, for their support and encouragement from the inception of the project to its conclusion. Equipment for geophysical and topographic survey and for aerial photography was provided through grants by the W. M. Keck Foun- dation to the Center for Remote Sensing. Autodesk Inc. gave the Nikopolis Project copies of its superb drawing program, AutoCAD, Version 12, for each of the three computer platforms used by the project: Macintosh, DOS, and UNIX. Trimble Navigation Company lent the project two Global Po- sitioning Systems for the 1994 season. In 1993, the Apple Computer Cor- poration contributed four computers to the project, two Quadra 950s and two PowerBook 160s, which served many of the computing needs of the project, both in Greece and in Boston. The Archaeometry Laboratory of the University of Minnesota, Duluth, provided substantial aid in personnel time and support for analyses. Finally, we thank Carol A. Stein, a member of the Nikopolis Project staff and Manuscript Editor at the American School of Classical Studies at Athens, for her congenial, thoughtful, and perceptive help in editing this volume and guiding it through the publication process. On behalf of the entire staff of the project, we acknowledge with deep gratitude the help and contributions of all.

James Wiseman Konstantinos Zachos

XVI

PREFACE AND ACKNOWLEDGMENTS

FRIENDS OF THE NIKOPOLIS PROJECT

BENEFACTORS

Lloyd Cotsen and the Neutrogena Corporation Dr. Martha Sharpe Joukowsky and Dr. Artemis A. W. Joukowsky James H. Ottaway Jr. Malcolm Hewitt Wiener

PATRONS

Ms. Betty Banks Elizabeth Buntrock Leon Levy Dr. Anna Marguerite McCann and Mr. Robert Taggart Professor J. P. Sullivant and J. L. Godfrey

SPONSORS

Anonymous Mr. James R. James Jr. Philip J. King Dr. William Ruf and Mrs. Elizabeth Ruf J. Robert Sewell

SUSTAINING MEMBERS

Anonymous Dr. Barbara Bell Doreen C. Spitzer Susan and Stephen Wiseman

CONTRIBUTING MEMBERS Dr. Patricia Anawalt Professor Apostolos Athanassakis Robert S. Carter Professor Marian B. Davist Ernestine S. Elster, Ph.D. Dr. Howard Gotlieb In memory of Stuart Haupt Professor G. L. Huxley Mr. Robert F. Johnston Michaelt and Susan Katzev Norma Kershaw Tom Lucia W. V. MacDonald

Katherine Nordsieck Leonard V. Quigley, Esq. Eleanor Robbins Susan Petschaft Rothstein

Jane Ayer Scott

Jane Dunn Sibley Judith P. Sullivan Professor and Mrs. Homer A.

Thompsontt Dr. George Udvarhelyi Elizabeth Lyding Will Donald and Rae Wiseman

XVII

CHAPTER I

THE NIKOPOLIS PROJECT:

CONCEPT, AIMS, AND

ORGANIZATION

by James Wiseman and Konstantinos Zachos

Human societies at all times and in all parts of the world interact with the

landscape they inhabit: it could not be otherwise, even if the interaction were somehow limited to the selective exploitation of natural resources. Human activities alter the landscape and the natural environment, often in dramatic ways; the alterations may occur as the result of human design, as in clearing a forest to plant crops, or may be incidental, as in the destruction (or reshaping) of a mountainside by Roman miners of precious metals. Conversely, humans at various times in the past have physically adapted to changes in their environment (especially in the distant past), or

responded to environmental change in a variety of other ways. Some of these responses, such as migration or technological innovation, have been drastic and revolutionary in their effect and are often recognizable in the

archaeological record, while other responses were more gradual, even subtle, and are more difficult to detect. To acknowledge the importance of the natural setting, of the environment at large, in studying change in human

society is not to deny the importance of intercultural relationships, or the role of the individual intellect or collective social conscience in the evolution of ethical, spiritual, or other sociocultural phenomena in human affairs. The point is that to understand and explain changes in human society over time, it is critically important to study society in relationship to the changing environment in which it existed. Through this approach to the past archaeologists are able to provide insights into the factors that underlie changes in human-land relationships, sometimes over a short time- span or even regarding specific events, but especially over the long term. And they can explore those intercultural relationships and sociocultural phenomena cited above, which themselves evolve within specific environmental settings and change.

We have sought to apply these concepts in the formulation and conduct of the Nikopolis Project, an undertaking in landscape archaeology focused on the human societies that inhabited southern Epirus in northwestern Greece from earliest times to the medieval period. More specifically, the project has employed intensive archaeological survey and geological investigations to determine patterns of human-activity areas, and


Figure 1.1. Map of Epirus and adjacent regions. The survey zone is indicated by crosses.

what the landscape and other features of the natural setting were like in which those activities took place, in an effort to understand and explain observed changes in human-land relationships through time.1

THE CHOICE OF SOUTHERN EPIRUS FOR THE STUDY

Southern Epirus was selected for this broad diachronic study in part because, at the time, it was only in Epirus and in Thessaly that there was material evidence for something approaching the full range of prehistoric periods. Palaeolithic stone tools, for example, were first attested in Greece in the Louros River valley of Epirus.2 The area is also topographically diverse, including coastal regions, marshy lagoons, inland valleys, high upland plains in rugged mountain terrain, and mountain passes,3 thereby providing a variety of environmental settings for different types of human activities that might be investigated by the project. What is more, prior to the Nikopolis Project there had been no large-scale, systematic, modern survey of the region, and most of the previous archaeological excavations were limited in a variety of ways.4 The Nikopolis Project thus could be expected to enlarge our knowledge of a region that was not well known archaeologically.

Another important consideration was the existence in the survey zone of Nikopolis, the "city of victory" founded by Augustus to celebrate his

1. This introductory section is an expanded version of the statement of aims set out in Wiseman 1995a, p. 1, and uses some of the phrasing of that earlier formulation.

2. Dakaris, Higgs, and Hey 1964; Higgs and Vita Finzi 1966; Higgs et al. 1967.

3. Etude gdologique. 4. See below, "Previous Archaeo-

logical Work in the Survey Zone."

2

THE NIKOPOLIS PROJECT

Louros River

Acheron River X f Voulista 1

,~~\ ~ ~ ( ~ ~Panayia

, Parga ,Kiperi Vouv taos

Phnr^ Ephyra Kastri Thesprotiko Ayios Yeoryios

(Ammoudia . 'N2manteion Kastro Rizovouni kinopilos , *Spilaion \

Aloaki. *Loutsa V

Ch .:- dio mLadiouros i;':: Palaiorophoros? Louros-Kastro Rogo,

-' - :, Cassope* / '

Arachthos

K' '"X ) * Strongyli \ River

Kastkrosykm Grammeno

Archan los

mian Sea *1o Chlts Nikopolis * SaIaor ^

Ormos Vathy y7."\

.Tmas Ambracian Gulf Prey ;:, .a

Actm

Figure 1.2. Map of survey zone with selected toponyms

0 5 10 15 20 25 KM

victory in 31 B.C. over Antony and Cleopatra in the Battle of Actium. The creation of the urban population by the officially encouraged migration or forced removal to Nikopolis of populations from other cities of Epirus, Acarnania, Leucas, Amphilochia, and Aetolia,5 and the long life of Nikop- olis as the metropolis of Epirus, raised a number of challenging problems regarding the relationship between the city and its territory to which the

project's research concepts were directly applicable. The project thus takes its name from Nikopolis, the best-known toponym in southern Epirus. Finally, there was an urgent need for interdisciplinary survey before certain types of evidence, including some of the cultural remains, vanished as a result of various activities: land reclamation near the coast, the growth of the modern town of Preveza and several other smaller communities, industrial and agricultural development, limestone quarrying, and other de-

velopment activities related to tourism. These activities had wrought ma-

jor changes on the regional landscape since 1950, and the pace of change in recent years had accelerated.

THE SURVEY ZONE

5. Kirsten (1987), Murray and Petsas (1989, pp. 4-5), and Purcell (1987) all discuss the founding of Nikopolis and cite the most important sources.

The survey zone (Figs. 1.1, 1.2), about 1,200 km2, includes the entire nomos (administrative district) of Preveza, a modern town on the Nikopolis peninsula, extending from the straits of Actium almost to the walls of the ancient city. On the east the survey zone extends into the nomos of Arta,

IC

*.. 11,- I I

..:. I .

3


so that the entire deltaic, lagoonal area of the Louros River after its exit from its gorge at the modern town of Philippias was included; not included was the course of the Arachthos, a larger river east of the Louros which flows through the city of Arta (the ancient Ambracia) before emp- tying into the Ambracian Gulf, also known today as the Gulf of Arta. It is the western part of the north coast of the gulf, therefore, that lies within the survey zone, from Salaora on the east to the southern tip of the Nikopolis peninsula. The other boundaries follow those of the nomos of Preveza. That is, the western boundary of the survey zone is the shoreline of the Ionian Sea, from the straits of Actium on the south, where the Ambracian Gulf is linked to the sea, extending north beyond Ammoudia Bay (= Phanari Bay), at the mouth of the Acheron River, to Parga. The northern boundary of the survey zone runs east from Parga, along the middle Acheron River, and across the mountains to the narrows of the Louros River gorge near the modern town of Kleisoura, below the ancient acropolis known locally as Voulista Panayia.

The geology and geomorphology of southern Epirus are discussed in detail in Chapters 3, 5, and 6, so comments here are limited to observations of an introductory nature, primarily focusing on features providing general constraints on communication and exploitation of resources. A series of north-south Mesozoic limestone ridges, 600-1,000 m high, extends across the region from the Louros gorge to the Ionian coast, alternating with Tertiary flysch basins at elevations of 150-600 m, so that the basins provide now, as they did in the past, corridors of varying conve- nience for traveling north-south; fortified town sites of Archaic, Classical, and Hellenistic times are situated along the routes. Access to these natural corridors on the south is via passes through or between a series of mountains along the Ambracian embayment: from west to east, Mts. Zalongo, Stavros, and Rokia (see Fig. 5.1). The Louros River valley was an important communication route from early prehistoric times to the present; the principal road from Arta to Ioannina, present-day capital of Epirus, still passes through the gorge. The next basin on the west is most easily entered from the south between Mts. Rokia and Stavros, and a traveler would pass near a fortified Classical and Hellenistic town site (Kastro Rizovouni) en route to the north and the passes that lead eventually into the valley of Dodona. The next basin to the west includes access to the upper Acheron River, and can be entered over a low ridge between Mts. Stavros and Zalongo. A bit further west, the natural route is over a ridge of Mt. Zalongo, by the Classical and Hellenistic town of Cassope, and from there through a winding pass to the modern town of Kanallakion in the eastern part of the plain of the lower Acheron River.

Agriculture is now practiced throughout the region, wherever it is possible to do so, in the upland valleys, along the courses of rivers and streams, and in the coastal areas. In the latter regions, especially around Ammoudia Bay and along the north coast of the Ambracian Gulf, swamps and marshy areas have been drained during the past half-century and flooding has been further controlled by the construction of canals, which also serve as conduits for irrigation of fields. Dams were built on both the Louros and Arachthos Rivers. There has been extensive work also in some of the

4


upland basins; for example, a small lake (Lake Mavri) was drained in the basin east of Kastro Rizovouni to provide more arable land, and the deep waters of Lake Ziros in the same area are now being tapped for irrigation. The whole lower Acheron and the valley of its chief tributary, the Vouvos (ancient Kokytos) River, as far as the modern town of Paramythia (outside the survey zone) are now lush with vegetation, including a variety of cash crops and orchards.

PREVIOUS ARCHAEOLOGICAL WORK IN THE SURVEY ZONE

6. A detailed account of previous investigations in southern Epirus is being prepared by K. Zachos.

7. Dakaris 1971, 1975b, 1977,1978, 1979, 1980, 1981, 1982, 1983.

8. Dakaris 1958, 1960, 1961, 1962, 1963, 1964, 1975a, 1975b, 1977, 1993; Wiseman 1998.

9. Dakaris, Higgs, and Hey 1964; Higgs and Vita-Finzi 1966; Higgs et al. 1967.

10. Bailey et al. 1983a, 1983b; Bailey, Papaconstantinou, and Sturdy 1992. The investigations in Epirus by G. Bailey and his colleagues, as well as other recent work somewhat further afield (e.g., by K. Petruso in Albania), are discussed, and additional publications cited, by Runnels and van Andel in Chapter 3.

11. Hammond 1967. 12. Dakaris 1971, 1972. 13. Papers presented at the

symposium were published in Chrysos 1987.

14. Wiseman 1987, p. 413.

The most significant archaeological activities in the larger region in earlier

years6 were excavations by Greek and German scholars at the ancient town of Cassope;7 Greek excavations at a site near the mouth of the Acheron identified by the excavator as the Nekyomanteion, the Oracle of the Dead;8 and investigations by British scholars of Palaeolithic sites in the Louros River gorge to the northeast of Nikopolis.9 Recently the British renewed their interest in some of Eric Higgs's early work at Kokkinopilos and its environs (e.g., Asprochaliko), and carried out limited survey for Palaeolithic remains along the coast.10 Little was known of Neolithic, Bronze Age, and

early Iron Age developments in the region, but the historical period was somewhat better represented in the scholarly literature. Important, useful studies of the region in antiquity were published by N. G. L. Hammond11 and by Sotirios Dakaris.12 Both authors included copious topographical observations in their books and their research involved some survey, which was, however, neither systematic nor intensive. Other archaeological in-

vestigations in the area have been limited to small-scale operations, usu-

ally involving salvage or preservation by the ephoreias, and have been briefly reported over the years in the annual Archaiologikon Deltion of the Greek

Archaeological Service.

BACKGROUND AND ORGANIZATION OF THE PROJECT

The Nikopolis Project had its origins in the First International Sympo- sium on Nicopolis in 1984.13 A paper presented by one of us (JW) focused on the need for the study of Nikopolis in its topographic setting, and suggested approaches to such a study. One specific recommendation, particularly relevant to the eventual development of the Nikopolis Project, was phrased as follows.

A survey both of the natural resources and the cultural remains of the region will be required if Nikopolis is to be studied in its regional context. What is more, the ancient topographic profile, including the changing coastlines, must be determined, along with climatic changes and the palaeoecology generally.14

5


Remote sensing, including geophysical prospection, and computer-aided analysis were discussed in the same presentation as useful tools to aid in such an undertaking, as well as in the investigation of Nikopolis itself. Geophysical prospection, in particular, was cited as an important methodology by which at least some parts of the city plan of Nikopolis might be established before any excavation was initiated. Symposium participants and organizers were deeply interested in the investigation and preservation of the great city itself, and a coordinated, multifaceted, long-term effort was formally declared by the symposium board to be a desirable outcome of the symposium.15

Continued concern for Nikopolis eventually led to the appointment in 1986 by the Greek Minister of Culture, Melina Mercouri, of a special Committee for the Preservation of Nikopolis, which was headed by Evangelos Chrysos (now based at the University of Athens), who was then Professor of Byzantine History at the University of Ioannina and one of the organizers of the symposium. The committee members represented the groups and organizations in Greece with concerns or responsibilities for Nikopolis, including the Greek Archaeological Service, the Archaeo- logical Society of Athens, the city of Preveza, the University of Ioannina, and others. Architects hired by the committee were given an office in the Town Hall of Preveza, and they began the important jobs of mapping all visible remains in Nikopolis and its periphery, and of documenting the ownership of all properties within the archaeological zone of Nikopolis. The committee was reconstituted occasionally in the 1990s to reflect political (both local and national) and institutional changes, but Chrysos retained the chairmanship throughout the permutations of the committee until the completion of the Nikopolis Project.

With the encouragement of Chrysos, Wiseman began discussions in 1988 with Angelika Douzougli, the newly appointed proistameni (director) of the 12th Ephoreia of Prehistoric and Classical Antiquities, and her husband, Konstantinos Zachos, senior archaeologist in the same ephoreia, regarding possible collaboration on a project in the Nikopolis region, which lies within the purview of that ephoreia. The 8th Ephoreia of Byzantine Antiquities, directed by Frankiska Kephallonitou, also became involved in the early planning, because Late Antique and Byzantine remains in the same region were among the responsibilities of that ephoreia. The decision was reached in 1990 that the two ephoreias, both based in loannina, and Boston University would jointly share the responsibilities of the project, so that the proposal for the project, when finalized, was for a joint undertaking, synergasia in Greek terminology. The directors of the two ephoreias and K. Zachos were codirectors of the project with Wiseman, the Ameri- can Principal Investigator, and other representatives of the ephoreias were also members of the staff. The project proposal was then submitted first to the American School of Classical Studies, as then required by Greek law for a project involving American sponsorship or cosponsorship.

There was for a time consideration of a collaborative project based on Nikopolis itself, working in cooperation with the group that would carry out the regional study, as envisioned at the Nikopolis symposium.16 The principal aims of work at Nikopolis would have been to determine at least

15. Chrysos 1987, pp. 417-418. 16. Wiseman 1987.

6


17. van Andel and Runnels 1987; Jameson, Runnels, and van Andel 1994.

the general outline of the city plan through geophysical prospection and other forms of remote sensing; photography from a tethered blimp both to help in detecting the town plan and to aid in the documentation of above- ground remains; and test excavations intended to provide a stratigraphic control for regional ceramics, an urgent need because there were then few published groups of well-dated ceramics. These plans were abandoned in 1991, as it became clear that there were too many conflicting and competing claims to archaeological rights at Nikopolis itself for any one group, especially a new one, to obtain the support of the Archaeological Council in Athens, the responsible body for approving permits for archaeological investigations of any kind in Greece. The proposal as finally submitted was for a combined archaeological and geological survey of the region, but not including Nikopolis, conducted in synergasia. For 1991, the project would involve mainly ground-truthing of satellite imagery and gaining greater familiarity with the landscape by the American staff, and finalizing the aims and methodology of the regional investigation. The subsequent permit was for three years, 1992-1994, during which the archaeological and geological investigations were carried out. There were study seasons in the summers of 1995 and 1996, when senior staff, based in Ioannina to study archaeological materials collected during the survey, were able to revisit the survey zone with staff reports in hand and to discuss project results and interpretations. Laboratory analyses and study both of the artifacts and the archives have continued since that time.

A number of scholars in Greece, the United States, the United King- dom, and other countries contributed to the eventual research design, including both specific research aims and methodologies adopted by the Nikopolis Project, especially those who have devoted so much of their time and effort as members of the staff. George (Rip) Rapp, a geoarchaeologist at the University of Minnesota, Duluth, with extensive field experience in Greece and other parts of the eastern Mediterranean, was one of the first scholars invited to join the staff; he organized and directed much of the project's geological survey, coring program, and shoreline studies. Curtis Runnels, an archaeologist at Boston University, brought his exper- tise in the early prehistory of Greece and in survey to the Nikopolis Project. He would lead the Palaeolithic survey, with the aid and cooperation of his wife, Priscilla Murray, Research Fellow in Archaeology at Boston Univer- sity, and Tjeerd van Andel, a geoarchaeologist formerly of Stanford Uni- versity, then (and now) of the University of Cambridge. Runnels and van Andel would now apply survey techniques they had jointly developed on projects in southern Greece to the investigation of early humans and hominids in Epirus.17 Their survey, which supplemented, but was conducted separately from, the intensive surface survey carried out by other staff, involved intensive geomorphologic studies in the detection of Pleistocene landscapes, which they then searched. Both would also join in other project responsibilities-Runnels, for example, in the analysis of prehistoric stone tools, and van Andel in geomorphology for all periods, as well as providing counsel and insight for all geoarchaeological concerns. Lucy Wiseman of Boston University's Center for Archaeological Studies was also a member of the staff from the beginning, serving both as project administrator

7


and registrar of artifacts. Three advanced graduate students in archaeology at Boston University were also part of the senior staff. Thomas Tartaron and Carol Stein were the primary team leaders in archaeological survey, and provided both supervision and guidance for others who subsequently became survey team leaders. Tartaron also developed a specific sampling strategy for the Acheron River valley and Ayios Thomas peninsula, re-

flecting the overall stratified sampling strategy of the project, and carried out a special study of the Bronze Age sites and materials, part of which was included in his doctoral dissertation.18 Melissa Moore oversaw the

study and registration of ceramics, and part of her research has been included in her Ph.D. dissertation.19 Other staff and consultants included

geologists, computer scientists, archaeologists, and specialists in various other fields; all staff and their affiliations during the Nikopolis Project are provided in Table 1.1. Students enrolled in a Boston University Archaeo-

logical Field School were invaluable members both of the field survey teams and the geological coring and survey units in 1992, 1993, and 1994. As a

part of their archaeological training, they participated in all activities of the project in Greece, including the processing of artifacts, data processing on computer, digitizing of maps, ground-truthing of satellite imagery, to-

pographical survey, geophysical prospection, aerial photography by tethered blimp, and other investigations. Their names and the institutions where

they were studying at the time are listed in Table 1.2.

SPECIFIC RESEARCH AIMS

Research aims, nested within the larger conceptual framework described above, relate mainly to specific time periods and include the following topics, phrased as questions, which much of the project's fieldwork was intended to answer.

1. What forms do the cultural remains of the earliest inhabitants of southern Epirus take, and how may we explain their distribution in the different periods of the Palaeolithic? What resources were exploited by the early humans and hominids, and what was the environmental setting?

2. What is the evidence for the shift from hunting/gathering groups to agricultural societies? Can that shift be related to changes in the landscape?

3. What was the nature of the contacts between peoples of this region in later prehistoric times, especially in the Late Bronze Age, and groups on the shores of the Ionian Sea, in other parts of Greece, and more generally in the eastern Mediterranean? Do these contacts differ in quality during fully historical times?

4. How are colonial activities of southern Greeks manifested in this region?

5. What were the effects of the development of political leagues and interregional alliances on settlement patterns, sizes of sites, religious centers, and resource exploitation in Classical and Hellenistic times?

18. Tartaron 1996. 19. Moore 2000.

8


6. What were the effects of the historically documented Roman intrusion into Epirus (which was also the earliest intervention

by Romans in Greek affairs) in the 3rd and 2nd centuries B.C.,

and how may they be identified in the landscape? How intru- sive into local society were the Romans, and what activities

(military, industrial, commercial, social, etc.) are indicated by the cultural remains?

7. What was the regional effect of the synoecism involved in the

founding of Nikopolis by Octavian, later Augustus, first

emperor of Rome? How are the new patterns of settlement and communication related to changes in the landscape itself?

8. What was the nature of the exploitation of the countryside in the Late Antique period (4th-6th centuries A.c.) and how was it related to the socioeconomic transformation into medieval times? More specifically, what was the economic basis of southern Epirus in late antiquity and in medieval times? When did the extensive exploitation of wetlands along the Ambracian Gulf begin, and when the deliberate reclamation of land from coastal lagoons?

METHODOLOGIES

20. A practice recommended in Sever and Wiseman 1985, pp. 70-71.

The research design called for the archaeological sampling by intensive surface survey of all environmental zones: coastal plains, inland valleys, mountainous terrain, and upland valleys. The large size of the survey zone precluded archaeological survey over the entire region. The selection of the areas to be surveyed within each environmental zone would be guided primarily by acquired knowledge of the region. Geological survey and other geomorphologic investigations provided important information, both negative and positive, influencing the selection of fields and transects to survey; fieldwalking teams, for example, could avoid areas of recent alluviation where remains (if any) of prehistoric-medieval times would have been covered over and not detectable. The location of early historical or even Pleis- tocene landscapes exposed by erosion, on the other hand, offered opportunities for survey with greater expectation of detecting archaeological remains. Even so, occasional surveys were conducted to test negative indications from geomorphology or satellite imagery,20 as when fieldwalking teams spent a day walking transects across the presumed relict coastlines of Ammoudia Bay that were formed by long-shore deposition in recent historical times. The negative results of the intensive survey confirmed the geomorphologic conclusions and the interpretations of imagery. The degree of visibility was recorded for all areas surveyed. Fields where vegetation was too dense for archaeological remains to be seen during preliminary reconnaissance were not selected for survey. This practice is an important consideration in evaluating the results of the survey, because in some other year, or some other time of year, those fields might be clear of vegetation, and might, of course, yield archaeological materials. On the other hand, in some instances fieldwalking teams were able to return to a region to survey fields that had been too densely covered for survey in a

9


TABLE 1.1. PROJECT STAFF AND THE YEARS OF THEIR PARTICIPATION

Name 1991 1992 1993 1994 1995 1996

CODIRECTORS

Angelika Douzougli/Konstantinos Zachos, 12th Ephoreia of Prehistoric and Classical Antiquities * * 0 0*

Frankiska Kephallonitou, 8th Ephoreia of Byzantine Antiquities 0 0 0

James Wiseman 0 .

ADMINISTRATION AND INVENTORY

Lucy Wiseman (registrar of artifacts, administration) * * * 0

Melissa Moore (registrar of ceramics, archaeology) * * * * e

Lia Karimali (lithics, survey) * * *

Dimitra Papagianni, University of Cambridge (lithics, survey) * * 0

Katerina Dakari, 8th Ephoreia of Byzantine Antiquities (survey, Late Antique ceramics) * *

Ricardo Elia (associate director, archaeology)

Asymina Kardasi, Athens (Byzantine ceramics) Stavroula Vrachionidou, 12th Ephoreia of Prehistoric and

Classical Antiquities (administration, survey) *

ARCHAEOLOGY, SENIOR STAFF

Timothy Baugh (remote sensing, ground-truthing) Brenda Cullen (survey, remote sensing) * * * *

Priscilla Murray (survey, drafting) * * *

Curtis Runnels (field director, Palaeolithic survey; lithics) * * * * *

Carol Stein (survey, remote sensing) * S 0

Thomas Tartaron (survey, ground-truthing) * * *

Stavros Zabetas, Greek Archaeological Service (survey)

GEOLOGY AND GEOPHYSICS

Mark Besonen, University of Minnesota, Duluth

(geological survey, coring) Richard Dunn, University of Delaware (geological survey, coring) Zhichun Jing, University of Minnesota, Duluth

(geological survey, coring) * *

Jon Jolly, Seattle, Washington (oceanography, instrumentation)

George (Rip) Rapp, University of Minnesota, Duluth

(geology, geoarchaeology) * * 0

Apostolos Sarris, Athens, Greece (geophysics) Marie Schneider (geology, survey) Tjeerd van Andel, University of Cambridge

(Pleistocene geology, geomorphology, geoarchaeology) * * *

Sytze van Heteren (geology) John Weymouth, University of Nebraska (geophysics) Li-Ping Zhou, University of Cambridge

(geology, thermoluminescence dating)

IO


TABLE 1.1-Continued

Name 1991 1992 1993 1994 1995 1996

COMPUTER SCIENCE

Robert DeRoy (computer science, remote sensing) Daniel Juliano (computer science, remote sensing) Rudi Perkins, Bangor, Maine (computer science)

PHOTOGRAPHY

Michael Hamilton (aerial photography, generalphotography) *

Eleanor Emlen Myers' (aerialphotography) J. Wilson Myers (aerialphotography)

TOPOGRAPHICAL SURVEY AND DRAFTING

Theodoros Chazitheodoros, Greek Archaeological Service, Athens (topographical survey, drafting) *

David Clayton (topographical survey, drafting) Athina Kotsani, Preveza (drafting) a

Kostas Papavasileiou, Preveza (architecture, drafting) a

Anne Van Dyne, Seattle, Washington (topographical survey, drafting) a

GENERAL STAFF

Stephen Agnew (ground-truthing) 0

Kael Alford (survey) Alesia Alphin (survey, inventory)

Betty Banks, Spokane, Washington (survey, inventory, data entry)

Mark Greco (survey) *

Cinder Griffin, Bryn Mawr (survey, inventory) a

Nikola Hampe, University of Miinster (survey) a

Alan Kaiser (survey) Petra Matern, University of Miinster (survey) 0

Michele Miller (ground-truthing, survey) * S

Lee Riccardi (survey, inventory) *

Katrin Vanderhuyde, University of loannina (survey) Elizabeth Wiseman, Littleton, Colorado

(photography, ground-truthing)

CON SULTANTS

Virginia Anderson-Stojanovic, Wilson College (ceramics) *

Evangelos Chrysos, University of loannina (Byzantine history) * * *

Harrison Eiteljorg II, Bryn Mawr (databases, AutoCAD) Panayiotis Paschos, IGME, Preveza (geology) * * *

Staff members listed without an institutional affiliation or city were from Boston University.

II


TABLE 1.2. FIELD SCHOOL STUDENTS

I992 Kael Alford, Boston University Alexandra Bienkowska, Boston University Anne Cockburn, Williams College Todd Gukelberger, SUNY, Albany Deborah King, Rensselaer University Dawna Marden, University of Southern Maine Thomas Matthews, Utica College of Syracuse University Richard Rotman, Boston University Bayleh Shapiro, Boston University Jane Sontheimer, Boston University Anita Vyas, Boston University Erika Washburn, Boston University

1993 Alessandro Abdo, Boston University Evie Ahtaridis, University of Pennsylvania Tracy Barnes, Texas Christian University Arlyn Bruccoli, Bard College Christina Calvin, George Mason University Scott deBrestian, Boston University Antonina Delu, University of California, Riverside Katherine Demopoulos, University of California, Los Angeles Cheryl Eckhardt, Boston University Jennifer Fisher, Boston University Lorena Freeman, University of the South

Stephani Kleiman, Loyola Marymount University Noah Koff, Boston University

AND THEIR HOME INSTITUTIONS

Natalie Loomis, Tulane University Michael Marton, Franklin and Marshall College Martin McBrearty, Furman University Scott McCrimmon, Boston University Sean Mulligan, Boston University Wendy O'Brien, Boston University Dena Pappathanasi, University of New Hampshire Rudolph Perkins, Boston University Jamie Ravenscraft, Duke University Jonathan Wood, Princeton University Kelly Younger, Loyola Marymount University

1994 Lisa Davis, Harvard University Mely Do, University of Pittsburgh Aviva Figler, Boston University Mike Gaddis, Princeton University Amy Graves, Miami University Leslie Harlacker, Boston University Karla Manternach, Loras College Joe Nigro, Boston University Anne Maxson, Duke University Kathy Montgomery, Boston University Jennifer Murray, SUNY, Buffalo

Stephan Papageorgiou, Versallius College, Brussels T. J. Reed, Cornell University Yasuhisa Shimizu, Boston University Alison Spear, Mount Holyoke College

previous year. The methodology of the surface survey is discussed in detail by Tartaron in Chapter 2, but it is important to note here that surface surveys included both transects within large regions and intensive sampling, or complete coverage, of human-activity areas ranging from small single-activity sites to extensive settlements. In addition, one fortified town site (Kastri, in the lower Acheron valley) was selected for intensive urban survey.

Geomorphologic studies formed part of the central core of the project, as required by the research concept. If we were to study the interaction between humans and their environment, we reasoned, one of the first steps must be to determine what that natural setting was-that is, what the landscape and other aspects of the environment were like over time. A number of investigations, therefore, were planned to provide the needed evidence. An extensive coring program was initiated in 1992 and continued through 1994 that was aimed at determining changes in shorelines over time both in the Ambracian Gulf and along the Ionian coast. The analyses of the cores, most of which were carried out in the Archaeometry Laboratory of the University of Minnesota, Duluth, also made it possible to establish a sequence of local change and, through radiocarbon dating, to determine the chronology of change. Cores also provided microfauna,

I2


21. See the discussions in Wiseman 1992b, pp. 3-5; 1993a, pp. 12-13.

22. The following brief account is intended mainly to explain what kinds of remote-sensing imagery were acquired and used by the project, and why they were used.

23. Wiseman 1996a, 1996b. 24. Stein and Cullen 1994;

Wiseman 1996a, 1996b.

macrofauna, and pollen for paleoenvironmental reconstruction. Geomor-

phologic investigations involved geological survey in all parts of the survey zone, and intensive work, including coring and mapping, at selected sites or regions. Geological survey and coring were coordinated as closely as

possible with the archaeological survey, so that field teams often com-

prised both geologists and archaeologists working together. We had planned offshore investigations to supplement the study of

shoreline change, and there was a promising beginning to that research. The Hellenic Navy dispatched a research ship, the Pytheas, to work with

project staff for two weeks in 1992. A Klein side-scan sonar and a Klein subbottom profiler were towed behind the ship both in the Ionian Sea and in the Ambracian Gulf, the former recording features on the surface of the sea bottom, the latter detailing the depth and nature of sediments below the sea floor. The survey, in perpendicular transects forming a grid pattern, produced data covering some 300 linear kilometers, which to this date have received only preliminary analysis21 because they were subsequently sequestered by another bureau of the Greek government.

Remote sensing from space was determined to be a potentially useful tool for our survey well before the initiation of the project, as noted above.22 We did not, however, expect remote-sensing imagery to play a significant role in the detection of archaeological sites because at that time most remote sensors were known to be unsuccessful in penetrating dense vegetation, which covered much of our survey zone.23 What is more, although the resolution of satellite imagery had been improved, the smallest picture element (= pixel) of available multispectral imagery was 20 meters to a side, too large to be helpful in detecting the small features and artifacts of most archaeological landscapes. It is an interesting sidelight on the devel-

opment of archaeological methodologies that remote sensing in the end

proved to be quite useful in detecting Pleistocene landscapes, which could then be located and searched by ground-truthing survey teams, and which resulted in the discovery of five prehistoric sites.24 Its greatest value, we

thought at the time, would probably lie in its ability to provide imagery of the entire region that would permit the classification and identification of

present-day land cover. It could, therefore, help in defining the environmental zones of the survey area; show current conditions that might affect the conduct of surface survey; and perhaps provide some insight into routes of communication among known (or subsequently discovered) ancient settlements. The imagery would also serve as a layer in the computer- aided GIS (geographic information system) maps to be generated by the project, and we hoped to develop spectral signatures-that is, a characteristic spectral response identifiable in the imagery-for features of archaeo-

logical interest. Both multispectral (MSS) and panchromatic imagery of the entire

survey zone was acquired from the French satellite company SPOT before the beginning of fieldwork in 1991. SPOT imagery was selected primarily because its spatial resolution was the finest available for general research at that time: MSS at 20 meters, panchromatic at an even finer 10 meters. The United States'Thematic Mapper (TM) satellite imagery, in contrast, has a resolution of 30 meters. Since spatial resolution on the ground is a

I3


Figure 1.3. Multispectral image (SPOT) of the northern part of the

survey zone

Figure 1.4. Multispectral image (SPOT) of the southern part of the survey zone. Leucas (lower left) and other regions south of the Ambracian Gulf lie outside the

survey area.

I4


function of altitude as well as the type of sensor, we could have achieved finer resolution from sensors mounted on aircraft, instead of spacecraft. The only airborne platform available to the project, however, was a tethered blimp, which, although excellent for individual sites and smaller areas, was not appropriate for such a large regional survey as ours because of the time and other logistical difficulties such coverage would require. Full

coverage of the survey zone required two images, both in MSS and panchromatic. The northern image (Fig. 1.3) included almost the entire sur-

vey zone, and the second (Fig. 1.4) added the southern part of the Nikopolis peninsula, along with areas outside the survey zone: Actium, Leucas, and other areas south of the Ambracian Gulf.

Multispectral imagery is particularly useful in showing different types of landcover because landcover types have a different reflectance value in each band of the electromagnetic spectrum. The combination of these numeric values in the bands used by the sensor (SPOT uses green, red, and near infrared) constitutes a spectral signature, which may be represented by a (false) color assigned in a multispectral image generated by the

computer. This assigning of colors, or classification of images, is a process whereby each land area having the same kind of cover receives the same (false) color in the image. The researcher, then, after identifying on the

ground at least once the class represented by a particular color as a particular landcover (e.g., class 12 = red = limestone outcropping), may reason-

ably expect other patches of red in that image to represent the same kind of landcover; in the example just cited, more limestone outcrops. In practice, however, the classification of an image may result in the combining of several signatures into a single class, or the subdivision of a signature into more than one class, depending on the number of classes the researcher chooses for the image and on other physical aspects of the landcover. Mak-

ing use of the facilities of the Center for Remote Sensing at Boston Uni-

versity, Carol Stein classified the MSS imagery of the Nikopolis Project into fifty classes, with all unclassified landcover assigned class 0. The number of classes was considerably larger than proved useful in the field because the fine distinctions the classification made possible resulted in the identification of many kinds of landcover that were irrelevant for our research. For example, there was no reason for us to be able to distinguish kiwi

plants from maize, which our classification enabled us to do. In retrospect, we now see that fewer landcover classes (say, fifteen to twenty) would have been preferable, because such a classification would have resulted in a ben- eficial lumping together of rock outcroppings, and would have created other continuous zones-as in fact they were-of barren land, instead of a number of separate units in the classified imagery. The finer distinctions involved in developing a spectral signature of an archaeological feature, or

archaeological feature combined with a particular vegetation, would still have been theoretically possible.

The relevant portions of the MSS images were then subdivided by Stein into twenty scenes, each representing about 100 km2 on the ground, and printed for field use. Transparent overlays at the same size were also

printed, five for each scene, each displaying ten of the fifty false colors of classes of landcover, so that field teams were able to use them conveniently

I5


Figure 1.5. The eroded landscape of , Z ~ ~ W . . . . . . _ ~~~~~ ~ ~ .......Kokkinopilos above the Louros

__ ............ .River gorge

to determine what on the ground was actually being represented by each false color; this kind of fieldwork is called "ground-truthing." The hard copy of the scenes and transparencies were at a scale of 1:50,000, so they could be used in conjunction with our topographical maps of the same scale; the transparencies could be used as overlays of the maps, just as they were on the printed scenes.

Ground-truthing, a focus of our fieldwork in 1991, required precise location of the observed landscape, so the field teams were also provided with copies of the panchromatic scenes, and even more detailed subscenes. Locations were marked on 1:5,000 topographical maps, and aerial photographs (scale: 1:20,000) also were used to help locate specific features in the landscape; both maps and photographs were obtained from the Geo- graphic Service of the Hellenic Army. Additional locational information was obtained by 1) global positioning systems (GPS), which provide UTM as well as longitude/latitude readings through communication with the navigational satellites (21 in number in 1991) that constantly orbit earth; 2) altimeter readings (more accurate at that time than GPS in determining altitude), when benchmarks are not readily available; and 3) readings by electronic laser theodolite, for still more precise location in three dimensions, as appropriate. These ground-truthing expeditions, which were led by Timothy G. Baugh during the first, preparatory field season, resulted in the identification of 27 of the 50 classes of landcover. An additional 12 classes were created for areas with distinctive features related to human activity whose spectral signatures might serve as guides to the location of other similar areas: e.g., quarries or ancient sites. One of those new classifications was the eroded Pleistocene landscape of Kokkinopilos (Fig. 1.5), which eventually led to the discovery of five other similar landscapes, and prehistoric sites, as mentioned above. The experience gained in using GPS, satellite imagery, and topographic maps in 1991 was invaluable in developing standard procedures for the survey teams of 1992-1994.

I6


25. Hemans, Myers, and Wiseman 1987.

26. Hemans, Myers, and Wiseman 1987.

What is more, the ground-truthing expeditions of 1991 provided several members of the staff with a fundamental familiarity with the Epirote landscape.

Another kind of remote sensing, aerial photography from a tethered blimp, was employed by the project to document some of the larger known ancient sites. Four sites were photographed with radio-controlled cameras in 1992 by field teams led byJ. Wilson Myers and Eleanor Emlen Myers: the fortified town of Kastro Rogon south of the Louros River gorge; Kastro Rizovouni, a fortified town in an enclosed plain north of Kastro Rogon; the Roman aqueduct near Ayios Georgios in the Louros River gorge; and Voulista Panayia, a Hellenistic site overlooking the narrows of the same gorge further north at Kleisoura. Michael Hamilton, who was the project's staff photographer, led the blimp-photography team in 1993 that photographed the large fortified Classical and Hellenistic site at the abandoned modern village of Palaiorophoros, north of the town of Louros. The use of this technique was limited by a number of factors. The necessity for permits from multiple civilian and military authorities resulted in numerous, costly delays and disruption of schedules (e.g., blimp photography in 1991 had to be cancelled and the 1993 season was severely curtailed). The expense was significant, and was greatly increased in 1993 when we decided, for safety reasons, to use helium in the blimp instead of less expensive, but highly flammable hydrogen. In addition, there were the normal delays and logistical problems imposed by the technique itself, such as the need to await favorable winds (that is, none or very light) and other climatic conditions. The photographic results of this technique, however, are highly useful, especially when, as on the Nikopolis Project, multiple cameras are used to provide coverage both in black and white and in color. A particular advantage of photography from a tethered blimp is that the views are vertical and so can be used in mapping, unlike the oblique views frequently gathered by cameras on aircraft. It is also possible in a single flight to obtain photos at a series of elevations up to a maximum of 800 m, thereby providing both close-ups and extensive coverage (see Fig. 1.6). The aerial photograph also can be scanned and then combined with the multispectral image of that area, a technique we used in the study of the fortified town site of Palaiorophoros.

The Boston University blimp-photography system was designed byJ. Wilson Myers, who modeled it on the system he had developed earlier, and is described in detail elsewhere.25 A multispectral video camera, successfully used on a tethered blimp in the Corinthia by a Boston University team in 1986,26 was not used by the Nikopolis Project, but could usefully be deployed in the future, since it can provide high spatial resolution in six bands of the electromagnetic spectrum.

Geophysical prospection of various kinds was carried out at a number of sites, primarily to provide data on possible subsurface features in areas where surface survey suggested significant human activity. Only limited prospection was possible in 1992 because of staffing and equipment problems, but successful programs of investigation were conducted in 1993 under the direction of John Weymouth of the University of Nebraska and

I7


Figure 1.6. Aerial view of the fortified town site at Kastro Rogon from an elevation of 400 m. Photo by J. Wilson and Eleanor Emlen Myers

in 1994, when Weymouth was succeeded by his protege, Apostolos Sarris. Instrumentation included a proton magnetometer, electrical resistivity meter, and electromagnetic conductivity meter, of which the first was most frequently used. Weymouth and Sarris are preparing a report on their in-

vestigations for volume 2 of this series, and the results are also being incorporated into reports on the town sites where geophysical prospection detected significant subsurface features such as probable kilns and buildings.

The permit of the Nikopolis Project was for survey, not excavation; indeed, under Greek law a single permit might cover only one or the other. As a result, the project had an arrangement whereby one of the cooperat- ing Greek ephoreias would perform excavation if a site was discovered by the project to be in need of emergency attention. The discovery at the Roman villa site of Strongyli, for example, that burials had been plundered by clandestine diggers and parts of floor mosaics had been exposed prompted excavations by the Greek ephoreia to ensure conservation.27 A similar situation arose at Frangoklisia, probably another Roman villa, on the Ionian coast near Loutsa.28 The project did carry out limited excavation in 1991 at the request of the Prehistoric and Classical Ephoreia in the Roman aqueduct below the village of Ayios Georgios, so that details of the water channels and the chronological sequence of aqueduct bridges across the Louros River might be studied and drawn (Fig. 1.7). Our work here resulted in, among other conclusions, the confirmation that the northern bridge was built and utilized for the aqueduct after the earlier, Au- gustan bridge had been damaged and abandoned.

27. Douzougli 1998a, 1998b. 28. Zachos 1998.

I8


Figure 1.7. Aerial view of the water i channel (right) and aqueduct bridges . t across the Louros River from an elevation of 320 m. Photo by a ' f J. Wilson and Eleanor Emlen Myers s

DOCUMENTATION

All team leaders and individual investigators kept a daily record of their activities and observations in bound, hardback notebooks, which also contained photographic prints and drawings, and were indexed upon completion. The notebooks were numbered sequentially. This permanent historical record, partially in narrative form, was supplemented by an array of printed forms that were filled out in the field or laboratory, as appropriate, providing detailed information on all aspects of the investigations, from surface survey to artifact inventory. These two kinds of written documentation were cross-referenced on a daily basis, but it was primarily the series of printed forms that provided the bulk of the information that was entered into the computer databases. I summarize below the principal databases of the Nikopolis Project. All forms were numbered by year and sequential accession within the year, e.g., 92-1. Databases marked with an asterisk are dealt with in greater detail in Chapter 2.

1. Ground-Truthing Form (GTF). A GTF was filled out at every location where ground-truthing was conducted to identify the landcover of classes in the satellite imagery.

*2. Tract (T). The tract, an area of arbitrary size, is the project's primary survey unit whether in the countryside or within a large site. The database includes location, size, description, conditions of the survey, total artifact counts, and summary results.

I 9


*3. Site/Scatter (SS). An SS is any location where there was a concentration of artifacts or that is marked by visible, in situ remains. This category includes any location from a small scatter of lithics to a fortified town. The database includes location, size, description, chronology, and survey data.

*4. Walkover (W). A W indicates a nonintensive survey or a visit either for reconnaissance or reexamination.

5. Sample. The sample database includes the description, counts, dates, and other details of all cultural material collected during survey. Sample numbers are identical to the numbers of the survey units where they were collected.

6. Inventory. Artifacts selected for inventory were catalogued and stored according to material/function. Selection criteria included, among others, significance for dating or functional analysis, or the likelihood of publication as a type artifact.

7. Special Analyses. This database provides a record of the context and nature of samples taken for laboratory analyses, from clay samples to geological cores.

8. Photo Inventory. A record of all black-and-white and color photographs taken by the Nikopolis Project, in the field, photo studio, or laboratory.

9. Drawing Inventory. A record of all drawings made by and for the

project.

Relational databases 1-6 were all created in FoxBase+ for Mac, which seemed to the staff, including the computer scientists and engineers, the most suitable at the time. Unfortunately, when the software was redesigned as FoxPro in 1993, databases in earlier versions of the software could not be upgraded; all windows for data entry would have had to be redesigned and the data reentered to use FoxPro, a duplication of effort we declined to do. The program, therefore, lacks some of the flexibility and ease of some of the more recent databases, but still has served the project well. The

design of the relational databases reflects the archaeological concerns and

experience of the senior staff, and there was much (both fruitful and lively) discussion between the archaeologists and the computer experts who put it all together.

The various forms and notebooks were supplemented by copies of

maps, primarily the 1:5,000 topographical maps, on which field teams marked survey locations and other observations. Each member of the staff also prepared a staff report at the end of each season, which summarized the activities each person performed, the forms and notebooks in which the records were kept, and whatever other comments the staff desired to make. There were numerous other logistical records, including logs to keep track of the forms assigned for field use, and extensive cross-referencing. We hold redundancy in archaeological records to be a virtue because it makes it possible to discover the inevitable recording errors that occasionally creep into databases, however carefully they are kept. All databases and other archives of the Nikopolis Project are stored in the Center for Archaeological Studies at Boston University.

20


POST-FIELDWORK ANALYSES

During study seasons in 1995 and 1996, materials collected during the surveys were reexamined and studied in Ioannina. The Byzantine Ephoreia made available for study space the secularized former mosque, Fetiye Dzami, located on the highest part of the fortress of Ali Pasha and adjacent to the new Museum of Byzantine and Post-Byzantine Archaeology. The glori- ous view from one side of the mosque included the lake of Ioannina and the Pindos Mountains, and there were trees nearby that offered shade for staff members who might be working outside. The staff is particularly grateful to the Byzantine Ephoreia for providing such a splendid place to study, and to the Prehistoric and Classical Ephoreia for permitting the survey material to be transported across town from the Archaeological Museum to the Kastro during two summers.

During each of the two study seasons, the senior staff also had the precious opportunity to revisit survey areas unaccompanied by survey teams to direct, and not burdened with surveys to conduct or detailed forms to fill out. The staff, then, were able to contemplate on the spot the observations of previous years, and had the leisure to discuss observations and interpretations with each other in the midst of the landscape we were studying.

PRESENTATION OF RESULTS

29. Wiseman, Zachos, and Kephallonitou 1996, 1997, 1998.

30. Wiseman 1991,1992a, 1992b, 1993a, 1993b, 1994, 1995a, 1995b, 1997a.

31. Rapp and Jing 1994; Runnels 1994; Stein and Cullen 1994; Tartaron 1994; Tartaron and Zachos 1999; Wiseman 1997a, 1997b; Wiseman and Douzougli-Zachos 1994; Wiseman, Robinson, and Stein 1999; reports by several staff members recently appeared in Isager 2001. Articles and abstracts in press have been omitted here.

32. Runnels and van Andel 1993b; Tartaron and Runnels 1992; Tartaron, Runnels, and Karimali 1999.

33. Papagianni 2000, which is based on her (1999) dissertation at the University of Cambridge.

34. Besonen 1997.

Preliminary reports of the Nikopolis Project appeared regularly in Greek in the Archaiologikon Deltion29 and in English in Context and the Nikopolis Newsletter, publications of Boston University's Center for Archaeological Studies.30 Papers by several members of the staff have appeared in full or in abstract form in the published transactions of the several conferences and symposia at which they were presented,31 and a few special reports have been published in journals and edited volumes of essays.32 In addition to the doctoral dissertations of Moore and Tartaron, which were based mainly on project results and have been cited above, a dissertation by Dimitra Papagianni also includes research on material from the Nikopolis Project.33 Chapter 5 in this volume, written by Mark Besonen, George (Rip) Rapp, and ZhichunJing, is based in part on Besonen's M.S. thesis.34

The present book is the first of two volumes of final reports. Chapter 1, by Wiseman and Zachos, provides a history of the Nikopolis Project, and discussions of the research aims, the interdisciplinary methodologies employed, the databases, and the organization of staff and responsibilities. In the second chapter Tartaron presents in detail the methodology of the diachronic surface survey and places both the methodology and the aims within the historical and theoretical context of survey archaeology, especially as that field has evolved in the archaeology of Europe. These two chapters, which constitute an introduction to the work of the project, provide a historical, theoretical, and methodological framework within which the results of the overall interdisciplinary project may be understood and evaluated. They are not intended to be summaries of the results them-

2I


selves, which are presented in the reports that follow in this volume and its

forthcoming companion volume. In Chapter 3 Runnels and van Andel present the results of the Palaeo-

lithic survey, which they conducted as a supplement to the diachronic sur-

vey. Their methodology, developed over some fifteen years of survey in southern and central Greece, was based first on the investigation of the

paleoenvironment, especially the geological history of Pleistocene sediments and other landforms. Their report thus deals comprehensively with the geomorphology and changes in the environment of southern Epirus in early prehistoric times, as well as the cultural evolution of its human inhabitants, from the Lower Palaeolithic to the Mesolithic. One of the most remarkable of the open-air Palaeolithic sites investigated by the project is Spilaion, an Early Upper Palaeolithic site near the current mouth of the Acheron River, where the ground surface was littered with an estimated 150,000 lithic artifacts. Runnels, Evangelia Karimali, and Brenda Cullen

report in Chapter 4 on their study of the Spilaion assemblage, including the results of a spatial analysis of the distribution of the artifacts.

Chapters 5 and 6 carry the discussion of the geomorphology of southern Epirus and its relationships to archaeological sites from the end of the Pleistocene to the present. Both reports are based on extensive geologic coring programs and intensive laboratory analyses of the cores, as well as other geomorphologic investigations in the field. ZhichunJing and George (Rip) Rapp document the changes over the past 10,000 years in the coastal

landscape of the Nikopolis peninsula and the area to its east, which com-

prises most of the north coast of the Ambracian Gulf. The locations of the important Classical, Roman, and medieval town sites in this region, and of human habitation generally, are related to the dramatic changes in the

landscape, which are themselves shown to result from a variety of environmental, geomorphologic, and cultural factors. Besonen, Rapp, and Jing report in detail on the post-Pleistocene geologic history of the lower Acheron valley, tracing the changing course of the Acheron River, the creation and demise of the Acherousian lake, and the gradual change over time of the deep embayment known to Strabo as the Glykys Limen, where large fleets of ships found anchorage both in Greek and Roman times, to the small bay of the present day at the mouth of the Acheron River. The historical implications of the coastal changes are also discussed. In a final chapter the editors comment briefly on the results reported in this volume.

Volume 2 of Landscape Archaeology in Southern Epirus, Greece will include a catalogue of sites/scatters and all tracts surveyed; reports on the pottery, lithics, and other artifacts; and a chronological presentation of the cultural remains in their environmental contexts.

22

CHAPTER 2

THE ARCHAEOLOGICAL SURVEY:

SAMPLING STRATEGIES AND

FIELD METHODS

by Thomas F Tartaron

1. Keller and Rupp 1983; Barker 1991; Cherry 1983, 1994.

2. Alcock 1993; Cherry 1994; Alcock, Cherry, and Davis 1994; Kardulias 1994a; Bintliff 1997.

3. Cherry 1994, pp. 92-95. 4. Binford 1964. 5. Fish and Kowalewski 1990;

Trigger 1989, p. 311. 6. Fish and Kowalewski 1990; but

see Alcock, Cherry, and Davis 1994, pp.137-138.

7. Kintigh 1990; Plog 1990. 8. Parsons 1990; Sumner 1990. 9. Fish and Kowalewski 1990.

Systematic surface survey has been practiced and refined in the Mediter- ranean region for more than a quarter century,1 and there is no longer serious controversy about the legitimacy of survey as a robust method-

ological tool for regionally focused research, or about the contribution it has made to the study of all periods of the Mediterranean past.2 Al-

though the many achievements of survey projects are self-evident and arouse much optimism,3 few would suggest that a state of disciplinary maturity has been attained. The developmental years have witnessed continuous and serious challenges to many of the theoretical and methodological foun- dations upon which surface survey rests, as archaeologists have increas-

ingly recognized the complexity of the surface archaeological record, and the inadequacy of many of our methods and conceptual frameworks for

analysis and interpretation. Vigorous debate continues on a range of theoretical and practical

matters. Recently, the validity of probabilistic sampling schemes and quantitative methods, once regarded as powerful means of characterizing entire

regions from carefully chosen samples,4 has been called into question. Ex-

perimental data suggest that such "samples" often fail to capture the true

variability present in the archaeological record, making suspect the notion that patterns discerned for a portion of a region are necessarily valid for the whole.5 Fish and Kowalewski are particularly vocal in advocating "total" regional coverage to offset the problem of sampling,6 but this approach fails to solve-and in some cases to address-a range of problems, which are well documented by Kintigh and Plog.7 Among these is the fact that

many of these "full-coverage" surveys ignore the powerful effect of survey intensity; thus, one project that employed a 30-m spacing interval between walkers, and another in which intensive and systematic coverage are treated as secondary concerns, hardly point the way forward to revealing the fullness of human activity upon a landscape.8 In more practical terms, while the general principle of covering a region, however narrowly or broadly defined, in its entirety would seem unimpeachable, the immense increase in costs entailed in such coverage must be justified by suitably enhanced results. In view of the cases presented by Fish and Kowalewski,9 we must at present conclude that sometimes they are, and sometimes they are not.

THOMAS F. TARTARON

At minimum, the critical parameters of intensity and systematic data collection must be integral, not independent,'? variables in full-coverage sur-

vey design. Most Mediterranean surveys, while acknowledging potential problems with sampling, have relied on some type of stratification of the

survey universe, typically incorporating samples of a full range of environmental zones with survey locations derived from known distributions of

archaeological remains.T"

Perhaps more disturbing is the fact that whereas archaeologists ask ever more expansive and complex questions of the archaeological record, the development of increasingly refined methods capable of providing the answers has failed to keep pace. Archaeologists have not been able to resolve a range of difficulties that stem, on the one hand, from the inherent

complexity of the surface record and, on the other, from an inability of

existing methods to record the scatters in a way that faithfully represents their distribution, density, and degree of clustering. The issues are both observational and analytical in scope.

The intrinsic complexity of the surface archaeological record has been measured in a number of recent studies. It is well understood that surface scatters of artifacts at a given location are constantly modified by diverse natural and cultural agents over time, as replication and experimental studies have clearly demonstrated.12 Ammerman's work in particular reminds us that the local circumstances and timing of an inspection strongly influence the results, and that repeated visits over a period of years may be necessary to capture the fullness of the archaeological record. (This became abun-

dantly apparent to us at locations such as Grammeno and Ormos Vathy; see below.) Yet the precise effects of erosion and deposition, water action, plowing, and other processes on forming and transforming the surface record are not always well understood. Simulation studies have provided a number of promising approaches,'3 but they seem not to have been widely applied, in part because it is difficult to control a broad range of variables in nonexperimental situations, and because for each survey, unique conditions pertain.

Some relationship between surface scatters and subsurface remains is

usually assumed, but rarely demonstrated.14 In recent years, however, sur-

vey archaeologists have developed a battery of techniques designed to measure the relationship between surface scatters and the subsurface remains with which they are presumed to be associated. These techniques not only evaluate our measurements of this relationship but improve upon them. Thus, the application of long-term replication studies,'5 geophysical remote sensing,16 phosphate studies,'7 and controlled collections followed by limited, targeted excavation'8 all contribute positively to the measurement of subsurface phenomena from surface or plowzone scatters, either

by identifying subsurface remains directly or by isolating the variables affecting the surface/subsurface relationship.i9 The most promising results have emerged when these techniques are practiced in combination. At one Fort Ancient site in southwestern Ohio, the patterning of surface material was found to supply information that was lacking or ambiguous from excavation, with the result that an anomalously early village of circular plan was recognized.20 The Laconia Survey applied controlled collection, phos-

10. Fish and Kowalewski 1990, p. 2. 11. Alcock, Cherry, and Davis 1994,

p. 138. 12. E.g., Ammerman 1981,1985,

1993; Shott 1995. 13. Odell and Cowan 1987; Shott

1995; Dunnell and Simek 1995. 14. Dunnell and Simek 1995,

pp. 306-307; Downum and Brown 1998, p. 111.

15. E.g., Ammerman 1981, 1985. 16. E.g., Weymouth and Huggins

1985; Sarris and Jones 2000. 17. E.g., Cavanagh, Jones, and

Sarris 1996. 18. E.g., Shott 1995. 19. See Dunnell and Simek 1995;

Odell and Cowan 1987; Shott 1995. 20. Hawkins 1998.

24

THE ARCHAEOLOGICAL SURVEY

21. Cavanagh, Jones, and Sarris 1996.

22. Downum and Brown 1998. 23. Downum and Brown 1998,

pp.119-120. 24. Wandsnider and Camilli 1992. 25. Bintliff and Snodgrass 1988a,

p. 506. 26. Alcock, Cherry, and Davis 1994,

p. 141. 27. E.g., Odell and Cowan 1987;

Stoddart and Whitehead 1991. 28. See especially the contributions

to Sullivan 1998.

phate analysis, and geophysical methods to a number of small rural sites in

southern Greece.21 A notable finding of this work was that the extent of habitation sites tends to be larger than the scatter of surface artifacts would

suggest. In a large cultural resource management (CRM) project in southern Arizona, certain artifact types were found to be more reliable predictors of subsurface remains than others.22 This study also found that in cases where post-depositional disturbances are great, subsurface remains

may have largely or completely vanished, making the surface assemblage the sole remaining source of information.23

Wandsnider and Camilli concentrated on the interface of survey de-

sign, survey performance, and the physical properties of the archaeological record, asking in effect what we are measuring with our survey methods, and how this impacts the investigator's aim to faithfully document the

archaeological record.24 Specifically, they sought to measure the disparity between the archaeological record (the total population of artifacts that is available to be found), and the document (the actual population of artifacts

discovered). In a series of controlled collections, they measured the effects that intensity and interval of transects, ground visibility, and the size, color, and shape of artifacts have on the document that is produced. Their results indicate that discovery is biased toward obtrusive and highly clustered ar-

tifacts; low-density scatters are acutely underrepresented because the typical CRM survey in the United States is not designed to detect them. Thus, the apparent clustering of surface material may be more an artifact of the measurement technique than an inherent property of the record itself. In Greece, however, where there exists a long tradition of intensive, nonsite

surveys, the data reveal striking regional variability in artifact density and

clustering. While the Boeotia survey reports an "almost unbroken carpet" of off-site pottery scatters,25 other intensive surveys have recorded a lower-

density, more discontinuous pattern in which artifacts tend to occur in discrete clusters with little intervening scatter.26

The success of the discipline in finding viable solutions to these chal-

lenges holds obvious implications for the validity of inferences from the surface record. The present state of progress toward that goal depends in

part on one's perspective (is our cup half empty or half full?), but it is also

important to recognize that all field situations are not equally amenable to the kinds of innovative approaches that appear with ever-greater fre-

quency in the literature. Thus, in the event of poor preservation of surface materials, less-than-ideal conditions of site visibility, or restrictive

permit regulations that preclude excavation and other complementary operations, rather pessimistic assessments of the utility of surface data can be expected.27 Yet a guarded but growing optimism is apparent in recent years-based on an increasing archive of successful applications in a wide range of settings-that careful research design and field methods can unlock the intrinsic interpretive potential of the surface archaeological record.28

In conclusion, rather than constituting a cause for alarm, the discom- fiture over the limitations and uncertainties surrounding surface survey reflects a phase of critical self-awareness in survey archaeology, and a will-

ingness to tackle the problems head-on rather than simply bemoaning

25

THOMAS F. TARTARON

them.29 The Nikopolis Project, mindful of a host of problems and potential solutions, sought to introduce certain refinements, in part responding to some of the issues raised above, and in part designed specifically for the unique conditions of the Epirote landscape. Though it is not the aim of this chapter to examine in detail the contingencies of archaeological survey, an attempt has been made to come to grips with many of them- sometimes successfully and sometimes not. Instead, its purpose is to explain in specific terms the principles on which the survey was designed, and the means by which data were collected.

THE NIKOPOLIS PROJECT AND REGIONAL STUDIES IN GREECE

There are compelling reasons that the sampling strategy and methods of data collection be described in explicit detail for each survey project. While it is certainly true that the range of research strategies and field methods must be flexible enough to respond to widely varying conditions of local topography, vegetation, and access, as well as past archaeological investigation and current research goals, it is nonetheless imperative that a framework be provided by which survey results can be evaluated on their own merits, and compared to those of other surveys.

The emergence and proliferation of systematic, intensive survey techniques in Greece provide the potential for such a framework by introduc- ing methods, using well-defined, quantifiable parameters, which form a basis for comparison of data among projects.30 While acknowledging the complexities of establishing objective criteria by which data can be evaluated and compared (and achieving that objectivity in one's fieldwork),31 it cannot be doubted that comparability of information, more than a com- mendable ideal, is in fact a matter of great urgency. Surface survey com- prises an ever-increasing proportion of archaeological research in Greece for several reasons, among which are the moderate cost and logistical complexity of surveys relative to research excavations; the perception that survey is a less destructive technique;32 and the growing interest in landscape archaeology and regional dynamics, which are best investigated using survey methods. As the pace of survey research quickens and wide tracts of the Greek countryside are explored, some thought should be directed to the legacy of information that is to be left to future generations as the combination of surface survey activity and modern development dimin- ishes the country's archaeological resources.33 The most ominous prospect of ending up with a patchwork of projects whose data are not comparable is that we shall never learn much about interregional, diachronic trends- precisely the sorts of issues about which regional archaeological survey ought to be informative.

The detailed publication of the major theoretical and methodological components of a systematic survey design-research goals, sampling schemes, and data collection methods-plays a key role in constructing a basis for evaluation and comparison. The considerable attention devoted

29. Cherry 1994, p. 105. 30. E.g., Cherry 1982; Bintliff and

Snodgrass 1985; Wright et al. 1990; Cherry, Davis, and Mantzourani 1991; Jameson, Runnels, and van Andel 1994; Wells and Runnels 1996; Davis et al. 1997; Mee and Forbes 1997.

31. Keller and Rupp 1983, pp. 43- 44; Bradley, Durden, and Spencer 1994.

32. Surface survey cannot be considered a nondestructive technique, however. Under certain circumstances, as in the case of a field that is plowed frequently, artifacts on the surface may be replenished, redistributed, or fragmented across the surface. But in many cases, the traces of human activity discovered on the surface have no corresponding subsurface sources, or the mechanisms for bringing additional material to the surface in the short term are lacking. In these instances, artifact collection may have the effect of permanently removing evidence. Another concern is the confounding effect of the piling up or scattering of artifacts left behind by archaeologists and others making surface collections. At the very least, this action adds a post-depositional stratum for which allowance will have to be made in all future research. On these issues, see Lloyd and Barker 1981, p. 390; Ammerman 1981, 1985, 1993; Cherry 1983, pp. 397-400 (discussion).

33. Runnels 1981.

26


34. E.g., Bintliff and Snodgrass 1985, 1988a; Wright et al. 1990; Cherry et al. 1991; Wells, Runnels, and Zangger 1990; Wells and Runnels 1996, pp. 15-22; Davis et al. 1997.

35. Cherry, Davis, and Mantzourani 1991, p. 53.

36. Jameson, Runnels, and van Andel 1994.

37. Cherry et al. 1988; Wright et al. 1990.

38. Bintliff 1985; Bintliff and Snodgrass 1985.

39. Wells and Runnels 1996. 40. See Chapter 1; Wiseman,

Zachos, and Kephallonitou 1996, 1997, 1998.

41. E.g., Gaffney and Tingle 1985; Bintliff 1985.

42. Wandsnider and Camilli 1992, p. 183.

to laying out field methods (and their essential linkages to research aims,

analysis, and interpretation), particularly notable in the reports of recent intensive survey projects,34 cannot help but encourage a replicative or self-

perpetuating effect. Methods that work well in the field and are of sound theoretical basis will be recognized, imitated, and refined, with the result that an evolution toward methods yielding statistically valid data amenable to comparison with other regions is set into motion. Recent experience has shown that it is both desirable and possible to devise such field

methods, even though the exact replication of methods from one survey to another is rarely practical and often undesirable.35

The Nikopolis Project surface survey, accomplished in three field seasons from 1992 to 1994, was, like any other, a particular response to a

unique set of research interests, environmental conditions, and logistical limitations. Methods and innovations that were developed in earlier

surveys in Greece and elsewhere were nonetheless incorporated, and

adapted for use in the context of southern Epirus. Particularly influen- tial were those employed in systematic, intensive surveys of recent

years: the Argolid Exploration Project,36 the Nemea Valley Archaeological Project,37 the Cambridge/Bradford Boeotian Expedition,38 and the Berbati- Limnes Archaeological Survey.39 By positioning our survey methodology squarely in this (now well-established) tradition, we acknowledged the

validity of the tradition, and sought to produce data that will be, as much as possible, directly comparable to those recorded in other regions of Greece.

Yet in putting together the methodological package described below, we made a conscious effort to address critically some of the shortcomings we perceived in previous survey practice, and to develop methods that would work well on the Epirote landscape, though perhaps not elsewhere. A first area of concern was to create a program of geomorphological investigation that was more closely integrated with the intensive survey than was

typical at the time.40 Because southern Epirus contains a high percentage of erosional landscapes, it was essential to establish control on the movement of soils and sediments so that we did not misunderstand the depositional contexts of cultural material we encountered. Coarse-scale geomorphological mapping of the survey area was supplemented by fine-scale

analysis of the contexts of many sites and other locations of interest. In cases where built features were known or suspected, geophysical sur-

vey often augmented the results of surface collections. Coastal geomorphology was studied in the lower Acheron valley and on the northern shore of the Ambracian Gulf to measure the change in coastlines over time.

We were also convinced that not enough emphasis had been placed on the resolution of quantitative density data from off-site locations, al-

though the benefits of high-resolution data collection were not unknown.41 We agreed with Wandsnider and Camilli's recommendation to decrease survey transect intervals and overall survey pace as a way to ensure that both the low- and high-density surface records are documented.42 To pursue this objective, we designed a method of close-interval survey with high-

27

THOMAS F. TARTARON

resolution data recording that could be used, with simple modifications, for discovery-phase reconnaissance, investigation of rural sites, and urban

survey.

THE PURPOSE AND PLACE OF INTENSIVE SURVEY IN THE NIKOPOLIS PROJECT

In many ways, the intensive survey of the Nikopolis Project was different in scale and purpose from the earlier surveys in which it found inspiration. First, the cultural landscape of southern Epirus was investigated by a number of means, of which intensive survey was but one. Other methods by which activity areas were discovered and investigated included 1) extensive survey, comprising systematic but nonintensive "walkovers" (see below), scouting, and the independent Palaeolithic survey; 2) a wide-ranging season of ground-truthing of satellite imagery in 1991, during which known sites were visited and unknown sites were noted (but not investigated); 3) geomorphological studies, in which natural processes affecting sites and

landscapes were examined, and unknown sites sometimes found; 4) aerial balloon photography; 5) geophysical survey; and 6) documentary research.43 A consequence of this full, multidisciplinary program was that crew members were shared among the teams listed above, and additionally assigned to laboratory and data input tasks. Furthermore, the Nikopolis Project was conceived and operated as a field school intended primarily for under-

graduate students, requiring the senior staff and graduate assistants to en-

gage in many hours of instruction, and requiring students to spend substantial time learning other components of the overall project. Whereas no apology is offered for our research design or educational model, in practice these circumstances placed limitations on available person-hours, and most notably on the total territory covered by the intensive survey, which was also constrained by the deliberately intensive methods we employed (see below).

The principal purposes of the intensive survey were to test its feasibil- ity in the Epirote landscape; to reveal the overall characteristics of the region's archaeological resources; and to examine rigorously certain locations of particular prehistoric or historical interest, such as the lower Acheron valley and the Ayios Thomas peninsula (Fig. 2.1). It was not known whether the southern Epirote countryside would be as well suited to intensive surface survey as had been the southern Greek mainland and islands, where most such surveys have taken place. In fact, the climate and topography did present unusual challenges. The southern Epirote climate, transitional between Mediterranean and temperate, is characterized by a much higher annual rainfall than that of southern Greece and, consequently, the vegetation is more lush. The terrain is, on the whole, more rugged and mountainous, although such topography is by no means lacking in the south. Furthermore, the land is less developed agriculturally, so there is less open terrain. A significant consequence of these conditions was that surveying in large, contiguous blocks of tracts became difficult, and at times impossible. 43. See note 40.

28


Louros River

Acheron River

Parga X

:...,/,-, , astEphyra (Kastri

Phanari Bay:.: - , (Ammoudia) .Lower Acheron Valley

*- t.

Figure 2.1. Map of southwestern Epirus, showing locations of places mentioned in the text

Ionian Sea

0 5 10 15 20 25

*':- Grammeno c Rive

Nikopolis a: *.d,. Ayios Thom ' 'a "

.?. .: ."...-Peninsula ..-. :.. Ormos Vathy <K!..\ '_.Ambracian Gulf

Actium . .

5 KM

SAMPLING STRATEGIES

44. Hammond 1967; Dakaris 1971, 1972; Dakaris, Higgs, and Hey 1964; Higgs and Vita-Finzi 1966; Higgs et al. 1967.

45. Redman 1973.

The sampling strategies adopted for the Nikopolis Project survey were

shaped by certain environmental and operational constraints specific to the region and to the project. The most salient of these constraints were the enormity of the study area; a lack of previous systematic exploration; characteristics of the local terrain and vegetation cover; limits on time and available manpower; and specific research interests.

The Nikopolis Project study area comprised some 1200 km2, corre-

sponding roughly to the nomos of Preveza. This large expanse was chosen with a broad research agenda in mind, and not to be manageable for an intensive survey (see Chapters 1 and 7). While portions of the nomos (notably the environs of the ancient city of Nikopolis) have long attracted the attention of archaeologists and other scholars, no systematic exploration of the countryside had ever been undertaken. A scattering of sites was known as a result of chance finds by locals and the explorations of individuals and small research teams.44 Most of the countryside in this relatively large nomos was, however, archaeological terra incognita in 1991. It was thus a daunting challenge to devise a scheme by which a meaningful sample could be taken in the space of a few field seasons.

Our initial response was to develop, prior to the actual fieldwork, a multistage sampling strategy in which each phase of field research informed the direction and focus of subsequent phases.45 The entire study area was first stratified according to general environmental and cultural criteria. In

Arachthos

,7- vL I

29

,..,,k,

er

THOMAS F. TARTARON

keeping with the broad aim of the Nikopolis Project-to explain "the changing relationships, from prehistoric through mediaeval times, between humans and the land and resources they exploited"46-the first stratum consisted of preliminary estimates of environmental parameters, using topographic and geological information. In practice, this process involved the classification of the surface of the study area by features of topography or terrain (e.g., floodplains, low foothills, high elevations, swamplands, etc.), as hypothetical correlates of distinct environmental zones. It was noted that these categories also encompassed considerable variation in soil type and modern land use. The theoretical underpinning of such a strategy- that the full range of human activity in, and exploitation of, the environment can only be captured by taking a meaningful sample of a variety of environmental zones and landscape settings-is widely recognized and

applied in modern survey archaeology.47 The classifications were generated from topographic and geological maps and in the course of ground- truthing satellite imagery in the summer of 1991, and served as general guidelines to the range of environmental zones available for study.

A second stratum comprised the many specific research objectives related to cultural and historical phenomena (see pp. 8-9). One example is the basic inquiry concerning the relationship over time of the city of

Nikopolis to the surrounding countryside. In order to test the full range of

core-periphery interactions, it was deemed essential to sample locations both close to and at a distance from the limits of the urban area. For lack of information about the city's hinterland, the locations of known sites and the results of previous research were incorporated into the sampling de-

sign, which also included survey tracts chosen at random. The integration of environmental variables, prior research, and spe-

cific research objectives in the sampling design created a mosaic of potential survey areas. Locations identified by the different methods frequently coincided in terrain suitable for tractwalking, providing a starting point for placing survey units. A season of nonarchaeological reconnaissance in 1991, aimed at studying environmental parameters and modern land use, permitted an initial familiarity with the landscape and the varied terrains that survey teams would encounter. During the first season of archaeological survey (1992), diverse environmental zones and landscapes were investigated, field methods were adapted and refined, and promising locations requiring further attention were identified. On the basis of this experience, detailed sampling plans were drawn up for intensive work in 1993 and 1994 in specific areas of interest, e.g., the lower valley of the Acheron River (Table 2.1)48 and the Ayios Thomas peninsula. Once the survey work began for a given season, information gathered on scouting trips and by other components of the project allowed us to add suitable locations, elimi- nate those that offered little hope for results (for reasons of geomorphology, access, or ground cover), and set priorities for the fieldwork. Sampling designs were subject to daily modification based on current information regarding vegetation cover and accessibility, geomorphology, new discoveries, and a host of other factors. Thus, within each environmental stratum, some survey units were placed using judgmental criteria, while others were positioned where terrain permitted, without prejudice of prior knowledge.

46. Wiseman 1991, p. 1. 47. Schiffer, Sullivan, and Klinger

1978, p. 12; Wright et al. 1990, p. 604; Barker 1991, p. 3.

48. Tartaron 1996, pp. 384-390.

30


TABLE 2.1. STRATIFIED SAMPLE AND SYSTEMATIC SURVEY COVERAGE, LOWER ACHERON VALLEY, 1992-1994

TotalArea ModifiedArea Systematic Coverage Percent of Topographic Class (sq km) (sq km)* (sq km) ModifiedArea

Floodplain 53.40 0 0.014 -

Coast 6.85 6.85 0.631 9.21 Swamp/Bay 8.55 0 0.035 -

Foothills (<100 masl) 17.64 17.64 0.315 1.79 High hills (>100 masl) 22.56 22.56 0.155 0.69

Total 109.00 47.05 1.150 2.44

*This figure reflects the subtraction of areas determined by geomorphological analysis to preserve no premodern materials because of deep burial by recent alluvium, a finding confirmed by archaeological testing.

49. As explained in Chapter 3, the Palaeolithic survey, directed by Curtis Runnels, was a separate entity of the Nikopolis Project, with a sampling design and field methods quite distinct from those of the intensive survey.

50. See Barker 1991; Cherry, Davis, and Mantzourani 1991; Barker and Mattingly 1999. Modern landscape archaeology draws upon several older traditions, including human geography, the "New Archaeology," and landscape studies such as those of the British Royal Commission on Historical Monuments (Keay and Millett 1991, pp. 129-131). A sharp distinction is sometimes made between "landscape approach," reflecting the concerns of processual archaeology with ecological and geological system variables (Rossignol 1992, pp. 4-5; Rossignol and Wandsnider 1992), and "landscape archaeology," often associated rather narrowly with the postmodernist's historical and contextual focus (Roberts 1987; Yamin and Metheny 1996; Ashmore and Knapp 1999). But the distinction appears artificial, as both perspectives may be profitably applied to the study of regions, and in our usage both may be subsumed under the term "landscape archaeology."

51. Plog, Plog, and Wait 1978; Cherry 1983, p. 387.

Close consultation with geologists, geomorphologists, and members of the separate Palaeolithic surface survey optimized the ongoing selection of sample locations. Geological coring and geomorphological observation identified areas in which erosion and redeposition were so extensive that the discovery of remains of human activity in primary context was not anticipated; those landscapes were relegated to testing to confirm or refute the geomorphological findings. Information was shared to great mutual benefit with the Palaeolithic survey team,49 whose small size and extensive methods allowed it to range over large portions of the study area. Indeed, many late prehistoric and historical sites were first discovered by the Palaeo- lithic survey team, while important evidence for Palaeolithic activity was detected during intensive archaeological survey.

SURVEY INTENSITY AND COVERAGE

In recent years, the practice of surface survey in the Mediterranean has been integrated into the broader pursuit of landscape archaeology, a diverse array of approaches that has come to embrace both tangible (topography, environment, artifacts and features) and intangible (social action, symbolism, perceptions of space and place) aspects of living in the world.50 Surveys adopting this perspective naturally concern themselves with the landscape in its entirety and in every sense of the term; thus, all tangible manifestations of human activity, from major settlements to tiny scatters of pottery or lithic debris in the countryside, form part of the greater landscape(s) and are deserving of analysis. Intensive field techniques, fea- turing systematic coverage of the surface by fieldwalkers at close spacing intervals, emerged particularly as a way to recover more of the archaeological record, and have flourished as ideal methodological extensions of the landscape archaeology perspective. It has been demonstrated conclu- sively that increasing intensity yields a corresponding increase in the number of sites and scatters discovered,51 and for this reason alone intensive methods are consistent with the goals of landscape archaeology. But the enhanced power of resolution has not arrived without a concomitant cost in terms of overall surface coverage. Survey intensity and areal coverage are inversely related, a fact that presents theoretical and methodological

31

THOMAS F. TARTARON

difficulties that must be confronted in the planning of any regional project. The choice of optimal levels of intensity and coverage involves compromise between legitimate desires for a representative sample and survey precision (i.e., the power of resolution).52

This conflict is particularly acute in the case of unusually large, poorly known regions like southern Epirus. The response formulated by the

Nikopolis Project integrated extensive and intensive survey methods. This

path was chosen in recognition of the need for basic information, on the one hand, and a desire to initiate an intensive study of the human land-

scape of southwestern Epirus, on the other. Three modes of survey were applied to the surface archaeological

record: extensive nonsystematic, extensive systematic, and intensive (systematic). Extensive nonsystematic survey involved scouting, geomorphological evaluation coupled with archaeological testing, and the work of the Palaeolithic survey team, which employed a judgmental sampling design based on a predictive site model. Extensive systematic mode refers to systematic search, carried out in most respects like intensive survey, except with fieldwalkers arrayed at intervals of greater than 15 m. The two modes of extensive survey were used to reveal the overall characteristics of the

region: the number of sites, their distribution, chronology, function, and

relationship to the environmental context.53 Simultaneously, a program of off-site, intensive survey was carried out in locations of particular research interest and across environmental zones in order to reveal patterns of human activity of every description over the entire landscape. A primary concern in designing the intensive survey was the acquisition of data comparable to those collected elsewhere in the Mediterranean area. Quite different

approaches have been taken by other projects facing similar circumstances.54 In the neighboring province of Aetolia, a recent survey employed extensive techniques that allowed for the plotting and dating of relatively obtrusive sites over a large area, but systematic collection of information on site densities and off-site phenomena was beyond the scope of the method.55 For that project, extensive methods were consistent with the stated research aim of gaining an overall impression of human settlement in the

region. In an intensive survey, total areal coverage is determined by a series of

variables, only some of which can be controlled.56 The principal determi- nants are labor input, measured by crew hours expended, and survey in-

tensity, measured by spacing between crew members. Other variables include crew experience and the complexity of field operations. If the study area is relatively small, or if a large crew can be assembled and maintained, the pressure to attain adequate surface coverage is eased considerably. But the amount of labor that can be dedicated to survey is often fixed by finan- cial constraints that cannot be closely controlled, with the result that survey intensity is frequently the critical consideration in decisions concerning areal coverage. In the case of the Nikopolis Project, neither the size of the study area nor the labor contribution could be dictated exclusively by the interests of the survey. Although surface survey was the activity to which the greatest commitment of resources was made, crew members were shared with teams engaged in geology, geomorphology, aerial pho-

52. Wandsnider and Camilli 1992, p. 170.

53. Cherry 1983, p. 393. 54. Rutter 1993, table 1, with

references. 55. Bommelje and Doom 1987. 56. Many of these variables are the

same ones regarded as factors influencing site discovery; see, e.g., Schiffer, Sullivan, and Klinger 1978, p. 4; Cherry 1983, p. 397; Barker 1991, pp. 4-5.

32


TABLE 2.2. TYPICAL DAILY WORK ASSIGNMENT, JUNE 28, 1994

SURVEY TEAM I: LOWER ACHERON

VALLEY

Tom Tartaron (team leader) Brenda Cullen (assistant)

Jennifer Murray Yaz Shimizu Anne Maxson Aviva Figler

SURVEY TEAM 2: KASTRI, LOWER TOWN

Carol Stein (team leader) Alan Kaiser (assistant) Alesia Alphin Karla Manternach Katerina Dakari Alison Spear T.J. Reed

TOPOGRAPHIC SURVEY: KASTRI

Theo Chatzitheodoros

Joseph Nigro

PALAEOLITHIC SURVEY TEAM

Curtis Runnels Priscilla Murray Tjeerd van Andel

57. This figure has been arrived at in experimental situations: Wandsnider and Camilli 1992; Bintliff and Snodgrass (1988b, p. 58) use a figure of 2.5 m for the Boeotia survey.

58. Whether such scatters are actually discovered depends on a host of variables that relate to intrinsic properties of the archaeological record itself, survey strategies, and human factors: Wandsnider and Camilli 1992.

GEOLOGICAL CORING: ACHERON VALLEY

Zhichun Jing Mark Besonen

Stephan Papageorgiou Amy Graves

ARTIFACT PROCESSING AND INVENTORY TEAM

Lucy Wiseman

Betty Banks Mike Gaddis Lisa Davis Leslie Harlacker

LITHICS: Lia Karimali

COMPUTER DATABASES AND DIGITIZING

Rudi Perkins (students from artifact processing team)

PHOTOGRAPHY: ARTIFACTS AND KASTRI

Michael Hamilton

GEOPHYSICAL PROSPECTION: KASTRI

Apostolos Sarris Dimitra Papagianni Mely Do

Kathy Montgomery

tography, artifact and data processing, and other operations (Table 2.2). The total labor pool from season to season depended largely on the number of undergraduate students in the Nikopolis Project field school (1992: 12; 1993: 24; 1994: 16), along with a few graduate students each

year from several different institutions. Ultimately, therefore, the balance of intensity and coverage could best be influenced by field procedures, and most directly by the spacing interval chosen and the complexity of field methods.

Variability in spatial coverage is a systematic and implicit consequence of the spacing interval that must be addressed in the framework of objectives for data collection and analysis. It is expected that each surveyor will have an effective visual range of 1-2 m on either side of the survey transect.57 At 15-m spacing, therefore, a swath of 11-13 m is unexamined, with the result that scatters or other findspots with axes of 10 m or less perpendicular to the survey transect may not be discovered; at this spacing interval, an average ground coverage of 17% (assuming teams of 4-6 surveyors) is achieved. An 8-m interval may preclude the loss of any manifestation with a perpendicular axis of 5 m or more, and yields an average ground coverage of 32%.58 In view of the availability of extensive reconnaissance as a comple- ment to systematic, intensive survey, it was decided that a spacing interval of 5 m would be adequate to provide precise and statistically quantifiable data concerning the location, density (both on- and off-site), chronology, and range of human activity within the study area. A suite of investigative

33

THOMAS F. TARTARON

procedures was then developed to measure these target variables. In three

seasons, averaging 25 field days and 15 crew members, approximately 2 km2 were investigated intensively, and an additional 3 km2 received extensive systematic coverage. The area covered by extensive nonsystematic in-

vestigation is estimated to be just over 100 km2.

FIELD METHODS

The surface archaeological record of southern Epirus preserves evidence of a wide range of human behavior, from minute, single-event loci to large, fortified cities and towns. To facilitate the effective treatment of such diverse phenomena, procedures were required that would permit survey teams to observe and record information at very different scales of complexity. 59 Three separate areal units of survey were developed, each reflecting a different purpose and a different level of intensity: the tract, the site/scatter, and the walkover. A unique field form was created for each to ensure the full recording of data, and to be compatible with the project's computer database.

TRACTS

The tract was the basic unit of investigation for off-site, intensive survey, in which landscapes of unknown, but presumably relatively high, site potential were explored. The tract was defined as a parcel of land of varying size, the parameters of which were determined by physical boundaries, such as field borders, fences, or roads; natural features, such as topographic contours; or by some maximum size guideline.60 The terrain of the tract

normally exhibited relative uniformity of vegetation, visibility, and modern land use.61 In open terrain, tracts were normally of rectilinear shape, but features of terrain and topography often imposed unusual outlines, as for example a doughnut-shaped tract wrapping around the lower slopes of a hill. Tracts were walked by small teams, typically of five or six members. Experience has shown that teams of smaller or larger size tend to be less efficient in gathering and reporting data. Each team included a team leader, a graduate student assistant, and three to five undergraduate field school students. The level of experience among the students varied

widely; some had participated in several archaeological projects, while others had no experience at all. Team leaders were charged with overseeing the survey process and gathering critical environmental and archaeological information, and therefore rarely walked survey lines themselves. For purposes of record-keeping, each tract was designated with a T for tract, followed by an accession number, sequential by the year in which the tract was walked (e.g., T93-20 refers to the twentieth tract of the 1993 season).

A tract was begun by lining up the crew at the desired spacing interval-the standard interval was 5 m, although owing to diversity of local conditions, intervals as small as 3 m and as large as 8 m were observed- and proceeding through the tract in parallel transects. Team members were

59. Considerable trial and error were required before field methods were solidified. The procedures described in this section were essentially in place by the middle of the second season: Wiseman 1993a, 1993b; Wiseman, Zachos, and Kephallonitou 1998.

60. Tract size guidelines for recent intensive surveys include the following: Cambridge/Bradford Boeotia Survey, 0.6-0.9 ha (Bintliff 1985, p. 201); Nemea Valley Archaeological Project, no more than 1 or 2 ha (Wright et al. 1990, p. 604); Berbati-Limnes Archaeological Survey, average 3.75 ha (Wells, Runnels, and Zangger 1990, p. 214). In the survey of northern Kea, tract size was determined by field walls enclosing groups of agricultural terraces (Cherry et al. 1991, p. 22). A size limit of 2 ha was observed in our survey, but the average unit size was ca. 0.50 ha.

61. It is recognized, however, that tracts defined by modern features may comprise multiple units of geomorphological deposition. The definition of survey tracts as discrete units of deposition is an innovation of the Eastern Korinthia Archaeological Survey; see description at http:// eleftheria.stcloudstate.edu/eks/ methodol.htm.

34


trained to call out to each other simple observations about artifacts encountered in their transects; calls such as "Sherd!," "Tile!," or "I have a concentration of lithic debris over here!" were typical. This information alerted the team leader and other crew members to the density and distribution of artifacts, and encouraged fellow surveyors to maintain a high level of concentration.

Each surveyor carried a tally counter or "clicker" on which to record the quantity of artifacts, clicking once for each artifact noted. Whereas all artifacts were counted, only those considered diagnostic were collected from the surface. Generally, this meant all pottery fragments except un- decorated body sherds (though fabric samples of the latter were collected if not represented among the diagnostic pieces), a sampling of bricks and tiles, and all other artifacts. Items were retained in plastic sample bags until survey of the tract was completed; a subset of the objects was then taken for inclusion in the project's contextual collection.

To provide close control over the spatial distribution of artifacts in the tract, artifact counts were recorded for each category of item (pottery sherds, bricks, tiles, lithics, other) by the team leader or the graduate student assistant at 30-m intervals. This simple expedient yielded grids showing variations in artifact density by artifact type, providing a wealth of comparative information for off-site locations. The task of keeping separate counts of different classes of artifact was complicated by the fact that tally counters display a single total, and was especially difficult when a wide range of material was represented in the tract. A few simple innovations were developed to preserve the integrity of the data, while decreasing memory demands on the crew. In the overwhelming majority of tracts, artifacts of a single material, i.e., pottery, architectural ceramics, or flaked stone ob-

jects, dominated the surface assemblage. In those cases, the usual procedure was to use the tally counter for a total artifact count throughout the tract, while simultaneously keeping a mental count, for each 30-m block, of all objects not falling into the dominant class.

A simple example will illustrate the implementation of this procedure. At the 90-m mark of a given tract with a preponderance of pottery sherds relative to other objects, the team leader first requests of each surveyor a total artifact count, which is read off the tally counter. Hypotheti- cal surveyor a, whose total count at the previous check point (60 m) was 45, reports 65 artifacts, or 20 artifacts in the current 30-m segment. Next, the team leader asks for a count by type of all objects other than pottery in the current 30-m segment only. Surveyor a reports 5 tile fragments and 2 lithic flakes. Simple subtraction reveals that 13 pottery sherds were counted by surveyor a between the 60- and 90-m points. The tally counter is not reset until the tract is completed, and those walking survey transects need only retain mental counts for 30 m at a time. Inevitably, tracts were encountered in which two or even three artifact types were abundant; typically, these were locations in which historical-period sherds, tiles, and bricks littered the surface. In those instances, a second method was employed in which mental estimates of the relative percentages of abundant artifact types were maintained and reported at 30-m intervals. To continue the example, surveyor a reports 120 total artifacts at 90 m, 40 more than the

35

THOMAS F. TARTARON

Rev. June 30, 1993

The Nikopolis Project Archaeological Survey Tract Form

Tract# Date I Recorded by I

1/50,000 Map I I 1/5000 sheet I I Elevation I

Tract Size I Visibility (1-10) - Spacing Interval (meters) l=poor; 10=excellent and Direction

Walking Order I IGPS Reading

Associated Site/Scatter #s Sample #s

ARTIFACT COUNTS: Other (specify): Sherds Rooftiles Flaked Stone Ground Stone Metal

Notebook Refs Photographs

Inventoried Artifacts

Sketch map of tract location. North should always be up. Include locations of adjacent roads, villages, known sites, contiguous tracts, etc. as reference points.

Figure 2.2. Archaeological survey tract form

80 reported at 60 m, and estimates that, over the past 30 m, 60% of the artifacts were pottery sherds and 40% fragments of rooftiles. Approximate values of 24 sherds and 16 rooftile fragments are thus obtained for a 30-m segment of a single survey transect. Obviously, the first method is preferable, and was applied wherever possible. These methods of recording artifact information at close intervals, while time consuming, were simple to implement, and allowed a remarkably full accounting of off-site artifact distributions.

Additional passes, executed in the same manner as the first pass, were required when the parcel of land selected for the tract was larger than the team could cover in a single pass. Once the walking of the tract was finished, the team assembled to perform critical documentation activities: the preprinted tract form was filled out, the tract plotted on a 1:5,000 topographic map, and the artifacts processed.

The information requested on the tract form (Fig. 2.2) summarizes the location and size of the tract, several environmental variables, details

36


of the survey procedures used, and the results obtained. Visibility of the ground surface, a critical variable in assessing survey results,62 was discussed by the crew, ranked from 1 (poor) to 10 (excellent), and entered on the form. Also at this time, the team leader plotted the tract on the appropriate 1:5,000 Greek army topographic map, black-and-white and color photographs of the tract were taken, and a locational reading (in UTM and latitude/longitude) was taken with a hand-held global positioning system (GPS). The data recorded by the GPS serve as a check on the manual plotting of the tract, and are ideal for use in geographic information systems (GIS) applications. Finally, the artifact samples were sorted by type and a subset was selected. It was often not desirable to retain all artifacts collected from the tract; for example, redundant rooftile fragments and duplicate fabric samples were frequently discarded. Once the sample was decided upon, sample bags with wooden tags recording contextual information were prepared for each of the various materials represented (pottery, tile, brick, etc.). The team leader supplemented all of these activities by including in the field notebook a multitude of observations regarding field conditions, survey procedures, geomorphology, artifact patterning and chronology, and so forth.

SITE/SCATTERS

62. See discussion in Ammerman 1993, pp. 369-371.

63. Wilkinson 1982; Dunnell and Dancey 1983; Bintliff and Snodgrass 1988a; Barker 1991, pp. 5-6.

64. Thomas 1975, pp. 62-63; Dunnell and Dancey 1983, pp. 271- 274; Wright et al. 1990, p. 603.

65. Gaffney and Tingle 1985, p. 68; Cherry et al. 1991, p. 21.

66. Doelle 1977, p. 202. 67. Thomas 1975; Foley 1981. 68. Cherry 1984, p. 119. 69. Hope Simpson 1984; Schofield

1991a.

In the course of walking the countryside, survey teams frequently encountered anomalously dense scatters of archaeological material, or isolated but recognizable architectural features, such as sections of wall or agricultural installations. The recognition, investigation, and classification of these concentrations reflect a project's theoretical orientation toward the spatial aspects of human behavior, and the ways in which behavior is preserved in surface deposits. The traditional concept of the "site," easily recognizable by a dense clustering of artifacts and definable spatial limits, has been found inadequate to encompass the full range of human activity. One con-

sequence of the reevaluation of the site concept was the emergence and

development of intensive survey techniques, which forced a rethinking of the spatial implications of human activity as ever smaller and less clustered loci came to light. Interest arose in the meaning of low density or "off-site" scatters,63 and ultimately the individual artifact, rather than the site, was designated as the basic unit of analysis.64 Sites were defined and redefined in relative terms (e.g., as density peaks against a background of artifacts

spread across the landscape);65 in absolute terms (e.g., number of artifacts per square meter);66 or rejected altogether.67 A great achievement of intensive survey has been the development of approaches to the study of the kinds of activities that took place largely outside the confines of traditional settlements, among them hunting, pastoralism, agriculture, and tool manufacture. It remains true, however, that identification of off-site scatters on the landscape can be as much an interpretive as an observational exercise,68 and that the inference of specific behaviors from them is usually problematic.69

Ultimately, each survey develops its own approach to the concep- tualization, recognition, and treatment of surface concentrations. Follow-

37

THOMAS F. TARTARON

ing Cherry and colleagues, we perceive the surface archaeological record of a given landscape as "a variable distribution of residue from past cultural activities, in some places dense, in others less so."70 While allowing wide berth to evidence for nonsettlement activity, this perspective does not es- chew the importance of the traditional settlement. The concept of the site/ scatter (read "site or scatter") formulated for the Nikopolis Project survey was similarly intended as a nonjudgmental way of referring to a wide range of material phenomena, and in using it we sought to avoid rigid, usually unworkable definitions of the "site." In our system, the term site/scatter referred to a spatially definable locus of past human activity characterized by high artifact density relative to the background material distribution, and embodied almost any identifiable, concentrated evidence of cultural activity, as the following examples demonstrate: a small scatter of flaked stone; a single, isolated olive mill; the entire ancient city of Nikopolis; or an exposed stratigraphic profile.

Picking meaningful artifact concentrations out of a background of scattered material can be difficult indeed. It requires an experienced team leader and crew attentive to subtle changes in the type, chronology, and density of material relative to the surrounding landscape. At times the recording procedures described above alerted the team leader to the presence of potentially significant patterns, but often the deciding factor was the ability of the team leader to monitor continuously the finds and their contexts of discovery. From time to time, potentially significant concentrations became apparent only after density data were assessed and finds were inspected. In such instances, the relevant tracts were earmarked for subsequent revisits. Although concentrations of very high density in absolute terms were almost always recognized as sites, lower-density scatters were evaluated relative to the surrounding landscape. This evaluation was done in recognition that certain behaviors produce fewer and less clustered remains (or remains that are not expected to survive in the archaeological record), and that the quantity of durable artifacts available for preservation varies significantly for different periods of the past.7"

Investigation of sites and scatters was guided by the observation that different scales and natures of surface concentrations require different methods of documentation; that is, survey teams must find ways to deal consistently and effectively with sites and scatters that range widely in size and complexity. Extremely small concentrations, for example localized patches of artifacts exposed by the bulldozing of a dirt road, may be measured, documented, and collected in their entirety. On the other extreme, large, multiperiod sites, such as the fortified citadels of southern Epirus, present very different challenges of scale and complexity; clearly, the application of total measurement and collection to large settlement contexts is not practical, and the examination of a sample of the total site area may be necessary. Because most site/scatters occupied points on a con- tinuum between these two extremes, methods falling between total collection and sampling portions of sites were desirable for the vast majority of occurrences.

70. Cherry et al. 1988, p. 159. 71. Millett 1991.

38


72. See Shott 1995.

SITE/SCATTER DOCUMENTATION AND COLLECTION

PROCEDURES

Site/scatters of moderate size (up to 100 m on the shortest axis) were normally documented immediately upon discovery, or in a matter of days thereafter. The most compelling justifications for prompt action are the uncertainty and potential loss of data that delay may introduce. It is never prudent to assume that a surface scatter will be available for study in its present condition next year, or even next week. A common problem concerns the growth of vegetation and crops during the field season. Artifact concentrations that are obtrusive on plowed, newly planted fields become invisible in a matter of a few weeks. The changes in vegetation that occur from one year to the next can be even more dramatic. The thorough, intensive surface study carried out in 1993 at Ormos Vathy (see Fig. 2.1), the probable main port of the city of Nikopolis, would have been impossible in the summer of 1994 as a result of heavy winter rains that fostered the growth of dense, impenetrable vegetation over much of the bay area. Smaller artifact scatters can be washed away or otherwise lost at any time if they are particularly fragile or ephemeral.

Human factors also argue for prompt documentation. Since designation as a site/scatter is often a rather subjective decision, it is preferable that the persons making the judgment also carry out the investigation on the basis of the patterns that are perceived at the time. Patterns that are clear when one is focused on a given landscape become blurry with the passage of time; revisits are especially difficult if key members of the discovery team are not present. Furthermore, given the constraints of time and the uncertainties of carrying out archaeological research, one must never assume that a concentration of surface artifacts will be available for study in the future; as a result, the amount of information obtained at the time of discovery should be maximized. This policy in no way diminished the importance of planned site revisits, which were employed to review and expand the original documentation.

Typically, as soon as the team leader suspected the presence of a site/ scatter, the tract was stopped, and preliminary investigation was begun. The team first determined the approximate, or notional, boundaries of the artifact scatter (bearing in mind that the limits of a surface scatter may change with each new document produced).72 This was normally accomplished by having all surveyors walk outward in different directions from an estimated "center" of the scatter, and place a flag in the ground at the point where the anomalous density of material appeared to cease. As this task sometimes required rather difficult judgments, the team leader and graduate student assistant scrutinized the work closely. As a general rule, we felt it was better to overestimate the dimensions slightly than to under- estimate them; quantitative data collected as part of the site/scatter may point to the need to rein in the boundaries, whereas data that are not documented as part of the site/scatter tend to blend with the off-site material, making it difficult to establish wider boundaries at a later time.

39

THOMAS F. TARTARON

Once the site/scatter was outlined, the tract procedure was modified somewhat. The tract was continued, including that part of it falling within the site/scatter, but within the area of the site/scatter, artifacts were counted as normal but not collected. The reasons for this modification are as follows. Artifact counts were continued at the normal intervals in order that the density data would be comparable to those of all other tracts. Artifacts were not collected within the confines of the site/scatter during the walk-

ing of the tract, however, because if they were, our ability to exercise spatial control over the site/scatter as a separate entity (in some cases, extending beyond the present tract) would be compromised. In essence, then, two

entities-separate but intimately associated-were created. The Nikopolis Project recording system ensured that associated tracts and site/scatters remained linked in the project's database.

Upon completion, the tract was processed as usual. If the site/scatter fell completely within the tract, it was then investigated in detail before other tracts were begun. If the site/scatter extended beyond the present tract, additional passes were normally walked until the site/scatter was cir- cumscribed. For very large sites, on which it was not practical or desirable to envelop the entire dense scatter of material in a single tract, several tracts were placed to encompass the site and the off-site territory sur-

rounding it. After finishing the documentation of the tract, the team turned to the

investigation of the site or scatter. Very small scatters were subjected to total measurement and collection, and the largest sites, usually known previously, were deferred for investigation using "urban survey" techniques (see below). For all but the smallest and largest concentrations, a method

very much like tractwalking, but more intensive, was applied. Surveyors were arrayed at closer intervals, usually 3 m, and walked survey transects

through the site/scatter. Artifact counts were recorded (by type) at 10-m instead of 30-m intervals, and diagnostic artifacts were now collected. In certain experimental cases in which spatial control was crucial, separate samples were taken for each 10-m (or even smaller) block, by team or even by individual surveyor. In all cases, descriptive information about patterns in the artifact distribution was recorded by the team leader in the field notebook. Detailed information was entered on a field form similar to that used for the tract, and the site/scatter was designated in a like manner (e.g., SS93-5 refers to the fifth site or scatter discovered in the 1993 sea-

son). The samples taken from the site/scatter were associated with, but also fully distinct from, the tract(s) from which they issued.

SPATIAL RELATIONSHIPS OF TRACTS, SITES, AND

SCATTERS

For several reasons, the spatial relationships between tracts and site/scatters may be simple or they may be quite complex: tract limits were arbi- trarily chosen, usually on the basis of modern features; the potential for sites in a given tract is usually unknown; and evidence for many forms and levels of concentrated human activity may be present in a small area. Hy- pothetical example A (Fig. 2.3) is the simplest relationship: a single site/ scatter appears completely within the confines of a tract. Moderately greater

40


A

Site / Scatter

Tract 1

B

Site/ Scatter

Tract 1 Tract 2

Figure 2.3. Examples of spatial relationships between tracts and site/ scatters: A) site/scatter falls completely within one tract; B) site/ scatter extends over more than one tract; C) locations of survey units at Ormos Vathy, showing several tracts and site/scatters within SS93-8, the Roman-period harbor settlement

S / .SS 93-3

SS93-2

SS 93-7 ? . . .. ..

. SS 93-17

.. . . .

.3

. SS^ ^ .-;:.-^.-.;- :.,

ss93-8. . : :

Roman -... ..15. ...

harbor town '

2 ::

::.

,. : : : . . . 0 200 400 r

complexity is introduced if, as in example B, one or more site/scatters extends over the surface of several tracts. Example C, showing the actual

survey locations at Ormos Vathy, illustrates the spatial complexity that

frequently exists between concentrations of cultural debris and the survey units that are superimposed on them. Our investigations at Ormos Vathy in 1993 confirmed the suspicion that this was the main port for the city of

Nikopolis in the Roman period. Through scouting and tractwalking, the

approximate limits of the main part of the port town were determined, and the entire enclosed area was designated a site (SS93-8). In the numerous tracts that were walked within the site, several smaller loci of important, concentrated activity were discovered, including collapsed architecture from buildings that stood on the water's edge (SS93-5), domestic

assemblages, a probable purple dye factory (SS93-23), and an inscribed tombstone (SS93-35). As a result, several site/scatters were designated within the larger site/scatter. We found this system of designation (i.e., site/scatters within site/scatters) a convenient way to portray the spatial complexity of human behavior, without having to create new terms or procedures.

URBAN SURVEY

73. Bintliff and Snodgrass 1988b; Alcock 1991.

In recent years, a suite of methods based on the tract concept has been

developed for surveying the surface of large, complex sites such as the urban areas of classical Greekpoleis.73 It was realized that a failure to sur-

vey the large settlements that often anchored regional systems amounts to an inversion of the tunnel vision for which traditional excavation has been justly criticized, and leads to regional studies that are similarly incomplete. The need for surface survey on large settlements that have been previously excavated is no less urgent; even the most extensive excavations usually

. .. . .

4I

THOMAS F. TARTARON

examine only a small percentage of the total site area.74 Intensive survey techniques similar to those developed for off-site contexts offer a means by which the complete extent of complex, multiperiod settlement sites may be examined for overall quantitative information and detailed patterns of internal periodization.75 An additional benefit is the potential comparability of the data to those obtained from off-site contexts.

Previous urban surveys in Greece have relied on the replication of tract procedures, with the addition of supplemental grab samples and/or controlled collections, as a way to effect representative artifact samples and near-total ground coverage.76 The methods adopted by the Nikopolis Project differ from these in that instead of adding supplemental collections, tract procedures were simply intensified, creating smaller sample contexts and resulting in enhanced control over artifact and density information. Tracts were laid out in the same fashion, and surveyors were arrayed at an interval of 5 m. Artifacts were counted and collected according to the same guidelines used for off-site tracts, with counts recorded at 30- m intervals. Unlike off-site tracts, however, a sample was routinely taken at each 30-m interval. This modification created sample cells of approximately 30 x 25 m (depending on the number of surveyors). Documenta- tion procedures were the same as those observed for tracts. The data obtained from urban survey permit the construction of density maps for each period of occupation, and offer the possibility of discerning localized functional attributes of cells or groups of cells.

Certain innovations were introduced to accommodate the overwhelming density of material often encountered in urban contexts. Surveyors were provided with two tally counters for recording quantities of the most common classes of artifact: pottery sherds and fragments of tile and brick. Since all examples of other classes of material were collected, they were counted when samples were assembled. Urban sites often preserve architectural features, including clear remains of houses and other buildings, wells and cisterns, and towers. In another innovation, team leaders were allowed to treat such features as unique tracts, regardless of their size or shape. They could then be investigated with the same methods used on other urban tracts or, if deemed appropriate, subjected to total collection or sampling at closer intervals. By separating such contexts, the integrity of artifacts that may form coherent and meaningful assemblages could be preserved.

Urban surveys were initiated at a small number of fortified town sites in 1994. The most comprehensive of these was at Kastri, in the floodplain of the lower Acheron River (see Fig. 2.1). Dakaris identified Kastri as the site of the Elean colony of Pandosia, mentioned by Strabo (6.1.5) and others.77 Intensive survey was carried out on 11 ha of this 33-ha site, representing most of the accessible and walkable terrain. More than 85,000 artifacts were counted, including over 15,000 pottery sherds and over 70,000 tile and brick fragments, yielding an average figure of more than 7,700 artifacts per hectare. A topographic survey carried out at the same time as the urban survey generated a new site map, incorporating several architectural features that were discovered in survey tracts. The results of the urban surveys at Kastri and elsewhere will be presented in volume 2.

74. Alcock 1991, pp. 422, 443. 75. Bintliff and Snodgrass 1988b,

p. 58. 76. Bintliff and Snodgrass 1988b,

pp. 58-59; Alcock 1991, p. 448. 77. Dakaris 1971, p. 164. For

further discussion of the identification of Kastri, see Chapter 6.

42


WALKOVERS

In our terminology, walkover was a multipurpose designation referring generally to any systematic reconnaissance that was conducted differently from regular tract or site/scatter practices. The main function of the walkover designation was to provide documentation in a standard format for activities that lay outside the methodology of systematic, intensive survey, in recognition of circumstances in which treatment in the manner of a tract was impractical or undesirable. Walkovers documented the following kinds of investigative activity: extensive systematic off-site survey, extensive nonsystematic survey (scouting), site revisits, and resurvey of tracts. Extensive systematic survey, defined as any investigation carried out at a

systematic spacing interval of greater than 15 meters, and exploration of territories by nonsystematic means (scouting), formed integral elements of the Nikopolis Project sampling design that allowed the rough estimation of local environmental and cultural features, often in advance of placing intensive survey units (see the discussion of sampling above). System- atic walkovers were mapped, photographed, and documented on a walkover field form similar to the tract form; nonsystematic walkovers were documented by narrative notes, with maps and photographs where appropriate. Additional visits to previously investigated units (e.g., site revisits and tract resurveys) were also designated walkovers. Tracts that were walked for a second or third time were documented as walkovers, even if tract

procedures were used; the artifacts and documents associated with the walkover were linked to all previous investigations of the tract.

SITE REVISITS AND RESURVEY

78. These teams are often known as "verification" teams, a term I find dissatisfactory because of the implication of a group of experts that functions in part to correct erroneous observations made in the discovery phase (though this is certainly not out of the question). In many cases, valid observations made during the initial visit will no longer be justified at the time of the revisit. An important role of the revisit team is therefore to measure the changes that have taken place in the surface scatter.

79. Ammerman 1981, 1985, 1993.

The information acquired in the fieldwalking phase was supplemented by comprehensive follow-up studies of selected tracts, site/scatters, and other locations of interest on the landscape. This second phase was normally initiated by revisit teams composed of archaeologists, geologists, and geomorphologists.78 The initial purposes of the revisit were to check the information supplied by the discovery team (e.g., surface visibility, dimensions, chronological periods represented); to describe the geomorphological setting; and, if necessary, to make additional observations and collections. Revisit teams also visited locations not designated as sites or scatters by the discovery team, but deemed worthy of reconsideration based on analysis of the density data and the artifact samples. These activities were documented as walkovers, and were accompanied by specialist reports as necessary. Additions and alterations were also made to the original tract and site/scatter documents.

Based on the findings of the discovery and revisit teams, certain units were selected for further analysis involving resurvey or geophysical prospection. Resurvey of a small percentage of tracts was a planned activity that allowed archaeologists to measure the effects of changing conditions of discovery brought about by natural processes (erosion, deposition, and vegetation growth) and human actions (such as agriculture and land clear- ance).79 Since conditions of access and surface visibility may exercise a

43

THOMAS F. TARTARON

Figure 2.4. General view of the site at Grammeno (SS92-6). View is from the east end of tract 93-1, the designation in 1993 for the same area surveyed the previous year as tract 92-39.

profound influence on the archaeological document that is produced for a given landscape, resurvey results served to qualify and inform inferences drawn from surface data.

Geophysical prospection was another integral component of comprehensive site investigation. Remote-sensing techniques, including resistivity, electromagnetic conductivity, magnetometry, and ground-penetrating radar, were applied to a number of sites, some previously known and others discovered by the surface survey. Geophysical surveys were often performed on sites with high artifact densities but no visible architectural remains as a means to test for subsurface structural features. In several cases, the likely existence of buildings, roads, and other features was established.

An example from the locality "Grammeno," near the modern village of Archangelos, illustrates the typical manner in which such detailed studies came together. In the course of walking tracts in the Grammeno area in 1992, a fallow field walked as tract 92-39 was found to contain massive amounts of pottery sherds and brick and tile fragments, as well as other materials, including ground stone, flaked stone, and bits of glass, limestone, and metal slag (Fig. 2.4). Over 17,000 artifacts were counted in the field, which measures ca. 1 ha in area; this quantity contrasted sharply with that encountered in the surrounding tracts. The field was immediately designated a site (SS92-6). Later in the 1992 season, SS92-6 was resurveyed, resulting in an additional collection of diagnostic artifacts. Pre- liminary analysis suggested a domestic assemblage of Roman times, indicating the presence of perhaps a farmstead or small villa.

In the winter of 1992-1993, it was discovered that the fallow field had been freshly plowed, bringing to the surface considerable quantities of new material. Consequently, the site was again surveyed at the beginning of the 1993 season, this time yielding a remarkable total of over 36,000 artifacts, among them many diagnostic types not previously collected there.80 It was noted that the dimensions of the definable scatter had not increased,81 but that the chronology and function of the site were clarified. Later in the

80. Although a portion of this twofold increase in artifacts is ac- counted for by the further fragmenta- tion of surface objects during plowing, the recovery of numerous new diagnostic ceramic, glass, and stone artifacts indicated clearly the richness of the subsurface deposits.

81. We cannot be certain of this finding, however, because the adjacent field most likely to harbor an extension of the site remained heavily overgrown throughout the period of our investigations.

P

44

X, 4.,

:'"': :':.. :: . ':': . . ? '

' * --

" ' ^ -

* * ; - *


field season, a geomorphological profile was made of the environs of SS92-6. The site was also selected for geophysical survey, owing to the dense, well-delimited scatter; the suspicion that broken bits of limestone derived from architectural blocks; and the general amenability of the site to geophysical techniques. Electrical resistivity, electromagnetic conductivity, and magnetometry surveys were performed on grids laid out over portions of the site. The preliminary results indicate the existence of one or more structures and at least one linear feature, and accord well with the inferences made from the surface remains.82

Very different conditions prevailed at SS92-6 in the summer of 1994. Heavy winter rains and the return of the field to fallow caused the ground to be obscured by scrubby vegetation. It was also noted that a substantial reburial of artifacts had taken place, probably as a consequence of earth movement during pluvial runoff. A resurvey of the tract in 1994, in circumstances of dramatically reduced visibility, produced a count of 3,000 artifacts, far fewer than in either of the previous two years.

The history of investigation at SS92-6 is instructive on many levels. From one perspective, it is a testament to the amount and quality of information that can be recovered by intensive, multidisciplinary study of surface phenomena. To another point of view, it must stand as a caution that the vagaries of time and nature can have a profound effect on our interpretations of surface patterns, and as an affirmation of the need for replication studies. Had we first discovered this site in 1994, our perception of it would undoubtedly be quite different.

CONCLUSION

82. Weymouth 1993. 83. E.g., Moore 2000; Tartaron

1996, 2001; Tartaron, Runnels, and Karimali 1999.

In response to a unique set of research conditions, we chose a strategy calling for both extensive and intensive reconnaissance methods. The former permitted us to acquire, on a coarse scale, information about the diverse landscapes of southern Epirus, and the distribution of cultural remains upon them. The latter were undeniably of high intensity in terms of spacing interval and the number of data measurements, and were intended to furnish high-resolution data about a smaller sample of locations within the survey area. We willingly sacrificed broad areal coverage in the belief that the total spectrum of approaches would provide a good initial understanding of the region's past through an exceptionally long expanse of time.

The success or failure of the intensive survey will become evident as analyses continue and the results are held up to the scrutiny of the scholarly community and those who follow us in the study of the Epirote landscape. Detailed interpretive studies, making use of the surface data obtained so meticulously, have begun to appear and will continue to do so, both in volume 2 of this series and elsewhere.83 We are reminded that decisions in survey design and execution are very much about compromise among competing priorities, and balancing opportunities against constraints. The final evaluation must consider, among other things, the degree to which the intensive survey served the overall aims of the project.

45

CHAPTER 3

THE EARLY STONE AGE

OF THE NOMOS OF PREVEZA:

LANDSCAPE AND SETTLEMENT

by Curtis N. Runnels and Tjeerd H. van Andel

INTRODUCTION

1. Dakaris, Higgs, and Hey 1964. 2. Higgs and Vita-Finzi 1966;

Higgs et al. 1967. 3. Bailey 1992. 4. For Boila and Klithi: Bailey et al.

1999; Kotzambopoulou, Panagopoulou, and Adam 1996.

The first extensive and methodical search for Palaeolithic sites in Greece was begun under the direction of Eric Higgs of the University of Cam-

bridge in 1962.' Palaeolithic finds had been made there before as the result of other activities, but Higgs's survey was aimed specifically at identi-

fying Palaeolithic sites and cast a wide net over nearly all of northern Greece from Thrace through Macedonia to Epirus. It was in Epirus that Higgs's team concentrated their efforts because of the large number of sites suitable for excavation (for place-names, see Fig. 3.1). Higgs tested three sites

by excavation, the cave sites of Asprochaliko and Kastritsa and the open- air site at Kokkinopilos, all in the 1960s.2 For various reasons these excavations were not published and research languished after Higgs's death. Work was resumed in 1979 by another Cambridge University team under the direction of G. N. Bailey (now at the University of Newcastle-upon-Tyne), with additional surveys, geological research, and an excavation of Klithi Cave in the Zagori, northern Epirus.3

These two projects established Epirus as the center for sustained Palaeolithic research and attracted other scholars to the field. A Greek team is now excavating the cave of Boila near Klithi, and the work re-

ported here is part of an international Greek-American effort, the Nikopolis Project, centered on the nomos of Preveza.4

The integrated evidence from geoarchaeology and archaeology relating to the long history of human settlement and land use is the unifying theme that connects the scholars working on the many different periods covered by the Nikopolis Project. In this context it was our aim to deepen and extend the work of our predecessors in order to obtain a detailed picture of the early prehistory of the region. By far the longest episode of human occupation of the region, the Palaeolithic and Mesolithic span the later Pleistocene and earliest Holocene and so provide a foundation for the study of later prehistoric and historical periods.

To accomplish this goal we concentrated our efforts on the investigation of the geological history of the red Pleistocene sediments that in this region are so closely associated with human artifacts, and extended the

CURTIS N. RUNNELS AND TJEERD H. VAN ANDEL

search for new sites to those parts of the Preveza nomos that lie west of the Louros valley toward the sea. In order to obtain the most complete picture possible of land and resource use, we paid special attention to the search for very small findspots marked by only a few artifacts in a limited area.

When we accepted the invitation of its directors to join the Nikopolis Project, we were especially interested in three topics. Could we extend our methods of studying Palaeolithic sites in their landscape contexts-methods developed in the course of fifteen years of surveys in the Argolid and

Thessaly-to the quite different landscapes of coastal Epirus? Second, could we, in the face of strong skepticism regarding the value of studying Palaeolithic open-air sites,5 demonstrate their essential role in developing hypotheses of Middle Palaeolithic land use? Finally, we wanted to test the

utility of our paleosol stratigraphy for the exploration for Palaeolithic sites and their relative dating.6

THE IMPORTANCE OF PALEOENVIRONMENTS

In prehistoric research it is essential to place all traces of past human activities in a framework of detailed, dated paleoenvironmental reconstructions. Humans have made an impact on the natural environment, chiefly through predation and the use of fire, from the time they arrived on the scene until they became a major determinant of the environment with the introduction of agriculture. In Palaeolithic and Mesolithic landscapes their

impact was still small, and it was the environment that played a central role in structuring human activity. To understand their movements and activities we must take into consideration all of the evidence for tectonic

activity, geomorphological processes, climate, sea-level changes, hydrology, and flora and fauna.

Like Pleistocene landscapes everywhere, those of southern Epirus were very different from today. At times, the climate was considerably colder and especially drier; glaciers capped the high ranges of the Pindos and the now submerged continental shelf was a large coastal plain exposed by lowered sea levels. Reduced runoff and, as the level of the sea fell, steepened river gradients greatly altered the cycles of aggradation and incision. The flora was drastically reduced to a shrub and sagebrush (Artemisia) steppe and most tree species were isolated in sheltered refugia in lowland valleys. Large herds of herbivores browsed the lower slopes, valley bottoms, and coastal plains, where bison, wild ass, antelope, and aurochs were vastly more numerous than their human predators.

At other times, the climate would swing to a warmer interstadial phase or even a full interglacial. Melting ice caps caused the sea to rise, drowning the prime grazing land of the coastal plain and reshaping the coastline. Trees recolonized valleys and plains, and deer, elk, pig, and other forest species flourished. Rivers carried glacial outwash, aggrading valley floors, building deltas, and burying or destroying sites of previous human occupation. Throughout, the cycles of cold and warm climates were accompanied by tectonic activity that might change drainage patterns, raise an area and so expose it to erosion, or cause another to sink and be turned into a lake. The nature and continuity of human exploitation were continuously

5. Bailey, Papaconstantinou, and Sturdy 1992.

6. van Andel 1998a.

48

EARLY STONE AGE OF THE NOMOS OF PREVEZA

Figure 3.1. Map of Epirus and 20km LEUCAS

surrounding areas I 1 /

affected by such environmental changes and can only be properly interpreted when they can be related to them.

Among Greek landscapes, the nomos of Preveza is distinctive for the widespread dominance of limestone landforms (karst), including numerous basins of internal drainage controlled by bedrock patterns and tectonics. Because they supply reliable sources of fresh water and associated resources, these basins have always been major factors in determining the pattern of human occupation in coastal Epirus. The explanation of the relations between karst features and Pleistocene archaeology is our most important finding, with the widest implications for understanding past human behavior.

GEOLOGICAL FEATURES AND PAST HUMAN BEHAVIOR

Consciously or subconsciously, the images that come to mind when con- templating Late Pleistocene landscapes inhabited by human beings are dominated by the action of glaciers, rivers, and the rise and fall of the sea. It was a landscape of floodplains and river terraces with a backdrop of

49


mountains and distant glaciers that informed much of the thinking of Eric

Higgs and his coworkers.7 Yet the landscape west of the Pindos from the Gulf of Corinth to the Albanian border and beyond is, with few excep- tions, not like that at all. Here the main influence on the landscape of the

past five million years has been the dissolution of its limestone bedrock

during uplift and subsequent peneplanation and the more recent creation of basins of internal drainage through renewed uplift accompanied by extensive faulting. The result is a classic karst landscape that contrasts sharply with the river landscapes ofThessaly, Macedonia, both sides of the Gulf of

Corinth, and parts of the Peloponnese. Being mountainous and tectonically active, the Epirus karst land-

scape is dominated by erosion. More than three-fourths of its surface is

being worn down continuously, mainly by dissolution, and only a few localized areas receive and preserve the sediments that record the years since human beings first walked on Greek soil. These sediments, produced by the dissolution of limestone and widely but loosely known as terra rossa, were deposited in depressions of the rugged karst plains and in the numerous closed basins created by recent tectonics. They are closely and fundamentally associated with the Palaeolithic inhabitation of the

region. Here and there are river-dominated landscapes, some as extensive as

the tectonic graben that underlies the Ambracian Gulf and has caused the

development of coastal Epirus's main rivers, the Louros and Arachthos, others minor, such as the lower Acheron valley. Thus riverine landscapes, always valued by ancient humans, are the exception not only in coastal

Epirus but also farther north along the Adriatic and in the Dinarides. What attracted Palaeolithic hunter/gatherers to these often-barren karst

regions and governed the pattern of their movements there? We shall show below that the swamps and lakes of the many basins scattered among the

precipitous, barren slopes of the mountains were the main attraction, to-

gether with the resources of the coastal plains when exposed by low glacial sea levels. The resources of the karst and lake environments were so im-

portant that the rationale for the distribution pattern of Palaeolithic sites in coastal Epirus can only be understood in the context of their paleoenvironmental reconstruction.

PREVIOUS RESEARCH

The first phase of Epirote Palaeolithic research (1962-1967) began under the direction of Eric Higgs and combined both survey and excavation. The survey involved the inspection of all caves, rockshelters, and redbeds to identify likely places for more detailed investigation and excavation.

Only preliminary reports on this research have been published and it is sometimes difficult to know exactly where in Epirus Higgs and his team went.8 It is nevertheless clear that Higgs identified two broad Palaeolithic phases, an earlier Mousterian chiefly represented by open-air sites associated with redbeds (e.g., Kokkinopilos, Louros, and Stephani), and a later Palaeolithic presence in small rockshelters and caves such as Asprochaliko and Kastritsa.9

7. Higgs and Vita-Finzi 1966; Higgs et al. 1967; Harris and Vita- Finzi 1968; MacLeod and Vita-Finzi 1982.

8. E.g., Dakaris, Higgs, and Hey 1964; Higgs and Vita-Finzi 1966; Higgs et al. 1967.

9. Bailey 1992.

50


10. Dakaris, Higgs, and Hey 1964, pp. 215-221.

11. Dakaris, Higgs, and Hey 1964. 12. The discrepancy between

radiocarbon years and calendar years, nearly always significant, is particularly large between ca. 10 and 50 kyr B.P.,

when 14C years lag up to 4,000 years behind the calendar (Laj, Mazaud, and Duplessy 1996). Given the rapidly oscillating paleoclimate of the interval, calibration is essential (van Andel 1998b). Therefore, we have calibrated all radiocarbon dates cited in the literature in calendar years as thousands of years before present (kyr B.P.). Each calibrated date is followed by the radiocarbon date, cited as b.p. and without the published confidence limits.

13. Higgs and Vita-Finzi 1966, p. 21.

At this early stage of research Higgs made a detailed study of one of the richest of the open-air sites at Kokkinopilos ("red clay" in Greek). The study pinpointed thirteen locations where Higgs thought he could identify "chipping floors," by which he appears to have meant assemblages of

flintknapping debris.10 Two of these chipping floors were tested by excavation (sites a and P, hereafter Alpha and Beta) with somewhat ambiguous results. Large numbers of worked lithic artifacts (ca. 800 in site Beta) were recovered and, although the flints were not associated with features or organic remains, Higgs regarded them as in situ occupation surfaces. Site Alpha produced a typical late Palaeolithic assemblage with backed blades, along with typical Middle Palaeolithic artifacts, and site Beta an

early Palaeolithic (Mousterian) industry. It is unclear what stratigraphic relationship the two sites with their different industries might have, especially because they were not placed on a plan or map and are difficult to

identify today. Nevertheless, the artifacts from the excavations comple- mented the unprovenanced surface materials, and it is reasonable to conclude that there are two periods of occupation at Kokkinopilos, early and late Palaeolithic. A typological analysis of the early Palaeolithic artifacts

by Mellars described the Kokkinopilos assemblage as a variant of the Mousterian with many varieties of side scrapers, Mousterian points, and bifacial foliates ("leafpoints")."1 He considered the Kokkinopilos assem-

blage to be a mixture of different industries. A metrical analysis of the flints allowed Mellars to compare the industry (or industries) with the

European Mousterian and he concluded that the Greek industry was distinct enough to be considered a separate entity, although it shares some characteristics with the typical Mousterian.

To clarify the stratigraphic position of the Mousterian, Higgs excavated a rockshelter at Asprochaliko, approximately 4 km northeast of

Kokkinopilos in the Louros River valley. The stratified deposits of the rockshelter are approximately 2-4 m deep. The Mousterian is found in

layers 14-18. After a stratigraphic hiatus, the Mousterian layers are overlain by layers rich in Upper Palaeolithic artifacts, dated by a series of radiocarbon assays to 17-29 kyr B.P. (thousands of years before present).12 The

majority of the Mousterian levels were too old to be dated by the radiocarbon method, but were assumed by Higgs to be older than ca. 39 kyr B.P. (at that time, the effective upper limit for detecting radioactive carbon). The

age of the earliest deposits of the Mousterian is unknown.

Higgs divided the Middle Palaeolithic into two units, the earlier "basal" Mousterian and a later Mousterian called "micromousterian" because of the small size of the flints. He related the basal Mousterian to the Kok-

kinopilos industry on the basis of the use of similar fine retouch, but he also noted many dissimilarities between the two assemblages, particularly the absence of leafpoints at Asprochaliko.13 The later Upper Palaeolithic layers at Asprochaliko have Gravettian and Epigravettian backed-blade industries. The industrial succession seen in the flints is reflected also in the animal bones. The Mousterian was found with an extinct Pleistocene megafauna, including Merck's rhinoceros, aurochs, bison, buffalo, antelope, and wild horse. The Upper Palaeolithic shows a marked change in the fauna, with a great emphasis on the hunting of horse, red deer, ibex, and chamois.

5I


Higgs conducted a third excavation at the site of Kastritsa, south of Ioannina. Here he found a long sequence of Upper Palaeolithic backed- blade industries, dating to ca. 13-24 kyr B.P. and similar to those from Asprochaliko, but there were important differences in the lithics and faunal remains.14 Because the sites were largely contemporary, these differences encouraged Higgs and Vita-Finzi to postulate that the sites were seasonal camps or specialized hunting stands along the route of faunal migration from coastal lowlands, used in the winter, to upland summer pastures.15 Their model was based on the different locations and elevations of the two caves, the differences in fauna and artifacts, and on a comparison with the pastoral transhumance activities of modern Sarakatsani shepherds.

In sum, Higgs established evidence for two phases of human occupation in the Middle and Upper Palaeolithic. The Middle Palaeolithic was found chiefly on the surface at undated redbed sites where it was not associated with animal bones. The Upper Palaeolithic from the rockshelters was interpreted with an innovative model of seasonal logistics based on following herds of migrant megafauna. This considerable advance in prehistoric knowledge put Epirus on the map, as it were, and the state of knowledge stood here for more than a decade.

Because Corfu was connected with the mainland of Epirus during much of the last glacial period, the two areas can be regarded as a single cultural region, and a research project undertaken in 1964-1966 by Augustus Sordinas revealed a similar sequence of cultures.16 Sordinas located and sampled a large number of Middle Palaeolithic surface sites associated with redbeds. He classified a Middle Palaeolithic industry as typical Mousterian, with the same regional features (e.g., leafpoints) as in Epirus. His excavation at Grava Cave brought to light an undated Upper Palaeolithic industry similar to those from Asprochaliko and Kastritsa.17 His most important discovery was a low tell at Sidari on the north coast of the island, which was excavated in the 1960s.18 Sidari produced a sequence of three major layers dated to the Mesolithic, the Early Neolithic, and the Early Bronze Age.

After the death of Eric Higgs, prehistoric research in Epirus was interrupted until 1979, when a new project was initiated by Bailey. This project had several goals. One was to reexamine the excavations and surveys of Higgs with a view to updating his conclusions about settlement and land use, specifically the model of transhumance.19 Another goal was a campaign of research in northern Epirus centered on the excavation of the rockshelter of Klithi in the Zagori near Konitsa. The results of this project have now been published in full.20

The excavation records of Asprochaliko, Kastritsa, and Kokkinopilos were inspected and checked against the remaining sections and the finds stored in the magazines of the Ioannina Archaeological Museum. Further work was undertaken at Asprochaliko to clean the section and extract samples and burned flints for dating analysis.21 A study of the Middle Palaeolithic artifacts was useful in correcting some of the earlier views on the Mousterian. It was shown that the earliest "basal" Mousterian (layers 16 and 18) made greater use of the Levallois technique for the production

14. Galanidou et al. 2000. 15. Higgs and Vita-Finzi 1966. 16. Sordinas 1968. 17. Sordinas 1969. 18. Sordinas 1970. 19. Bailey et al. 1983a. 20. Bailey 1997. 21. Huxtable et al. 1992; Bailey,

Papaconstantinou, and Sturdy 1992.

52



23. Huxtable et al. 1992; Bailey, Papaconstantinou, and Sturdy 1992.


25. Bailey, Papaconstantinou, and Sturdy 1992, p. 140.

26. Bailey, Papaconstantinou, and Sturdy 1992, p. 142; cf. pp. 91-95, below.

27. Bailey et al. 1983a, pp. 76-77. 28. Bailey 1992, 1997.

of large blades or bladelike flakes, while the later Mousterian of layer 14 made less use of the Levallois technique and was characterized by the use of Mousterian points and small, atypical pseudo-Levallois points called Asprochaliko points. The later Mousterian, called "micromousterian" by Higgs, was shown by reanalysis to differ little from "basal" Mousterian, and the two assemblages are now regarded as similar in composition and typology.22 The conclusion is that there are not two types of Mousterian at

Asprochaliko, and the different labels have been dropped. New dates indicate that the Mousterian at Asprochaliko spans a considerable period of time. The Mousterian of the lowest layers may be as much as 90-100 kyr B.P., while the Mousterian of the upper layers, while not precisely dated, may date to 39 kyr B.P. or later.23

Bailey's reexamination of the excavation sites at Kokkinopilos was accompanied by an attempt to date the redbed sediments. The conclusions reached by Bailey's team differ from those proposed earlier by Higgs.24 The artifacts from sites Alpha and Beta had been considered by Higgs and his colleagues as part of in situ chipping floors associated with Pleistocene red clay deposits that had accumulated relatively rapidly as the result of erosion and aeolian dust deposition. Bailey and his colleagues, in contrast, concluded that the red earth deposits in the region were much older than the Middle-Late Pleistocene and were perhaps Pliocene in age.25 In their view, the deposits were formed as a result of dissolution of limestone, a process that they believed to have ceased before the Pleistocene. They concluded that the lithic artifacts were much younger than the redbeds and were incorporated accidentally into the deposits at a later time ("reworked" in their terminology), perhaps washed from the surface into gullies as a result of tectonic uplift, slumping, ponding, and gully erosion.

As a result of this analysis Bailey concluded that the open-air sites were essentially fortuitous admixtures of lithic artifacts, often of greatly differing ages, with the redbeds. Apart from the fact that the lithics were associated with the redbeds in spatial terms, Bailey attached little analytical value to them because they were unstratified and could be dated only in units of 100,000 years or more.26

Much attention was given to the new excavations at Klithi, which Bailey and his colleagues placed at the center of a new interpretation of the Higgs model of Upper Palaeolithic transhumance. At first Bailey proposed that Klithi was probably a "home base" or base camp at the center of a hierarchical settlement pattern consisting of smaller seasonally occupied shelters and specialized activity camps spread across a large and geographically diverse region.27 This hypothesis had to be abandoned when the excavations revealed Klithi to be a small specialized hunting camp.28 Klithi is located in a river gorge in the Zagori, at the head of the Konitsa plain, and was occupied only during the summer over a period of a few thousand years following the last glacial maximum. In that time the nearby peaks were relatively unglaciated and the camp was used for hunting ibex. Highly fragmented ibex bones and horn cores, along with many thousands of backed blades (from projectiles) and end scrapers (for cutting meat and processing hides), indicate the short but intense specialized hunting activities.

53


This evidence, when combined with a reanalysis of the fauna at

Asprochaliko and Kastritsa, allowed Bailey's team to conclude that there was a degree of specialization in hunting and that the three excavated rockshelters were hunting stands.29 In recent years, noting that the exploitation territories of the late glacial were very large and embraced the coastal

plains, Bailey's team proposed a modified version of Higgs's model to ex-

plain the data. The new model relies heavily on the study of bedrock, flora, and the distribution of animal populations.30 The model posits "topographic barriers," resulting from active tectonics and composed of rock supporting a flora of unpreferred species, that bounded discrete zones where the bedrock supplied preferred browse for horse, deer, chamois, and ibex. Human

predators did not "follow" the herds, as in Higgs's formulation, but pur- sued a round of residential mobility, shifting periodically from one rockshelter camp to another. Each camp was strategically located on the

edge of one of the zones of preferred browse, taking advantage of the ten-

dency of animals to concentrate in these zones. The caves gave the hunters access to water and shelter just out of sight of the herds. Following the

seasons, animals were drawn to the mountains in the spring and summer, and the hunters shifted their camps accordingly in a periodic seasonal fashion, only to return to the coastal plain in autumn and winter. This logistical strategy of land use took full advantage of the available resources and sustained hunters over a period of 10-20,000 years. The maximum extent of the activity, however, appears to have been in the millennia immediately following the last great glacial advance between 20 and 25 kyr B.P. These modifications to the Higgs model advance our understanding of the period considerably, and the new model incorporates a rich body of paleoenvironmental data derived from the study of the geomorphology of the Klithi region and two major pollen cores from northern Epirus.3

THE LATE QUATERNARY LANDSCAPE OF WESTERN EPIRUS

GEOLOGICAL HISTORY

Epirus is located at the meeting point of three tectonic plates (Fig. 3.2) whose rapid convergence (10-15 mm/year) is causing widespread deformation of the whole region west of the Pindos, including subsidence of the Ambracian Gulf.32 The deformation is regarded by Bailey and coworkers as a major force which affected the Palaeolithic landscape by creating, at a rate perceived on a human timescale, "topographic barriers" that had a

large impact on Palaeolithic resource availability and use.33 The required rates of uplift, however, seem excessive and to evaluate this proposition we examine the tectonic state of the region in some detail.

From the Early Mesozoic to the Late Eocene, Epirus formed part of a vast midocean plateau covered by shallow marine Jurassic limestones overlain by deepwater limestones of Cretaceous-Eocene age.34 In the Oli- gocene the eastern portion of the Pindos range began to rise and sand and silt were shed westward, covering the limestones with flysch deposits. In- tense deformation with a northwest trend, still visible in today's landscape,

29. Bailey et al. 1983b. 30. Bailey, King, and Sturdy 1993. 31. Bailey et al. 1990; Bailey 1997;

Willis 1994. 32. Kahle et al. 1993; Le Pichon et

al. 1995. 33. Bailey, King, and Sturdy 1993,

fig. 5; King, Sturdy, and Whitney 1993; King and Bailey 1985.

34. Aubouin 1959; Jacobshagen 1986.

54


Figure 3.2. Tectonics of northwestern Greece and the Ionian Sea. Offshore from the Ambracian Gulf the subduction of the Mediterranean plate under the Peloponnesian continental margin changes to a collision between the Italian/Apulian and Greek continental blocks. The change is marked by a strike-slip fault (opposing arrows), the

Cephalonian transform fault. After Le Pichon et al. 1995, fig. 1

35. E.g., Etude geologique; Schrbder 1986.

36. Clews 1989; Sorel 1989; Underhill 1989.

37. Piper, Kontopoulos, and Panagos 1988.

38. Etude geologique; Waters 1994, fig. 1.4:b, plans 1,2.

39. Sorel 1989; Underhill 1989; Waters 1994, pp. 168-211.

40. Bailey, King, and Sturdy 1993.

ended this phase. In the Miocene, the emergence of western Epirus created a land floored predominantly with limestone that was eroded down to a broad peneplain (a rough planar surface) during the Pliocene.35 Raised remnants of this peneplain can still be discerned from the widespread uniform elevations of high plateaus, for instance around Loutsa southeast of the lower Acheron valley.

Late in the Pliocene, the collision with the Apulian block reinitiated the deformation of the region west of the Pindos front from the Gulf of Corinth to Albania.36 The compression, continuing today, resulted in regional uplift of the Pliocene peneplain, while cross-faulting created young grabens in the Ambracian Gulf,37 Kalamas delta, lower Acheron valley, Doliana basin, and elsewhere (Fig. 3.3).

The region is thus seismically active and geological maps show many strike-slip, thrust, and normal faults (Fig. 3.3).38 For many of those faults present activity has been assumed, but hard evidence is sparse and many faults are relics of the mid-Cenozoic mountain-building phase and now inactive.39 The best evidence for present tectonic activity are the fresh striae on fault planes, identified with detailed fieldwork by Waters (Fig. 3.4).

The high uplift and subsidence rates (up to 100 m in historical times) proposed by Bailey and his group in support of their paleoenvironmental resource models for Late Quaternary Epirus are not locally documented; they rest mainly on an assumed similarity in seismic activity between Epirus and other active but actually tectonically quite different regions, such as California, New Zealand, Japan, and the Middle East.40 How valid are the proposed magnitudes?

55


Waters, assuming that deformation began about 3.5 million years ago, showed that prominent fluvial surfaces now raised above the present river level in the Botzara and Epirus synclines indicate that the synclinal axes are rising at 10 m/100 kyr.41 The crestal elevations of the Pantokrator and

Parga anticlines suggest uplift at 25 m/100 kyr, and regional uplift of 15- 35 m/100 kyr is implied by remnants of the Pliocene peneplain. During the same interval, the Ambracian Gulf has widened at ca. 2 cm/yr.42 Val- ues of 0-16 m/100 kyr, indicated by coastal deposits of the last interglacial, fall within the same range (see below, pp. 76-83). These rates do not suf- fice to alter the Epirote landscape perceptibly in the last 100-200,000 years and consequently Bailey's models lack a credible foundation.43

Figure 3.3. Possibly active (Late Quaternary) tectonic features of western Epirus. The structure of the Ambracian, Kalamas, and lower Acheron grabens is much simplified. Major gorges were incised by rivers during Pliocene uplift of the peneplain. After Waters 1994, fig. 5.10

Figure 3.4. Present tectonic activity in western Epirus as indicated by fresh striae on fault planes. Compare with presumed active faults shown in

Figure 3.3. After Waters 1994, fig. 5.9

41. Waters 1994, pp. 208, 213. 42. Kahle et al. 1993. 43. Cf. Bailey, Papaconstantinou,

and Sturdy 1992.

56


Figure 3.5. Simplified bedrock map of western Epirus showing the prevalence of Mesozoic to Eocene limestone west of the Pindos front (PF). After Etude g6ologique

Q y '" ....... "

" : -!.....

I Quaternary **

Miocene-Pliocene .. A

" Oligocene

|[|| Mesozoic-Eocene limestone

KARST LANDSCAPES

44. Ford and Williams 1989. 45. Ford and Williams 1989,

pp. 396-412. 46. Ford and Williams 1989,

pp. 31-34, 96-114.

Landscapes are shaped principally by two forces, the internal dynamics of the earth that deform its crust and the external forces of climate, water, wind, and vegetation (and recently also human beings) that modify the relief generated by tectonics. The impact of the external forces is strongly influenced by the kind of bedrock on which they operate. Because nearly all large continental regions display a mosaic of diverse rock types, most landscapes have been created by the work of rivers, ice, and wind, modified by zonal climates. In contrast, areas where limestones dominate, such as Epirus (Fig. 3.5) and much of former Yugoslavia, are shaped mainly by the dissolution of limestone by CO2-charged rain, runoff, or groundwater.44

A weathered limestone land surface, called a karst, drains mainly downward through cracks and fissures into extensive subterranean conduits, rather than horizontally at the surface by rivers. Characteristic for karst landscapes are thus a lack of rivers, and numerous sinkholes or dolines (Fig. 3.6), round, steep-walled pits, usually flat-floored and up to a few hundred meters across, that derive from collapse of subterranean channel roofs.45 Typical dolines can be seen along the road from Loutsa to Strounga on the plateau south of the lower Acheron valley. Another diagnostic karst feature are blind valleys, former stream valleys deprived of surface runoff by the formation of a subterranean drainage system.

Rather than by a whole panoply of processes ranging from weathering to river incision, karst surfaces are molded almost entirely by limestone dissolution and minor slope wash. Solution of limestone removes all calcium carbonate (CaCO3), leaving behind a small, fine-grained, insoluble residue that is iron-rich and hence red. Because the rate of solution varies with small-scale variations in the properties of the bedrock, a rough,jagged surface forms, which is randomly dimpled by depressions.46 There water may collect, stand for a while, and gradually dissolve the rock, thereby

57


Cross section of a doline doline

widening and deepening the ponds. On gentle slopes the residual mantle, called terra rossa in the Mediterranean, slowly thickens and, being fine-

grained, reduces infiltration, allowing sheet wash to transfer the weather-

ing residue into the depressions (Fig. 3.7: Time 1). In the absence of major tectonic activity, the result is a peneplain dotted with red patches of fine-

grained, redeposited terra rossa. Over time, both primary and redeposited terra rossa thicken and spread, but because dissolution rates are slow and the insoluble residue constitutes a mere few percent of the pure limestone of Epirus, millions of years may pass before redbeds cover the whole

region.47 Recent deformation and uplift of western Epirus have reshaped the

Pliocene peneplain into many synclinal troughs, initially not connected with each other or the sea by rivers, and separated by steep, lofty anticlinal mountain ranges. The uplift has accelerated the downslope transfer of the terra rossa by slope wash, especially on fault scarps (Fig. 3.7: Time 2). Where exposed faults cut across subterranean channels, springs have turned

many closed depressions into lakes drained by swallowholes, called "katavothres" on Greek maps, which draw off excess water into subterranean channels.48

Consequently, enclosed flat-floored basins and others raised and in the process of being eroded dot the landscape of western Epirus (Fig. 3.8; Table 3.1). Common in Yugoslavia also, they are known by the word "polje." The poljes of western Epirus range from ca. 1 to 6 km in length and up to 1 km in width, exluding a few large ones (Table 3.1). Until recently, many had permanent or seasonal lakes (Fig. 3.9), but today most of those have been drained for agriculture. Large, permanent lakes have accumulated thick calcareous deposits; an example is the Ioannina basin, which has

preserved a record of paleoenvironmental change going back to the Early Pleistocene.

Western Epirus is not the only part of Greece where large karst basins with lakes are conspicuous landscape components.49 Boeotia has its Lake

Kopais and, in the Peloponnese, the Stymphalos, Orchomenos, Manti- neia, and Tegea basins are of similar origin. In most of those regions deformation ceased or became insignificant some time ago, whereas the continuing tectonic activity in Epirus has lent its karst basins a special character.

Figure 3.6. Formation of a doline (sinkhole) by roof collapse or dissolution in a subterranean drainage system. The doline shown here has a subaqueous exit in a lake or sea, but subaereal outlets are also common.

47. Spate et al. 1985. 48. Ford and Williams 1989,

pp. 428-432. 49. Pfeiffer 1963; Sweeting 1985.

58


Figure 3.7. Diagram of the genesis of loutses and poljes on a karstic peneplain. Time 1) The karstic peneplain develops a mantle of dissolution residue called terra rossa, which is washed down and deposited in shallow, closed karst surface depressions called loutses. Time 2) Elongate basins of internal drainage called poljes develop as a result of renewed tectonic activity, which raises ridges along normal or reversed faults. Accelerated weathering and slope wash fill poljes with secondary terra rossa. Time 3) Continued uplift raises the polje and surface streams cut back into its deposits by headward erosion, often along weakened fault zones.

50. Nicod 1992. 51. E.g., Gams 1978; Sweeting

1993. 52. Budel 1973. See also, e.g., Gams

1973; Rathjens 1960.

Time 1 - Peneplanation and dissolution

terra rossa LOUTSA redep. terra

tera ro assa

I I I I I I I I I I I ! J I I I I

I [ I I i I I I I I i I I 1 i [ I !I I I I I I I I I I I , I I

Time 2 - Faulting, uplift, polje formation

redep. terra ACTIVE rossa POLJE

! I I' I If I I5 ~.' - '......-_:::::::::::::: ',' 1ww Xwe @ @evZ*-{ | w X I. .

I I I I

I I

I I

I I I I I .I IN .

I i I I I I I I I I

i I 1, f i I i 1,

i, I , I

-1-

I

I

I I .I I I I I I I

I' i I I 1 II :I :I:I ;/,I,I,I ,I, I I,

Time 3 - Uplift and/or headward stream erosion, redep. terra polje dissection

iLOUTSA /rossa DISSECTED ^LOUTSA /, \ y POLJ

Compared with the basins named above, most Epirote poljes are small and many have been raised recently, sometimes to great heights, and are now subject to erosion. Examples are two deeply eroded, fault-bound hanging poljes perched on opposite flanks of the Thesprotiko valley, Kranea (Fig. 3.8:25, 300 masl) and Galatas (Fig. 3.8:27, 200 masl).50

Is it of the essence of poljes that they have structural origins, or are simple solution basins in the karst surface also a class of poljes?51 The solution basins tend to be shallow (Fig. 3.10) and small and are fed by winter and spring runoff rather than by springs. Their value as a resource in the Palaeolithic context is therefore seasonal and differs much from that of true poljes. For this reason we have adopted for use in this paper the separate term "loutsa" (pl. loutses), the Greek name for a seasonal pond or wet sink.

Budel and others have suggested that poljes are fossil elements of the landscape that are no longer being formed.52 In a tectonically active area, however, nothing is permanent. While deformation constantly creates new

poljes, continued uplift permits small streams to cut back upstream and capture former poljes, draining the lakes, drying the surfaces, and exposing the stratigraphy in stream incisions (Fig. 3.7: Time 3). Epirus west of the Pindos front contains poljes in every stage of evolution from recent birth (e.g., Valtos Kalodiki, Figs. 3.8:6, 3.9), to old age (e.g., Cheimadio, Figs. 3.8:24, 3.11), to stream incision and removal (e.g., Kokkinopilos, Fig. 3.8:30).

.,:';- , ], , , I . . I

i

I

59


32 w

/ Cheimadio I %~~~~~~~~~4

/04/

o Doline

F

F Fault

Coastal or river plain

Active or drained polje

Raised, dissected polje or loutsa

Figure 3.8. Poljes and loutses in western Epirus. The deposits dominate the Quaternary landscape except for the GulfofAmbracia graben and its feeder river valleys. Sordinas (1983) also reports widespread redeposited terra rossa on

Corfu and adjacent islands. The upper Acheron valley may conceal a large polje or its surface runoff may have been lost to underground drainage, making it a "blind valley." Numbers refer to Table 3.1.

[ .i. . .. , .]

60


TABLE 3.1. DIMENSIONS AND ELEVATIONS OF POLJES AND LOUTSES IN WESTERN EPIRUS

Length Width Number Name (km) (km)

Ayia

Kokkinographos Domia Gouri Valtos Kalodiki Mavradis Mazaramia Mavrikambos

Arapanitos Saita

Nerotopos Paramithia Kalaboukia Mesovouni

Morphi Lamma

Kalosykies Tsapela Kordeli Alonaki Loutsa

Pyra Cheimadio Kranea

Thesprotiko Galatas

Kalyvia-Kraneas Lake Mavri

Kokkinopilos Gymnotopos

Tsoukka Ioannina basin

1.2 2.0 0.7 2.2 1.0 3.8

10.5 3.0 3.5 1.8 2.0 5.0

18.0 2.0 2.4 3.2 3.5 2.5 1.6 1.3

2.6 1.6 2.7 1.0 6.0 0.6 3.0 4.0 1.6 1.3 2.8 1.1

20.0

0.5 0.4 0.4 0.8 0.5 1.3 1.7 1.5 1.8 0.6 0.8 2.5 6.0 1.1 0.6 1.9 2.0 2.0 1.2 1.3

1.0 0.7 0.5 0.6 1.8 0.5 1.4 1.5 1.0 0.7 0.7 0.9 6.0

Elevation

(masl)

ca. 300

ca. 520

400 380 440 110

110

ca. 130 ca. 250

160 ca. 160

100

80 260 290

ca. 160 100

50 30 20

10-15 ca. 200

400 400

ca. 300 40-60

ca. 200

ca. 30

20-30 120

ca. 300 ca. 500

ca. 1400 ca. 400

Type

Raised loutsa Raised loutsa Raised loutsa Raised loutsa

Active polje Active polje, drained Active polje, drained Active polje Raised loutsa/polje Raised loutsa/polje

Active polje Raised loutsa/polje Raised loutsa/polje Raised polje Raised loutsa/polje Blind valley? Blind valley? Blind valley? Subsided loutsa? Raised active? loutsa Raised active? loutsa Active polje, drained

Hanging paleopolje? Active polje, drained

Hanging paleopolje? Active polje Active polje, drained Raised dissected polje Raised polje/loutsa Raised polje/loutsa Raised polje/loutsa Active polje

Number = location number in Figure 3.8. Sites 4-15,17,18, and 31-33 were studied only on topographic (scales 1:50,000 and 1:5,000) and geological (scale 1:50,000)

maps.

TERRA ROSSA: DERIVATIVES AND LOOK-ALIKES

Terra rossa, once common in western Epirus and elsewhere in the limestone regions of Greece and the Mediterranean, has been defined formally as the weathering residue of limestone in a warm climate that is commonly assumed to have been that of the Late Miocene and Pliocene;53 strictly speaking, terra rossa is a Pliocene paleosol (Fig. 3.12). Other red Mediterranean sediments of different origin exist and mature Mediter-

53. Foucault and Raoult 1980. ranean paleosols are usually also red (see below). This diversity of"redbeds"

1 2 3 4 5 6 7 8 9

10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33

6i


has been insufficiently appreciated and the resulting loose usage of the initially precise term "terra rossa" has robbed it of most of its meaning. This is regrettable because, although redbeds are widely regarded as in- tractable in terms of their genesis, stratigraphy, and geochronology,54 they are, with their vivid colors and their close association with Palaeolithic artifacts, a welcome guide in Palaeolithic field studies in Greece, and surely elsewhere in the Mediterranean as well.55

In practice there are three main types of red deposits, each with a different genesis and a different role in Palaeolithic site preference: 1) primary terra rossa, the in situ limestone weathering residue covering the original rugged karst surface; 2) polje and loutsa redbeds that are terra rossa redeposited in low places (Fig. 3.12:a); and 3) colluvial redbeds formed by slope creep, debris flows, and small, braided ephemeral streams on alluvial fans.56 The third group differs from the others because the currents involved are capable of carrying coarse material up to sand and chert gravel size (Fig. 3.13), so making them lithologically very distinct.

Paleosols often are as red as terra rossa (Fig. 3.12:b) and to confuse the two is easy. Red sediments, however, bear the imprint of their environment and time of deposition, whereas paleosols record the duration and intensity of the weathering of a stable surface; a clear distinction between red sediments and red paleosols is thus critical for stratigraphic and paleoenvironmental investigations.

In Greece the Quaternary is mainly a time of erosion, assisted since the Neolithic by human deforestation and land use.57 Primary terra rossa is therefore rare, although small pockets remain in the rugosities of karst surfaces. Stripped from steep limestone slopes by sheet erosion and redeposited in wet poljes and loutses, terra rossa retains its diagnostic fine grain size, but the typical red color is lost by reduction in wet environments. In alluvial fans, which are usually dry, redeposited terra rossa retains its color, but is mixed with coarse components.

STRATIGRAPHY AND SEDIMENTOLOGY OF REDEPOSITED

TERRA ROSSA

Almost all terra rossa in western Epirus is now on secondary location, some of it in colluvium and alluvial fans, but most in loutses or poljes. The color of primary terra rossa ranges from red to dark red (10R 4/6 to 2.5YR 3/6 and 4/6) because of abundant hematite.58 Because groundwater pref- erentially dissolves hematite59 and because active loutses and poljes are seasonally or permanently wet, the colors of redeposited terra rossa tend to be paler, more yellowish in hue (5YR to 7.5YR).6? The reducing effect of a varying groundwater level is also evident in discoloration (gley) to gray and yellowish gray (10YR to 2.5Y 4/6-7/0) as horizontal bands, or mottling (Fig. 3.12:c).The same process is responsible for manganese coatings on fractures and bedding planes and the formation of manganese-iron nodules.61 Discoloration of subvertical stripes, probably due to the decay of root systems, is common. The reduction process can be reversed as the sediment dries during long intervals of drought or by uplift.

54. E.g., Schneider 1977. 55. Dakaris, Higgs, and Hey 1964;

Higgs and Vita-Finzi 1966; Runnels and van Andel 1993a, 1993b.

56. Bull 1972, 1977; Coussot and Meunier 1996; Innes 1983.

57. van Andel, Zangger, and Demitrack 1990.

58. Boero and Schwertmann 1989; Mirabella and Carnicelli 1992.

59. Goethite is more stable in wet conditions; see Schwertmann 1971.

60. Yassoglou, Kosmas, and Moustakas 1997.

61. Boero and Schwertmann 1987, 1989; Mirabella and Carnicelli 1992.

62


Figure 3.9. View of Valtos Kalodiki, an active polje partly used for farm- land and partly still in its original state, illustrating the natural resources available to prehistoric hunter-gatherers

Figure 3.10. The eponymous loutsa on the raised peneplain south of the lower Acheron valley (Fig. 3.8:22). Although currently being dissected by headward erosion from small streams on its northern edge, Loutsa still ponds water in winter and spring, attracting vegetation and wildlife.

Figure 3.11. The polje of Cheimadio, at an early stage of incision and drainage by streams at each end

63


a

Figure 3.12. Red sediments and

paleosols: a) redeposited terra rossa in a small raised polje at Kranea; b) mature paleosol (Bt horizon) in the raised polje of Kokkinopilos; c) banded terra rossa, alternating between reduced (pale) and oxidized

(red), near the base of the redeposited terra rossa in the loutsa of Ayia; d) red oxidized terra rossa overlying (with a sharp boundary) yellow reduced terra rossa redeposited in lacustrine conditions in the raised

polje of Kokkinopilos b

64


C

d

65


deposit

primary terra rossa I I

slump/landslide

Redeposited terra rossa has a characteristic bimodal grain-size distribution with a sharp upper limit at ca. 0.04 mm (Fig. 3.14) and, except for some limonite micronodules (Table 3.2), it lacks particles larger than 0.064 mm. Together with poor disaggregation during the preparation for grain- size analysis, these nodules probably explain the 20% sand noted by Dakaris and his colleagues.62

Terra rossa, primary or redeposited, displays two size frequency modes, a silt mode between 0.010 and 0.040 mm making up 5-30% of the whole and a clay mode below 0.002 mm ranging from 50% to more than 90% (Fig. 3.14; Table 3.3). In a large number of cases the distinction is sharp, but the 0.002 to 0.010 fraction is variable and may obscure the fine limb of the silt mode.

Mineralogically, the clay mode varies somewhat from one site to another, but is consistent within sites (Tables 3.4 and 3.5). The mode consists mainly of illite, except at Kokkinopilos where kaolinite and illite occur in roughly equal amounts. Smectite (chlorite and vermiculite) is a minor accessory and quartz is rare. The clay mineral spectra agree well with those of Yassoglou and of Barbaroux and Bousquet, who concluded that the influence of Mesozoic and Cenozoic source rocks was limited and attributed the variations mainly to differences in the weathering history of the source deposits.63 In contrast, the composition of the silt mode is uniform to the point of monotony; it consists almost entirely of quartz with rarely a little feldspar, the latter exhibiting a variable orthoclase/plagioclase ratio (Tables 3.4, 3.5). An exception is the Rodaki stony red deposit, which is feldspar-rich and, as will be discussed below, should not be classified as terra rossa.

The bimodal grain-size distribution of redeposited terra rossa was first noted by Tippett and Hey, who suggested a long-distance aeolian origin

Figure 3.13. Terra rossa may be redeposited in fan complexes that consist of colluvium, debris flows, slumps, and landslides, or on alluvial fans deposited by small ephemeral streams. Because fan formation is intermittent, paleosols may form on temporarily stable surfaces.

62. Dakaris, Higgs, and Hey 1964, fig. 15.

63. Yassoglou, Kosmas, and Moustakas 1997; Barbaroux and Bousquet 1976, fig. 3.

66


TABLE 3.2. COMPOSITION OF THE FRACTION >0.064 MM IN REDEPOSITED TERRA ROSSA

Sample Description Remarks

VA93-01 Poorly disaggregated sediment Paleosol Bt VA93-05 Poorly disaggregated sediment Paleosol Bt VA93-02 Limonite concretions to 3 mm and

poorly disaggregated sediment VA93-04 Poorly disaggregated sediment VA93-03 Small (0.5-3 mm) limonite nodules and

poorly disaggregated sediment VA93-06 As above VA93-07 As above VA93-08 Limonite nodules, 2-10 mm in diameter Paleosol Bt VA93-09 Limonite nodules, 2-6 mm in diameter Paleosol Bt VA93-11 Limonite nodules, 2-6 mm in diameter and heavily

Mn-stained, plus poorly disaggregated sediment Paleosol Bt VA93-12 As above VA93-13 Poorly disaggregated sediment VA93-16 As above

20- . -, r?"

-, A

VA93-04 Kokkinopilos

0

20-

10I!

0 , , I

VA93-08 Kokkinopilos

VA94-19 Galatas Go CD

I F F I F I F

VA94-23 Galatas

20- VA94-04 Loutsa

- o

10 -

Figure 3.14. Typical grain-size frequency diagrams of terra rossa redeposited in poljes and loutses. Material coarser than silt size (>0.064 mm) is very rare. Percentage of clay (<0.002 mm) is shown on the left. Approximate area of the silt mode is shaded.

20- a,"P VA94-12 Alonaki

- Cs (D

. . ..... ...............:...................

ioj, ,''''',"'"' ,F T .

0 2 10 100

VA94-14 Galatas

VA94-07 Cheimadio L0

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

0 2 10 100

t % 2-4 microns size (log microns)

._

a -

f I I I i I I ........ .

--

67


TABLE 3.3. GRAIN-SIZE DISTRIBUTION OF REDEPOSITED TERRA ROSSA

Silt Clay (<0.002 mm) Silt/Clay Depth below

Sample Size (mm) % % Ratio Surface (m)

ALONAKI (#21: subsided loutsa?) VA94-12

AYIA-I

VA94-11 VA94-26

0.010-0.035 17 62 0.274

(#1: loutsa) 0.008-0.035 0.010-0.040

AYIA-2 (#1:

VA94-27 VA94-29 VA94-30 VA94-31

loutsa) 0.010-0.040 0.008-0.035 0.008-0.035

CHEIMADIO (#24: polje) VA94-07 0.010-0.030 14 65 0.215

GALATAS (#27, south section: polje) VA94-14 0.009-0.035 12 VA94-15 0.010-0.035 9 VA94-16 0.008-0.040 16 VA94-16 0.010-0.040 11 VA94-17 0.010-0.040 13 VA94-18 0.009-0.040 13 VA94-19 0.009-0.040 15 VA94-20 0.008-0.035 10 VA94-21 0.010-0.035 18

GALATAS (#27, west section: polje) VA94-22 0.010-0.040 5 VA94-23 0.009-0.040 13 VA94-24 0.006-0.030 14

KOKKINOPILOS (#30: polje) VA93-01 VA93-05 VA93-02 VA93-03 VA93-04 VA93-06 VA93-07 VA93-08 VA93-09 VA93-11 VA93-12 VA93-13 VA93-16

LOUTSA VA94-04 VA94-36

0.010-0.035 0.010-0.035 0.010-0.035 0.008-0.035 0.008-0.040 0.009-0.040 0.010-0.035 0.008-0.040 0.010-0.035 0.010-0.035 0.010-0.040 0.008-0.035 0.009-0.040

(#22: loutsa) 0.010-0.040 0.009-0.030

12 9

11 9

11 8

13 19 14 23 22 15 13

11 11

72 74 65 71 67 73 68 73 65

0.167 0.122 0.246 0.155 0.194 0.178 0.220 0.137 0.276

79 67 74

72 78 72 73 78 65 64 56 65 57 60 68 73

70 73

0.063 0.194 0.189

0.167 0.115 0.153 0.123 0.141 0.123 0.203 0.339 0.215 0.403 0.367 0.220 0.178

-0.50

-0.15 -1.00 -1.90

-1.90 -2.85 -3.20 -3.60 -4.70 -5.30

-0.60 -1.10 -1.60

-0.50 -0.50 -2.50 -2.50 -2.50 -4.00 -6.00 -7.00

-10.50 -14.00 -17.00 -22.00 -30.00

5 10

90 73

9

14 27

0.055 0.137

0.122 0.206 0.529

-0.50

-0.50 -0.80

-1.20 -2.50 -4.00 -7.00

74 68 51 96

0.157 surface 0.151 -0.50

68


TABLE 3.3-Continued

Silt Clay (<0.002 mm) Silt/Clay Depth below

Sample Size (mm) % % Ratio Surface (m)

MORPHI (#16, section 1: polje) PP1-19 0.008-0.040 17 52 0.327 -3.00 PP1-17 0.010-0.030 7 73 0.096 -4.00 PP1-15 0.008-0.030 17 51 0.333 -11.50 PP1-02 0.008-0.040 18 50 0.360 -16.80 PP1-03 0.008-0.040 13 65 0.200 -17.00 PP1-01 0.009-0.040 14 62 0.226 -19.50

RODAKI (coastal marsh)

VA94-06 0.010-0.045 12 66 0.182 surface VA94-35 0.010-0.040 15 59 0.254 surface

Numbers in parentheses refer to locations in Table 3.1 and Figure 3.8. Grain-size analysis of fraction <0.064 mm (64 m) with a Micromeritics Sedigraph

5000ET and 5100 V3.07 with computer interface (Jones, McCave, and Patel

1988) after disaggregation with 0.2% Calgon. Fraction >0.064 mm (not always present) consists of limonite micronodules and poorly disaggregated fine

sediment, except for samples VA94-06 and VA94-35, which contain detrital sand and gravel.

64. Dakaris, Higgs, and Hey 1964; MacLeod 1980; Yaalon 1987.

65. Higgs and Vita-Finzi 1966; Harris and Vita-Finzi 1968; MacLeod and Vita-Finzi 1982.

66. Rossignol-Strick 1983. 67. Kubilay et al. 1997; Pye 1992;

Yaalon 1997.

for the silt of the Kokkinopilos redbeds as did MacLeod and Yaalon.64 Others, citing interbedded stream gravels at the periphery of the Kok- kinopilos deposit, which we have been unable to confirm, regarded it as an alluvial fan complex.65 Following MacLeod's brief but perceptive study, we consider the high sorting and fine grade of the silt mode as conclusive evidence for a long-distance windblown origin of this component of the sediments. During glacial conditions, strong south- westerly to southerly winds during winter and spring and easterlies in the summer were probably at least as common as they are today.66 The sorting of modern North African dust, collected at many sites in Crete and elsewhere, resembles closely our own and a North African source seems possible, but the Epirus dust is slightly coarser and may instead

belong to the attenuated southern fringe of the central European loess belt.67

Whatever the sources of terra rossa silt may have been, its distant origin and complete bleaching during aerial transport make the material suitable for luminescence dating. Moreover, aeolian dust fall is likely to have been more constant than the local flux of weathering residue. The variations in the ratio of the two components might thus be useful to correct bulk sedimentation rates.

Uniform as polje and loutsa sediments are, stratigraphic sections of raised and dissected poljes suggest depositional histories that differ from place to place. Best surveyed is Kokkinopilos (Fig. 3.15), studied and re- studied by the Higgs and Bailey teams and by ourselves in collaboration with Panayiotis Paschos from the Institute of Geology and Mineral Ex- ploration (Preveza branch).

69


TABLE 3.4. MINERAL COMPOSITION OF REDEPOSITED TERRA ROSSA AT KOKKINOPILOS

Clay (<0.002 mm) Silt (0.010-0.064 mm) kaol ill sm + ve ve/ch feld qtz qtz abund

Sample (%) (%) (%) ratio (%) (%) (mm)

VA93-01 38 47 15 xx 5 95 65 VA93-02 48 31 21 xxxx - 100 30 VA93-03 46 52 2 xxxx - 100 34 VA93-04 53 40 7 xxxx - 100 80 VA93-05 48 52 - xx - 100 32 VA93-06 43 49 8 xxx - 100 47 VA93-07 46 50 4 xxx - 100 85 VA93-08 45 41 15 xx - 100 80 VA93-09 48 39 13 xx - 100 59 VA93-11 47 48 5 xx - 100 46 VA93-12 42 50 7 ? - 100 60 VA93-13 40 60 - x - 100 47 VA93-16 40 46 14 xx - 100 7

kaol = kaolinite; ill = illite; sm + ve = smectite and vermiculite; ve/ch ratio = vermiculite/chlorite ratio, from low (x) to high (xxxx), based on higher of two 14.2 A peaks in XRD trace relative to 10A peak; feld = feldspar, orthoclase and

plagioclase combined; qtz = quartz; qtz abund = quartz abundance as peak height of 4.23 A quartz line.

The Kokkinopilos polje is separated from the Louros valley by a limestone fault block. A paleosol, now stripped down to a mature Bt horizon, is preserved along its edges and spottily in the slightly bowl-shaped center, where it forms the foundation of the vents for a subterranean section of the Roman aqueduct leading to Nikopolis.68 Locally within the paleosol numerous flint artifacts occur in situ.69

Kokkinopilos is being eroded rapidly by ephemeral streams, which are responsible for its spectacular badland topography and have exposed deep sections throughout the sediment body (Figs. 3.16,3.17). A mature paleosol tops a red zone (C),70 which locally preserves faint, thin inter-bedding with subhorizontal, gray, bleached layers. In places, gray, subvertical stripes, probably root channels of the existing pine woodland, are seen. A thin but conspicuous desiccation zone and an immature paleosol, indicating a brief hiatus in deposition, separate it from the more yellow (5YR-7.5YR) zone B. Both layers can be traced throughout the central area.

In zone B, careful cleaning reveals fine, subhorizontal laminations indicating subaqueous deposition almost everywhere, and diffuse gray gley zones mark fluctuations in groundwater level during and after deposition. The deposition took place mainly under water, but two moderately mature, truncated paleosols indicate breaks in the deposition and dry surfaces exposed for several thousand years. Both paleosols, located at 10 and 14 m below the top of the sequence, are associated with thin (10-30 cm), discontinuous gravel lenses rich in small flint fragments, many of them Palaeolithic artifacts.

68. For further discussion of paleosols, see below. The well-preserved bridges for the Roman aqueduct (see Fig. 1.7) lie just east of Kokkinopilos.

69. Runnels and van Andel 1993b. 70. Dakaris, Higgs, and Hey 1964.

7?


TABLE 3.5. MINERAL COMPOSITION OF REDEPOSITED TERRA ROSSA FROM POLJES AND LOUTSES IN WESTERN EPIRUS

Clay (<0.002 mm) Silt (0.010-0.064 mm) kaol ill smec qtz qtz feld or/pl

Sample (%) (%) (%) (%) (%) ratio

ALONAKI (#21: subsided loutsa?) VA94-12 28 66 7 - 93 7 1.1 VA94-32 28 66 6 - 93 7 1.1

AYIA (#1: loutsa) VA94-11 19 63 18 VA94-26 95 5 - VA94-27 24 67 9 + 95 5 0.3 VA94-29 22 62 16 + 95 5 0.6 VA94-31 20 67 6 - 100 - -

GALATAS (#27: polje) VA94-15 25 74 5 VA94-17 12 68 20 + VA94-19 14 54 32 + VA94-21 25 56 19 + 100 - -

LOUTSA (#22: loutsa) VA94-04 32 49 19 VA94-36 37 60 3 - 93 7 1.7

RODAKI (coastal marsh) VA94-06 65 35 2.8 VA94-35 22 73 5 - 58 42 2.2

Numbers in parentheses refer to locations in Table 3.1 and Figure 3.8. kaol = kaolinite; ill = illite; smec = smectite (chlorite and vermiculite); qtz = quartz

(+ = trace); feld = feldspar, orthoclase and plagioclase combined; or/pl = ortho-

clase/plagioclase ratio (height of 3.19 A peak over height of 3.24 A peak).

Figure 3.15. The raised polje of Kokkinopilos

71


Figure 3.16. Badland erosion at Kokkinopilos

Zone B overlies the deep red (10R-2.5YR) zone A, again with signs of desiccation and, here and there, minor erosion at the boundary. The zone itself is in part mottled with gray and rests on a karst surface. Just upstream from Tsiropolis, the basal portion is interbedded with layers of very fine, white sand, up to 30 cm thick, that consist of clear, well-formed calcite crystals with a few percent quartz, indicating precipitation as a playa evaporite combined with dust fall.

Notwithstanding the evidence for breaks in deposition and for many alternations between wet and dry conditions, the uniform grain-size distribution convincingly argues for the same restricted source of sediments throughout.

The raised, dissected polje at Morphi (Fig. 3.18) resembles that at Kokkinopilos. The sequence begins with a thin modern soil resting on a truncated mature paleosol that grades into a yellow zone similar to zone B at Kokkinopilos.71 The sequence contains a few thin, distal debris flows and four moderately mature truncated Bt horizons that mark hiatuses of significant duration. The yellow zone rests on a thick (ca. 2 m) tephra deposit derived from an eruption in central Italy and dated by Ar/Ar methods to 374 + 7 kyr B.P. Underneath, several meters of leached, gray (7.5YR 6/8) polje deposits rest on yellowish red (5YR) silty clays that are separated from the underlying karst surface by a very mature truncated Bt horizon.

Figure 3.17 (opposite). Cross section

through the incised polje deposits of Kokkinopilos. Left: Grain-size frequency diagrams for the size range 0.000-0.070 mm (silt mode shaded); the percentage of clay (<0.002 mm) is indicated in the upper left corner of each diagram. Center: Silt/clay ratio variation with time; data from Table 3.3. Right: Geological section based on Runnels and van Andel 1993b, fig. 6; depths in meters above sea level (masl) by altimeter survey. Zone labels from Tippett (in Dakaris, Higgs, and Hey 1964, pp. 221-225).

71. Pyle et al. 1998.

72

EARLY STONE AGE OF THE NOMOS OF PREVEZA 7

%/Phi unit

paleosol

-140

paleosol

-paleosol

paleosoll -:130

fine flint gravel

"handaxe"

-120 'paleosol

-pale gray silt 11-

paleosol

- 110 masl karst surface

0 10 log microns

/

f.0

( 0

I'a

c~o Nc%

(0

I0

Nz >

0

Ico

lw0 0

100

73

I m


Figure 3.18. Morphi polje outcrop with paleosols forming hard, protruding benches

AYIA composite profile

Bt (M2)

mottled terra rossa - VA94-27

Bt(M5) - VA94-28 gravel (debris flow)

- VA94-29

mottled terra rossa

- VA94-30

stratified mottled terra rossa

- (lowest stone tools)

- VA94-31 massive dark red terra rossa

Bt (>M5?)

limestone karst

Figure 3.19. Composite profile of Ayia loutsa showing the lithological sequence and paleosols. Palaeolithic stone tools occur throughout the section from ca. 333 to 339 masl. Paleosol maturity codes (MS 2, MS 5) from Table 3.8.

Occasional bands of fine to medium stream gravel were laid down by small ephemeral streams during a brief period of flooding of the polje floor or as thin debris flows produced by catastrophic failure of the slope mantle or a fan (Fig. 3.18). These bands testify to brief invasions of a high- energy regime, probably during times of exceptionally high rainfall because they are too thin and sparse to indicate major climate changes of stadial/interstadial or glacial/interglacial rank.

*) *? **

5 5 5 5

5 5 51 $5

339 -

335-

330 -

masi

74


Figure 3.20. (above) Stratified lower section of the Ayia loutsa looking

sions a truncated, very mature Bt horizon as at Morphi. Elsewhere the

west. Mousterian artifacts are embedded in the exposed section where the figure is poin ting; (below) detail of the Mousterian artifacts in situ.

The history of the deeply dissected loutsa at Ayia is simpler (Fig. 3.19). A rugged karst surface with a relief of 50-200 cm locally retains in depressions a truncated, very mature Bt horizon as at Morphi. Elsewhere the oldest deposit is a pale (5YR 6/8), finely crystalline dolomite sand analo- gous to the basal calcite sand at Kokkinopilos. The loutsa fill itself is a well-bedded, red and gray mottled deposit (Fig. 3.20), interbedded with layers bearing stone tools and near the top a debris flow. A little higher, a mature, truncated paleosol is overlain by a modern soil.

In summary, even without its telltale red color, redeposited terra rossa

unfailingly discloses its origin by its bimodal grain-size distribution.

Temporary or permanent wet conditions of deposition are indicated by soil features (mottling, gley). Dry conditions are revealed by color-band- ing, desiccation zones, and thin beds anomalously rich in fine quartz that

represent periods when mainly dust was being deposited. The fine mm- scale horizontal stratification seen associated with artifact scatters at

75


mcoas:t.al;pa-:n_ | . ^^HE^y ^^::^ v^^^RFigure 3.21. The Adriatic Sea during coastal plain Fi the last glacial maximum, 20-18 kyr B.P., when sea level was over 100 m

,-. present coast ' lower than today. River courses

lonian Sea across the emerged coastal plain are extrapolations. After van Andel and

J____________________________________________________________________ Shackleton 1982, fig. 4

Kokkinopilos and Ayia, and common elsewhere, is a result of deposition in very low energy conditions, far too low to entrain even the smallest flint

debitage. As the fine stratification is not easily seen, failure to observe it in the past has led to faulty stratigraphic interpretations, such as the view that at Kokkinopilos, and by implication in other redbed sequences, the artifact assemblages are on secondary location. Occasionally, artifacts are associated with debris flows or small-scale stream gravels, but those are rare brief incidents in the history of poljes.

Ultimately, the relevance of our knowledge of the genesis and history of Epirote poljes to our understanding of its Palaeolithic inhabitants de-

pends on our ability to fit a time dimension to them. The debate about the age of Kokkinopilos, so far the only polje viewed with an age perspective, still includes those who regard all redbeds as Pliocene in age (except for portions reworked by recent erosion) and others who see them as belonging to the Late Quaternary. We shall return to this subject below.

SEA LEVELS AND COASTAL PLAINS

All but the narrowest Mediterranean shelves are the flat surfaces of sediment wedges, which, when exposed at lowered sea level, may form wide coastal plains (Fig. 3.21). Often well watered and bordering today's rugged coastlines over long distances, they offered major wildlife resources 72. van Andel 1989; van Andel and and convenient migration paths for early humans.72 Shackleton 1982.

76


E 50-

I - t X, w -

100-

0 20

Figure 3.22. Global sea-level variations for the past 140,000 years, reflected by two oxygen isotope records based on bottom-dwelling

deep-sea foraminifera (Shackleton 1987; Labeyrie, Duplessy, and Blanc

1987) and calibrated with raised coral reef data. Dots: U/Th dates on corals

(Bard, Hamelin, and Fairbanks 1990; Stirling et al. 1995); lozenges: recent U/Th dates from Huon Peninsula coral terraces in New Guinea

(Chappell et al. 1996). Numbers at the top indicate oxygen isotope stages.

73. Imbrie et al. 1984; Martinson et al. 1987.

74. van Andel, Zangger, and Perissoratis 1990.

D 80

AGE (kyr)

MIDDLE AND LATE QUATERNARY PALEOSHORELINES

Traditonal Quaternary stratigraphic names, such as the Wirm or Weichsel

glacials, have no meaning in the Mediterranean. Therefore, we use here the global chronostratigraphy based on oceanic oxygen isotope stages (OIS) of Imbrie and Martinson.73 During the last 140,000 years, the interval of interest to us, global sea level was twice at a low glacial stand (ca. -120 m in OIS 6 and 2) and twice at an interglacial level slightly above (OIS 5e) or at (OIS 1) its present value. It remained at either extreme for only five to ten millennia, but occupied intermediate levels for roughly 100,000 years, from the climatic decline following the last interglacial (OIS 5d-e) throughout most of the subsequent pleniglacial.

Global glacial and interglacial sea-level positions can be estimated from oxygen isotope ratios (180/160) of bottom-dwelling microfossils that record the volume of seawater stored in ice caps. To obtain a true picture of sea level against time, the 180/160 curve must be calibrated with past sea-level

positions deduced from raised reefs and coastal terraces or from shore features submerged on continental shelves (Fig. 3.22). In the absence of offshore seismic reflection data for the Epirus shelf, past shores can be determined only by appropriate present bathymetric contours (Table 3.6), but Late Quaternary sediments tend to be thin on Greek shelves and errors of position are within the limits of precision of the isobaths.74

Over the past 140,000 years, the width and area of the emerged coastal plain in Epirus have varied a great deal (Fig. 3.23; Table 3.6). Except during the two brief glacial maxima, a total of some 20,000 years, the coastal plain, although continuous, was narrow. If the resource potential of an environmental zone is assumed to be roughly equal to its area, most of the time the coastal plains were at best equal in potential to the combined area of all poljes.

77


TABLE 3.6. APPROXIMATE PALEOSHORELINE DEPTHS AND COASTAL PLAIN WIDTHS, 140 KYR B.P. TO PRESENT

Interval Shoreline Coastalplain OIS (kyr B..) Event depth (m) width (km)

6 >135 Glacial maximum -130 10-20 5e 130-117 Interglacial peak 0 to +10 0

5d-a 117-74 Run-up to glacial -20 1-4 4 74-59 First glacial maximum -80 to -90 5-15 3 59-24 Mild phase -60 to -70 5-7 2 24-20 Main glacial maximum -120 10-20 2 18-15 Early deglaciation -110 to -100

1/2 15-8 Main deglaciation -90 to -20 1 11 Mesolithic starts -40 1-5 1 9 Mesolithic ends -20 1-2

OIS = Oxygen isotope stage. Shore depth from Figures 3.22 and 3.24. Coastal plain width is a representative range, indicating distance beyond the present

shoreline. (Note: at times the plain between Corfu and the mainland was

considerably wider.)

Whenever the sea stood above -80 m, the shelf between Corfu and the mainland was largely flooded. Given a present least depth between -45 and -50 m, however, the two were joined by a land bridge during all of OIS 6 and from 90 kyr B.P. to 10 kyr B.P. This persistent connection between Corfu and the mainland may have been a key point in strategies for hunting migrating herds of large herbivores.

In the Ambracian Gulf, which has a shallow sill, the -20 and -50 m isobaths show that between 10 and 105 kyr B.P. it was occupied by a lake. The glacial sediment load of the Louros and Arachthos Rivers, the only major sediment-carrying rivers in the area, was dumped there in the form of a delta complex very similar to the present one.75

Because of the rapidity of climate change and sea-level rise during the decline of the last glacial maximum, and its importance for the latest Palaeolithic and Mesolithic occupation in western Epirus, we need a more precise sea-level curve for that interval. This requires compensation for glacio- and hydro-isostatic effects, for which we may use Lambeck's corrected curve for Kavalla because that area is at the same distance from the northern European ice edge as Epirus.76 The corrected curve (Fig. 3.24) shows that the sea began to rise slowly some 18,000 years ago, accelerated rapidly around 14 kyr B.P. and continued through the Mesolithic to reach about -10 m 6,000 years ago.

Lambecks isostatically compensated curve reads time in radiocarbon years. If we convert the deglaciation sea-level history to calendar years by using the U/Th-dated curve of Bard, the deglaciation rise begins earlier, the Mesolithic shorelines are shallower (-30 m at the start and -15 m at the end of the period), and the coastal plain is proportionally narrower.77

75. Piper, Kontopoulos, and Panagos 1988.

76. Lambeck 1995, fig. 6:e; 1996. 77. Bard, Hamelin, and Fairbanks

1990.

78


Figure 3.23 (above). The emerged coastal plain off Epirus at six key moments between the maximum of '

the OIS 6 glacial to the present 0 present sea level

interglacial (OIS 1), according to - Table 3.6 and Figure 3.22. Isobaths E _ representing paleoshores are based E/ on nautical and topographic charts ) / and are highly generalized. The Bard et a. (1990) / Mesolithic shore corresponds with CZ -50- the Mesolithic interval in calendar 0, _-' years (Fig. 3.24). c) O / Figure 3.24 (right). Two sea-level rise / curves for the deglaciation interval of / Lambeck (1995) late OIS 2. Bard, Hamelin, and 0o / Fairbanks 1990, in calendar years, is a) -100 / based on U/Th dates of submerged _ - / coral terraces in Barbados; Mesolithic (cal BP) Lambeck 1995, in radiocarbon years - --

(Fairbanks 1989), is based on the' I . . I , , , i I , , ' , same samples and has been used to 20,000 15,000 10,000 5000 date the isostatically compensated local sea-level history. Age in years before present

79


THE COAST OF EPIRUS IN THE LAST INTERGLACIAL

(OIS 5)

For all but one of the periods in the interval OIS 1-6, the associated

paleoshores are now below sea level and, in the absence of high-resolution seismic reflection studies, their nature and true position cannot be known

accurately. The exception is the 10,000-year long peak of the last interglacial (OIS 5e), with observed sea-level positions that in stable areas range from 0 to +10 masl,78 although most of those elevations do not differ significantly from the present level after a correction for glacio- and hydro- isostatic effects has been applied.79

In Greece the paleoshores of the last (Tyrrhenian) interglacial have in

many places been raised above their original levels by coastal tectonics.

They are marked by a distinctive warm fauna of corals and large robust mollusks, including the warmwater gastropod Strombus.80

In coastal Epirus raised shore deposits are in evidence at several points (Fig. 3.25). They have generally been regarded as Late Pleistocene or earliest Holocene in age, notwithstanding the high uplift rates that the low sea levels of that interval would imply. A large outcrop exposed at Anavatis, on the other hand, has been mapped as Pliocene on the basis of shallow- water agglutinating foraminifera of little stratigraphic value.81

The Anavatis complex, now at approximately 40 masl, is exposed in the south wall of a sand and gravel pit where it consists of thick, massive to thin-bedded layers of unconsolidated, white to pale yellow, fine, well-sorted sand (Fig. 3.26:1, 3). The sands are interbedded with thin (2-20 cm) layers of gray-black, finely (1-10 mm) laminated silt deposits with the characteristic grain-size distributions of marsh or tidal flat silty clays (Fig. 3.26:2, 8). Locally, sand-filled channels are cut into underlying beds. Some of the

moderately calcareous sand beds contain abundant coastal marine mollusks such as Cerastoderma, while shell debris is common in burrows in the laminated silts. There are also a few lenses of rounded, well-sorted fine (1- 5 cm) gravel. There can be no doubt that this is a coastal or very shallow marine deposit.

In the opposite north wall of the pit, a thick series of unconsolidated medium-coarse sands and fine-coarse gravels is exposed, traversed by many faults of small displacement. The coarse strata are interbedded with thick (5-20 cm) lenses of fine sand or grayish green silty clay, perhaps formed in pools on a braided, low-angle fan. Grain size (Fig. 3.26:4, 5) and chaotic bedding point to braided or torrential streams. At ca. 48 masl the sequence is topped by a paleosol. The maturity level of this paleosol (MS 4/5)82 and extensive frost-shattering of the finer gravels indicate deposition during the cold Late Pleistocene pleniglacial (OIS 4-3). The paleosol contains an early Mousterian industry (see below).

The torrential unit, although apparently deposited in a low-lying area, is too close to the coastal unit of the opposite scarp to be contemporane- ous. Moreover, if the coastal sediments were late glacial in age, the low sea level of the time would require an uplift rate of some 4 m/kyr, quite in excess of other tectonic rates in the region (as discussed above). More probably, they underlie the torrential unit and so are of interglacial age. Sea

78. Bard et al. 1993; Chen et al. 1991; Edwards et al. 1987; Ku, Ivanovich, and Luo 1990; Stirling et al. 1995.

79. Lambeck and Nakada 1992. 80. E.g., Kelletat 1974; Kelletat et

al. 1976; Keraudren and Sorel 1987; Schr6der and Kelletat 1976.

81. Etude geologique. 82. For a discussion of paleosol

maturity stages (MS), see below, pp. 86-89.

80


Figure 3.25. Locations of raised paleoshore deposits of the last interglacial (OIS 5e and OIS 5c) in coastal Epirus

Figure 3.26 (below). Cumulative grain-size distributions of coastal sediments of the last interglacial and early Holocene from the Anavatis sand pit (1-5, 8) and Alonaki Beach (6, 7): 1) VA94-2, shallow marine or dune sand; 2) VA94-3, silty laminated marsh clay; 3) VA94-37, shallow marine or dune sand; 4) VA94-39a, torrential stream sand; 5) VA94-39b, same; 6) VA95-2, Holocene dune sand; 7) VA95-3, same; 8) VA94-38, marsh or tidal silty clay. Sedigraph analyses of the fraction 4.000-0.002 mm.

100

-- z

0

Q _

- 50-

LU

D

0-

0

.5 .25 .125 .062 .031 .016 .008 .004 .002 .001

GRAIN SIZE (mm)

8i


level for that time ranged from about +10 to -20 m, yielding a reasonable

uplift rate of 0.4-0.6 m/kyr for the Anavatis area, as luminescence dates confirm.83

Other paleoshore deposits occur at several points along the coast west of Preveza. At Alonaki Beach and Tsarlambas (Fig. 3.27), a few meters of well-consolidated, horizontally bedded, low-angle cross-bedded series of fine to medium, well-sorted sands topped by a truncated very mature

paleosol (MS 5) are exposed at the base of the coastal cliffs. The sand is

moderately calcareous and contains many small, thick-shelled gastropods and a few corals. We regard these bench-forming deposits, now located ca. 3 masl, as a Tyrrhenian paleoshore which can be traced intermittently as far as the cape at Mytikas. Thick-shelled gastropod fragments, probably Strombus, also occur in boreholes in a down-faulted sequence at the entrance to the Pantokrator suburb of Preveza.84

A similar sand complex crops out east of Preveza at Ayios Thomas on the flank of a coastal hill. Topped by a red (10R 6/6) mature paleosol, the well-sorted, subrounded medium sands with lenses of rounded gravel may represent a raised Tyrrhenian coastal fan.

At Rodaki, south of the mouth of the Paliourias River, a raised coastal

complex is exposed consisting of weakly consolidated, low-angle, cross- bedded, fine, pale yellow (10YR 8/2) sand with thin layers of coarse sand and stringers of small pebbles. It is topped by a dark red (2.5YR 4/4-3/4), mature (MS 5) truncated Bt horizon. Because the complex is located 8-20 masl, we regard it as another Tyrrhenian beach and coastal dune deposit. Like all other coastal deposits, and in stark contrast to the terra rossa, the sand contains abundant feldspar (Table 3.7). Nearby is an important site of consolidated red, thin-bedded sand and gravel containing a Middle Palaeolithic industry (see below), but because of

complex active faulting and poor outcrop conditions its relation to the assumed interglacial shore deposits is unclear. Similar coastal deposits with characteristic Tyrrhenian fauna occur at 30 and 10-12 masl on Corfu.85

Figure 3.27. The raised Tyrrhenian beach at Tsarlambas, visible as a low, rocky coastal deposit on the right

83. An infrared stimulated luminescence (IRSL) date of ca. 128 ?+ 23 kyr B.P. places the deposit within the main phase of the last interglacial (OIS 5d- e). Another IRSL date of ca. 188 ? 30 kyr B.P. is questionable for technical reasons and because it places the interbedded marsh deposit in the OIS 6 glacial maximum when sea level was low. See below and Table 3.10.

84. P. Paschos (pers. comm.). 85. Sordinas 1983, p. 343, table 1.


TABLE 3.7. MINERAL COMPOSITION AND LAST INTERGLACIAL COASTAL WESTERN EPIRUS

OF MODERN SANDS IN

Sample Quartz (%) Feldspar (%) Or/PI Ratio Remarks

ALONAKI BEACH (Holocene dunes) VA95-02 53 47 3.2 VA95-03 93 7 1.2 fair amount of calcite

ANAVATIS (coastal deposits) VA94-02 57 43 2.6 VA94-37 56 44 3.0

RODAKI (coastal deposits) VA94-05 54 46 2.1 fair amount of calcite VA94-06 65 35 2.8 VA94-35 58 42 2.2

Or/Pl Ratio = orthoclase/plagioclase ratio.

86. Bottema 1974, 1994; Tzedakis 1993.

87. Willis 1994; Culiberg and

Sercelj 1996; van Andel and Tzedakis 1996.

88. Willis 1992, 1994; Turner and Sanchez-Gofii 1997.

89. Smit and Wijmstra 1970. 90. Bennett, Tzedakis, and Willis

1991; Tzedakis 1993.

At Ormos Odysseos, on the south side of the Acheron River valley, the remains of a thin alluvial fan sequence, its top at 4 masl, occur on a low, north-dipping karst surface. It consists of strongly consolidated, dark red (2.5YR 3/4), coarse sand and red clay, overlain by a definitely mature, Pleis- tocene coastal dune paleosol, now a little above present sea level and of last interglacial age. Associated Middle Palaeolithic findspots are discussed below.

VEGETATION HISTORY AND CLIMATE, 140-10 KYR B.P.

Our understanding of the Late Quaternary vegetation and climate history of western Epirus rests mainly on long cores from Lake Ioannina, first studied by Bottema and more recently byTzedakis (Fig. 3.28).86 The cores, supplemented with other data from Greece, Italy, and the Balkans, reasonably reflect the long-term climatic history of northern Greece, but afford little insight regarding the diverse local conditions of the mountainous terrain of western Epirus with its largely orographic climate conditions.87 Cores in lowland and highland lakes are beginning to provide some detail for local areas and for a range of elevations, but because they cover only later phases of the deglaciation period and the Holo- cene, the results are not directly applicable to the long interval from 60 to 25 kyr B.P.88

During the penultimate glacial of OIS 6, a discontinuous steppe vegetation of sagebrush (Artemisia), chenopod species (indicative of aridity), and grasses predominated in southern Europe. Cold-stage pollen from Tenaghi Philippon in Macedonia contained Eurotia ceratoides and Kochia

laniflora, species found today in the central Asian steppe and semidesert that point to a cold, arid climate.89 In sheltered spots of the western Balkans and mountains of Italy, however, scattered temperate tree populations sur- vived in refugia where temperature variations were not extreme and precipitation was sufficient, thus enabling a swift return of the woodland when the climate improved.90

83


LATE DEGLACIATION (I2-IO KYR B.P.)

Climate warming, moister; oak woods mixed with warmth-

loving species develop; coastal regions have open Mediter- ranean woodland of pine, evergreen oak, wild olive, and

pistachio.

L- A' LAST GLACIAL MAXIMUM (OIS 2)

Dry, cold climate; sagebrush and chenopod steppe, open deciduous woodland on south-facing slopes or middle elevations benefiting from orographic rains.

MID-GLACIAL MILD INTERVAL (OIS 3) Milder, moister climate; steppe with open deciduous woodland in favorable places.

FIRST GLACIAL EXPANSION (OIS 4)

Cold, dry climate; steppe gains on open woodland; trees con- \tract into refugia toward final phase.

TRANSITION TO GLACIAL (OIS 5D-A)

Cool and warm periods alternate between Mediterranean mixed evergreen and deciduous woodland and cold, dry steppe.

LAST INTERGLACIAL (OIS 5E)

Climate a little warmer than now; deciduous oak/elm forest followed by Mediterranean wild olive and evergreen oak woodland; maximum summer insolation.

PENULTIMATE GLACIAL (OIS 6) Cold, arid climate; chenopod and sagebrush steppe; refuigia for temperate trees.

When the OIS 5 interglacial began, trees spread outward from the

refugia in a vegetation succession beginning with deciduous oak (Quercus) and elm (Ulmus), followed in southern Europe by a major expansion of Mediterranean forest characterized by evergreen oak and values of wild olive (Olea) even higher than in the Holocene.91 This was the time of maximum summer insolation (12-13% above present value) of the last

interglacial and indeed the last 150,000 years. In the eastern Mediterranean, the climatic oscillations that led from

the end of the full interglacial (OIS 5e) to the first large ice advance in OIS 4 produced alternations between the cold, dry chenopod and sagebrush steppe and returns of the Mediterranean mixed evergreen and deciduous woodland.92 These interstadial landscapes were more open than in OIS 5e, however, and semidesert plant communities were present even

during warmer phases.93 Because their refugia were close, tree populations expanded rapidly in each interstadial, but the gradual climatic deteriora-

Figure 3.28. Climate and vegetation changes during the last two glacial- interglacial cycles (OIS 1 through OIS 6), illustrated by the variation of the arboreal pollen sum. Based on Tzedakis 1993, 1994

91. Tzedakis 1994. 92. Bottema 1994; Tzedakis 1994;

Wijmstra 1969; Wijmstra and Smit 1976; Wijmstra, Young, and Witte 1990.

93. Cheddadi and Rossignol-Strick 1995.

E

c 40- .

80-

0 100 Tree pollen (%)

84


tion and increased aridity are evident in ever larger expansions of chenop- ods and sagebrush. Still, the cold, arid steppe took over only toward the end of OIS 4, driving the warmth-loving tree populations into refugia even in northwest Greece.94

OIS 3 is marked by several warmer intervals during which a mixed deciduous woodland with beech (Fagus), oak, elm, hazel (Corylus), and lime (Tilia) partly covered northern Greece; southern Greece was sparsely repopulated by deciduous and evergreen oak, pine (Pinus), and juniper

(Juniperus) woodland.95 Although OIS 3 was a good deal milder than is

usually assumed, the Mediterranean woodland at the time was open in character; highest tree densities are recorded in only a few places with optimal soil conditions and sufficient moisture, such as northwestern Greece.

During the latest Pleistocene a chenopod and sagebrush steppe typical of a dry, cold climate covered most of the Balkans and Greece.96 A low but persistent level of tree pollen suggests, however, that the monotony of the steppe may have been relieved by patches of very open deciduous woodland. This woodland would have been concentrated on favored south-fac-

ing slopes in middle elevations that benefited from orographic rains, precipitation being a more important limitation than temperature.97

This vegetation type vanished around 11 kyr B.P. and was replaced in northern Greece by a deciduous oak forest mixed with more warmth-lov-

ing species such as hop hornbeam (Ostrya) and pistachio (Pistacia).98 In coastal regions the Mediterranean woodland of evergreen oak (Quercus ilex), pine (Pinus halepensis), Phyllyrea, wild olive, and pistachio took over

slightly later. The impact on animal populations was considerable: wandering herds

of herbivores, such as wild ass (Equus hydruntinus), bison, and perhaps Saiga antelope, vanished from the cold coastal and inland plains and were

replaced by the more diverse but far more dispersed wildlife of the forest, dominated by red deer and wild boar.99

In northern and western Europe, sharp oscillations between warm and cold climates, of which the Younger Dryas (12.9-12.5 to 11.6-11 kyr B.P.) was the last, marked the deglaciation period.100 Whether such oscillations had any real impact in southeastern Europe and the Near and Middle East is in doubt; neither Bottema nor Willis find convincing evidence for them in southeastern Europe during the deglaciation.'10

94. Tzedakis 1993. 95. Wijmstra 1969; Tzedakis 1994. 96. van Zeist and Bottema 1982;

Willis 1994; Willis et al. 1995. 97. Willis 1994. 98. Bottema 1974, 1978. 99. Jameson, Runnels, and van

Andel 1994, pp. 331-338; Miracle 1995.

100. Bard and Kromer 1995; Kromer et al. 1995.

101. Bottema 1995; Willis 1994. 102. Bailey, Papaconstantinou, and

Sturdy 1992. 103. van Andel 1998a.

CHRONOLOGY OF THE LATE QUATERNARY OF WESTERN EPIRUS

Open-air sites are notoriously difficult to place in a chronological context.

Bailey, in expressing doubt regarding the utility of Palaeolithic open-air sites, had this difficulty very much in mind.102 We have approached the

dating problem in two ways: 1) by paleosol stratigraphy, designed to ar-

range sites in stratigraphic order by means of paleosol maturity levels;103 and 2) by the use of thermal luminescence (TL) and infrared stimulated luminescence (IRSL) to obtain calendrical ages for the aeolian silt fraction in redeposited terra rossa.

85


PALEOSOL STRATIGRAPHY IN GREECE

Mediterranean soils evolve in a summer-dry, winter-wet climate that is relatively uniform over large areas. In time the soils mature, forming chronosequences with time-dependent characteristics.104 They are therefore valuable for the identification of Palaeolithic surfaces, stratigraphic correlation across diverse bedrock lithologies, and temporal sequencing of findspots.105 The paleosols discussed here are the alfisols typical for the extensive regions of Mesozoic and Paleogene limestone and flysch and of the Quaternary alluvium derived from those terranes. On different sub- strates, other kinds of paleosols are found that may also be red or brownish red, such as the rendzinas on Late Tertiary marls in the Peloponnese, but they are not considered here.

In using paleosol chronosequences we have limited ourselves to traditional descriptions of soil horizons based on field characteristics that allow the assignment of paleosols to six maturity stages (Table 3.8).10?6 Chemical methods can refine the definitions of the stages, but have not yet been used widely in Greece.107

In a typical Mediterranean soil profile, winter rains percolating down from a dark organic A horizon leach a pale E horizon and precipitate solutes in a yellow-brown to red Bt horizon, which becomes progressively enriched in iron oxides that intensify the color with time (Table 3.8). The Bt horizon has an internal structure evolving from small granular aggre- gates to ever larger blocks and prisms called peds; accumulating illuvial clay particles form shiny clay films on ped surfaces.108 Wherever CaCO3 is present in the substrate or the groundwater, dissolved CaCO3 precipitates to form a calcareous Bk horizon below the Bt. Underneath, the soil grades into the unaltered or only little altered substrate, the C horizon.

The Bt horizon expresses its increasing maturity by means of changes in color, structure, and the thickness and abundance of clay films.109 The Bk horizon similarly develops as a sequence of precipitated CaCO3 stages.110 Both horizons ultimately reach a point where no further maturation can be detected, unless erosion or renewed deposition terminates the process and sets a new sequence in motion.

The sequence of maturity stages is shown in Figure 3.29, using dates based on superimposed archaeological sites, 14C-dating of organic sediment particles, and luminescence dates of silt grains. All of these dating methods estimate the time of deposition of the substrate and hence the onset of soil formation. The time needed to form a given paleosol can be determined from U-series dates of calcareous paleosol nodules.111 Digests of all described paleosol Bt horizons in western Epirus are listed in Table 3.9 with their maturity stages.

Paleosol stratigraphy has worked well in the Peloponnese, Thessaly and Macedonia, and in the Pindos region of Epirus,112 but western Epirus raises problems of its own. Because the redeposited terra rossa is often initially red, clay-rich, and CaCO3-free, the abundance and thickness of clay films on ped surfaces, the remaining Bt diagnostics, are the only maturity criteria. Color is of no value except where reduction in a water- logged depositional environment has bleached the sediment and started the process of soil formation.

104. Birkeland 1984; Vreeken 1975. 105. Holliday 1989; Morrison 1976. 106. Birkeland 1984, app. 1;

Retallack 1988. 107. Fitzgerald 1996; Harden 1982;

Harden and Taylor 1983; McFadden, Ritter, and Wells 1989; Smith, Nance, and Genes 1997.

108. Birkeland 1984, p. 16; Retallack 1990, p. 40.

109. Birkeland 1984, figs. 1-6, 8-10, tables 1-4, 8-2.

110. Birkeland 1984, fig. A-2; Machette 1985.

111. Ku et al. 1979; Ku and Liang 1983.

112. Pope and van Andel 1984; Runnels and van Andel 1993a; Woodward, Macklin, and Lewin 1994.

86

87 EARLY STONE AGE OF THE NOMOS OF PREVEZA

TABLE 3.8. MATURITY INDICATORS OF THE B HORIZON OF GREEK

QUATERNARY PALEOSOLS

BtHorizon Bk Horizon

Stage Color Structure Clay Films CaCO3

>2 kyr B.P

MS 1 10YR, medium gray-yellowish brown granular none none

>4 kyr B.P

MS 2 10YR-7.5YR, yellowish to reddish brown subangular blocky thin, few I-II

ca. 10-15 kyr B.P

MS 3 7.5YR, reddish to dark brown subangular blocky thin, common II

ca. 40 kyr B.P.

MS 4 5YR, yellow red to reddish brown angular blocky thin, many II-III

ca. 80 kyr B.P

MS 5 2.5YR, reddish brown to red angular blocky to small prismatic thick to continuous III-IV

ca. 110-200 kyr B.P MS 6 2.5YR to 10R, red-brown to red medium to large prismatic or platy thick, pervasive IV

MS = maturity stage; Color = Munsell color chart. For Bt horizon diagnostics, see Birkeland 1984, app. 1. Bk horizon character-

istics after Birkeland 1984, fig. A-2. Boundary ages (from Fig. 3.29) are approximations.

I I . . I I I I I I

Dating method

+ archaeological

o calibrated radiocarbon

* U/Th disequilibrium

* TL/IRSL

X stratigraphic M

Figure 3.29. Maturity stages and

approximate ages of the Mediterra- nean paleosol chronosequence. Note that paleosol maturity asymptotically approaches a final stage beyond which no change can be observed.

Age scale is logarithmic; no vertical scale. Dates are from Demitrack 1986; Pope and van Andel 1984; Pope, Runnels, and Ku 1984; Runnels and van Andel 1993b; Zangger 1993; and Table 3.9 (below). After van Andel 1998a, fig. 5

MS2

~~~~~~MSI ~~~~~-- . .

|~~~~~~~~~~~~~~~~~~~~~~~~~~~i' io i

... ..

.....

*+*

+ .

......:: ii!

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

~~~~~~~~~~~~~...:. +S *iii

..+.... .. ~,-.. :' r,.

+e ~ ~~ ~ ~~~ is '; ii

MS6 MS5

MS4

.... :: ::

1S3 I-

100,000 years BP

. * f * II I

, I 0

10,000 1,000


TABLE 3.9. SHORT DESCRIPTIONS AND MATURITY STAGES OF PALEOSOL BT HORIZONS AT KEY SITES IN COASTAL EPIRUS

Site Sample Maturity Stage Color Structure Clay Films

P2

P3topBt P3baseBt P4 VA94-12a VA94-13 P5

BEACH

Pltop, VA95-01

Plmid, VA95-02 to 4 Plbase

MS 4/5 MS 4 MS 5 MS 5 MS 5 MS 4 MS 5

MS 1 MS 2 MS 5

2.5YR 5YR 2.5YR 2.5YR 2.5YR 5YR 2.5YR

5YR 5YR 10YR

ang blocky ang blocky ang blocky ang blocky ang blocky ang blocky ang blocky

Bt on sandb Holocene dune sandb Bk on sandb

thick, many thick, many thick, abundant

thick, abundant massive

thick, many thick, abundant

(CaCO3 stage II)

AMMOUDIA SS92-21 SS92-21

P1

P2top P2base

MS 3/4 MS 3/4 MS 5

5YR 5YR 2.5YR

ang blocky ang blocky ang blocky

medium medium

thick, abundant

ANAVATI S

SS94-16

AYIA

SS93-9.1 SS93-9.2 SS93-9.2 SS93-9.2

Pltop

AYIOS THOMAS

MS 5

top, VA94-27

base, VA94-28 to 31

base

2.5YR

MS 4 MS 2 MS 4/5 MS 6

MS 2

2.5YR 10YR 2.5YR 2.5YR

7.5YR

med ang blocky

ang blocky med granular ang blocky platy

med granular

thick, abundant

(no information) few

thick, abundant massive, pervasive

thin, few

CHEIMADIO

SS94-2 SS94-18

VA94-01 VA94-07

MS 5 MS 3/4 MS 3

2.5YR 5YR 7.5YR

blocky, prism f ang blocky ang blocky

thick, abundant thin, few thin, pervasive

Ptop, VA94-14 to 24 MS 4 2.5YR ang blocky medium, abundant

KOKKINOPILOS

SS91-3 VA93-05

LOUTSA

SS94-12 VA94-04

MS 5

MS 4

2.5YR

2.5YR

med ang blocky

med ang blocky

P4

Top, VA94-05, 34

Base, VA94-06, 35

MS 5 MS 4 MS 5

2.5YR 5YR 2.5YR

ang blocky Bt on sandb Bt on sandb

thick, abundant

Sample = Sample number (preface VA) or paleosol profile number (preface P); Color = Munsell soil color chart; Structure = f(ine), med(ium) ang(ular) blocky (Birkeland 1984, app. 1).

aFrom same stratum as "chipping floor" industry. bColors on sand or sandstone tend to be 1-2 hue values lighter than on clay-rich sediments. cAlong main coastal highway from Preveza to Albania.

ALONAKI SS92-23.2 SS92-22.7 SS92-22.7 SS92-22.7 SS92-22.7 SS92-22.7 SS92-22.1

ALONAKI SS94-23 SS94-23 SS94-22

GALATAS SS92-13

RODAKI

Above E55c SS92-15 SS92-15

thick, many

thin, pervasive

88


Erosion followed by deposition produces complex soil sequences. An

example can be seen on the Tyrrhenian beaches west of Preveza at Alonaki where the interglacial beach sand is topped by a truncated, light red (10R 6/6), very mature Bt horizon overlain by 2-4 m of similar but unconsolidated dune sand (Fig. 3.26:6,7) containing two paleosols. A lower immature (MS 2) Bt horizon of early Holocene age has Mesolithic finds on top of and within its upper 50 cm. It was later truncated and covered by well- sorted dune sands, which locally show an uppermost, very immature

paleosol (MS 1). Dune migration and local deflation are continuing today.

DATING OF EPIRUS SEDIMENTS AND FINDSPOTS

113. Dakaris, Higgs, and Hey 1964; Higgs and Vita-Finzi 1966; Higgs et al. 1967; Higgs and Webley 1971.

114. Huxtable et al. 1992. 115. Bailey 1992. 116. Bailey, Papaconstantinou, and

Sturdy 1992; Huxtable et al. 1992. 117. Huxtable et al. 1992. 118. Galanidou et al. 2000. 119. See, e.g., Wintle 1996.

The first attempt to obtain a chronology of the Palaeolithic and Mesolithic in Epirus was undertaken by Cambridge University teams beginning in 1962. Higgs obtained a series of radiocarbon assays on bone, charcoal, and other materials from Asprochaliko and Kastritsa.113 Kastritsa was dated to approximately 10-23 kyr B.P. (10,000-20,000 b.p.), and the Upper Palaeo- lithic deposits at Asprochaliko evidently began to accumulate somewhat earlier, ca. 29 kyr B.P. (26,000 b.p.), but otherwise overlapped the Kastritsa

deposits in time.114 The Middle Palaeolithic levels at Asprochaliko proved to be beyond the effective range of the radiocarbon technique (at that time, ca. 39 kyr B.P.) and Higgs found nothing datable at Kokkinopilos.

The later work of Cambridge University teams has added to the chronology. Radiocarbon dates from Late Upper Palaeolithic Klithi (10,420 b.p.-16,490 b.p.)15 fall between the glacial maximum (18-20 kyr B.P.) and the last cold event before the onset of the Holocene, the Younger Dryas of 11-12 kyr B.P. (10-11 kyr b.p.). New dates are also available from Aspro- chaliko, based on 14C assays and TL analyses of sediments and burned flint.116 The new dates place layers 16 and 18 (basal Mousterian) at ca. 98.5 kyr B.P. (TL), and layer 14 (upper Mousterian) at ca. 39 kyr B.P. (37,000 b.p.). Uncertainty remains, however, in part because the TL dates lack the detail necessary to evaluate them. An attempt to date sites Alpha and Beta at Kokkinopilos with optically stimulated luminescence was inconclusive, suggesting only that sediments at the test sites might be older than 150 kyr B.P.117 New dates from Kastritsa, placing the beginning of occupation somewhat earlier, range from 27 to 16 kyr B.P. (24-13 kyr b.p.).118

This program, although adequate for the study of the stratified deposits in the rockshelters, is of little use for dating open-air sites. Most open-air sites are too old to be dated by 14C, even though reliable dates are now being obtained up to 45,000 B.P., and substances suitable for K/Ar or U/Th dating, such as tephra or flowstone, are lacking. Relative dating is difficult in the absence of floral or faunal remains, and comparisons of lithic industries are useless in the absence of stratified deposits with a succession of lithic types.

LUMINESCENCE DATING OF SEDIMENTS

Thermoluminescence dating of sediments has been practiced with varying success since 1979.119 Since 1985 it has been possible to date sediments using optical dating methods in which a light-sensitive lumines-

89


cence signal is measured.120 While several light-sensitive signals have been

used, infrared stimulated luminescence (IRSL) is the preferred method for dating loess and colluvial sediments derived from loess.121 Our confirmation of a suggestion byTippett and Heythat the silt in the Kokkinopilos redbeds might be the result of long-distance wind transport,122 a suggestion rejected by Bailey, made this component an attractive target for luminescence dating, notwithstanding an earlier, inconclusive attempt by Debenham.123 For this purpose, a suite of samples was collected in sealed, foil-wrapped plastic tubes under conditions of total darkness. TL and IRSL

dating were carried out by Li-Ping Zhou in the Godwin Laboratory at

Cambridge University and IRSL dating by Andreas Lang at the For-

schungsstelle fur Archaometrie, Max Planck Institut fiir Kernphysik, in

Heidelberg, Germany (Table 3.10).124

During long-distance aeolian transport, silt-sized quartz grains will have been fully bleached before deposition. After deposition the grains become covered with more grains and are exposed to radiation from natural sources of radioactivity in their environment. Dose rate estimates de-

pend on uranium, thorium, and potassium concentrations determined by alpha counting, X-ray fluorescence, and neutron activation analysis. In the current study, two methods were used to determine the radiation received

by the samples since deposition, known as the equivalent dose (DE). In the additive method (a) a series of laboratory doses is given to sample disks in order to increase the luminescence signal. This produces a luminescence growth curve that, when extrapolated back to a base level provided by a bleached sample (bleaching: 180 minutes forTL, 60 for IRSL), allows the natural signal accumulated since the last exposure to light to be converted to a measure of the equivalent dose. With the regeneration method (r), f3 doses are given to sample disks after exposure to light. A match of the natural luminescence signal with the regenerated one then allows the determination of the DE. The application of this method is ultimately limited by the long-term stability of the signal and by reaching a dose level at which the luminescence signal no longer increases with further applied doses. Thermal instability will result in an underestimation of the true age, whereas saturation of the luminescence signal will permit estimation of a "greater than" age.

For samples of nonwindborne sediments, bleaching of the earlier geological signal may be incomplete. This will result in an overestimation of the TL age, and possibly the IRSL age, if the laboratory bleaching is more effective at reducing the signal than the original light exposure. For IRSL, the signal can be reduced to 3% of its initial value by exposure to one minute of bright sunlight, whereas 1,000 minutes are required for the TL signal from the same grains.125 Therefore, for nonwindborne deposits, ages obtained using the IRSL data sets are preferred.

With the TL and IRSL dates (Table 3.10) and the paleosol maturity stages described above (Table 3.9), we compiled a chronological diagram of the last two glacial/interglacial cycles in the Preveza region that for the first time seriates many open-air sites (Table 3.11). Its "golden spikes" are the confirmation of the existence of an older Middle Palaeolithic between 60 kyr B.P. and the end of the last interglacial, the identification and dating

120. Huntley, Godfrey Smith, and Thewalt 1985.

121. Lang and Wagner 1996. 122. Dakaris, Higgs, and Hey 1964. 123. Bailey, Papaconstantinou, and

Sturdy 1992. 124. Zhou, van Andel, and Lang

2000. 125. Wintle 1997.

9o


TABLE 3.10. THERMOLUMINESCENCE AND INFRARED STIMULATED LUMINESCENCE SEDIMENT DATES FOR WESTERN EPIRUS

Sample Method

VA93-05 TLr VA94-27 IRSLa

IRSLr VA94-29 IRSLa

IRSLr VA94-30 IRSLr VA94-32 TLa

TLr IRSLa

VA94-36 TLa TLr

IRSLr VA94-37 IRSLr

TLa VA94-38 IRSLr VA95-01 TLa

IRSLa VA95-02 TLa

IRSLa VA95-03 TLa

IRSLa VA95-04 TLa

IRSLa

DE

385 ? 79

27?+2

31?4 275 ?22 354?40 376 ?49

56?+3

56?+2

51?2 320 ?24

278 ?16

294 ?41

289?49

443 + 34

445 ?+ 65

14.8 ?0.7 18.3 ?1.4 34.4+ 2.6 31.3 ? 8.8

6.2 ?0.7

7.1 ?0.7 9.3 ?+0.9

11.0?1.8

D/R

4.22 4.47

4.16

4.52 5.57

5.45

2.18

2.23 2.03 1.82 3.32 2.82 1.57 1.42 1.97 1.71

Age (kyr B.P.)

91+?14 6.1?0.6

7?1 65.5 ? 6.8

84?11 83.1? 12

10?2

10?2

9?2 59 ? 9 51?+8

52?+8

<128 ?23

<185?28

<188?30 7?1

11.1?+1 10.4? 1.6 10.5 ? 3.0

3.9 ?0.6

4.6 ?0.4

4.7?0.7 5.8 ?0.6

Remarks

Kokkinopilos, paleosol (MS 5) Ayia, upper paleosol (MS 2)

Ayia, lower paleosol (MS 4)

Ayia, lower paleosol (MS 4/5) Alonaki, redeposited terra rossa (MS 3)

Loutsa, surface paleosol (MS 4)

Anavatis, coastal sand

Anavatis, coastal marsh Alonaki Beach, early Holocene paleosol (MS 2)

Alonaki Beach, early Holocene paleosol (MS 2)

Alonaki Beach, Holocene dune

Alonaki Beach, Holocene paleosol (MS 1)

IRSL = infrared stimulated luminescence; TL = thermoluminescence; a = additive method; r = regeneration method; DE = equivalent dose; D/R = dose rate.

of coastal deposits of the last interglacial, and the presence of Palaeolithic industries belonging to or predating the interglacial. The many sites not dated byTL or IRSL have been arranged by the maturity of the paleosols with which they are associated. As usual, the dates are not as numerous as one might wish and the chronostratigraphic units are therefore long. The

findspots included in each unit are neither necessarily synchronous nor can they be placed in chronological order. Except for sites on alluvial fans, whose lithic assemblages may be in secondary location, all Palaeolithic and Mesolithic findspots are associated with former, level depositional surfaces of poljes, loutses, and, in a few cases, coastal plains.

LATE QUATERNARY AND PALAEOLITHIC CHRONOLOGY

Paleosols are common in stratigraphic sections of the raised poljes and loutses of western Epirus, and many of these landforms are capped with very mature paleosols that allow the sequencing of Palaeolithic finds (Table 3.9). The approximate paleosol-based dating has been augmented with thermal and optical luminescence dates (Table 3.10) to construct a provi- sional chronology of the Palaeolithic and Mesolithic periods (Table 3.11).

The Palaeolithic sequence in Epirus begins approximately 200 kyr B.P., perhaps even earlier. The Mousterian is underway before or during

9 I


TABLE 3.11. CHRONOSTRATIGRAPHIC DIAGRAM FOR ARCHAEOLOGICAL SITES, SEDIMENTS, AND PALEOSOLS IN THE PREVEZA REGION

Age Site, Sediment, or SedimentAge Paleosol Maturity Stage (kyr B.P) OIS Paleosol (cal kyr B.p) (estimated age range)

Holocene Alonaki Beach: dunes Alonaki Beach: upper soil

Ayia: upper paleosol Alonaki Beach: site Alonaki: young paleosol Tsarlambas: site

4.6 ?0.4

5.8 ?0.6

6.1 ? 0.6-7? 1 9.4+ 1-10.5 + 3.0

10?2-9?+2

MS 1 (1-3 kyr B.P.)

MS 2 (4-10 kyr B.P.)

MS 2 (4-10 kyr B.P.)

MS 2/3 (10-30 kyr B.P.)

Glacial maximum

Kastritsa

Asprochaliko Klithi

Galatas paleosol Asprochaliko

Loutsa paleosol

MS 4 (30-70 kyr B.P.)

39-25a

51 ?8-59?9 MS 4 (30-70 kyrB.P.)

4

5 d-a

5e

6

First major ice advance

Ayia Kokkinopilos Asprochaliko

Tyrrhenian/Eemian interglacial Anavatis Ormos Odysseos Alonaki: site

Kokkinopilos handaxe?

65-85 91 ?14

96-98a,b

MS 4/5 (60->100 kyr B.P.)

MS 5 (70->100 kyr B.P.)

128 23 MS 5? MS 5

ca. 200c

aBailey, Papaconstantinou, and Sturdy 1992 b Huxtable et al. 1992 c Runnels and van Andel 1993b

the last interglacial (115-130 kyr B.P.) and continues into the Late Pleis- tocene (10-29 kyr B.P.). The Upper Palaeolithic (ca. 13-34 kyr B.P.) terminates the sequence at the end of the Pleistocene. The Mesolithic is dated to the early Holocene (7-10.5 kyr B.P.). This chronology can be supplemented with dates for the Palaeolithic-Mesolithic sequences from elsewhere in Greece, including the Southern Argolid, the Argive Plain, Thes- saly, Corfu, and Franchthi Cave.126

In an archaeological context what do the ages and maturity values of Table 3.11 really mean? The TL and IRSL dates tell a fairly simple story: they give the ages of the sediments upon which Palaeolithic or Mesolithic tool assemblages were left. But the meaning of the paleosol maturity values, which estimate the length of time a deposit was exposed at the sur-

126. Southern Argolid: Jameson, Runnels, and van Andel 1994, pp. 325- 340; Pope, Runnels, and Ku 1984. Argive Plain: Reisch 1980. Thessaly: Runnels 1988; Runnels and van Andel 1993a. Corfu: Sordinas 1969. Franchthi Cave: Perles 1987.

1

12

25 2

3

13-25a 13-25a 16-10a

59

74

115

130

190

92


face, is less clear. The many findspots buried deep in loutses and poljes are

either associated with paleosols of low maturity or not associated with

paleosols at all (Figs. 3.16, 3.18), indicating that the floors of poljes and loutses were occupied during brief dry periods or even while they were

active, in keeping with the exploitation of their resources. Most findspots, however, especially those at or near the present land

surface, are associated with Bt horizons of great maturity that originally formed some meters below the original surface. Allowing for the stripping of a few meters of fragile A and E horizons, these truncated Bt horizons

imply that old polje and loutsa surfaces were stable for long periods. Since it seems unlikely that Palaeolithic humans preferred eroded Bt horizons for their campsites, are their tools all on secondary locations, removed by erosion and redeposited somewhere else and hence of no value at all, as has been claimed by Bailey?127And perhaps most importantly, how did so many lithic suites become incorporated within the Bt horizon rather than rest-

ing upon it? We may dispose of the pessimistic view first. The tool assemblages are

always embedded in extremely fine-grained sediments (see above, pp. 63-

76), easily mobilized by weak currents or winds powerless to move even small flint chips. The artifacts tend to be matrix-supported, implying that flint and matrix were not deposited simultaneously by the same agent. The fine stratification that is in most places visible to the careful observer confirms that only very low-energy transporting agents were involved. Debris flows may incorporate stone artifacts, or they may be left on gravel beds

deposited by flash floods or small ephemeral streams in distal alluvial fans at polje margins, but those are lithologically distinct events (Figs. 3.13, 3.19) that are rare in the very low-energy environments of poljes and loutses.128 We have been unable to confirm claims by Bailey that the Mous- terian at Kokkinopilos was redeposited by late streams in gullies incised in

pre-Middle Pleistocene or Pliocene redbeds. Because the strong currents required to move objects of gravel size are

rare in polje and loutsa basins, the stone tool assemblages must have been lowered onto, or more often worked into, the Bt horizon by gentle removal of the overlying A-E horizons, or the Bt horizon must have crept upward over time to engulf and protect them. Gentle erosion is possible even in the low-energy polje and loutsa environments, but its universal action to explain the position of all stone tool suites on top of mature Bt

paleosols beggars belief and it is incapable of inserting them within the Bt. Artifacts may work their way down during the dry season in clays that

alternately swell and crack, but this would disperse the assemblage verti-

cally, perhaps sending the smaller ones deeper down. This is not what we observed in Epirus where the artifacts, small and large, tended to stay to-

gether in clusters or thin lenses, often giving the impression of being in situ.

Another process seems required, for which we present the following working hypothesis. If the land surface is raised at a very slow rate, the top

127. Bailey, Papaconstantinou, and of the Bt horizon itself would move gradually upward as a result of slow

Sturdy 1992. deposition, so engulfing any artifacts laid down on former land surfaces 128. Runnels and van Andel 1993a. above it. This process depends on the relative rates of sedimentation and

93


DEPOSITION > SOIL FORMATION

A A

debris flow

'"li,A"l", r

Bt(M1)

SOIL FORMATION > DEPOSITION

II 1 11 II

, 4

EXAMPLE: KOKKINOPILOS

ni',' Bt(M5)

,iii,,,, Bt(M2)

4

AIc.4944^ debris flow

A A

A A

1811111

AA Bt(M1)

3

. Bt(M1)

.A I

.,,,.,*

Bt(M4)

3

Bt(M3)

2 111111111

"""""""""' Bt(M1)

kst 1 karst

||l||Is1|||||| Bt(M2)

karst

mIIII paleosol (Bt) or desiccation zone

Bt(M1)

flint gravel

"handaxe"

Bt(M2)

Bt(M1)

karst

A A Palaeolithic stone tools

Figure 3.30. Relationship between paleosol maturity, terra rossa deposition rate, and Palaeolithic stone tool age in poljes and loutses. Stone tools are deposited on old surfaces, the age of which is defined byTL or IRSL dates. Left, sequence 1-4: Deposition rate exceeds rate of soil formation; immature paleosols are associated with old Palaeolithic material. Center, sequence 1-4: Rate of soil formation equals or exceeds deposition rate; maturing Bt horizon grows upward until it engulfs stone tool assemblage. Right: Profile from Kokkinopilos, incorporating both phenomena.

paleosol formation. In a landscape where the slope mantle has not yet been destroyed, terra rossa deposition rates in an active polje may be as

high as 10-15 cm/kyr.129 In contrast, soil formation is slow.130 If deposition is significantly faster than soil formation, either no soil

or an immature one will form when deposition is temporarily interrupted by a period of drought. Stone tools may then be left on desiccation surfaces, on thin and immature paleosols, or in and on marginal fan deposits (Fig. 3.30, left).

If, on the other hand, deposition raises the land surface more slowly than the Bt and Bk horizons form, the horizons will thicken upward into the overlying sediment, a process that is common in the lower floodplains of small rivers in the semiarid climate of the Peloponnese and Thessaly.131 When the polje approaches old age, the rate of soil formation begins to equal or exceed the rate of deposition. In the now raised polje at Morphi, for instance, a volcanic ash dated at 374?+ 7 kyr B.P. and located 12 m below a very mature paleosol estimated to be ca. 100,000 years old implies an average deposition rate of a mere 4 mm/kyr.132 The maturing Bt horizon

129. Kukal 1990, pp. 101-103; Runnels and van Andel 1993b.

130. Spaargaren 1979; Williams and Polach 1971; Magaritz, Kaufman, and Yaalon 1981; Demitrack 1986; McFadden and Weldon 1987; Harden et al. 1991; Bockheim, Marshall, and Kelsey 1996.

131. Pope and van Andel 1984; Jameson, Runnels, and van Andel 1994. See also Birkeland 1984, figs. 8-10.

132. Pyle et al. 1998.

94

liilPI


then grows upward until it engulfs any stone tools, protecting them from soil and wind erosion (Fig. 3.30, center).

The Kokkinopilos section (Fig. 3.30, right) demonstrates both processes. During its youth, the polje accumulated a thick sediment sequence with scattered immature paleosols and widely spaced stone tool assemblages; in its old age, however, when the dissolution of limestone was slower than its removal, the depletion of the slope mantle sharply reduced the sediment supply. Moreover, uplift eliminated most of the runoff from springs and initiated headward stream incision. This dropped the rate of deposition below that of soil formation and produced a thick, consolidated Bt that incorporated stone tools and kept them safe until Neolithic, Bronze Age, or in this case, post-Roman soil erosion exhumed them.133

A different example is Ayia (Fig. 3.20), a shallow loutsa with only 8- 10 m of fill, now almost entirely removed by recent erosion. A composite profile (Fig. 3.19) shows a lower, red-and-gray mottled sequence with distinct centimeter-scale stratification formed in an alternatingly dry and wet environment. Palaeolithic stone tools are intercalated in this sequence, which, near the top, was interrupted once by a debris flow underneath a mature (MS 5) paleosol. The thin, much younger upper zone has an immature (MS 2) Bt horizon.


GOALS AND PROCEDURES

133. Bailey, Papaconstantinou, and Sturdy 1992; Runnels and van Andel 1993b; van Andel, Zangger, and Demitrack 1990.

The survey of prehistoric sites took place between 1991 and 1995. Special attention was given to the red sediments (paleosols and redeposited terra rossa) because of their known association here and elsewhere with Palaeo- lithic artifacts, and the structure and characteristics of paleosol horizons were investigated to establish a rough chronology of the archaeological finds. In practical terms we used the available geological and topographic maps as a rough guide. Our goal was to produce a complete picture of Palaeolithic activity, in as wide a variety of geographic contexts as possible, within the time available for searching. Fifty-seven days of fieldwork were undertaken by a specialist team devoted entirely to the search for Palaeolithic and Mesolithic sites and consisting of three to four persons at all times, with the addition of student volunteers to assist. Two strategies were pur- sued. The first strategy was to locate and search all occurrences of Pleis- tocene soils and sediments in the study area (Fig. 3.8). The second strategy was intended to increase the coverage of the surface by inspecting nonredbed surfaces, such as dunes, alluvial fans, bare hillslopes, and remnants of the old peneplain (e.g., in the vicinity of the village of Loutsa).

The search for Palaeolithic and Mesolithic materials was also part of the general diachronic survey. The general survey teams, consisting of three or four field school students and two experienced graduate students as leaders, were trained to identify and collect all lithic artifacts before walking survey tracts. This point should be emphasized: in order to minimize selection bias, fieldwalkers were taught to recognize and collect all lithic

95


artifacts regardless of size, raw material, type, or date. The quantity of lithic artifacts collected (ca. 13,000 pieces) and the range of periods represented (Lower Palaeolithic to modern gunflints) are evidence that this training was effective. This procedure was useful in two ways. Large tracts of presumed post-Pleistocene surface were inspected and the usually negative results helped to confirm our assumption that we were not missing any sites in these areas. The teams also walked Pleistocene deposits not inspected by the specialized team (e.g., the Ayios Thomas peninsula), dis- covering important Palaeolithic findspots which materially increased our confidence that a reasonably comprehensive picture of the preserved archaeological record had been obtained. The tract finds collected by these teams were inspected on a daily basis by one of us (CR), and in cases where general survey teams brought in lithics from tracts, walkovers, or site/scatters that were of interest, the specialist prehistoric survey team revisited the area to make a separate inspection and collect and record additional samples as walkovers.

At the time when Palaeolithic or Mesolithic artifacts were discovered by the specialist prehistoric survey team, the following procedures were employed (a more detailed description of collection procedures used by general survey teams is given in Chapter 2). The first concern at all times was to determine the source of flints found on the surface. Our working model of site formation processes was based on the assumption that flints were associated with redeposited terra rossa, and to test this hypothesis each findspot was searched carefully for a source of the lithics. At a number of sites (e.g., Alonaki, Galatas, Kranea, and Kokkinopilos), flints embedded in the sediments were associated with paleosols (Bt horizons) of various maturity stages, formed when the original surface had been exposed for a sufficient quantity of time. At Ayia, on the other hand, fresh unweathered flints were found in unbroken, "mint" condition within uninterrupted, consolidated, horizontally bedded deposits at a depth of 3 m below the modern surface, where they must have been deposited during brief, perhaps seasonal, dry intervals. There can be no question that these flints are part of the redbed deposit and must be considered in situ in a geological sense and not later intrusions. The evidence for later reworking of redbed deposits at Kokkinopilos, cited by Bailey as proof that the flints are accidental intrusions,134 is based on excavations situated in gullies in the northeast margins of the site; these gullies were probably subject to local reworking that did not affect the entire polje fill.

The number of flints exposed in a paleosol horizon was limited, and the sample for each findspot was supplemented by collecting flints from the surface that were deemed to be derived, or highly likely to be derived, from the paleosol. This element of subjective judgment was based on extensive experience, and was justified by the close spatial association of materials within the outcrop and just below its weathered face. The flints are often stained with red clay from the paleosols or have fragments of Bt material adhering to them. The collections include all retouched artifacts, cores, complete flakes, blades, and flaking debitage with typological characteristics (e.g., core rejuvenation pieces). Only incomplete and typologi- cally unclassifiable fragments were discarded on-site.


96


Our purpose was not to collect samples for analyses that require large numbers of pieces for detailed typology or spatial analysis of activity areas. With the exception of one site (Spilaion; see Chapter 4), we did not attempt to map the spatial distribution of materials. Recent erosion of the

paleosols was responsible for extracting flints and scattering them on the surface and in erosional gullies that dissected the redbed sequence. In our view only excavations of the paleosols would reveal culturally meaningful spatial patterns. The samples collected from the surface were intended

only to provide sufficient information to assign the findspot to a cultural

period and to compare it with other findspots in the region. Papagianni has undertaken a more detailed technological analysis of the Mousterian from our collection, utilizing essentially all Mousterian artifacts found in

Epirus since 1962.135 The treatment of artifacts collected on the surface was simple. The

lithics were soaked in water to clean them before they were bagged for

storage in PVC bags labeled with provenience data. All samples were recorded in field notebooks and on printed recording forms, which permitted the samples to be tracked through cleaning and storage, and the information transferred to the project's computerized database.

Once cleaned, the lithics were described and assigned to typological categories according to the system of classification originally developed by Fran9ois Bordes, with certain modifications that have become accepted in recent years.136 Selected specimens were pulled from the samples for drawing and photography. These selected specimens were given separate inventory numbers, in addition to their sample numbers, to aid retrieval.

ARCHAEOLOGICAL SITES IN THEIR GEOLOGICAL

SETTING

135. Papagianni 1999. 136. Bordes 1992; Debenath and

Dibble 1994; Mellars 1996, pp. 169- 192.

Our survey was carried out in the territory west of the Louros River valley, with an emphasis on the coast from Parga to Preveza, and produced evidence for human activity from the early Palaeolithic through the Mesolithic. The majority of sites are coastal with the exception of those in the Thesprotiko and Cheimadio valleys. An interesting finding was a number of smaller, perhaps specialized, sites that may include quarry sites and

flintknapping areas, which supplement our picture of the regional settlement pattern.

Our program discovered or confirmed forty-four prehistoric findspots called "Site/Scatters" and designated "SS," followed by the year and number of the site recorded in that season (e.g., SS92-22 for Alonaki in the Acheron valley, the twenty-second findspot recorded in the 1992 season; see Appendix). Approximately 4,600 lithic artifacts were collected from these findspots and were used to assign them to general periods. Of these findspots, four produced Lower Palaeolithic materials, thirty produced Middle Palaeolithic, six produced Upper Palaeolithic, and six Mesolithic.

We supplemented our survey with data from an extensive program of augering in the Acheron River valley and the Louros delta (see Chapters 5 and 6); a geochronological program of radiocarbon (14C), thermolumines-

97


TABLE 3.12. EARLY STONE AGE CHRONOLOGY

Period Calendar Years (kyrB.P.)

Mesolithic 7 to 10.5

Upper Palaeolithic 13 to 34

Middle Palaeolithic (Early Palaeolithic) >31

Lower Palaeolithic (Early Palaeolithic) > 100

cence (TL), and infrared stimulated luminescence (IRSL) dating of alluvial sediments and sand dunes; and laboratory sedimentological analyses. A chronological summary is given in Table 3.12.

THE EARLY PALAEOLITHIC

The traditional terms of "Lower" and "Middle" Palaeolithic have been

questioned in recent years primarily because they refer both to chronostratigraphic units and lithic typology, which overlap and are not congruent. The Lower Palaeolithic was once regarded as a Middle Pleistocene sequence of Acheulean industries with the handaxe-cleaver complex as type fossils. The Middle Palaeolithic was a Late Pleistocene flake industry (the Mousterian) with Levallois technology. It was also thought that the Acheulean was associated with Homo erectus and the Mousterian with Neanderthals or other archaic Homo sapiens.137 All of these assumptions have proved to be unreliable. The Acheulean, with handaxes, continues until the last interglacial (ca. 115-130 kyr B.P.) in many places, while new finds have placed the beginning of the Mousterian at more than 100,000 years before the last interglacial (ca. 200-250 kyr B.P.). There are significant overlaps in chronostratigraphic terms, and the Acheulean and Mous- terian also share the use of the Levallois technique, flake tools, and handaxes (or "bifaces" in formal typology).

It is thus no longer possible to correlate lithic technocomplexes and hominid grades. Some authorities question whether Homo erectus was responsible for the European Acheulean, which might also be attributed to archaic Homo sapiens, and both Neanderthals and anatomically modern Homo sapiens are associated with classic Middle Palaeolithic in the Near East. Neanderthals in western Europe and perhaps the Balkans are responsible for industries that are similar to and contemporary with industries ascribed to the Early Upper Palaeolithic (EUP). It is unlikely that traditional lithic industrial identifications can be other than labels of con- venience, permitting us to discuss problems and describe newly discovered materials but which in no way imply either chronological position or cultural affinities. In these cases, Rolland recommends calling the traditional Lower and Middle Palaeolithic "Early Palaeolithic" to avoid the problems inherent in the earlier classification.138 We will follow that suggestion in this report, although we also use the older terminology when greater chronological or typological precision is required.

Localities with the earliest materials (on stratigraphic and chrono- metric grounds) are found at Kokkinopilos, Alonaki, and Ayios Thomas. Kokkinopilos (SS91-3) is the most important of these, and has been de-

137. Mellars 1996, pp. 2-4. 138. Rolland 1986.

98


Figure 3.31. Palaeolithic site/scatters in the Thesprotiko valley, showing the location of findspots associated with redbeds. Kokkinopilos is a major site. Galatas and Kranea are presumed to be specialized activity sites. Small findspots are found on the margins of the polje deposits and at the entrances to the valley. Hatched areas are modern settlements.

139. Runnels and van Andel 1993b. 140. Dakaris, Higgs, and Hey 1964.

scribed in detail (Fig. 3.31).139 A pointed biface (handaxe) of late Acheulean

type was found stratified within a zone of interbedded subaerially weathered but mainly subaqueous polje or loutsa deposits (Fig. 3.17) ca. 17 m below a paleosol containing a later Middle Palaeolithic industry dated to ca. 90 kyr B.P. (Table 3.11), close to the present center of the polje and near its thickest deposits. Three undisturbed immature paleosols mark the interval between the Middle Palaeolithic paleosol and the handaxe zone, which is almost entirely sterile except for a thin (ca. 50-cm) bed of matrix- supported fine flint gravel a few meters above the handaxe. At about the same or slightly higher stratigraphic levels, other localities roughly to the south and southwest of the findspot produced heavily patinated artifacts of large size. In 1991 we observed numerous artifacts eroding from the sediments in the northwest part of the deposit and perhaps similar to the "chipping" floors described by Higgs in the northeast part of the site some 300 m away.140 The artifacts consisted of large flake tools, non-Levallois in technique, with denticulate and notched edges. We were prevented from making a collection of these artifacts and from sampling this layer for geochronological dating by a post-issue alteration to the Nikopolis Project's research permit; we are thus unable to give details about the lithics from the handaxe layer or to directly date the handaxe. Our best estimate of date (150-200 kyr B.P.) is derived from extrapolation of plausible sedimentation rates based on the dating of the very mature Middle Palaeolithic

paleosol at the top of the sequence, obtained on a sample taken before our research was stopped (Table 3.10; VA93-05).

99


A second area of interest that produced material of Early Palaeolithic character is Alonaki in the Acheron valley (Fig. 3.32). Alonaki (SS92-22, SS92-23) appears to be a loutsa-type karst depression filled with alluvial/ colluvial, redeposited terra rossa. An extensive outcrop was inspected and found to have at least two distinct Bt horizons. The upper Bt has a maturity of MS 4 and the lower Bt MS 4/5 or 5 (Table 3.9). The sequence has a total depth of more than 3 m below the surface, and appears to contain more than one Palaeolithic industry. Large flake artifacts were found throughout the deposit, including the lowest part, exposed in a clay ex- traction pit. Although our ability to correlate the industries with outcrops of different depths is limited, it appears that a conventional Middle Palaeolithic Mousterian is found on or near the upper Bt horizon and a significantly earlier large flake industry in the lower Bt horizon.

A curious feature of the Alonaki deposits is the presence of dense concentrations of angular stones mixed with lithics, which occur in discrete units 1-3 m in diameter and ca. 0.30 m thick (Fig. 3.33). They are unsorted and matrix-supported, but their sharp boundaries (at top, bottom, and laterally) against redeposited terra rossa argue strongly against an origin as a stream channel or debris-flow deposit. These features resemble in some ways the "stone clusters" identified at Early Palaeolithic sites as far afield as Hoxne in England, which are sometimes described as artificial in

origin.141 The Alonaki "stone clusters" are associated with the artifacts of the lower Bt horizon, large flakes with wide, thick platforms and large bulbs of percussion and equally large cores of non-Levallois type (Fig. 3.34). These materials were recovered from the bottom of a shallow ero-

Figure 3.32. Palaeolithic and Mesolithic site/scatters in the Acheron valley. Alonaki is a major site. Smaller findspots (e.g., Skep- asto, Loutsa, and Valanidorrachi) are specialized activity sites. Other findspots (e.g., Ayia Kyriaki, Tsouk- nida, Ammoudia), perhaps temporary camps, are located on the edge of the valley. Two Mesolithic sites (Ammoudia, Loutsa) are found near the coast, and Tsouknida was located on the edge of an ancient lake or embayment. Hatched areas are modern settlements.

141. E.g., Singer, Gladfelter, and Wymer 1993, p. 124.

IO0


Figure 3.33. View of a stone cluster: at Alonaki illustrating the sharply defined edges of the feature ,i -i

sional gully that cuts through the site as well as in the modern clay extrac- tion pit. In a few cases they were prized from within the structures found in the lower Bt horizon, and must clearly be regarded as in situ within the paleosol. In the surface collections some mixing with later materials is unavoidable, but in the lower levels of the deposit, wherever in situ artifacts were observed, they were always of the non-Levallois big flake type.

These large artifacts differ in raw material, technique, and retouched tool typology from the Mousterian and consist chiefly of core-choppers and flakes (Fig. 3.35). The raw material is a dull dark brown, fossiliferous chert, derived from Eocene limestone, that contrasts with the glassy blu-

ish-gray nodular flint without macroscopic fossils that is derived from the Mesozoic Pantokrator limestone and was widely used to manufacture Mousterian artifacts in Epirus. The Eocene chert has been worked by hard- hammer direct percussion. Flakes have large broad platforms and well- defined, swelling bulbs of percussion. The size of the platform and the pronounced swelling of the bulb are indications that considerable force was used to detach each flake from its core. Cores include core-choppers (Fig. 3.36) and large cobbles from which flakes were removed from one face using a broad plain striking surface (Fig. 3.37). The resulting flake scars are wide and deep.

There is not enough material for a metrical analysis or a study of the complete reduction sequence, but the characteristics we can observe seem to point to the production of broad flakes from boulders and large cobbles as the chief goal. There are other characteristics that separate the lower Bt industry from the Mousterian. Retouch is confined to direct, invasive retouch, and large notches were created by a single inverse or direct blow (Clactonian technique). Notched and denticulated edges are common, and typical Mousterian forms, such as points and side scrapers, are lacking in this material.

IOI


Figure 3.34. Early Palaeolithic artifacts from Alonaki:

1) double convergent side scraper; 2-4) notched pieces/denticulates; 5) convex side scraper. Scale 1:2

Figure 3.35. Early Palaeolithic

choppers from Alonaki. Scale 1:4

1 2

3 4

5

102


Figure 3.36 (above, left). Early Palaeolithic core-choppers from Alonaki. Scale 1:2

Figure 3.37 (above, right). Early Palaeolithic core from Alonaki. Scale 1:2

Also from this area is a small biface (handaxe), found near Ormos

Odysseos (W94-20), about 500 m to the west of Alonaki (Fig. 3.38). Here a thick mantle of Pleistocene red clay, sand, and gravel (and possibly in-

volving a paleosol as well) covers a limestone karst surface inland from a coastal paleosol superimposed on a sand dune of probable interglacial age (SS92-25) which also contains Palaeolithic artifacts (Fig. 3.39). The inland deposit is today overgrown with bushes, but goats have worn trails

through them and the tracks have eroded down to bedrock exposing out-

crops up to 3 m thick. The handaxe was found in one such ravine (Fig. 3.40). The area is essentially level and the handaxe could not have been

transported very far. Other artifacts were observed in the deposit, which

may be of the same general age as the lower Bt horizon at Alonaki. The sand dune, which overlies the Bt deposit at its northwestern corner, is

nearly at present sea level, but is definitely of Pleistocene age and hence can only belong to the high sea level of the last interglacial period (or an even earlier interglacial).

In our estimation the preponderance of the evidence the high ma-

turity of the Alonaki paleosols and the overlying interglacial coastal suite-

places the lower Bt paleosol with its associated artifacts before the last

interglacial, or more than 130,000 years ago, close to the lower age suggested for the Kokkinopilos handaxe (150-200 kyr B.P.).

Other Early Palaeolithic materials are found in the southern part of the survey area near the town of Preveza. Tracts walked on the Ayios Tho- mas peninsula at the northern end of the Ormos Vathy (T93-17, T93-3, T93-4, T93-5) recovered large numbers of Mousterian pieces from a

Io3


Figure 3.38. Early Palaeolithic biface (handaxe) from Ormos Odysseos. Scale 1:2

Figure 3.39. Interglacial sand dune (SS92-25) at Ormos Odysseos, looking southwest

Figure 3.40. Ormos Odysseos, biface

findspot (W94-20), looking south

I04


Figure 3.41. Early Palaeolithic biface / ' \ - or bifacial core from Ayios Thomas. Scale 1:2

paleosol associated with marine deposits of Eemian age. Among these materials are large flakes of Eocene chert similar to the Alonaki lower Bt artifacts, including a rough amygdaloidal biface or bifacial core (Fig. 3.41). Very few outcrops exist in this area and the exact source of these materials could not be pinpointed. The deposits are bedded horizontally and the material cannot have been transported far. Although no age assignment is possible, we suspect that these materials are of the same general age and

type as those from Alonaki.

Except for Kokkinopilos, Early Palaeolithic materials are found only on the present coastline, and it is for this reason that they have not been noticed before. An inspection of Higgs's collections in the Ioannina Ar- chaeological Museum showed that they contain no artifacts similar to the lower Bt materials at Alonaki.

THE MOUSTERIAN (MIDDLE PALAEOLITHIC)

Most Early Palaeolithic artifacts in the Preveza nomos are Mousterian in type. In his pioneering survey, Higgs found large numbers of Mousterian artifacts on surface sites in Epirus, including thousands from Morphi, Karvounari, and Kokkinopilos.142 Specialist prehistoric survey identified thirty findspots (site/scatters in database terminology); although this number could be easily multiplied by additional fieldwork, we believe it includes a representative range of site types and habitats. Our study collection from Mousterian findspots includes more than 1,500 artifacts, and about 10% of the 13,000 lithics collected by the general survey teams are also Mousterian.

The abundance of the Mousterian may be attributed to several factors. Geological contexts of the appropriate age are more common than those of earlier periods or those immediately following. Mousterian sites are also more conspicuous ("obtrusive" in survey terminology), and the preferential selection of redbeds for their camps makes them easy to find:

142. Dakaris, Higgs, and Hey 1964. the large, often heavily patinated, artifacts stand out as white spots on a

Io5


red background. The erosion affecting many of the redbeds has no doubt contributed to the obtrusiveness of sites by deeply incising gullies and de- nuding the deposits of vegetation.

Besides Epirus, the Mousterian is found in the Argolid, Corfu, Elis, Kephallinia, Messenia, Thessaly, and many other localities.143 The preference for open-air settlements is one reason for this ubiquitousness. While Upper Palaeolithic occupation of caves and rockshelters is common in the last glacial (OIS 2, ca. 25-12 kyr B.P.), Mousterian occupation of rockshelters or caves is rare: Asprochaliko, Kephalari, and Franchthi are at present the only published examples. Site selection strategy may be another major factor, and climate also played a role. The Mousterian is found in the early glacial period (OIS 4-3), a time marked in western Europe by numerous climatic oscillations from nearly full glacial to warm conditions, some of which (e.g., the Hengelo interstadial at ca. 40 kyr B.P. and the Denekamp interstadial at ca. 36-32 kyr B.P.) were quite mild.144 If these conditions also prevailed in southeastern Europe, this would suggest that Mousterian settlement was encouraged by, or limited to, the warm phases.

The identity of the makers of the Mousterian is a contentious problem, but the European evidence indicates that Neanderthals were the pro- ducers of this industry, a working hypothesis we accept.145 Like the Euro- pean finds, the Mousterian in Epirus similarly shows relatively little variation. That of the basal layers (16 and 18) at Asprochaliko, dated to ca. 98.5 kyr B.P., is characterized by the frequent use of the Levallois technique to prepare cores to produce numerous large lamellar flakes.146 In addition, Levallois points, large convex side scrapers (racloirs), and other elements are typical. This Mousterian belongs to OIS 5 (ca. 115-74 kyr B.P.) and perhaps continues into OIS 4 and 3 (ca. 74-59 kyr B.P.), the early glacial. The Mousterian of Asprochaliko's upper level (layer 14), poorly dated by radiocarbon assays ranging from 29 kyr B.P. (26,000 b.p.) to >39,900 b.p., 47is quite similar, but makes less use of the Levallois technique and is rich in Mousterian points and small side scrapers in a wide range of types. It was once described as a diminutive facies called the Micromousterian, but this designation has been questioned for Aspro- chaliko because the difference in size between the basal and upper (or late) Mousterian is not great enough to warrant the qualifier "micro" for the latter.148 There is no question, however, that the late Mousterian differs from the preceding Levallois-Mousterian in some typological, technical, and metrical characteristics and that it is younger.

Mousterian artifacts found on the surface can be placed in chronological order only with great difficulty (Table 3.11). There are abundant surface sites with Mousterian finds: Kokkinopilos, Ayia (SS93-9), Alonaki (SS92-22 and SS92-23), Kranea (SS92-14), the Anavatis quarry (SS94- 13 and SS94-16), Skepasto (SS92-20), and Valanidorrachi (SS91-4), among others. Late Mousterian artifacts are found at Kokkinopilos, Ayia (in its upper levels), Alonaki (SS92-22), Galatas (SS92-13), Loutsa (SS93-31, SS94-12), and some smaller sites. A possible source of chronology for the later Mousterian comes from the Southern Argolid and Thessaly where similar Mousterian artifacts are dated by radiocarbon and U/Th series from 55 to 30 kyr B.P.149

143. Bailey et al. 1999; Runnels 1995.

144. van Andel and Tzedakis 1998. 145. Mellars 1996, pp. 1-8. 146. Bailey, Papaconstantinou, and

Sturdy 1992; Huxtable et al. 1992. 147. Bailey, Papaconstantinou, and

Sturdy 1992, p. 138 148. Bailey, Papaconstantinou, and

Sturdy 1992; Huxtable et al. 1992. 149. Pope, Runnels, and Ku 1984;

Runnels 1988; Runnels and van Andel 1993a.

io6


150. Dakaris, Higgs, and Hey 1964.

A major interest of our survey was the reconstruction of Mousterian

paleoenvironments, settlement, and land use. Our discussion of this topic is divided into two parts in accord with the twofold division of the Mous- terian, although the many elements of continuity should be stressed. The decisive feature governing land use and settlement patterns is the distribution of karst features such as poljes, loutses, and dolines that served to attract and concentrate animal, plant, and mineral resources and permitted and encouraged a seasonally scheduled, partially logistical strategy of land use. Throughout this discussion we refer to strategies that make struc- tured, planned, and repeated use of a landscape as logistical or partially logistical land-use strategies.

The earlier Mousterian is found in abundance in the redbeds of Epirus, particularly at Kokkinopilos. The variety of locations showing evidence of

early Mousterian activity is perhaps the best picture of partial logistical land use. The largest concentrations of artifacts are found at Alonaki, Kokkinopilos, and Ayios Thomas (Ormos Vathy). A fourth findspot is in the vicinity of Morphi in Thesprotia, where large numbers of Mousterian artifacts were collected by Higgs.150

These larger sites are supplemented by a series of small sites at Ayia, Kranea, Loutsa, and Anavatis. Still smaller sites, possibly specialized in character, are found at Skepasto and Valanidorrachi, located near flint out-

crops where quarrying, flintknapping, and testing of nodules were the main activities. The site of Rodaki may have been occupied by Neanderthals utilizing coastal resources, although the lack of faunal remains here and elsewhere makes this hypothesis difficult to evaluate. The principal characteristics of the known smaller sites are these. All presented easily available surface water, which ponded on the clay surfaces of the loutses in late winter and spring and slowly evaporated in summer. We found standing water and evidence of recent wet conditions on these sites to the end of June and into earlyJuly. Loutses mainly depend on winter rain rather than on major springs and are found as shallow depressions in exposed localities where they dry out early. They are more exposed to the elements than

poljes, which are located in deep basins that offer more sheltered conditions. Large poljes, like the modern Valtos Kalodiki, retain water much longer, or permanently in the form of shallow lakes, swamps, ol marshes. There were found reeds, aquatic plants, willows, and stands of trees in well-watered side valleys together with a varied wildlife. Camps were placed along the margins of the poljes-partly to be on well-drained ground and partly to avoid scaring off the game-but near springs or the inlets of winter or spring streams. Locations were probably shifted often. If groups returned on a seasonal basis over a long time, the spread of artifacts from overlapping camps would make spatial analysis difficult and may account for the large quantities of artifacts.

Loutses and poljes were magnets for animals and humans in this glacial landscape. Rivers in the summer carried some meltwater, but they had incised their channels to reach lower shorelines. Away from the rivers, springs were sources, but in the karst landscape the limestone bedrock has no surface water in the dry season. The poljes preserved water when it would be least available, in the late summer and autumn months, and they

IO7


offered predictable places to find food as well. The predictability of these spatially concentrated resources may explain the partially logistical settlement pattern seen in Mousterian times. In a fully logistic land-use pattern, the smaller outlying activity areas would have served very specialized functions, as is thought to be the case with the ibex and chamois hunting camp at Klithi in the late Palaeolithic,151 but there is no evidence for this degree of specialization in the Mousterian.

In contrast to the modified logistical pattern postulated here, some scholars have developed a picture of Neanderthals as opportunistic foragers who moved about the landscape in search of food.152 Such a model of residential mobility also postulates a repeated use of scheduled seasonal stops at particular sites, and this pattern of land use may have graded into various kinds of logistical foraging that depended to a greater or lesser degree on a few base camps.

In our model of modified logistical land use, the different types of sites offered different-sized and differently-timed packages of water, plants (for food, handles for tools, shelter, fuel), and animals. There was also a good chance of finding useful quantities of flint for toolmaking in most locations. Ayia is a typical example of the smaller "loutsa" site, consisting of sometimes large numbers of artifacts associated with small red deposits, typically no more than 300 or 400 m in diameter and located at some distance from larger polje sites such as Morphi and Kokkinopilos (Fig. 3.42). The findspots near Loutsa and Kranea are also examples. These sites are found in remote mountainous locations at elevations up to 400 masl or more. The lithics from Ayia include flintknapping debris and retouched tools indicating a wide range of activities at the site (Figs. 3.43, 3.44).

Even smaller sites show more specialized activities. Skepasto and Valanidorrachi in the Acheron valley appear to be flintknapping sites. At Skepasto there are many worked and unworked nodules of flint, some weighing up to 15 kg, eroding from limestone; associated with these nodules are numerous test-cores, Levallois and other cores, Levallois blades and flakes, and rare finished artifacts (e.g., two Levallois points). The Anavatis quarry sites, with paleosols containing both cores and finished tools, appear to be small encampments on a torrential stream fan (Figs. 3.45, 3.46).

An interesting example of the more specialized type of site is Rodaki, a red deposit located at the present mouth of the Paliourias River (Fig. 3.45). A large number of artifacts were found stratified in a complex sequence of paleosols, separated by a normal fault from a thick sequence of marine deposits of probable interglacial age (Fig. 3.47). The artifacts are in situ in a stony red paleosol that is capped by a layer (ca. 2 m thick) of stone-free dune or coastal sand. The lower, stony paleosol with artifacts is nearly completely buried, but the upper 30 cm of its thickness is exposed. This part is rich in artifacts that are difficult to classify (Fig. 3.48). They resemble materials from the island of Zakynthos at the site of Vassiliko where Sordinas found them in red sediments interstratified with marine deposits.153 The Zakynthos artifacts are undated, but appear to be a specialized variant of the Mousterian.154 In this respect they bear a resem-

151. Bailey 1997. 152. Mellars 1996, pp. 356-365. 153. Sordinas 1968. 154. Kourtesi-Philippakis 1996.

Io8


Figure 3.42. The Palaeolithic site of Ayia and its setting. The site is located in a small loutsa at an elevation between 300 and 400 masl. Hatched areas are modern settlements.

155. Kuhn 1995, pp. 46-72. The Italian sites are dated to the early to mid-glacial (ca. 110-35 kyr B.P.).

156. Cf. Kuhn 1995, pp. 95-97.

blance to the Pontinian of Italy, a littoral Mousterian rich in small side scrapers found both in caves and on open-air sites.155 It should be noted that these rather simple tools made on small pebbles are not very diagnostic and may reflect similar choices of raw materials for toolmaking rather than similar cultural traditions. It is notable, however, that this type of Mousterian is found only in coastal localities, where larger sizes of raw materials are also available, suggesting that the similarity of the industries may in fact be significant. The Rodaki artifacts are small in size and made from pebbles collected from the nearby riverbed. The most characteristic types are small core-choppers,'56 transverse convex scrapers, bladelike flakes, and rare end scrapers and notched pieces. This industry does not use the Levallois technique and we regard it as a specialized coastal facies of the Mousterian.

Io9


3

1 2

4

5

6

Figure 3.43. Middle Palaeolithic (Mousterian) artifacts from Ayia: 1) double convergent side scraper; 2) Levallois flake; 3) end scraper; 4, 6) blades; 5) core on a flake. Scale 1:2

Figure 3.44. Middle Palaeolithic

(Mousterian) artifacts from Ayia. Scale 1:2

IIO

EARLY STONE AGE OF THEOF THE NOMOS OF PREVEZA

Figure 3.45. Palaeolithic findspots in the vicinity of Kastrosykia. In this area of heavy vegetation, paleosols and redbeds are exposed only sporadically. Findspots represent small concentrations of lithics, probably remnants of ephemeral campsites. Individual Palaeolithic artifacts were noted in tracts and walkovers throughout the area, indicating that many more findspots exist. Hatched areas are modern settlements.

Figure 3.46. Anavatis site/scatter 94-13, located in the middle of the

photograph on the leveled area, looking northeast

III


Other veM sma sites are__ :u i bot. Figure 3.47. View of Rodaki (SS92-

Ammoud!a,!l -ou'/ta:s Ui Kyr!i~'~i " ";ak~-'~5.Pleistocene redbed (left) and

sites ~_, or. ; :raw . . . . '' ' .-__I a marine deposits (fight) are separated

: .~ i,~~ 6 -, ;by a vertical normal fault. The _smfall! ;i Z * | numbers Palaeolithic artifacts were found in a

and~~~~_~ . = i 3-ot *..- Pleistocene deposit in the fore-

ground, which is overlain by the

side scraper; below, bladelike flake.

Scaleistocene redbed.1:2

Ot her very small sites are found in both the Aheron (e.g., Tsou knida,

Thesprotiko (Iliovouni, Romia, Mesaria, and Galatas) valleys that are can- didates for specialized sites, perhaps hunting stands, seasonal camps, kill

imall numbers o f Mous terian artifacts, typically fewer than 20 sp ecimens,

all within a few hours walking distance from major sites such as Alonaki, Morphi, and Kokkinopilos.

movement of Mousterian people. Scattered flakes were found on former islands in Lake Mavri (Thesprotiko) and in the remote mountainous polje of Cheimadio. It is difficult to interpret these small findspots, which may be the disturbed remnants of now vanished sites, but it is reasonable to

suppose that most of them represent ephemeral episodes of activity in the

II2


157. Cf. Runnels 1996 for a similar pattern.


159. Kuhn 1995. 160. Mellars 1996, pp. 193-244. 161. Runnels 1988; Runnels and

van Andel 1993a. 162. Mellars 1992.

landscape. Individual Mousterian points were collected as stray finds by the general survey teams at Eli, at several places on the Ayios Thomas peninsula (up to five points were noted), at the mouth of the Ambracian Gulf, and at Kastro Rizovouni in the Thesprotiko valley. These points are clear evidence of off-site human activity, probably representing hunting losses.157

The distribution of existing sites must be used with caution, as most are found in redbed deposits at relatively stable points in the landscape, and are consequently the only places where sites would be preserved. In- tervening areas show much evidence of erosion and disturbance. This problem is most acute toward Preveza, where the surface is covered with modern vegetation and offers very limited opportunities for observing the Late Pleistocene surface.

Sites such as Kokkinopilos show much activity in this period, and others (e.g., the two sites at Loutsa, Galatas) were probably occupied only at this time. Our chronological control is not sufficiently precise to determine whether other sites (e.g., Kranea) went out of use. Stratigraphic excavation evidence is available only from Asprochaliko where the Moust- erian continues perhaps to the beginning of the Upper Palaeolithic, ca. 29 kyr B.P. (26,000 b.p.).158 The late Mousterian, found in layer 14, perhaps represents a shorter period of occupation or less intensive activity. It is based on a different pattern of core working, one that deemphasizes the Levallois technique and makes greater use of disk cores to produce short, pointed flakes. Tool types change also, with Mousterian points and side scrapers, particularly transverse and convex types, becoming the most common. Although the size differences have been exaggerated in the past, there is a small shift in the direction of smaller tools. A similar change is noted elsewhere in Europe, occurring after 60 kyr B.P. Kuhn has shown that this change occurs in the Pontinian Mousterian of Italy with a transformation in settlement pattern, lithics, and presumably land use.159 It seems to be a reasonable hypothesis that Neanderthal foraging strategies would change in the face of global climate change.

In Epirus, the less abundant evidence for the later Mousterian, especially on the smaller, more dispersed specialized sites, may reflect a response to climate change. Late Mousterians are concentrated on the larger, perhaps more reliable, poljes of Kokkinopilos and Morphi, and the Alonaki loutsa. All three were also particularly well positioned near river valleys with access to larger plains. The greater variety of specialized hunting equipment seen in the Mousterian points and leafpoints may hint at an increased reliance on hunting.160

The available evidence suggests that the last Mousterian was widespread in Greece during OIS 3. There is another noteworthy feature of the later Mousterian. In the stratified Thessalian sites this industry shows signs of contact with and borrowing from the EUP (Aurignacian) tradition, present in the neighboring regions of the Balkans ca. 30-45 kyr B.P.161 Such mixed industries, clearly derived from local Middle Palaeolithic traditions, overlap chronologically with the Aurignacian in the Balkans and westward into France and Spain (e.g., Chatelperronian) where they are regarded as the work of late Neanderthals.162 It is particularly difficult to

II3


sort out these industries in the absence of controlled, well-dated, strati-

graphic excavations, but the Greek version, as seen in Thessaly, is clearly of Middle Palaeolithic character.163 It has Mousterian points, leafpoints, and side scrapers, sometimes made on Levallois flakes and worked with typical Mousterian oblique, scalar, steep retouch. The EUP elements are end scrapers, carinated burins, and marginally retouched blades, all made from the same raw materials as the associated Mousterian pieces. It is noteworthy that this industry is still found at the top of the sequence of river deposits where typical Mousterian artifacts were discovered in a paleosol with an associated date of 31 kyr B.P. (28,000 b.p.).'64 If this late Mousterian was a

product of Neanderthals, it is an indication that they continued in existence, in Greece at least, for some time after they had been replaced by anatomically modern humans elsewhere in the Balkans and central- western Europe. As in Thessaly, the Mousterian continued at Asprochaliko until quite late.165 It is possible that the splintering of Neanderthal populations into isolated refugia by the intrusion of modern humans through the heart of Europe may have contributed to their eventual demise by preventing interbreeding and disrupting ancient patterns of migration and communication.

There is another possibility for European Neanderthals. If the dates for the earliest Aurignacian in the Balkans get pushed back farther in time to ca. 45 kyr B.P. or more, the overlap with the late Mousterian peoples becomes greater. Movements of modern humans, the presumed makers of the Aurignacian industry, into territories once occupied exclusively by Nean- derthals could have caused the displacement of the latter.166 The Neander- thals may have been confined to less favored reaches of Greece as a consequence of finding more northerly parts of the Balkans too cold or already occupied by anatomically modern Homo sapiens. Sometime after 31 kyr B.P., the Mousterian and thus the Neanderthals were gone from Greece.

THE UPPER PALAEOLITHIC

A small number of sites belonging to the Upper Palaeolithic are known in Epirus, primarily from excavations by Higgs, Bailey, and a team from the Ephoreia of Caves and Paleoanthropology. Stratified Upper Palaeolithic (UP) sequences in Epirus, including Asprochaliko, Kastritsa, Klithi, and

Boila, and one site in Corfu (Grava Cave) have sequences of UP layers with Gravettian and Epigravettian industries.167 Radiocarbon dates indicate that the Upper Palaeolithic began before ca. 34 kyr B.p.168

Curiously, evidence for the initial stages of the Early Upper Palaeolithic is very rare in Greece.169 In her review of the evidence Perles noted only a small sample of EUP artifacts from the basal layer at Franchthi Cave that may date to more than 30 kyr B.P., and she drew attention to possible Aurignacian elements in the unpublished sites of Arvenitsa, Kephalari, and Ulbrich in the Argolid.170 In recent years, new EUP finds have been forthcoming: in Thessaly, Aurignacian artifacts occur in a late Mousterian industry found in sites in the Peneios River valley west of Larisa;71 a similar industry was investigated at two open-air sites in the northwestern Peloponnese near Patras;172 and Koumouzelis has reported a late Aurig- nacian level with an estimated age range from ca. 34-24 kyr B.P. from a

163. Runnels 1988. 164. Runnels and van Andel 1993a. 165. Bailey, Papaconstantinou, and

Sturdy 1992. 166. Mellars 1992. 167. Bailey 1992, 1997; Bailey et al.

1999; Kotzambopoulou, Panagopoulou, and Adam 1996; Sordinas 1969.

168. Bailey et al. 1983b; Bailey, Papaconstantinou, and Sturdy 1992.

169. Runnels 1995. 170. Perles 1987. 171. Runnels 1988; Runnels and

van Andel 1993a. 172. Darlas 1989.

II4


Figure 3.49. Early Upper Palaeolithic end scrapers from Spilaion. Scale 1:1

173. Koumouzelis et al. 1996; Kozlowski 1999.

174. Adam 1989, p. 253.

rockshelter site in the Kleisoura Gorge in the Argolid.173 In the nomos of Preveza, EUP artifacts are extremely uncommon. They are lacking in

Asprochaliko,174 but a few artifacts of EUP type (chiefly end scrapers) were collected at findspots in the survey area (e.g., Galatas and Vouvo-

potamos). Definite EUP artifacts of Aurignacian type were found in abundance

at only one site, Spilaion, which is located on a limestone ridge between the Early Palaeolithic site of Alonaki and a former channel of the Acheron River. Spilaion has a large and dense accumulation of lithics on its southeastern slope. The extraordinary abundance of lithics on the surface (ca. 150,000 pieces) permitted a detailed spatial analysis, including a controlled collection from gridded sample sites and computer-assisted analysis of the distribution and association of the lithics (see Chapter 4). The finds are typical Upper Palaeolithic, including carinated and nosed end

scrapers, burins, and retouched blades (Fig. 3.49). Spilaion is undated, but the site was occupied long enough to accumulate numerous concentrations of flintknapping debris marking positions of prehistoric activity. The site is strategically located at a point where routes to the north (via the parallel valleys from Preveza to Parga) cross those running east-west, from the coastal plain to the interior. The EUP people seem to have had little interest in the poljes and loutses that determined Middle Palaeolithic settlement.

The concentration of activity at Spilaion suggests that we are dealing with a base camp. The sheer density of artifacts, including concentrations of debitage suggesting episodes of flintknapping, and the scarcity of retouched tools are the best evidence for a degree of sustained and repeated activity. The analysis of the retouched tools and the flintknapping concentrations suggests a range of activities that are expected in a base camp, shown by the more or less complete reduction sequence of stone debitage (cores, cortical pieces, blanks, tools, and debris).

In the Early Upper Palaeolithic there seems to have been little interest in caves or rockshelters, here or elsewhere in Greece, and the absence of

Aurignacian deposits in the stratified and excavated sites in Epirus com-

plicates the task of interpreting settlement patterns. It is nevertheless clear that the Aurignacian site at Spilaion represents a complete break with the

Early Palaeolithic pattern of dispersed settlement and land use based on the exploitation of loutses and poljes.

II5


Figure 3.50. Late Upper Palaeolithic backed blades. Scale 1:1

With increasing frequency Late Upper Palaeolithic (LUP) backed- blade industries (Gravettian and Epigravettian) dating to 29 kyr B.P. (26,000 b.p.) and after are found in Greece. Backed-bladelet industries are found

principally in caves or rockshelters, several of which have been tested by excavations. Asprochaliko, Kastritsa, Klithi, Boila, Grava, Franchthi, Keph- alari, Kleisoura, Seidi, Theopetra, Ulbrich, and Zaimis are the chief ex-

amples, and other sites have been tested in Boeotia, Elis, and Thessaly.175 Intensive systematic surveys in the Argolid, Berbati, Nemea, and Messenia, however, have produced surprisingly little evidence for LUP open-air sites. To take but one example, the surface survey in the vicinity of Franchthi Cave brought to light only a handful of LUP artifacts; the three or four sites with small geometric tools and backed blades were all small, seriously disturbed by subsequent erosion, and undated.176 Site E81, for instance, had only a single backed blade and other, isolated finds of backed blades were made in tracts, perhaps lost as the result of Upper Palaeolithic hunt-

ing activity. A similar situation was noted in the Berbati-Limnes survey, where the only LUP materials were scattered backed blades or end scrapers, found in the course of tractwalking and no doubt left by the hunters who occupied the Kleisoura shelters.177

In the Nikopolis survey, small numbers of LUP materials of Gravettian or Epigravettian type (Fig. 3.50) were noted at two sites near the village of Loutsa, at Galatas inThesprotiko, and at Kokkinopilos.178 A typical findspot of this period is located near Lake Pogonitsa on the Ayios Thomas peninsula, where a scatter of half-a-dozen artifacts was found in small pockets of sediment in cracks in the karst limestone.

The difficulty in relating these scattered finds to a pattern of land use and settlement is compounded by two factors. The lower sea level at the time of the last glacial maximum created an extensive coastal plain that

greatly enlarged the useful territory. It was also a habitat supporting biota not found in the highland interior or represented by any existing habitats on the mainland today.179 A second problem, noted by Bailey, is that late

glacial foragers required large exploitation territories, while today we see

only a small portion of this territory in the small areas covered by surface

surveys.'80 The LUP settlement pattern was probably hierarchical, with a network of sites serving as home bases and special activity sites. This hier-

archy could extend from an ibex hunting camp in the mountains (e.g., Klithi), to seasonal bases in the upland basins (e.g., Kastritsa), to winter

175. Bailey 1992, 1997; Bailey et al. 1999; Kotzambopoulou, Panagopoulou, and Adam 1996; Koumouzelis et al. 1996; Kyparissi-Apostolika 1996; Perles 1987; Runnels 1995; Sordinas 1969.

176. Jameson, Runnels, and van Andel 1994, pp. 335-340.

177. Runnels 1996. 178. Dakaris, Higgs, and Hey 1964. 179. van Andel 1989. 180. Bailey et al. 1983a.

II6


base camps at lower elevations (e.g., Grava, Asprochaliko). Without tak-

ing the large size of this territory into consideration, the small scatters in

any one position are unintelligible.'81 The only candidate for a LUP site of any consequence in the nomos

of Preveza is Asprochaliko, where faunal remains of ibex, deer, elk, and aurochs are evidence of the chief prey of local foragers. But the small number of artifacts and less dense deposits at this site when compared with Kastritsa in the Ioannina basin are evidence that Asprochaliko was nevertheless a highly specialized site in a larger, hierarchical system.'82 Whether or not Asprochaliko was a residential base camp or a way station between the plains and the mountains, it is still the most likely focus for the scattered survey materials, which may be remnants of small, specialized camps or hunting stands utilized in the exploitation of the territory. The population in the entire nomos of Preveza at this time was probably limited to a single small band of ca. 25 to 75 persons resident on a seasonal basis. The near invisibility of the Late Upper Palaeolithic in the survey area can be explained by the small size of the human population, the limited territory investigated archaeologically, and the evident shift to a settlement pattern centered on residential base camps in caves and rockshelters, some of which were located outside the limits of the study area.

POST-PLEISTOCENE SETTLEMENT HISTORY

181. Bailey 1992; Bailey et al. 1983a.

182. Bailey et al. 1983b. 183. Bailey 1992. 184. Higgs and Vita-Finzi 1966. 185. Sordinas 1969. 186. Sordinas 1970. 187. Petruso et al. 1994. 188. F. Harrold and J. Wickens

(pers. comm.).

The evidence suggests that the principal LUP sites in Epirus were abandoned at the end of the last glacial (ca. 10-13 kyr B.P.). Klithi, Grava, and Kastritsa have no record of occupation in the immediate post-glacial period.183 Higgs noted a disturbed and mixed upper layer at Asprochaliko that is sometimes called "epipalaeolithic" in the literature, but this undated

level, described as having "backed blades and geometric microliths," is just as likely to be Late Upper Palaeolithic as Epipalaeolithic.184

There are, however, two excavated Mesolithic sites near Epirus. Sordi- nas excavated a Mesolithic coastal site at Sidari (Corfu) dated to ca. 8.5

kyr B.P. (7.8 kyr b.p.), and he noted the many differences in lithic technol-

ogy, raw material, and subsistence activity at that site when compared with the backed-blade industries of Upper Palaeolithic Grava Cave on the same island.185 Sidari is an open-air coastal midden site characterized by extensive use of marine resources and by a microlithic industry based on atypical trapezoidal fragments of flakes.186 Sordinas regarded Sidari as a new settlement by people arriving on the island by sea. The site of Konispol Cave (near the southern border of Albania) is close to the Epirote and Corfiote sites, none of which is more than 70 km from another. The exca- vators of Konispol found traces of LUP occupation, above which is a Meso- lithic deposit up to 0.90 m thick dated from 8-8.4 kyr B.P. (7-7.6 kyr b.p.).187 The Mesolithic industry consists of small flakes and blades with intensive minute retouch. There are many composite tools and also denticulates, notches, pergoirs, and end scrapers. Microlithic trapezes are

present, made by retouching segments of flakes and blades. The fauna include ibex and lesser quantities of aurochs, elk, and pig. The site was

apparently a temporary shelter used by seasonal hunters on an episodic basis.188 It is unclear how the Konispol Mesolithic industry compares with

II7


Figure 3.51. Palaeolithic and

!^j^^^-, <ii -^. .'";:,. 'Mesolithic site/scatters in the Preveza area. Palaeolithic materials

. ?o 2). 20a ^- 7 are found throughout this area, but recent erosion makes it difficult to

? O. Vathy ,), 20 .Vt identify specific sites. The Middle

". ' : ^?)fi|| ̂ ^^^^^ ^ DPalaeolithic symbols mark areas

PREVEZA : / where materials are most abundant.

jTsariambas d _i/ .A) One small findspot near Lake Site s cat

. Thomasi Pogonitsa (T93-12; W93-2) may be

Site/scatter '*l1-7 Alonaki / '

. t Beachc Lake Pogonitsa Upper Palaeolithic in age. Three

.* Meso..t..ic ..* . . Mesolithic sites (Tsarlambas and two

x U.? Palaeolithic Alonaki Beach sites) are located on

* M. Palaeolithic Coastal plain 2km the coast west of Preveza. Hatched areas are modern settlements.

that of Sidari, where a somewhat different lithic technology is found. The chief difference is that the Mesolithic industry of Sidari makes greater use of flakes than blades, a characteristic also of the contemporary upper levels at Franchthi Cave.

Six sites in the Preveza nomos are possibly Mesolithic (Figs. 3.32, 3.51). One site may be earlier than the others, corresponding to the Lower Mesolithic at Franchthi Cave;189 the other five are later. An IRSL date of ca. 7-10.5 kyr B.P. for Alonaki Beach (Table 3.10) is in general agreement with the dates for the other Mesolithic sites in Greece. A total of 731 lithics were collected from these sites by the same method employed for

sampling Palaeolithic sites, namely retrieval of all cores, blades, retouched tools, and complete flakes, discarding only fragments of debris without

recognizable features (namely, platforms, bulbs of percussion, or retouch). Three sites are located at the western end of the Acheron valley (Fig.

3.32). The site of Ammoudia (SS92-21) is on a low limestone ridge north of Ammoudia Bay, directly on the present shoreline. Tsouknida (SS92-8) is located in a Pleistocene alluvial fan (ca. 50 masl) on the edge of the valley overlooking what was in antiquity a marsh or lake. The site of Loutsa (SS93-32) is on a limestone bluff overlooking Tsouknida and the Acheron shoreline north to Parga. All three lie directly on thin Pleistocene soils, and the artifacts on the surface show evidence of burning and weathering suggesting that they have been exposed on stable surfaces for some time. Artifacts are abundant on all three sites. The chief reasons for regarding them as Mesolithic are the characteristic typology of the lithics (e.g., bladelets, end scrapers, bifacially-retouched pieces, pieces with silica gloss, and trapezes) and the lack of sherds.

The site of Tsouknida produced a few Middle Palaeolithic flakes in the upper part of the alluvial fan where it is dissected by ravines. The Mesolithic artifacts are small in size and, while patinated, are not as heavily weathered as the Palaeolithic pieces (Fig. 3.52:1-6). The assemblage of 189. Perles 1990.

II8


190. Perles 1990. 191. Koumouzelis et al. 1996;

Runnels 1996.

nearly 200 pieces consists of small flakes, often retouched, struck from globular cores (ca. 20% of all pieces). Bladelets are present in small numbers. The predominant tool type is a short, steep convex end scraper, sometimes created opportunistically by retouching only a part of one edge of a flake. There are many truncated and notched pieces and one small ovate bifacially retouched piece. Many of the artifacts have one or more edges modified by minute nibbling retouch.

In general the industry resembles the Lower Mesolithic at Franchthi Cave, where the transition from the Upper Palaeolithic is marked by a large reduction in the percentage of backed blades (from as much as 60% to about 3%) and the disappearance of microliths made by microburin technique, which are replaced by flakes with minute retouch.190 There are also end scrapers and notched pieces in the Franchthi Mesolithic. A similar industry dated to ca. 10 kyr B.P. has been published from rockshelters in the Argive Kleisoura.'91

The two other Acheron sites, Ammoudia and Loutsa, are quite different from Tsouknida. At Ammoudia artifacts, many of which are less than 5 mm in size, are found in a small area, ca. 50 x 25 m. Retouched tools (ca. 10% of the total) are made on snapped and broken fragments of bladelets and flakes that have been shaped by fine, nibbling retouch (Figs. 3.52:7-11, 3.53). The flintknapping technology is similar to Tsouknida, but with few end scrapers. Retouched tools at Ammoudia include trapezes, a single possible microburin, backed blades, and small numbers of notches, denticulates, and retouched pieces. A total of 173 artifacts were found in a paleosol ca. 50 cm thick overlying the bedrock and cut by coastal erosion. A curious stone structure is found on the site (Fig. 3.54). It consists of a semicircular foundation, 8 m in diameter, with a wall ca. 1-2 m thick. The structure is associated with pieces of burned daub with impressions of cane or wood, but without excavation it is difficult to determine if it is an ancient or modern feature.

The Mesolithic site of Loutsa is found in a small (70 x 50 m) area of residual terra rossa, less than one meter thick, resting on a karst limestone surface at an elevation of 200 masl. The terra rossa has been eroded down to bedrock on its edges but has been protected until recently by a cover of scrub vegetation. Erosion was accelerated in the last five years by the construction of a road on its southern edge. Artifacts are eroding from the sediments and the assemblage of 135 pieces has a large number of cores (16.8%). The number of retouched pieces is high (ca. 21%), with notched and denticulated pieces, composite tools, end scrapers, trapezes, and per,oirs in the collection (Fig. 3.55). There are four retouched flakes with silica gloss on their edges. Pieces of burned daub with impressions of cane or wood were found in the same area as the lithics.

The remaining Mesolithic sites are on the coast west of Preveza and can be considered together (Fig. 3.51). They are found directly on the present shoreline in rubefied sands, the Bt horizon of a moderately mature paleosol (MS 2) overlain by younger active dunes (Fig. 3.56). Develop- ment of summer housing and dense vegetation limited the search to the exposed sea scarp and areas between construction zones; sites were detected by looking for lithic artifacts embedded in the top 0.20-0.80 m of red sandy paleosol (Fig. 3.57). The underlying fossil dunes were only vis-

II9


3

9

Figure 3.52. Mesolithic artifacts from Tsouknida (1-6) and Ammoudia (7-11): 1,2,4,5) end

scrapers; 3) end scraper and perfoir on a retouched flake; 6) bifacially retouched piece; 7) bifacial straight truncation and left oblique bec on a flake; 8) truncated flake with small burin (microburin?); 9) flake with

abrupt truncation; 10) backed bladelet with double oblique truncation; 11) trapeze. Scale 1:1

Figure 3.53. Mesolithic trapeze from Ammoudia. Scale 1:1

1

4

2

5

8

11

7

10

I20

I?

goS


Figure 3.54. View of Ammoudia, go feature visible at left (middle ground) .... .....

ible where the overlying sand had been removed. The paleosol crumbles quickly when it is attacked by wave action and rain, exposing the flints embedded in the upper part of the soil horizon. It should be emphasized that flints were seen at other localities in the sea scarp where the surface was not obscured by dunes or buildings. As dunes are found from Mytikas to the Paliourias River, that area is likely to contain additional sites.

Tsarlambas (SS94-19) is the most disturbed of the sites and consists of a small wave-cut paleosol scarp. The other two sites, at Alonaki Beach (SS94-22, SS94-23), are richer. One of them (SS94-22) produced two bifacially retouched projectile points in addition to a good trapeze with a retouched truncation (Fig. 3.58:7-9). The artifacts are found in a semi- consolidated Bt horizon of a paleosol (MS 2). At least five sherds were noted at SS94-22 that may be prehistoric, and it is likely that there is some mixing of periods on the surface of the old Bt horizon. TL and IRSL dates of 7-10.5 kyr B.P. for this site were obtained from samples taken from within the Bt horizon.

The second site at Alonaki Beach (SS94-23) had only recently been exposed by deflation, and samples for TL and IRSL dating were taken from the paleosol that contains the lithics (Fig. 3.58). One trapeze and one piece with silica gloss were collected from the site and two other geometric pieces were noted in a subsequent visit. Although this was the smallest findspot we noted (ca. 25 x 25 m), the materials are very similar in

typology and technique to those from Tsarlambas and the other Alonaki Beach site, and the three sites may be contemporary.

One other small accumulation of flakes was detected by a general survey team on the summit ofTourkovouni, at the southeast extremity of the Ayios Thomas peninsula. A large sample (65 artifacts) was collected from the bare karst surface, but the artifacts are difficult to classify with any confidence. All are small plain flakes of light brown flint. One has been retouched to form a minute end scraper. No other prehistoric materials were noted on this peak during this or subsequent visits and the curious

I2I


2

I I

3 4

6 7

Figure 3.55. Mesolithic artifacts from Loutsa: 1, 2) cores; 3-5) core and flakes with silica gloss; 6) trapeze; 7) small point on basally retouched flake. Scale 1:1

1

5

I22


looking southwest, with rounded l I -

modern dunes overlying Bt horizon j . *

exposed in the foreground ' t..

artifact scatter near Preveza (SS94-

.and age. The raw materials and very small size of the flakes bring to mind..... ........................

the identical specimens from Mesolithic layer D at Sidari92 and the end

scraper is similar to those e coastal sites west of Preveza. The

............ . .. ,. : ;

absence of cores and paucity of retouched tools, however, prevent us from

The transition from the Mesolithic (post-Pleistocene sites without

villages with use of domesticated species) cannot be documented with our

may have continued down to 8 kyr B.P. or after, overlapping chronologi-

they appear to have had no contact. At some point the f Nikopolis

. ... . .... .

habitation are archaeologically invisible, and-until the very latest Neolithic

or Early Bronze Age brought new settlement-the long prehistoric record ~~~~192~ ~.. Sordinas 1970. of .occupation was at an end.

exposed by removal of modem sand . .. ... ..... . . . . . dunes ? . a Mt ... .. ..... . .

scraper is similar to those found at the coastal sites west of Preveza. The

making a definitive attribution and t his site is not included in our final

The transi tion from th e Mesolithic (post-Pleistocene s ites without evidence of domesticated plants or animals) to the Ne olithic (permanent

villages with use of domesticated species) cannot be documented with our

survey d ata. Of unknown duration, the M esolithic in northwest Gre ece

may ha ve con tinued down t o 8 kyr B.P. or after, overlapping chronologically with the Early Neolithic settlements of eastern Greece, with which they appear to have had no contact. At some point the area of the Nikopolis survey appears to have become uninhabited, or at least the traces of such habitation are archaeologically invisible, and-until the very latest Neolithic or Early Bronze Age brought new settlement-the long prehistoric record

192. Sordinas 1970. of occupation was at an end.

I23


2 3

5 6

! I I ,

7 8 9

10 11 12

13 14

4

I24

I I

J -


CONCLUSIONS

Figure 3.58 (opposite). Mesolithic artifacts from the Preveza area. Alonaki Beach, SS94-23: 1, 2) truncated flakes forming piercing tools; 3) flake with minute nibbling retouch. Alonaki Beach, SS94-22: 4) flake with minute nibbling retouch; 5-6)perfoirs; 7) tranchet arrowhead; 8) tanged arrowhead; 9) trapeze; 10) truncated flake; 11) bladelet; 12, 14) microliths; 13) blade fragment with silica gloss on right edge. Tsarlambas, SS94-19: 15) obsidian core. Scale 1:1

193. Runnels 1995. 194. Runnels 1988; Runnels and

van Andel 1993a. 195. Runnels 1995, 1996. 196. van Andel, Zangger, and

Demitrack 1990. 197. Runnels and van Andel 1993a.

The physiographic characteristics of Epirus, particularly the wide variety and quantity of karst features such as poljes, dolines, and loutses, have created conditions that made the landscape particularly useful to early humans and served to preserve their relics to the present day. The ten-

dency of karst depressions to collect sediments from the surrounding land-

scape and act as traps for aeolian dust is the key to understanding the fossil cultural landscape investigated in this survey.

Ever since the survey of northern Greece by Eric Higgs, Epirus has been recognized as being unusually rich in Palaeolithic remains. It is no

longer possible to attribute the Palaeolithic abundance here to a lack of

systematic investigations elsewhere. A number of topographic and site

surveys have been conducted since the 1970s, several of which targeted early prehistoric periods, and those efforts leave no doubt that some portions of Greece have little preserved evidence for Palaeolithic activity.193

An example of the disparity in numbers can illustrate this phenomenon. The Larisa district of Thessaly, roughly the size of the nomos of Preveza, has been investigated periodically for Palaeolithic materials by German, Greek, and American teams from 1959 to 1991.194 Despite the

intensity of the survey methods, particularly in the 1987-1991 survey, the total number of findspots is only around thirty and fewer than 1,000 lithic artifacts were collected. A more dramatic comparison with the Nikopolis survey, where a similar number of sites produced lithics 100 times more numerous, can hardly be imagined. Surveys of smaller areas in southern Greece (Argolid, Nemea, Berbati, Pylos) produced equally small findings, typically fewer than five sites per region, each producing fewer than 250 artifacts.195 The disparity in numbers is surely to be explained by the vari-

ety of geological contexts present in each region that have affected the preservation of sites and artifacts and perhaps a priori the density of settlement too. The karst depressions in the mountainous tracts of Epirus have concealed and preserved large numbers of Palaeolithic materials, while active erosion in the Peloponnese and central Greece, at times caused or accelerated by human activity, has destroyed many sites.196 In northeastern Greece the cycles of aggradation and incision of the great rivers complicates the picture by reworking, removing, and burying older sites.197

The karst features of Epirus created an attractive environment for early humans. The poljes and loutses filled with sediment and those depressions supported marshes, swamps, and lakes. The swamps and lakes may have been permanent, seasonal, or episodic, and sometimes they disappeared as uplift forced new stream systems that drained them temporarily and eventually for good. The Epirote system of poljes and loutses is dynamic, creating a mosaic of small environments that concentrated important resources at precise and predictable locations. The most important of these resources was water, which in turn supported lush vegetation and wildlife. In the dry season, which extended for six months from spring to autumn, these reser- voirs of water attracted birds, terrestrial animals, and humans. Lastly, as a consequence of the dissolution of the surrounding limestone, larger quan-

I25


tities of flint, often in the form of high-quality nodules, were concentrated near the red deposits or in the streams leading to and from them.

The evidence from Greece so far tends to resemble the "late entry" model of European colonization.198 A fossil cranium of an early sapient from Petralona may date to ca. 200-400 kyr B.p.199 and a lithic industry, perhaps of the same general age, has been found at sites near Larisa in

Thessaly.200 The Thessalian lithics, and artifacts discovered recently in Macedonia by a team from the University ofThessaloniki,201 are similar to the industry excavated in Yarimburgaz Cave near Istanbul, which probably dates to ca. 300-350 kyr B.P.202 We can reasonably conclude that humans were present in Greece by 300-400 kyr B.P. in sufficient numbers to leave a detectable archaeological signature. The occurrence of core-

chopper industries (often called Clactonian or Tayacian) with the Acheul- ean industry, which differs principally in having handaxes (bifaces), is

typical of western Asia and Europe where they are stratigraphically inter-

spersed.203 The Epirote discoveries agree generally with this picture of late colo-

nization, yet there are many unanswered questions that will require additional research. Are the heavy flake tools from Alonaki part of the late Acheulean technocomplex or are they a variant of early Mousterian? The question of the co-occurrence, here and much farther afield, of flake tool industries and industries with handaxes has still not been satisfactorily answered, and the presence of handaxes in Epirus serves only to remind us that we are dealing with a very small set of data. We cannot determine finally whether the differences in typology reflect temporal, environmental, or functional differences.

Our conclusion with regard to the earliest Palaeolithic is that these hunter-gatherers appear to have been the first humans to appreciate the rich environmental possibilities offered by the poljes and loutses of Epirus. The archaeological evidence for human activity in the Middle Palaeolithic period is very rich. Our survey added as many as 20 new findspots to the list compiled in the 1960s by Higgs and his students. As we have shown, these findspots reflect specific activities that appear to have been carried out repeatedly in the same place in the landscape. Many, but not all, of these findspots are associated with fossil Bt horizons (paleosols) of considerable maturity and hence age. Luminescence dates obtained from aeolian silt grains in these soils indicate that they may be on the order of 85-90,000 years old and the Palaeolithic artifacts that are contained in them may be older still.

This pattern of repeated logistical use of the same locations in the landscape is of considerable interest. A similar pattern of partially logistical land use in the Middle Palaeolithic was detected in the Argolid, but the small number of sites and artifacts hampered interpretation.204 The land-use evidence in Epirus is of much higher quality, partly because of the better preservation and partly because the fixed positions and long duration of the karst features helped shape the structure of human activity. Moreover, our search methods-designed to detect very small scatters of materials so that no findspot, however small, would be overlooked-allow us to reconstruct the use of the landscape in considerable detail. In our view the Mousterian sites appear to conform to a modified or partial lo-

198. Roebroeks and van Kolfschoten 1995.

199. Darlas 1995. 200. Runnels and van Andel 1993a. 201. K. Kotsakis and S. Andreou

(pers. comm.). 202. Arsebuk 1993, 1996; Kuhn,

Arsebuk, and Howell 1996; Stiner, Arsebiik, and Howell 1996.

203. Bar-Yosef 1998, p. 268. 204. Jameson, Runnels, and van

Andel 1994, pp. 325-335; Runnels 1996.

I26


205. Mellars 1996, pp. 366-391. 206. Mellars 1996, pp. 245-268. 207. Papagianni 1999. 208. Bailey et al. 1983b. 209. van Andel and Tzedakis 1996;

van Andel 1989. 210. Bailey, King, and Sturdy 1993.

gistical pattern. Specialized sites (e.g., for toolmaking, quarrying, and butch- ering) were distributed across the interior uplands and repeatedly revisited on a seasonal basis by foragers, who were Neanderthals, or more generally speaking, archaic Homo sapiens.

The question of Neanderthal behavior has been hotly debated, and scholars are divided in their views about how "modern" Neanderthal behavior should be regarded.205 Some scholars believe that they were essentially opportunistic foragers who had little ability to plan future activities or to structure their movements in the landscape to take advantage of predictable resources. Other scholars are willing to admit a limited faculty for logistical behavior. This issue is a large one and here is not the place to discuss the larger theoretical issues involved, but we believe that we can detect an essentially "modern" cast to the Mousterian pattern of land use.206

The basis for this conclusion is the repeated use of the poljes and loutses over long periods of time. At Kokkinopilos, Alonaki, Ayia, and the Anavatis quarry sites (as well as on Corfu), the large number of artifacts is an indication of the sustained level of past activity. If Neanderthals had foraged opportunistically across the landscape, we would expect to find materials spread far and wide and not associated with specific locations. There is additional evidence in the existence of very small sites that appear to have had specialized functions. In the Acheron valley a findspot at the eastern end (Skepasto) is a specialized quarry/flintknapping site, an interpretation that may also apply to two other sites (Valanidorrachi and Ormos Vathy). An analysis of the intersite variability of the Mousterian in Epirus by Papagianni supports this view.207

In a strictly logistical land-use pattern, base camps would be used as staging sites for specialized activities and located across the landscape, but no base camps have yet been identified in Epirus. Asprochaliko, the only excavated Mousterian site, is best regarded as a seasonal camp or hunting stand rather than a home base.208 The high incidence of retouched pieces at almost every site indicates specialization. The lack of preservation of faunal and floral remains and the absence of recognizable features, such as burials or fireplaces, remains a problem.

During much of the Middle Palaeolithic, shorelines were displaced seaward and the coastal plains provided additional space for winter base camps. At times, however, the coastal plains were greatly reduced in extent,209 and it would be unwise to try to explain the lack of base camps solely by reference to the submergence of the continental shelf. It is also possible that base camps are unrecognizable or did not exist. If the Mous- terian land-use pattern was a modified logistical one, the Neanderthals engaged in "residential mobility," that is, they moved about the landscape to maximize foraging opportunities, but were forward-looking enough to plan their moves to take advantage of water sources, flint, and game that they had learned from experience were located at certain specific sites which they visited on a seasonal schedule. This pattern of residential mobility resembles that documented for the Late Upper Palaeolithic,210 but differs from the LUP model by being centered on a set of predictable water holes in the loutses and poljes, rather than on vegetation communities governed by bedrock. The more specialized LUP model was perhaps typical of Late Pleistocene foragers.

I27


Kuhn has developed a model of Mousterian foraging for Latium (Italy) that may also be applicable to Epirus.211 In Kuhn's model the different

provisioning of raw materials, which is reflected in the Mousterian industries in a number of ways, can be used to reconstruct the degree of residential mobility of Mousterian foragers. Kuhn argues for rational logistical behavior on the part of Neanderthals,212 and describes a pair of strategies that may have been in use at different times during the Middle Palaeo- lithic.213 One strategy emphasized scavenging over hunting and required greater mobility of individuals over their territory. Mobile individuals were

provisioned with retouched tools and Levallois flakes as supports for trans-

ported artifacts. In this mode, cores are efficiently used and tools are heavily retouched and often recycled. In the second strategy, hunting of live game permitted groups to maintain longer residences in base camps where raw materials could be stockpiled in the form of nodules and cores and worked on site as needed. The result would be large concentrations of flintworking debris at relatively few sites, with the artifacts and cores showing little use of the Levallois technique and less intensively worked. These different

strategies of foraging are interchangeable. Kuhn nevertheless detects a definite chronological pattern in the Italian data, with a tendency for the

"hunting strategy" of intensive use of a smaller number of residential sites to predominate after ca. 55 kyr B.P.214 and both strategies to be tied to fluctuations in sea level and the resulting availability of resources.

Kuhn stresses that this pattern, while evident in the data from Latium, need not apply to the rest of Italy, much less the Mediterranean. But we believe that his general conclusion-that the Mousterian people "behaved over the long term in an economically rational manner, adjusting patterns of tool manufacture and use to fit problems inherent in different patterns of land-use"-has wider relevance.215

The Mousterian period in Epirus is of unknown duration but appears to extend from at least the last interglacial (ca. 115-130 kyr B.P.) to a point late in the late glacial (ca. 31 kyr B.P.). Although climatically it spans a warm interglacial, with a long slow decline to a first, rather modest glacial maximum (between ca. 70 and 60 kyr B.P.), followed by an equally long intermediate interstadial climate, it was evidently a time of cultural stability and continuity that may be divided roughly into two distinct phases, similar to those defined by Kuhn. In the first phase, the Mousterian was based on the production of large flakes and blades, often by means of the Levallois technique, which were modified into points, scrapers, denticulates, and other tools, and would correspond to Kuhn's scavenging strategy. In the second phase, which may have begun around 60 kyr B.P., there was a shift to smaller less carefully worked flake blanks struck from non-Levallois cores (disk and Mousterian cores), corresponding to Kuhn's hunting strategy. The flakes were used to produce a wide range of side scrapers and points.

Stiner and her colleagues have noted that the evidence for population density in the early Middle Palaeolithic suggests that Neanderthals and archaic human populations were very small and dispersed,216 making them very hard to detect archaeologically. The same authors also observed abrupt population density increases in the late Middle Palaeolithic and in the Late Upper Palaeolithic-Epipalaeolithic (Mesolithic), patterns which we

211. Kuhn 1995. 212. Kuhn 1995, pp. 174-180. 213. Kuhn 1995, pp. 36-37. 214. Kuhn 1995, pp. 157-183. 215. Kuhn 1995, p. 182. 216. Stiner et al. 1999.

I28


have observed also in Epirus. The evidence for these increases comes from the analysis of faunal remains from excavated sites that point to growing populations which were compelled to intensify food collection and to pursue lower-ranked prey and other food sources. We cannot test their hypothesis with faunal data from Epirus, but demographic pressure might be one explanation for the concentration in Epirus on the karst features with their dependable (if low-ranked) resources such as fish, birds, mollusks, turtles, and reptiles.

Our data from surface sites cannot be compared with that from a large number of excavated stratified sites, and chronologically meaningful patterns of land-use strategy are thus difficult to support. The incidence of Levallois Mousterian and large numbers of retouched tools at some small

outlying loutses (e.g., Ayia, Kranea, and the Loutsa sites), however, may reflect periods of high mobility, perhaps linked with scavenging and the use of scattered small-scale resources such as shellfish, aquatic animals and birds, turtles, and the like.217 On the other hand, sites like Kokkinopilos, Alonaki, and Morphi have enormous quantities of flintknapping debris that may reflect longer-term residence at preferred central sites with a lesser

degree of mobility and greater emphasis on hunting. No doubt there was a mixture of these patterns that resulted from the adjustment of the Mous- terian people to the large, often sharp oscillations of climate experienced over the 70,000 years or more of the Middle Palaeolithic period.218

The principal results of our analysis are confined to the elucidation of the spatial land-use patterns that appear to have been broadly synchronic. The lack of a detailed chronology presents important difficulties, in particular the inability to determine whether any sites are truly contemporary.

THE TRANSITION FROM THE MOUSTERIAN TO THE

EARLY UPPER PALAEOLITHIC AND THE ORIGINS OF

MODERN HUMANS

217. See Kuhn 1995, pp. 150-151, 168.

218. van Andel and Tzedakis 1998. 219. Runnels and van Andel 1993a. 220. Harrold 1993; Mellars 1996,

pp. 392-419.

Asprochaliko Cave is the only site in Epirus where the transition from the Mousterian (Middle Palaeolithic) to the Upper Palaeolithic can be seen. There is a stratigraphic hiatus between the Middle and Upper Palaeolithic levels in this site. A similar disconformity is seen in Thessaly where the Peneios River sites have a recognizable Mousterian that persists until ca. 31 kyr B.P., after which time it abruptly ceases.219 An interesting feature of the Thessalian Mousterian is the admixture of Mousterian tool types (side scrapers, points) and techniques (Levallois) with EUP types (end scrapers, burins, and retouched blades). This kind of mixed industry is found widely in the Balkans and westward in Europe (e.g., the Chatel- perronian) where it is interpreted as evidence of the incorporation by Ne- anderthals of tool types and techniques as the result of contact with anatomically modern Homo sapiens. The existence of this late Mousterian has many implications for the debate surrounding the origins of modern humans.220 The abrupt disappearance of the Mousterian throughout Greece is evidence for their ultimate extinction because this disappearance comes after a period of overlap of the Mousterian peoples with anatomical moderns. Greece was evidently one of several isolated geographic refuges for late Neanderthals. Here, as in Iberia (Zafaryya Cave; Lapedo) and

I29


Italy (Ulluzian), Neanderthal sites continue as late as 31 kyr B.P. or even later, indicating the survival of Neanderthals long after anatomically modern humans had come to be the only human species in the rest of Europe.221

The evidence from Epirus may support the replacement hypothesis of modern human origins. Spilaion, at the mouth of the Acheron River, is a credible candidate for a home base with its abundant flintknapping debitage. The site is a large and very rich EUP site providing evidence for a totally different pattern of land use. Outside of this site, EUP artifacts are found only as small scatters at poljes and loutses, suggesting that these karst features were little used.

THE LATE UPPER PALAEOLITHIC AND THE MESOLITHIC

The uncertainty in dating the transition from the Early to Late Upper Palaeolithic, before the establishment in the stratified cave sites of the better-known Gravettian and Epigravettian industries rich in backed blades, is an unsolved problem. The occurrence of Gravettian and Epi- gravettian industries in Epirus coincides with a move from open-air sites to rockshelters and caves and a concomitant change in land-use strategy. The change in resource exploitation has been studied in detail by Bailey and his colleagues who have developed a model of seasonal exploitation of big game animals.222 In their model, the cave sites of Asprochaliko and Kastritsa functioned as hunting camps, located strategically to control points of access to limestone plateaus where the largest numbers of animals would feed. This pattern is fully logistical, a fact that can be supported by the composition of faunal remains and lithics in the cave deposits that demonstrate the specialized activities taking place at each site. Bailey and his colleagues are surely correct when they argue that the LUP exploitation territories were very large and included coastal plains exposed during the last glacial maximum. They locate the home bases on the continental shelf and relegate the known mainland sites to the status of subordinate seasonal special-activity sites in a regional settlement hierarchy, as Higgs had concluded earlier.223 The change in exploitation strategy to the pursuit of large game (by monitoring, close following, and ambush tactics) and the inclusion of coastal plains in the territory covered can together explain the sharply marked differences between the Middle and Upper Palaeolithic.

The Late Upper Palaeolithic is a very short phase in Epirus and it appears to have ended sometime between 10 and 13 kyr B.P.224 The only firm evidence for the latest Palaeolithic in the Nikopolis survey area comes from Asprochaliko where the upper layers are much disturbed and have not been fully published. The lithics from the uppermost excavated layer (as reported) agree with the latest radiocarbon dates and indicate an aban- donment of the cave at the end of the Pleistocene, a conclusion supported by the most recent excavations at Klithi in the Zagori.

The transition from the Pleistocene to the post-Pleistocene is difficult to interpret in Greece because of the small database. The Nikopolis survey has contributed six new sites to those known for that period in Greece, all located close to the early Holocene shoreline, and it is probable

221. D'Errico et al. 1998. 222. Bailey, King, and Sturdy 1993. 223. Bailey, King, and Sturdy 1993;

Higgs and Vita-Finzi 1966; Higgs et al. 1967.

224. Bailey 1992.

I30


that additional sites are still to be found along the western coast of Epirus. The new sites are found in two areas, the Acheron River valley and among the coastal dunes west of Preveza. Despite extensive searching of the poljes and loutses, and other areas, no Mesolithic artifacts have been identified in these contexts.

The Acheron sites have the same lithic industry as the Preveza dune sites and resemble the coastal midden site of Sidari on Corfu. Four of the Epirote sites are in geographic settings similar to that of Sidari. Ammoudia sits on a low hill overlooking the sea at the mouth of the Acheron River and the three sites west of Preveza are directly on the modern shoreline among dunes situated in an ancient complex mosaic of streams, swamps, coastal lagoons, and dunes. Tsouknida and Loutsa are in quite different settings. Tsouknida, probably the earliest of the sites, is located on a ridge that protrudes into the river valley and may have looked over an area of marshes and swamps, while Loutsa sits on a limestone ridge high above Tsouknida with a commanding view of the western end of the valley. The lithics point to hunting as the main activity at all sites: the trapezes could have been used to tip arrows, while small multipurpose tools were useful for maintaining and repairing hunting equipment or working reeds to build simple huts or "papyrella"-type boats.

A small core, evidently of obsidian, from Tsarlambas (Fig. 3.58:15) is a reminder that these small bands of hunters and fishers were aware of a larger world.225 To this may be added the coastal orientation of the Epirote and Corfu sites, suggesting that seafaring was part of the Mesolithic adaptation.

THE CONTRIBUTION OF THE PREHISTORIC SURVEY TO THE NIKOPOLIS PROJECT

225. The source of the obsidian is unknown, but the visual appearance of the material resembles the "snowflake" obsidian from Italian volcanic fields such as Lipari.

Southern Epirus has a rich archaeological record and its karst morphology preserves the lineaments of a fossil cultural landscape in the sense that the surface has traces of human activity associated with still extant portions of the landscape. The pattern of ancient land use is seen at more than one scale, from the small but rich site of Spilaion, where the flintknapping activity of the Early Upper Paleolithic is registered in the static structure of the lithic scatter, to the scale of the entire province, with its network of poljes, loutses, and sites.

In the end, how do we account for the richness of Epirus? There are three major factors. The first is the prevailing weather pattern that brings westerly rains to the province and supports a relatively lush flora. To this factor we can add the patchwork of runoff-collecting karst depressions that owe their existence to the present active tectonics of the region. Thus the Epirote landscape is dotted with marshy basins offering a wide variety of essential resources. Lastly, Epirus is separated from eastern Greece by the Pindos Range and from central and southern Greece by the Ambracian and Corinthian Gulfs. As a result, the inhabitants of the region have often been relatively independent and in cultural terms oriented more to the Balkans and the Italian peninsula than to the rest of Greece or the Aegean.

I3


The Balkans are easily reached by following the northwest-southeast trending valleys of the central massif, and Italy was connected by wide coastal plains during cold periods and by the sea at other times. A sea passage to southern Italy via Corfu has always been safer than the more perilous voy- age around the Peloponnese to the Aegean.

These factors have acted together to draw human migrants to Epirus in many periods, by land and sea, from Italy and southeast Europe. The relative geographical isolation and rich ecosystem of the region encouraged stable human adaptations during the Palaeolithic period. We may add to this the redbeds that have preserved a record of early cultural activity, and the archaeological riches of Epirus can be explained without in- voking different intensities of research or survey methods.

How does the prehistory of Epirus contribute to the understanding of the larger issues confronted by the Nikopolis Project? Perhaps the most striking feature of the early prehistory is the apparent decline of the human presence in the Upper Palaeolithic and Mesolithic. This tendency continues into later times, with the possibility that the nomos of Preveza was largely uninhabited during part of the Neolithic. The break in the chain of human presence may have had consequences for later periods by depriving the region of native inhabitants in possession of long-standing traditions. This hypothesis is too uncertain to pursue further, but another may be more to the point. The record is clearly an indication of the tendency in all periods for the inhabitants of this province to look first to themselves for support and inspiration, and secondly toward southeast Europe and Italy. This last feature looms large in the subsequent history of the region and is returning to prominence in the present age.

One last point can be mentioned. Almost all the later historical sites that rise to prominence after the end of the Bronze Age and the beginning of the Iron Age, to say nothing of modern times, are situated in or near those areas rich in water and soil and which are also those that have yielded the richest prehistoric finds. The emphasis found in this report on the properties of the landscape that helped to shape human behavior is surely useful for students of these historical periods.

I32

EARLY STONE AGE OF THEOF THE NOMOS OF PREVEZA

ACKNOWLEDGMENTS

A great many people participated in the Nikopolis Project and made contributions to our fieldwork and analysis, and others contributed to the

preparation of this report. To all of them we are deeply grateful. We wish to thank especially James Wiseman and Konstantinos Zachos for inviting us to participate in the project and for giving us aid and encouragement. We wish to thank Panayiotis Paschos of the Preveza IGME office, who was always generous and helpful in the field. We also thank Gillian Fore- man, Chris Jeans, and Richard Powys at the University of Cambridge for the sedimentological analyses, and we owe a special debt of gratitude to

Li-Ping Zhou, Andreas Lang, and Ann Wintle, who gave us much valuable advice in the preparation of the section on Late Quaternary chronol-

ogy. Our thanks are due to the many Boston University field school students who joined us from time to time in the field. James Wiseman and Dimitra Papagianni read early drafts of this paper and their thoughtful comments were very helpful. Our greatest debt, and hence our deepest appreciation, is due to Priscilla Murray, who labored with us in the field from first to last and who contributed useful advice, penetrating insight, and some of the best finds.

I33


APPENDIX: PALAEOLITHIC AND MESOLITHIC SITE/SCATTERS IN THE SURVEY AREA*

Lower Middle Upper Name SS/T/WNo. Palaeolithic Palaeolithic Palaeolithic Mesolithic

Alonaki

Alonaki Beach

Ammoudia Anavatis quarry

Ayia Ayia Kyriaki Ayios Thomas Cheimadio

Eli Galatas

Gymnon Iliovouni

Kokkinopilos Koumasaki Kranea Lake Pogonitsa Loutsa

Mesaria Ormos Odysseos Rizovouni Rodaki Romia

Skepasto Spilaion Stephani Tsarlambas Tsouknida Valanidorrachi

Vouvopotamos

SS92-22 SS92-23 SS94-22 SS94-23 SS92-21 SS94-13 SS94-16 SS93-9 n/a T93-3-5,17 SS94-1 SS94-2 SS94-18 SS92-19 SS92-13 SS92-17 SS92-9 SS91-3 SS92-24 SS92-14 T93-12, W93-2 SS93-31 SS94-12 SS93-32 SS92-16 SS92-25 SS92-11 SS92-15 SS92-10 SS92-20 SS92-37 SS92-18 SS94-19 SS92-8 SS91-4 SS92-12

*Only datable site/scatters are listed here.

x

x

x

x

x x x x x x x x x x x x x x x x

x

x

x

x

x

x x x

x

x

x

x

x

x

x

x

x

x

x

x

x

x

x

x

x

x

I34

CHAPTER 4

EARLY UPPER PALAEOLITHIC

SPILAION: AN ARTIFACT-RICH

SURFACE SITE

by Curtis N. Runnels, Evangelia Karimali, and Brenda Cullen

The site of Spilaion, a high-density lithic scatter located near the mouth of the Acheron River (Fig. 4.1), is a good example of a surface site that

archaeologists often find difficult to interpret.1 Spilaion does not have a cultural context in the usual sense and belongs instead to a class of sites that exist entirely on the surface with no extant stratification or cultural

deposits. While it has long been recognized that some sites may be strictly confined to the present surface with no relationship to a stratified deposit, such sites are often dismissed as of relatively little value for interpreting the past. We believe that it is possible to extract valuable information about past cultural activities from two-dimensional spatial associations preserved on such sites. Our analysis of Spilaion's high-density artifact distribution demonstrates that these surface sites may retain substantial evidence of spatial distributions created by past cultural activities.

The discovery and study of surface sites of all kinds is at the heart of intensive regional survey, regardless of whether sites are the manifestations of buried strata or phenomena existing only on the present-day surface. Careful study of high-density and low-density artifact scatters has demonstrated that they may preserve information about human activities in the past.2

As archaeologists have come to appreciate the richness of the archaeological record, various problems associated with explaining how concentrations of artifacts on the surface were formed have been recognized. One major issue is that archaeological materials are not always distributed on the surface in well-bounded clusters, but may exist in patchy and largely

1. We want to thank the codirectors of the Nikopolis Project, James Wise- man and Konstantinos Zachos, for their support and encouragement. We benefited also from discussions with Kenneth Kvamme in connection with the spatial analysis and with Janusz Kozlowski, Ofer Bar-Yosef, and Paul Mellars in connection with the lithic industry. We wish especially to thank Kael Alford, who oversaw the collecting

of the random samples in 1992. Tjeerd van Andel and James Wiseman kindly read an early draft of this chapter and made many useful suggestions that we have endeavored to address.

Very large concentrations of artifacts are found directly on the surface in many parts of the Mediterra- nean world, where they are sometimes called scatters, findspots, or high- density distributions. We prefer to use

the traditional term of "site" in its sense of "place" or "location" as a general term for any surface scatter of artifacts that has a well-defined spatial concentration of cultural materials, but without the unnecessary connotation of "settlement" or "habitation" that this term sometimes carries.

2. Alcock, Cherry, and Davis 1994; Cherry et al. 1988.

C. N. RUNNELS, E. KARIMALI, AND B. CULLEN

Louros River

Acheron River

rp

.1. v Spilaion

Arachthos _ River

Ionian Sea 3

~7,': Ambracian Gulf

4;: ̂ -^Î

0 5 10 15 20 25 KM

Figure 4.1. Map showing the location of Spilaion at the mouth of the Acheron River

discontinuous spatial patterns.3 This observation raises serious questions regarding the traditional concept of "site,"4 and underscores the need to refine field techniques and explanatory hypotheses that account for the formation of artifact scatters.5 The number of artifacts or the relative den-

sity of artifact concentrations is usually employed to define intra-site

density thresholds and to convey the degree of a site's discreteness or boundedness,6 but there is a consensus that the explanation of any surface site is a complicated issue that demands an understanding of the geological and cultural processes active at the site.

This problem has been faced in various ways. One approach is to define new sampling and quantitative methods that will increase the accuracy of the recovery techniques.7 Another is to draw on ethnographic analo-

gies to explain how artifacts are distributed on the surface (e.g., field

manuring).8 Yet another attributes surface scatters to short-term cultural episodes, such as flintknapping,9 the off-site storage of equipment by agri- culturalists or pastoralists, or isolated activities associated with animal folds, milking pens, or dumps.10 Other forces that are invoked to account for low-density distributions of artifacts include natural or anthro- pogenically induced processes such as soil erosion, deflation, or downslope movement.1l

In one case study carried out in the Southern Argolid in the early 1980s, soil erosion was documented in every valley investigated and was cited as one agent for moving artifacts from higher elevations to low-lying parts of valleys and spreading them over the plains.'2 Evidence for the

3. Cherry et al. 1991; Wright et al. 1990; Wells, Runnels, and Zangger 1990.

4. Fotiadis 1992. 5. Cherry et al. 1991; Bintliff and

Snodgrass 1988b. 6. E.g., Cherry et al. 1988, 1991. 7. Kvamme 1996. 8. Alcock, Cherry, and Davis 1994. 9. Kvamme 1996. 10. Murray and Kardulias 1986. 11. Wells, Runnels, and Zangger

1990; Whitelaw 1991. 12. Jameson, Runnels, and van

Andel 1994, pp. 172-194, 325-414; Pope and van Andel 1984.

M /+- I /

??? ?

I36

e.

I JR

*f\

-Y

EARLY UPPER PALAEOLITHIC SPILAION

rearrangement of valley watersheds by anthropogenic erosion is also found in the Berbati and Limnes valleys in the northern Argolid, where it is cited as a major agent in shaping the archaeological record.13 In the Corinthia, tectonics and manuring are cited as of equal importance to an-

thropogenic erosion in the formation of off-site scatters.14 From these examples, and the others cited above, it is evident that a

diverse array of factors is responsible for scattering cultural materials across the surface of Greece and creating both low-density and high-density distributions. Although some surface scatters of artifacts are explained as the result of physical processes, we take the position that some surface sites, especially high-density concentrations of artifacts, preserve cultural information in the form of spatial patterning, typically of a kind that has only recently engaged the full attention of archaeologists. On the most basic level, if artifacts were once part of a cultural matrix that has been destroyed, their location in the landscape is nevertheless the result of cultural choice, and is of some use in reconstructing prehistoric settlement and land-use patterns.

TAPHONOMY OF SURFACE SITES

13. Wells 1994; Wells, Runnels, and Zangger 1990; Zangger 1992.

14. Wright et al. 1990. 15. Schiffer 1987. 16. Rick 1976. 17. Efstratiou 1985.

The study of site formation processes has received serious and widespread attention since the 1970s and covers those natural and cultural operations that are responsible for the patterning of materials found in the archaeological record. Although a systematic and comprehensive survey of all processes involved in the formation of sites is far from complete,15 much has been learned that is useful for interpreting high-density artifact concentrations on the surface.

The downslope shifting of artifacts is one factor affecting surface sites on steep terrain. This process was studied in the Andes by Rick,16 who found that the absence of topographic barriers or heavy vegetation on one such site permitted heavier artifacts to move downhill through time as the result of gravity and slope wash, leaving only small artifacts in situ at the top of the slope. Rick concluded that in places where slopes are steep the operation of gravity will tend to sort artifacts by size and mass, and that only smaller artifacts will preserve cultural patterning.

Another factor affecting the spatial distribution of artifacts on the surface is deflation, a physical process by which wind removes the fine- grained materials from a site, leaving the larger, heavier artifacts exposed on the surface. In a related process, wind and water can move surface artifacts from one place to another horizontally through sheet erosion. Coastal sites may be affected by rising sea or lake levels. Recent investigations of submerged sites have shown that the horizontal patterning of cultural materials may be retained even after the stratification of the site has been destroyed. At Ayios Petros, for example, a Neolithic site in the Aegean Sporades, Efstratiou demonstrated that the spatial distribution of pot- sherds and other artifacts in the settlement was preserved, despite exposure to the full force of winter waves as the sea passed over the site with the eustatic rise in sea level in the early Holocene.17

I37


Support for the hypothesis that surface scatters may preserve evidence of spatial associations comes from a long-term study of an Archaic period site in the desert southwest of the United States, a site similar to Spilaion in terms of its high-density distribution of lithics.18 Assuming that the spatial distribution of the lithics was at least partly determined by cultural patterning, Kvamme subjected the site to detailed mapping, employing a Geographic Information System (GIS) to analyze the distribution. By using GIS in such an analysis, Kvamme was able to assess the distribution of individual typological classes of lithics in relation to changes in topography, vegetation cover, slope, and other geographic features. The combination of detailed mapping and spatial analysis detected a well-preserved pattern of cultural activity, with numerous concentrations or "hotspots," each consisting of a ring of large pieces of debitage on the margins and small-sized flakes in the center. Kvamme explains this structure as the result of the sorting of debitage by size that occurs during flintknapping, which he regards as a fundamental formation process in high-density lithic distributions. He supports his theory with computer modeling and observations derived from hard-hammer flintknapping experiments. The structure of the lithic distribution led Kvamme to conclude that the artifacts were deposited by humans who repeatedly visited the ridge, where they camped and engaged in flintknapping and toolmaking.19

It is thus probable that the natural and cultural forces responsible for the formation of sites can be identified and distinguished from one another on the basis of geomorphological observations and artifact analysis. Even in those cases where a site has been seriously affected by erosion, downslope movement, deflation, or submergence, there is reason to believe that the horizontal patterning of the artifact distribution may preserve some culturally significant associations. Our spatial analysis of the prehistoric lithic site of Spilaion is intended to demonstrate the value that artifact-rich surface sites may have for archaeology.

THE SITE OF SPILAION

The site of Spilaion (SS92-37) is a large, dense scatter of artifacts distributed over the surface of a low hillock at the western end of the Acheron River valley (Figs. 4.2, 4.3). The site was discovered in 1992 in the course of the study of Palaeolithic and Mesolithic sites (see Chapter 3). The name Spilaion (or "cavern" in Greek) refers to a sinkhole on the northern side of the hillock, which is in fact a limestone outcrop made up of a highly weathered karst surface. The site is now situated approximately one kilometer from the present coast. In the Pleistocene, however, at times when the sea level was lower, the site was much further inland and overlooked a gently sloping valley where the paleo-Acheron River flowed down to a coastal plain. At times of higher sea level, in the Pleistocene and the Holocene, the shoreline was often closer. The elevation of the site ensured that at all times it had a commanding view of the entire area. The present karst surface of Spilaion has a thin covering of scrub vegetation (evergreen oak and 18. Kvamme 1996; K. Kvamme olives) rooted in crevices in the rock and small patches of sediments, chiefly (pers. comm.). relict Pleistocene deposits or terra rossa (Fig. 4.4). 19. Kvamme 1996.

I38


Figure 4.2. Map of Spilaion showing topographic contours. The artifact / scatter is found on the south- southeast slope. The square marks the location of the 60 x 50 m sample grid, which is located in the approximate center of the scatter. . c

Figure 4.3. View of Spilaion, looking southwest. The artifacts are found in the open areas visible in the foreground. J - ..

The site consists of a dense, continuous, and extensive scatter oflithic

implements on the south-southeast slope of the hillock. We estimate that ca. 150,000 artifacts or fragments of artifacts are present within an area more than one hectare in size, with a density that averaged ca. 15 artifacts/ m2. We measured the number of lithics on the site by counting the lithics

present in 100 sample units (5 x 5 m each) and using this count to estimate the density of artifacts on the entire site. The limits of the site and the variable density of lithics were investigated by repeated walkovers to examine the surface visually. During the walkovers, the number of artifacts

was tabulated by three persons with handheld counters, stopping to record counts at one-meter intervals.

The majority of the Spilaion lithics are uniform in typology, technol-

ogy, and raw material, and appear to belong to a single cultural component. A small number of Late Bronze Age sherds were found in some of the sample units, but there are several reasons for believing that the major-

I39


Figure 4.4. View of the rugged karst surface on the southeast slope of Spilaion at the time of collection

ity of the lithics are not of the same date as the sherds. An inspection of lithics from Bronze Age sites at other locations in the Acheron valley permits us to distinguish lithics typical of the Bronze Age from those of the Palaeolithic.20 Bronze Age lithics are unpatinated, smaller in scale, made of different raw materials, and include retouched tool types that are different from those of the Palaeolithic in general and from those found at

Spilaion in particular. Small numbers of possible Bronze Age lithics were collected at Spilaion (much less than 1% of the total) and were easily dis-

tinguished from the earlier materials on the basis of their morphological characteristics.

The Spilaion lithics were manufactured from a good quality, fine-

grained, blue-gray flint available in local deposits of Cretaceous limestone. The artifacts exhibit a considerable degree of weathering manifested in the form of a thick white patina, an indication that the artifacts have been

exposed on the surface for some time. At present we have no means of

ascertaining how long artifacts must be exposed to acquire such a patina. There has been little study of the phenomenon of patina formation with

regard to lithic artifacts in Greece, but our observations of materials from other sites may provide a few guidelines.

Flints found on the surface and exposed to weathering are usually

patinated. This weathering process can begin at two points: either at the time of deposition, before the artifact is buried in a deposit of some kind, or once the artifact has been removed from whatever matrix it was in. Middle Palaeolithic artifacts from the nearby site of Alonaki have an estimated age greater than 50 kyr B.P. (thousands of years before present) and are sometimes weathered completely through, rendering the original flint into a material resembling chalk. Other sites in the Acheron valley have artifacts of Bronze Age type (e.g., tanged and barbed arrowheads) that are

unlikely to be more than 6,000 years in age and are not deeply patinated. These facts provide reasonable criteria for estimating the amount of time

necessary for accumulating a patina on flint artifacts, although we must allow for much variability and uncertainty. The Spilaion lithic artifacts are rather heavily patinated when compared with Bronze Age lithics, and are

20. Tartaron 1996; Tartaron, Runnels, and Karimali 1999.

I40


Figure 4.5. Sample grid on the southeast slope of Spilaion during 3 . collection

generally patinated to the same degree or slightly less than the Middle Palaeolithic ones. Thus we conclude that they have been exposed to weath-

ering for much more than 6,000 and somewhat less than 50,000 years. The numerous lithic artifacts and the technological and morphologi-

cal uniformity exhibited by the tools and the other debitage categories justifies the use of the term "site" for this scatter. The initial survey convinced us that this site required detailed study to determine if cultural

patterning was present in the spatial distribution of the artifacts. The large number of lithics, which may have been exposed directly on the surface for

many thousands of years, offered an unusual opportunity to test the effi-

cacy of spatial analysis for investigating the forces responsible for the site's formation.

To document the spatial distribution of the Spilaion lithic scatter, a

systematic program of sampling was undertaken in 1993. Although the site is more than one hectare in extent, the slope of the ridge is irregular and has steeper patches of nearly bare rock interspersed with flat areas that

trap fine sediment and artifacts. The patches of sediment (Pleistocene terra

rossa) are no more than 0.15 m in thickness. Taking into account the large size of the site and the irregularity of the terrain, a small sample grid (60 x 50 m) was established on the part of the slope where the surface was not obstructed by vegetation or interrupted by outcrops of limestone (Fig. 4.5). The grid was subdivided into four large squares (T93-100, T93-126, T93-127, and T93-128), each 25 m on a side, with a 10-m wide strip separating the northern and southern squares. Each square was then subdivided into twenty-five smaller squares measuring 5 m on a side. All of the lithics on the surface were collected from eighty of the 5 x 5 m

sample units (one large grid square, T93-126, was subsampled with a random sample of five squares). The density of artifacts in the sample area was low, and there were many small, undiagnostic fragments. A total of 3,218 identifiable artifacts were selected for the analyses reported here. Portions of the site not covered by the sample grid were investigated by walkovers to verify that the lithics collected are representative of the site as a whole.

I4I


THE LITHIC ASSEMBLAGE

On the basis of a rather limited range of tool types-chiefly carinated and nosed end scrapers on retouched blades and flakes, notches and denticulates, and rare burins-and a lack of typical Aurignacian bladelets (Dufour bladelets) and microretouched points, the lithic assemblage from Spilaion most closely resembles the Typical Balkan Aurignacian.21 Detailed study of the lithic artifacts was carried out in 1994 and 1995 with three aims: to

identify diagnostic tool types; to understand the technical and morphological characteristics of the assemblage in terms of the recognized lithic reduction stages (chaine operatoire); and to assess the spatial distribution of the by-products of the different stages in the reduction sequence, especially cores, cortical flakes, plain flakes, and retouched tools. The Spilaion lithics have suffered from their long exposure on the surface, and heavy patination, breakage, and scarring make many of the pieces difficult to classify. Although the sample of retouched artifacts is relatively large (n = 131), the absence of stratigraphic context, the possibility of a mixture of different phases of activity, and the difficulties of classifying the dam-

aged surface materials make our conclusions regarding the cultural affinities of the assemblage somewhat tentative.

The lithics were sorted on the basis of five technical categories: cores, cortical flakes, plain flakes, blades, and retouched tools (Table 4.1). Dif- ferent attributes were recorded for each category, including size, raw material, technology, and presence of cortex. Maximum length, width, and thickness were measured for each artifact. The maximum widths of platforms and flake scars on the faces of cores were recorded when possible. Other characteristics that were noted include patina, the state of preservation of the artifacts, and the identification of the flintknapping techniques employed. The classification of flakes was based on the following categories: primary (>75% cortical), secondary (25-75% cortical), and plain ("tertiary") flakes (<25% cortical).

The majority of the artifacts were manufactured from bluish gray flint, but flint of other varieties is also present, ranging in color from grayish orange to moderate reddish brown. Pieces in other colors, however, are usually found with no patina and exhibit different technological characteristics, thus we attribute them to a later period (possibly the Bronze Age). Many artifacts show evidence of burning, such as crazing, potlid fractures, and a reddish color.

Cores comprise 6.5% of sample (Table 4.1) and are derived primarily from cobbles gathered from the river rather than nodules extracted directly from outcrops. These cores have traces of a thick cortex, which was usually removed with the first series of flakes (25.7% of the cores have cortex remaining on their surface). The flake cores are variable in size and shape, and they include globular or polyhedral (13.8%), conical (6.7%), flat (2.8%), spherical (1%), and rectilinear (0.5%) types. The majority of cores are irregular in shape (37%) or are fragmentary (32%). Two Middle Palaeolithic Levallois cores were also noted in the sample. 21. Kozlowski 1999.

I42


TABLE 4.1. CATEGORIES OF FLINTKNAPPING DEBITAGE

Type n %

Cores 209 6.5 Cortical flakes 399 12.4 Plain flakes 2,419 75.2 Blades 60 1.8 Retouched tools 131 4.1

Total 3,218 100

22. Kozlowski 1999, p. 106.

Given the great variety of forms, the cores may be divided into two broad categories: rounded cobbles (core-choppers) and unshaped nodules.

Core-choppers preserve vestiges of unidirectional and, more rarely, bidirectional flaking. In the case of bidirectional flaking, the blows were directed to platforms positioned at right angles on the core surface. Un-

shaped nodules were heavily flaked in different directions and usually retain little cortex. Some of these specimens are small discoids with irregularly placed striking platforms resulting from multidirectional flaking. In many cases several platforms were created on a single core, although only a small number of flakes were struck from them. The same platform was com-

monly used for detaching flakes from more than one face of the core. These

strategies resulted in continuous flaking along the periphery of the core. Core rejuvenation was accomplished by flaking off a piece of the core car-

rying the old platform. Four conical cores collected from one tract (T93-128) range from 16

to 56 mm in length and from 15.7 to 37.7 mm in thickness. These cores have a single plain flat platform from which semiparallel or irregular flakes were detached by direct percussion. Blade cores are represented by two

complete and eight fragmentary specimens, and were used for striking elongated and irregular blades/flakes by direct percussion. The best ex-

ample of a complete blade core (65.6 mm in length and 34 mm in thickness) is patinated and has traces of large, irregular blade scars originating from one plain platform.

Core-processing activities at Spilaion show that lithic strategies were expedient and opportunistic in terms of goals, techniques, and the quality of the final products. The knappers employed simple direct percussion to remove blanks, and there is no evidence for the use of preparatory techniques such as cresting or platform faceting. The relatively high frequency of plain noncortical flakes with little evidence of use indicates that core testing and preparation took place repeatedly at the site.

Given the expedient flaking procedures at Spilaion, the goal of the chaine operatoire followed at the site is unclear. It appears that flaking was intended to extract flakes of different sizes suitable for deliberate retouch. Negative flake scars preserved on the cores indicate that flakes ranged from 8 to 15 mm in length. Blades were evidently produced in much the same manner as flakes. The rarity of typical Aurignacian bladelets is a characteristic of the Typical Balkan Aurignacian (and the earlier Bachokirian)22

I43


TABLE 4.2. TYPES OF RETOUCHED TOOLS

Type n %

End scrapers Simple end scrapers 7 5.3

Simple end scrapers on blades 6 4.6 Carinated end scrapers 3 2.3 Nosed end scrapers 8 6.1 Denticulated end scrapers 13 9.9

Atypical end scrapers 5 3.8 Side scrapers 3 2.3 Denticulates

Simple denticulates 29 22.1 Denticulates forming a bec 12 9.2 Denticulates forming a perfoir 5 3.8

Denticulate, backed 1 0.8 Notched pieces 8 6.1 Retouched flakes 25 19.1 Burins 3 2.3 Pieces esquillees 2 1.5 Arrowheads (FN type) 1 0.8

Total 131 100

and probably of the Aurignacian in Greece as well.23 The eastern Mediter- ranean Early Upper Palaeolithic (EUP) assemblages from the Balkans eastward to Turkey24 and the Near East are similar in this respect.25 Another feature of the knapping techniques employed at Spilaion is the retouching of core fragments to manufacture end scrapers and denticulates, often on the same blank.

A total of 131 retouched tools (4.1% of the sample) were identified and classified into sixteen types (Figs. 4.6-4.11; Table 4.2). Denticulated and notched pieces (42%) and end scrapers (32%) dominate the assemblage (see, e.g., Figs. 4.6, 4.7). Simple flakes with irregularly retouched margins (e.g., Fig. 4.10:6) comprise the next largest group of tools (19%), and there are small numbers of side scrapers (e.g., Fig. 4.7:6), burins, and pieces esquillees, as well as an arrowhead of a later (Final Neolithic or Bronze Age) date. The end scrapers include simple, carinated, nosed, and denticulated types (Figs. 4.10:3-5, 8 and 4.11). The majority were made on plain flakes (85.7%) ranging in size from 22 to 70 mm in length and from 6 to 30.5 mm in thickness. Only six end scrapers were made on blades. The nosed end scrapers were made on both thick and thin flakes.

Notched pieces and denticulates (e.g., Fig. 4.10:7) were made on flakes of all sizes, sometimes by the Clactonian technique (i.e., the notches were created by the removal of a single flake) and sometimes by retouch. The denticulates were often placed on blanks with other tools, such as becs or

perfoirs. Large primary cortical flakes appear to have been preferred blanks for denticulates. There are two carinated burins on flakes. There is also one small fragment of a bifacial foliate.

23. Darlas 1989, p. 157; Koumouzelis et al. 1996; Perles 1987.

24. Kozlowski 1992, 1999; Kuhn, Stiner, and Gule9 1999.

25. Clark 1994; Olszewski and Dibble 1994.

I44


1 2

J

3 4

Figure 4.6. Lithic artifacts from Spilaion: 1) raclette; 2-3) denticulates; 4-6) end scrapers. All artifacts are flint. Scale 1:2

Figure 4.7. Lithic artifacts from

Spilaion: 1-5) end scrapers (2 and 4 are on blades); 6) convex side scraper on a primary cortical flake. All artifacts are flint. Scale 1:2

4

5 6

1 ' 2 1 2

U

5

6

I45

-: i


Figure 4.8 (left). Lithic artifacts from Spilaion: 1) large blade with an end

-:::':~~~ ] --xtoM ^-' /''''* scraper and abrupt lateral retouch;

2-3) end scrapers. All artifacts are flint. Scale 1:2

2 Figure 4.9 (right). Lithic artifacts from Spilaion: 1-2) end scrapers;

3)perfoir; 4) fragment of a blade core; 5-6) notched pieces; 7) typical flake. All artifacts are flint. Scale 1:2

4

2

Figure 4.10 (below, left). Lithic

?^ j )< ^6 artifacts from Spilaion: 1-2) notched and pointed pieces (becs/perfoirs); 3-

5, 8) end scrapers; 6) retouched flake with small end scraper; 7) denticu-

3 late. All artifacts are flint. Scale 1:2 7

Figure 4.11 (below, right). End

scrapers from Spilaion. All artifacts are flint. Scale 1:2

2

4 3

61~~ ~ ~ ~~~~2 3

6 7 4 5

I46

f

I


26. Kirkby and Kirkby 1976. 27. Rick 1976.

GIS MODELING AND THE INTERPRETATION OF SURFACE SITES

The special taphonomic conditions pertaining to extensive open-air sites such as Spilaion shaped the objectives of our spatial analysis. The processes that affect open-air sites are quite different from those associated with enclosed habitation sites (e.g., caves and rockshelters). In addition, the absence of stratified deposits at Spilaion prohibited three-dimensional

analysis and limited the investigation to issues of horizontal patterning and site-formation.

One goal of our study was to determine if the lithics were distributed over the surface by natural forces or cultural activities. We considered three

hypotheses of site formation that may have played a role in shaping the site. Two of these hypotheses focus on cultural activities and the third

presumes natural processes. We began the analysis by considering the possibility that artifacts were deposited directly on the surface where they are found today (i.e., directly on the bedrock) as the result of short-term cultural activities unassociated with the deposition of sediments. The cultural activities may include flintknapping (e.g., core testing and flaking) or the

secondary disposal of artifacts that have been removed from their use locations (i.e., materials that were dumped). As an alternative hypothesis, we considered the possibility that human occupation of Spilaion over a considerable period of time resulted in the accumulation of sediments that formed the matrix of a stratified deposit. In this hypothesis, the artifacts ended up on the bedrock after the sediments containing them were removed by erosion or deflation (the "concentration effect").26 In these two hypotheses we assume that cultural patterning is preserved in the associations of artifacts of different type, and that size or weight has relatively little part in shaping the distribution. Thus, associations between different artifact classes (e.g., cores and flakes, flakes and tools) will be statistically significant despite differences in the dimensions and masses of the artifacts in question. In other words, if cultural processes are major factors shaping the spatial distribution, artifacts of different sizes and weights will have no statistically significant association with the slope of the site as they would if some natural force such as downslope movement was at work.

A third hypothesis for the artifact distribution at Spilaion holds natural forces rather than cultural activities as being chiefly responsible for shaping the lithic scatter. According to this hypothesis, artifacts would have been transported from another location by a physical process such as erosion or downslope movement and redeposited on the surface where they are found today. If this hypothesis is correct, surface artifacts should cluster on the basis of their size and weight (and, perhaps to a lesser degree, their shape) rather than by their type. The key factor in shaping the distribution is the gradient of the slope. In cases of steep gradients with low vegetation cover, artifacts are expected to scatter in a predictable manner, with heavier artifacts such as cores working their way downslope leaving smaller artifacts behind.27 We assume that the association, or lack thereof, between the frequency and size of artifacts and the topography of

I47


the site will be decisive for evaluating this hypothesis. Of course, it is probable that both cultural and natural forces are at work on the site thus blur- ring the original pattern, but it is important to determine first if natural forces are the most important agents of accumulation.

In a case where natural forces are determined to be negligible, evidence of horizontal patterning in the artifact distribution may be sought. The need to search for patterning in the debitage and tool types at Spilaion is corroborated by two additional factors: 1) the remarkable typological and technological uniformity of the lithic assemblage and the high degree of consistency in flaking techniques; and 2) the high degree of integrity in the assemblage's material composition (i.e., the dominance of stone tools), pointing to the occurrence of a single cultural activity (i.e., flintknapping), rather than to a mixture of activities on the site.

Emphasis is given here to the relationships between components of succeeding or interrelated stages of the reduction sequence, made possible by the application of the chaine operatoire approach, which assigns lithic specimens to different stages in their life cycle (i.e., production, use, or deposition). We considered the possibility that artifacts may derive from primary deposition of by-products (e.g., core testing and flaking taking place on the site), or the secondary disposal of artifacts that have been removed from their use locations (i.e., materials that were dumped). In the case of primary deposition, meaningful horizontal patterning refers to the degree and nature of association (e.g., overlapping clusters, statistical correlation) of artifact groups linked in the production chain (e.g., cores and cortical flakes linked in the decortication stage, blanks and tools linked in the retouching stage). In the case of secondary deposition, redundant disposal patterns are expected to result in mixed deposits of artifacts in different stages of their use-cycle (e.g., products of the earlier stages, non- retouched flakes, tools, recycled tools, artifacts with postdepositional scarring). Unfortunately, due to the considerable degree of weathering (patination) of the artifacts assigned to the Upper Palaeolithic period, any information on tool recycling, use-wear, or postdepositional scarring that could be used to distinguish groups of artifacts has been lost. Information regarding common morphological characteristics (e.g., color, texture) of flint categories was also lost, preventing the refitting of pieces. Neverthe- less, based on the obvious technological uniformity and integrity of the material, we endeavored to discern any type of meaningful spatial association between classes of artifacts related in the same sequence, despite their differences in dimensions and masses and their relation to the slope of the site.

SPATIAL ANALYSIS OF THE HIGH-DENSITY ARTIFACT

DISTRIBUTION

To test the three hypotheses, we undertook a spatial analysis of the Spilaion lithics that comprised a range of data sets and analytical techniques. Be- fore discussing each technique in detail, we provide a few notes regarding the variables and the units of analysis chosen. Because of the technological uniformity exhibited by the lithic assemblage and the lack of a stratigraphic

I48


dimension, the whole surveyed grid surface (60 x 50 m) was treated as a single unit with grid cells of 5 x 5 m. In addition, data summary plots were calculated on the basis of counts of artifacts per grid cell. Finally, artifact size was calculated as a product of length and width.

DOWNSLOPE MOVEMENT

In order to discern the role of natural agents in shaping the artifact distribution at Spilaion, particularly in detecting downslope movement of larger pieces, we used techniques developed for use in spatial mapping/GIS analysis. These techniques offer the possibility of manipulating the different variables to detect significant associations in spatial patterns. As a foundation for this analysis, topographical data (e.g., slope and configuration of bedrock) were entered and processed in our GIS database, GRASS.

As the first step in our analysis, we attempted to evaluate the effects of visibility (rated as either 25, 50, 75, or 100%) on the accuracy of our artifact counts. Calculations based on artifact counts and visibility measurements from all cells pointed to a moderate positive correlation (0.430) between these two variables, indicating that vegetation and soil cover may have affected visibility somewhat in the sample units at Spilaion.

A test of the strength of the correlation between artifact size and the gradient of the slope showed a small negative correlation (-0.173), indicating a very weak tendency for artifacts to be sorted by size. The largest artifacts were not concentrated at the foot of the hill, as one would expect if there was significant downslope sorting by gravity. A subsequent analysis, however, revealed a slightly positive correlation (0.229) between slope and the total number of lithics, suggesting that the gradient of the slope did play a limited role in the distribution of artifacts at the site. This observation is confirmed by spatial mapping, which revealed that clusters of plain and cortical flakes tend to concentrate at the southern end of the grid (see below).

The results of these analyses seemed to indicate that, while the gradient of the slope was a factor in shaping the lithic distribution at Spilaion, it was a small factor and did not fully explain the distribution of artifacts across the site. The low gradient of the slope, which did not exceed the "angle of repose" for large artifacts, and the distribution of artifacts of different dimensions and masses across the slope suggest that the materials were not carried by erosion from some higher, more distant source. Evi- dently, the gentle slope of the site, the irregular and highly weathered karst surface, and the thin covering of scrub vegetation prevented the continuous shifting of artifacts by erosion or downslope creep.

SPATIAL CORRELATION

Spatial correlation between different classes of artifacts, irrespective of their relative masses, followed. The classification of artifacts into meaningful groups is a necessary prerequisite of any correlation study, and artifact classes were defined on the basis of the typology of the presumed reduction sequence (i.e., core decortication, blade and flake production, retouch of blanks to create predetermined tool types).

I49


TABLE 4.3. DEGREE OF ASSOCIATION BETWEEN PAIRS OF CLASSES OF FLINTKNAPPING DEBITAGE

Debitage Classes Correlation Coefficient

cores/blades 0.254 retouched tools/blades 0.312 cortical flakes/blades 0.414 cores/retouched tools 0.456

plain flakes/blades 0.518 cores/cortical flakes 0.578 cortical flakes/retouched tools 0.594

cores/plain flakes 0.642

plain flakes/retouched tools 0.650 cortical flakes/plain flakes 0.836

Comments: there are relatively strong correlations between the spatial patterning of cortical flakes and plain flakes with flakes and tools, but the correlation is weaker between cores and retouched tools or blades, suggesting that these elements were

spatially segregated. The relatively strong correlations between cores and flakes is evidence that the artifacts of different masses have not been sorted by natural

processes such as downslope movement. The weakest correlations are found between the blades and all other classes of debitage except plain flakes, suggesting that the blades were spatially segregated.

Spatial associations between artifact classes were assessed visually from distribution plots produced by the mapping program SURFER (Figs. 4.12, 4.13). This visual analysis was supplemented with a correlation study, the results of which are summarized in Table 4.3.

Interestingly, there is a significant correlation between artifacts that

belong to successive stages of the reduction sequence (e.g., cortical and

plain flakes, plain flakes and flake tools). The most prominent correlation revealed by the study, and confirmed by visual inspection of the distribution plots, is that between cortical and plain flakes (0.836). Overlapping clusters of these two artifact groups can be seen in the southwest corner of the southeast quadrant. No doubt, the overlapping concentration in the partially mapped southwest quadrant is part of the same tendency. Thus, it can be concluded that cortical and plain flakes tend to form overlapping clusters of relatively high concentrations (n = >30 and n = >80, respectively) in the southern part of the grid.

A spatial overlap between cortical/plain flakes and tools was also noted (0.594 and 0.650, respectively). As apparent from the distribution plots, tools tend to overlap with the hot spots of cortical flakes and plain flakes in the southeast and northwest parts of the grid (Fig. 4.12). Numerically speaking, however, these clusters represent relatively low concentrations of tools (n = <6), showing that the latter are rather dispersed across the site.

In contrast, cores show only modest spatial correlation with these categories (i.e., cortical flakes, plain flakes, and tools). Cores tend to form distinct clusters in the northern and the southeastern parts of the grid (Fig. 4.12). Only in the southeastern portion of the grid do cores, and in

particular cortical cores, overlap with cortical and plain flakes. Although this is the least intensively surveyed area of the grid, the overlapping cluster of cores observed here is spatially segregated, and does not continue

I50

'5'

0 '. 0o "t o, r\. _> _

*r . 0_ j, o

a wl, .- : ,mWTV,,.... .- - 7 . '

v\.- '.

w1 w i~, _., _- 0

L~ ~ 4

en

_

rli II

_

.._

,,,...

,-a

"l

F... ..

C)

,...,

_G _

,-

E'

II C _e

is? ;

:?'

_,

E _r

wN rs

4J

--~_.. '--~U :~:.d

4 , M 4 4 4

-I H1UO 0

j ms - n I

I I f11 * ,

f 48 i Pf do - 0

A U^ i I

*144808o

cC 0

H

0

C) U

0 a

I t

0

Ct

U Ct

0 U

OCt U

o

* - U * --';s 0- 021

I-

1K ot * - 4-' ,flcC

q U U

'4-4

o *

r U,

U

cC0 4U

- U

,-.

,-1.

bl- I I

fL4

k,.;

:-

,...._

^

--

1-

^3

< *

- w

cr,

1-

:ri

I_I

ac

~rj C:

I-- I.- -.1.

If

bca

U4 C)

i1a .

',r'

._

EN1) SCRAPERIS (A I.,I, TY1'E1S) (n=361 EN1) SCRAP 'ERS (ON B1I,ADES) (n n=6}

-.., L .

........ . . .

k...~.........

i X

J ; w

.

I) I' NI' C ' I, A T ES (n=47)

&S ml

NMI ii .. ..

0 a -?.^ I , . b .

i f-T-~ I s -7

t tr .. 0 ! ? 5 .1 O ; 1; J y i 40 4 S O

j A 5 : -

a * 1 ' :: ,4 : i A ,.7 i A I AI fT j,i

N) TC I HED) FL AKES (n=8)

s, .

'9- 1 > 0 tj 20 30$ 1

09 f t r 4 ,,

LASTr1^

RETO C IIE !E) FLAKES (n=25)

3>

3,

t

00

4.

IC-0 '

-t tf -- * .... 0 o I 0 4 : J.0 0 I 3S 4*0 J3 40

As r T ;

Figure 4.13. Spatial distribution of individual categories of retouched tools. Each box represents the total area covered by the collection grid. The blank areas were not collected. Densities are indicated by the colors shown in the keys.

tn ki

> * \ r f L A

I r* r

z,

.'!-

;>.1

A.,

.1

1-

" ! t

- . t T -

9

1 1 .


toward the east. Lastly, blades comprise a separate category, since they are limited in number. Blades are rather more dispersed, although they tend to overlap spatially with blade cores.

Characterization of these concentrations as primary or secondary is the most difficult task. As already noted, recording of the state of preservation of artifacts sharing comparable technological characteristics (i.e., Palaeolithic) was impossible at Spilaion due to the obliteration by the patina of all vestiges of use-wear, recycling, and resharpening. In cases of secondary deposition, one would expect cortical flakes, products of an earlier phase, to be found mixed with tools, the final products of the sequence. Although this pattern is discerned at Spilaion, it may be explained by the opportunistic character of blank production and the selection of any type of blank (cortical flakes, exhausted cores, etc.) for retouch. There are no distinctive or clear-cut stages of reduction to be found at this site; reduction evidently proceeded in intermingled stages, succeeding each other on the basis of immediate needs and material restrictions.

If the lithic scatter at Spilaion is a primary deposit, the spatial clustering between cortical and plain flakes may indicate that core decortication and flake production were never two distinct stages of the sequence at this site. Given the rather incomplete decortication of minimally prepared cores, these two steps were interrelated parts of a continuous phase comprising partial cleaning of the core and immediate blank detachment (i.e., flakes were detached from the cleaned surface immediately after its decortication). The correlation between cortical/plain flakes and tools is better understood if we take into account the fact that most of the retouched tools were made on cortical and plain flakes. In contrast, the dispersal of cores and their grouping in distinct clusters suggest that they were transferred and flaked at any spot.

Turning to the distribution plots of single categories of retouched tools (Fig. 4.13), our aim was to detect any associations between individual tool types and their blanks (e.g., flakes and cortical flakes). Of major interest is the correlation of end scrapers with the hot spots of cortical and plain flakes, since the latter comprise the blanks from which scrapers were made. Another significant association is that between plain flakes and retouched flakes. All other tool types (i.e., end scrapers on blades, denticulates, notched flakes) tend to form partially overlapping clusters in the center of the northern grid. Generally, tool concentrations do not consist of large numbers of artifacts, but there are some significant correlation patterns arising between some tool types and the debitage categories on which they were formed.

In sum, our analysis indicates that natural processes were not the most significant factors in shaping the spatial patterning at Spilaion. Although natural processes are often significant in cases of dense open-air distributions of lithics, such as at Spilaion, our analysis shows no significant correlations between artifact size and slope, a relationship necessarily present if significant disturbance by natural processes had taken place. The only indication of natural processes shaping the distribution is the slight positive relationship found between the number of artifacts and the slope. The rather strong tendency of cortical and plain flakes to cluster at the southern side of the grid may be partially the result of downslope movement.

I53


Spatial mapping and correlation analysis yielded comparable results as to the degree of spatial association of different debitage categories linked in an operational chain (e.g., the reduction sequence of blank production and toolmaking). The strongest associations produced by both analyses are between 1) cortical and plain flakes; 2) cortical/plain flakes and some tool groups made on these blanks (i.e., end scrapers, retouched flakes); and 3) blades and blade cores.

The small size of the sample (ca. 2.1% of the total lithics on the site) and the considerable degree of patina on the majority of the artifacts dis- courage us from drawing more detailed conclusions from the shape of the artifact distribution. The most difficult problem posed by the analysis is whether the large number of artifacts at Spilaion were the result of primary or secondary deposition. If these associations are the result of primary deposition, they highlight the opportunistic character of flintknapping at Spilaion. The primary aim of flake detachment on this site was to create immediate blanks for tool production. Thus, there were no clear-cut stages of production, as cores could be partly decorticated and reused at a different spot for flake detachment. In spatial terms, this resulted in the dispersal of flake cores in all areas and the formation of overlapping clusters with debitage types linked to succeeding stages of production.

It is also difficult to determine the duration of the activity required to accumulate the large number of artifacts, whether these activities occurred over a long period of time or consisted of a few, short intensive episodes of flintknapping. We were equally unsuccessful in determining if stratified sediments once existed at Spilaion, the removal of which would have concentrated the artifacts on the bedrock. Yet, our technological study suggests that the site was utilized primarily in one period, the Early Upper Palaeolithic, a conclusion supported by the uniformity of types, materials, and techniques.

CONCLUSIONS

Spilaion is a high-density scatter of lithics with prodigious quantities of flintknapping debitage organized in discrete activity areas, presumably in culturally determined spatial associations. The artifact typology points to the Early Upper Palaeolithic (Aurignacian) as the main period of use of the site, and the "hotspots" may thus be as much as 30,000 years old or even more. The site was evidently not used extensively in other periods. Scattered and highly eroded artifacts of Middle Palaeolithic, Neolithic, and Bronze Age type account for less than one percent of the total sample, and can be discounted in the analysis.

The Spilaion assemblage is classified as Aurignacian on the basis of tool typology and flintknapping technology. The rarity of typical Aurig- nacian retouched blades and the absence of Dufour bladelets and microretouched points, types typical of the Italian and west European Aurig- nacian, are notable features of the Typical Balkan Aurignacian,28 but otherwise the assemblage conforms to the general pattern of Aurignacian assemblages in Greece.29

28. Kozlowski 1999, p. 106. 29. Darlas 1989; Koumouzelis et al.

1996; Kozlowski (pers. comm.); Perles 1987.

I54


30. Zilhao and D'Errico 1999, p. 43. 31. Kozlowski 1999. 32. Kozlowski 1999. 33. Zilhao and D'Errico 1999, p. 43. 34. Darlas 1989; Koumouzelis et al.

1996; Perles 1987; Runnels 1988, 1995. 35. Perles 1987, phase lithique I. 36. Perles 1987, p. 96. 37. Koumouzelis et al. 1996. 38. Darlas 1989. 39. Runnels 1988; Runnels and van

Andel 1993a. 40. Allsworth-Jones 1986; Runnels

1995. 41. Runnels 1988; Runnels and van

Andel 1993a. 42. Koumouzelis et al. 1996;

Kozlowski 1999, p. 114. 43. Kozlowski 1999, p. 108. 44. Kozlowski 1999; Kuhn, Stiner,

and Giile9 1999; Olszewski and Dibble 1994, p. 70.

An attempt has been made recently to deny that the Early Upper Palaeolithic of the Balkans (termed "Bachokirian" and found at Bacho Kiro and Temnata Caves) is in fact Aurignacian,30 but this view is not accepted by those most familiar with the assemblages in question.31 The issue is partly one of nomenclature. The Bachokirian is unrelated to the underlying Middle Palaeolithic industries at Bacho Kiro and Temnata and is un-

doubtedly Early Upper Palaeolithic in character,32 but Zilhao and D'Errico wish to reserve the use of the label Aurignacian strictly for those EUP industries having Dufour bladelets, numerous burins, and bone and ivory points.33 By their definition the Spilaion assemblage is not Aurignacian but Bachokirian. We believe that this distinction does not help to clarify matters and serves only to confuse the reader. For the present, we shall continue to refer to the EUP industry in Greece as Aurignacian.

Aurignacian sites similar to Spilaion are rare in Greece. Surface sites are found in Elis and Thessaly, and the cave sites of Kephalari, Kleisoura, and Franchthi in the Argolid also contain Aurignacian materials.34 The lithics from the earliest Upper Palaeolithic layer at Franchthi Cave35 exhibit typological traits of the Aurignacian (carinated and nosed end scrapers) but they were found in extremely small numbers and therefore cannot be taken as certainly Aurignacian.36 Finds from a rockshelter in the Klei- soura Gorge near Argos exhibit a similar preference for end scrapers on flakes and short blades.37 The surface sites in Elis38 and Thessaly39 produced industries of mixed character, combining Mousterian and Aurig- nacian elements (e.g., carinated and nosed end scrapers, marginally retouched blades and burins, along with Levallois flakes and Mousterian side scrapers), and similar Middle Palaeolithic or Early Upper Palaeolithic industries with this mixed character are known in the Balkans.40

The age of the Greek Aurignacian has not been precisely determined. It was apparently present at sites exposed in the banks of the Peneios River in Thessaly between 45 and 30 kyr B.P., as determined by radiometric dates.41 The recently excavated Kleisoura shelter has a rather late Aurignacian, dated to ca. 34-22 kyr B.P. (uncalibrated).42 We cannot say where in this long period Spilaion is to be placed, and can only give a rough chronological range of ca. 45-22 kyr B.P. for the cultural activity at the site. Outside of Greece, the Spilaion assemblage can be compared with the assemblages from Bacho Kiro (layers 9-11) and Temnata Cave (layers 3-4) in Bulgaria, where the Aurignacian layers have been dated from 45 to 28 kyr B.P.

(uncalibrated).43 The Spilaion assemblage is thus similar to the Aurignacian and other EUP assemblages of the eastern Mediterranean sensu lato.44

If we are correct in assigning the majority of the Spilaion lithics to the Early Upper Palaeolithic, this one site has more than 250 times as many artifacts as are found on the other EUP sites in Greece. Thus Spilaion is perhaps the largest lithic site in Greece. It is extraordinary even by local Epirote standards. The entire lithic collection from the rest of the Nikopolis survey, which is based on the total collection of lithics from all tracts, is less than 15,000 pieces. The largest Middle Palaeolithic sites in the Preveza region (e.g., Kokkinopilos), which are certainly among the richest lithic sites in the country, have less than one-tenth the number of lithics visible on the surface at Spilaion. The size and preservation of the EUP lithic

I55


scatter at Spilaion, therefore, presents a rare but important opportunity to study a site of this period, despite the complete absence of stratified

deposits. Artifact-rich surface sites are common in Greece, but there has been

some debate about their value for archaeology.45 We acknowledge that such sites cannot be studied by means of traditional excavation techniques, but we believe that the study of spatial patterning permits archaeologists to make greater use of them. If the quantities of artifacts preserved are large enough, spatial analyses can be useful in interpreting past cultural activity, even where stratigraphic associations have been lost or were never present. The number of these artifact-rich sites has increased greatly as the result of intensive surface reconnaissance on a regional scale. Such sites are not

exclusively prehistoric or marked only by scatters of lithics. We believe that the methods detailed in this report can be applied successfully to historical sites and to sites with rich concentrations of sherds, rooftiles, and other cultural materials. The identification of patterns in the artifact dis- 45. This debate is summarized in tribution at Spilaion should serve as an incentive for the continued study Cherry et al. 1988 and Alcock, Cherry, of surface sites in Greece and throughout the Mediterranean. and Davis 1994.

I56

CHAPTER 5

THE COASTAL EVOLUTION OF

THE AMBRACIAN EMBAYMENT

AND ITS RELATIONSHIP TO

ARCHAEOLOGICAL SETTINGS

by Zhichun Jing and George (Rip) Rapp

Coastal landscapes are a sensitive interface for environmental change. In the past 10,000 years, the Ambracian embayment and its vicinity have witnessed dramatic landscape changes in response to Holocene eustatic sea-level rise, sediment infill, erosion, and tectonic movement (Fig. 5.1). The changing landscape in this area, utilized since the Lower Palaeolithic

period,1 has affected both the spatial and temporal distribution of archaeological remains. Thus, the pattern of prehistoric and historical settlement must be understood in the context of the evolving coastal landscape. Paleoenvironmental reconstruction associated with archaeological investigation in Epirus has focused on the Palaeolithic period,2 and no investigation has been conducted to examine the Holocene environmental context of settlements based on the subsurface stratigraphy. A limited number of studies based on geologic or sedimentary perspectives have been undertaken to address the evolution of the coastal landscape during the Holocene. Although these studies revealed sea-level and coastline changes, they provide no essential data for the interpretation of archaeological settings in terms of either the temporal or spatial scales dealt with in the investigation of settlements in the embayment of Ambracia and its vicinity.3

In this chapter, we describe the changing landscape in the Ambracian embayment during the Holocene based on an analysis of the subsurface stratigraphy, and we establish the environmental context of various prehistoric and historical settlements. In order to reveal the subsurface stratigra-

1. Hammond 1967; Dakaris 1971; Runnels and van Andel 1993b; Bailey 1997.

2. Bailey, King, and Sturdy 1993; Dakaris, Higgs, and Hey 1964; King and Bailey 1985; Sturdy and Webley 1988; Sturdy, Webley, and Bailey 1997; Turner and Sanchez-Gofii 1997; Vita- Finzi 1978, pp. 139-158.

3. Piper, Panagos, and Kontopoulos 1982; Piper, Kontopoulos, and Panagos 1988; Poulos, Lykousis, and Collins 1995; Tziavos 1997. Both Poulos, Lykousis, and Collins (1995) and Tziavos (1997) studied the Quaternary subsurface stratigraphy through the analysis of 3.5-kHz seismic reflection profiles across the Ambracian Gulf,

providing some critical information on the formation and development of the basin during the Late Pleistocene and early Holocene. Tziavos (1997) carried out some drilling in the coastal plain north of the gulf aimed at studying the paleogeographic evolution of the basin during the Quaternary period.

ZHICHUN JING AND GEORGE (RIP) RAPP

_ _ - _ _- ,- I P7I I I

A 0 < . ? - ? . . . -

^ v-^ i~~~ ~..r.. .... I. i / Li (7' Nj I-^^ S^'-^~~~~~~~"''i ':1i'I T^ i ... ',;' "

3 '" 4 - . I I ..m:%,'

/, '? ,'/ /

(. ._j .. .....L~ m. ' m

~],, ..&~.m . V' .--V~.~ ',,qm . m-

? .. Acheron R. ',' ~ ...... . , / ... / - ' , . ~ . ..

N 26 T I~~~~~~~~~~~~~~~~~-. v-'...

I

V % 1 ------; - /-! .

'2 ,2' \.' ***' --/ \-^ --/t/. 2) / Stavros L2S n- Rokia

'. ' " " :; '

:.-^

' 2 /

\ i K <a " t i r SI>, , 7,?

N ....

4^\' -<- . ... . ... .: ? **:-^ ;-\ r, , / 'i '- ,. ... ',- ) /21 .. v.... . ;

V Vi < . v\ {- ,

Ion \ii, 'Sea Al; .7: ...Salaora

.the Ambracian Embayment and its Vicinity . ,. - a . .Anbracian

/2___ . 1__ 6, ._ 1.... KM ,.. .7 IV ;/. //d4 / .K V /N , Gulf-....

0T^-I Nlesoioic ',;KJ,, "I rtiarx ["*- IIHoloc' Alluvi." . . '. '

' .. , " ? . / . ,-" { '. :k'15-=[/ '

m.. L . Fvsch ,,'" ', .lisocL.. .- Hoi,cene Swamp \ , \. 4,.--.'

'v":

~1. M a z~o u ~m II. (in r.n I. en o 1. P ..A. T o a V. \l iV. . .A che ' ,~( . . . . . . . ?'?.. .,,.v -/1 co -C " - -

1. Actium 2. Anactoriurn 3. Pantokrator 4. Margarona 5. Nikopolis

6. Comaruis 7. Ormos Vathy 8. Michalisi 9. Archangelos 10. Kastrosykia

I I. Cassope 12. Riza 13. Palaiorophoros 14. Cheimadio 15. Kastro Rogon

16. Strongyli 17. Vigla 18. Kastro Rizovouni 19. Nckyomanteion 20. Koumasaki

21. Kastri 22. Kanalaki 23. Ephyvra 24. IThesproliko Ancient Hlarbor

S

/0' II,

1 6 fly

'.1.*

,; . - _2_ .. ......

/I . : :....

7 7 1 " ' " " ' " "

phy, we took a series of geologic cores using a hand-operated auger. Our

drilling was limited to the western part of the embayment. A total of 35 cores were drilled with a maximum depth of 13 m in different parts of the

lagoon-swamp-coastal delta plain along the Ambracian Gulf (Fig. 5.2). In addition to drilling in areas of geologic importance, we drilled a large number of cores around historical settlement sites to better understand their paleogeographic setting, particularly their relation to shorelines and

possible harbors. These sites include Nikopolis on the Preveza peninsula, the Roman harbor town site surrounding Ormos Vathy on the Ayios

Figure 5.1. Geology and geomorphology of the Ambracian embayment and its vicinity

Figure 5.2 (opposite). Locations of

geologic cores and cross sections. For legend, see Figure 5.1. For unnumbered cores, see Figures 5.3, 5.4, and 5.14.

158

---~~~~~~~~~~~~~~~~~~~~.

COASTAL EVOLUTION OF THE AMBRACIAN EMBAYMENT

Thomas peninsula, and Kastro Rogon on the limestone hill south of Mt. Rokia.4

4. Each core was described and logged in terms of lithology, color, structure, consistency, plant and faunal debris, cultural inclusions, stratigraphic position (depth and thickness), and other observable soil and sediment properties such as the presence of calcium carbonate and pebbles. The terminol-

ogy used follows Folk 1980 for sediments, and Soil Survey Staff 1975 and Birkeland 1999 for soils. Sediment and soil samples from the cores were taken for laboratory analyses, including grain size, microfossils (particularly ostracoda and foraminifera), organic matter, and calcium carbonate content. A total of

seventeen AMS radiocarbon dates were determined on core samples at the University of California, Riverside, and Peking University (see Table 5.1). Six of these dated samples are marsh grass and wood debris assumed to have grown in or near contemporary sea- level positions.

I59


GEOLOGY AND GEOMORPHOLOGY OF THE AMBRACIAN EMBAYMENT

The Ambracian embayment is a major tectonic depression (post-orogenic graben) in southwestern Epirus, situated in the so-called Ionian zone between the Hellenide mountain chain (Pindos Mountains) and the Ionian coast (Fig. 5.1).5 The embayment includes the marine gulf itself and a

low-lying lagoon-coastal delta plain to the north. At present over half of the embayment, including the Ambracian Gulf and three lagoons, is under water. The Ambracian Gulf measures ca. 35 km from east to west and 10 km from north to south. The lagoon-coastal delta plain to the north is ca. 10 km from north to south. The gulf itself is relatively deep with a maximum water depth of ca. 65 m in its southern part. The average tidal

range is only 5 cm, with a maximum recorded range in a single tidal cycle of 25 cm. The gulf floor shows a gentle gradient on the north due to sediment deposition from the Arachthos6 and Louros Rivers, while it drops off quite rapidly (to 40-65 m) on the other sides of the gulf. The gulf is sheltered from wave processes of the Ionian Sea by both the Preveza peninsula and the sandy spit at Actium; thus, secondary waves formed within the gulf create a train of littoral transport that is responsible for the formation of sandy barriers along the north shore. Most of the gulf is floored with rather uniform olive-gray silty sediments composed of 65-75% clay. The gulf is connected to the Ionian Sea by a narrow channel (Preveza Strait; 600 m wide) north of Actium.7

The Ambracian embayment is bounded by bedrock to the east, south, and discontinuously to the west. The southern flank of the gulf is bordered by a sharp cliff incised into Mesozoic limestones. A Tertiary flysch fringes the embayment in the east. To the west and southwest of the embayment is the Preveza peninsula, cut off from the mountainous limestone platform and flysch basin in the north by Mt. Zalongo. The low, hilly peninsula is composed mainly of interbedded mudstone, sandstone, marlstone, and pebbly conglomerate that formed in shallow marine, alluvial delta plain, and fan environments during the Pliocene and Pleistocene.8 In contrast to Mesozoic limestones, these Pliocene and Pleistocene sediments are easily eroded. As a result, the littoral transport of sandy sediments has created a relatively broad sandy beach along the peninsula's Ionian coast.

To the north, the embayment is bordered by a series of bedrock mountains that consist of alternating Mesozoic limestone and Tertiary flysch formations. The limestone ridges have elevations of more than 600-1000 m, while flysch mountains have relatively low elevations (150-600 m) and usually constitute the basins between Mesozoic limestone platforms. These alternating limestone platforms and flysch basins strike N25W; and they disappear underneath the Pliocene to Quaternary sediments in the south along an east-west structural feature represented by several east-west striking limestone mountains including Zalongo, Stavros, and Rokia. Along the Ionian Sea, the Mesozoic limestone ranges have relatively low elevations (500-700 m) and drop abruptly to the sea, forming a steep and nearly harborless coast in the northwest part of the study area. The steep coast is broken only by two small bays: Phanari Bay, into which the Acheron River flows, and a second small bay located slightly to the south (Fig. 5.1).

5. Monopolis and Bruneton 1982. 6. The Arachthos River is east of

the area shown in Figure 5.1; see Figure 6.1 for its location.

7. Piper, Panagos, and Kontopoulos 1982; Piper, Kontopoulos, and Panagos 1988.

8. Doutsos, Kontopoulos, and Frydas 1987.

I6o


9. Piper, Kontopoulos, and Panagos 1988, p. 285.

10. Currently, the Louros River emerges from the deeply incised valley in the north and unexpectedly turns to the west along the foot of Mt. Rokia. The river then turns back toward the south and enters the gulf near Micha- litsi. Aerial photos show a series of abandoned channels in the swampy area to the north of Rodia Lagoon, suggesting that the Louros River might have entered the lagoon in the past. The current flow pattern of the Louros River indicates that there might have been a major channel diversion in antiquity. This channel diversion was most likely made to drain the western part of the low-lying lagoon-coastal plain and expand the farmable area.

11. King, Sturdy, and Whitney 1993.

12. Clews 1989; Etude geologique; King, Sturdy, and Whitney 1993.

13. King, Sturdy, and Whitney 1993; Papazachos and Comminakis 1971.

14. King, Sturdy, and Whitney 1993, p. 157; Pirazzoli et al. 1994; Underhill 1989; Stiros et al. 1994.

The Ambracian Gulf is separated from the northern delta plain by a

swamp, the Salaora barrier, and three lagoons (Rodia,Tsoukalio, and Loga- rou), and from the lowland north of Nikopolis by Mazoma Lagoon. The Salaora barrier is relatively narrow, with a large projection to the north. It is approximately 8 km long, and is believed to have been formed by littoral

transport of sand and gravel sediments eroded from the Preveza peninsula and from those derived from the mouth of the Louros River. The sandy barrier bordering Logarou Lagoon may be associated with the main abandoned channels of the Arachthos River.

Both the Arachthos and Louros Rivers enter the embayment from the north and they have provided the majority of sediments to the Ambra- cian Gulf and its coastal plain. The Arachthos River, with an estimated

average annual discharge of 80 m3/s, is one of the largest rivers draining the high Pindos mountains of Epirus. It has been dammed and regulated since 1980. The Louros River, dammed since 1963, drains the mountains to the west of the Arachthos River. In terms of water discharge, the Louros River is much smaller, with an estimated average annual discharge of 30 m3/s.9 The Arachthos River is responsible for the formation of the eastern portion of the coastal plain, and the Louros River is the primary con- tributor to the development of the western portion. The eastern part of the coastal plain shows a much more developed alluvial morphology than the western part. This circumstance may be attributed to the much larger discharge of the Arachthos in the east. The Louros River and other streams emerging from mountains to the north and west have relatively small dis-

charges. As a result, the three lagoons occur only in this part of the embayment, and much of the area surrounding the lagoons is swampy (Fig. 5.1). Part of the swampy area, particularly north and west of Mavrovouni ridge, has been drained and the land reclaimed for agricultural purposes.

Bathymetric contours along the north shore of the Ambracian Gulf indicate several southward extending protrusions representing prodelta platforms that have developed near the mouths of the rivers (Fig. 5.1). There are two southward extending prodelta platforms in the western part of the gulf. The first one, projecting southeast, is relatively small and is associated with the current mouth of the Louros River. Another, near Salaora, protrudes southward. A series of abandoned channels exists to the east of the current Louros River channel, from Kastro Rogon southward to Tsoukalio Lagoon, suggesting that the Louros River may have flowed southward directly into the Ambracian Gulf after emerging from the mountain valley.10

Tectonically the Ambracian embayment and its surrounding area are a triple junction between the Ionian, Aegean, and European plates, showing a relatively complex pattern of local tectonism.11 The present morphology of the embayment was shaped by Oligocene-Miocene compressional folding and faulting (north-northwest to north-northeast) followed by extensional faulting (west-northwest to east-southeast) during the Late Pliocene and Quaternary.12 Continuing tectonic activity makes this region one of the most active seismic areas in the world.13 The embayment itself has been subject to continuous tectonic subsidence since the Pliocene- Pleistocene, but the Preveza peninsula to the west has been uplifted as indicated by anticlines.14


SUBSURFACE STRATIGRAPHY AND PALEOGEOGRAPHIC RECONSTRUCTION

Our approach to reconstructing coastal landscape change is to determine the vertical and lateral sequences of the subsurface stratigraphy that record

past geographic change from various processes such as eustatic sea-level

change, deposition, erosion, and local tectonism.15 Walther's Law of correlation of sedimentary facies constitutes the conceptual framework for our

analysis of changing coastal environments.16 The single most important component of coastal landscape reconstruction is to determine "paleo-time depositional surfaces."17 The depositional surface may be a deltaic flood-

plain above or near sea level, a coastal barrier above sea level, a lagoonal deposition surface below sea level, or a coastal marsh or swamp near sea level. Using paleo-time depositional surfaces, one can draw the shorelines and the spatial patterns of various lithosomes for different periods in the past. These sedimentary concepts have been successfully applied to the study of coastal environmental change in archaeological contexts in Greece.18

PREVEZA PENINSULA

Nikopolis is located on the middle portion of the Preveza peninsula that separates the Ambracian embayment from the Ionian Sea. Specifically it is located on a low Pliocene-Pleistocene ridge near the southwest coast of Mazoma Lagoon (Fig. 5.1). Between Mazoma Lagoon and the Ionian Sea is a lowland that dissects the Preveza peninsula composed of interbedded mudstone, sandstone, and pebbly conglomerate of the Pliocene to Pleistocene periods. The lowland, here referred to as the Nikopolis isthmus, is ca. 2.3 km long and 250-1000 m wide and is covered with alluvial and slope-wash sediments of the Holocene period. The highest portion, in the middle of the lowland, is ca. 16 m above current sea level (Fig. 5.3).

According to Strabo, there were two harbors near Nikopolis during the Roman period.

Comarus, the nearer and smaller of the two, which forms an isthmus of sixty stadia with the Ambracian Gulf, and Nikopolis ..., and the other, the more distant and larger and better of the two, which is near the mouth of the gulf and is about twelve stadia distant from Nikopolis.19

The smaller harbor, Comarus, is situated on the Ionian Sea (Fig. 5.1). Both Hammond20 and Dakaris21 interpreted Strabo's measurement of 60 stadia (12 km) as the length of the isthmus from the smaller harbor, Comarus, on the Ionian Sea, to Nikopolis, lying on the Ambracian Gulf. However, the length of the isthmus is only ca. 2.3 km (11.5 stadia) at present and it was even shorter during the Roman period due to marine transgression. If the location of the Comarus harbor is correctly identified, Strabo's measurement of 60 stadia must be wrong. Both Hammond22 and Leake23 placed the second harbor at Ormos Vathy, about 12 stadia (2.4 km) from Nikopolis. Ormos Vathy is situated at the junction between the Preveza peninsula and the Ayios Thomas peninsula (Figs. 5.1, 5.4).

15. Kraft and Chrzastowski 1985; Kraft, Kayan, and Aschenbrenner 1985; Rapp and Kraft 1994.

16. With Walther's Law one is able to reconstruct ancient sedimentary landscapes through time and space by establishing the three-dimensional stratigraphic shapes of coastal sedimentary lithosomes or the shapes of sedimentary bodies deposited in discrete coastal sedimentary environments (Middleton 1973).

17. Kraft 1985. 18. Kraft and Aschenbrenner 1977;

Kraft, Aschenbrenner, and Rapp 1977; Kraft et al. 1987; Niemi 1990; Zangger 1991, 1993, 1994.

19. Strab. 7.7.5 (C 324), trans. H. L. Jones, Cambridge, Mass., [1924] 1954.

20. Hammond 1967, p. 48. 21. Dakaris 1971, p. 6. 22. Hammond 1967, p. 48. 23. Leake 1835, I1, pp. 195-196.

I62

COASTAL EVOLUTION OF THE AMBRACIAN EMBAYMENT I63

Figure 5.3. Map of the Nikopolis isthmus showing the location of geologic cores and cross sections

Figure 5.4. Map of Ormos Vathy showing the location of geologic cores and cross section


Archaeological survey conducted by the project in 1993 and 1994 shows

that Roman and Late Antique sites are scattered along the flanks of the ancient bay. Among these the most important is a Roman site, ca. 250 m wide and 900 m long, along the western shore of the bay. It is believed that this site must have served Nikopolis as a harbor town. Leake24 presents a sketch map showing that the bay extended further to the north in the early 19th century than it does today. Geologic cores drilled on the Nikopolis isthmus and in the area north of Ormos Vathy provide the data for deter-

mining the paleogeographic setting of the city and its associated harbor town.

NIKOPOLIS ISTHMUS

On the Nikopolis isthmus, eight cores, with a maximum depth of 13 m, were drilled from the present shore of Mazoma Lagoon to ca. 1.2 km inland (Fig. 5.3).25 Five stratigraphic cross sections based on these cores show the relationships of the marine and alluvial deposits across the Nikop- olis isthmus. Two cross sections are parallel to the axis of the isthmus

(Figs. 5.5, 5.6), and the other three are perpendicular to the axis (Figs. 5.7-5.9).26

Cross section D-D' is based on three cores (92-03, 93-03, and 92-04) and extends ca. 1.2 km from Mazoma Lagoon (Fig. 5.5). The lowest sedi-

mentary unit found along this traverse consists of deposits in a marine

embayment or lagoon. The unit consists mainly of bluish gray (5BG 4/1)27 reduced mud containing marine and brackish gastropoda (Monodonta,

Cyclope), bivalves (Cerastodema), ostracoda (Loxoconcha, Cyprideis, Basslerites,

Leptocythere), and foraminifera (Ammonia beccarii, Elphidium, Trocham- mina). The marine or paralic unit is covered by olive (5Y 5/6, 5Y 4/3) and

light olive-brown (2.5Y 5/4,2.5Y 5/6) silt, sandy silt, mud, and sandy mud of alluvial or slope-wash origin. Their contact is sharp and not transitional. Inland (in core 92-04) the lower boundary of the alluvial unit lies at a depth of 7.0 m while near the shore (in core 93-03) it is found at a depth of ca. 4.85 m, showing an increasing thickness away from the shore. The

overlying alluvial or slope-wash deposits are separated into two parts by a

relatively well-developed paleosol, which is characterized by its olive (5Y 5/4) and olive-yellow (5Y 6/6) colors, a carbonate-enriched layer (Bk), crumb and blocky structure, and carbonate-enriched remains. Paleosols

represent periods of landscape stability because they only form on stable surfaces over a relatively long span of time.28 A few ceramic fragments of red color were found in the top part of the paleosol in core 93-03. A very thin (0.35 m) massive dark gray (2.5Y 4/1 to 5Y 4/1) mud is found in core 93-03 overlying the paleosol, most likely deposited in a mudflat environment as seen near the shore today.

Cross section E-E' (Fig. 5.6) is located northeast of cross section D- D'. It exhibits generally the same vertical stratigraphic sequence but with some variation. The top alluvial or slope-wash sediments are superimposed on a bluish gray mud deposited in a lagoon or marine embayment environment. In the alluvial or slope-wash deposit there is a buried paleosol as

24. Leake 1835, I, p. 187. 25. Two additional cores, C92-05

and C92-06, were attempted on the sandy barrier that separates Mazoma Lagoon from the Ambracian Gulf, but they penetrated only 1.35 and 1.75 m, respectively, due to the very loose nature of the barrier sand.

26. Elevations are based on the 1:5,000 Greek Army topographic maps and measurements using an electronic total station.

27. The color index is based on the Munsell Soil Color Chart.

28. Birkeland 1999.

I64

200 400 600 800

sand-"~~ "' s ii A carbonate . sand I ̂ ? shell ? nodule

l.?:S gravel ceramic plant " ?- I 11~0 fragments remains

1000 1200 1400 1600 m

soil 4 bivalve a foraminifera

radiocarbonte gastropod I d I ostracod

Figure 5.5. Stratigraphic cross section D-D', parallel to the axis of the Nikopolis isthmus. For core locations, see Figure 5.3.

m

12

10

8

6

4

2

0

-2

-4

-6

-8

-10

0

clay

t - - I mud

On

m 93-04

12 - : '

:-'-4-'-. t Tectonic Uplift SE

10

0 x - .- ...'' .0. 4.'0 -0 0 10 10 1 1

8i? . a'd

|u ,- d L 0 " allu'. , ' \

? * _ _ O_ _ % 'x * -. 'x " * * * _ _

- - - .. .~ ~..x.. 93-0

// \ / // \- \/ -was -\ \ \ \ \ \ "" ~- "02 - '.

2 1 \ N- ̂^^ ^ ~ d \ _ la^nr^^^^^ ^T^^^flat^^90 Mazoma

onI/ / ////. ///\/,////baym/ /t\ \ ." ':' " - -0 - 2 o Lagoon'emaen ////^ ^ // 7 \ . ' .. ~ -~ ~ - ... -..- ̂ .alluvium

I |'// ////X4-/////////////..- ' ' X, - slope/ wsh ^ ;

-10 - 2 9 3

0 200 400 600 800 1000 1200 1400 1600 m 0 200 400 -600 800 1000 ~~~~~1200 1400 1600m

Figure 5.6. Stratigraphic cross section E-E', parallel to the axis of the Nikopolis isthmus. For core locations, see Figure 5.3; for legend, see Figure 5.5.

O\ 0',


described in cross section D-D'. The upper boundary of the marine or

paralic unit lies at a depth of 9.1 m in core 93-04,5.3 m in core 92-02, and 3.8 m in core 93-01. The marine or paralic unit is more than 6.2 m thick in core 93-01, and ca. 1.5-2.0 m thick in both cores 92-02 and 93-04.

In core 93-04 the marine or paralic unit overlies a layer of black (N2/) and very dark gray (N3/) silt and mud that is very firm and contains many black nodules (possibly manganese oxide). The black or very dark gray silt and mud, 0.8 m thick, is underlain by a very firm olive-gray (5Y 4/2) mud containing common calcium carbonate nodules (0.1-1.5 cm in diameter). Both of these firm deposits are believed to represent a Pleis- tocene paleosol. The top black and very dark gray part is an A horizon, and the lower olive-gray part is a B horizon.

In contrast, the marine or paralic unit in core 92-02 rests on an olive (5Y 5/4 or 2.5Y 6/4) silt and sandy mud containing many calcium carbonate nodules, some pebbles, and some terrestrial gastropoda (snails). This olive layer shows a moderately developed blocky structure but it is much more friable than the bottom sediments seen in core 93-04. Based on its color, structure, inclusions, and consistency, the low-lying olive silt is interpreted as a paleosol developed on slope-wash or alluvial sediments of the early or middle Holocene.

Four radiocarbon dates were obtained from the marine or paralic unit (Table 5.1). Two radiocarbon dates are on samples from core 92-03, located on the landward shore of Mazoma Lagoon due east of Nikopolis. The whole column of the core consists of interbedded very dark gray (5Y 3/1 or N3/) and bluish gray (5BG 4/1) sandy mud, gravelly, muddy sand, and mud containing abundant marine and brackish mollusks (gastropoda and bivalves) and microfauna (ostracoda and foraminifera). The layer at a depth of 6.0-7.55 m yielded many cultural remains including charcoal, bone, burned wood, ceramic sherds, and plaster fragments. Among these remains are some diagnostic Early Roman sherds from a depth of 7.1-7.3 m. A charcoal sample from 6.0 to 6.2 m (5.9 m below modern sea level) gives a calibrated radiocarbon date of 650-440 B.P., much younger than the diagnostic sherds from the layer, which suggests that the culturally altered layer is a secondary rather than a primary deposit. Given the geomorphic position of the core, the cultural remains might have been washed away from the hill slope at Nikopolis. A relatively old date of 5320-4840 B.P. was obtained from a shell sample from a depth of 5.8-5.95 m (5.7 m below modern sea level). Inversion of younger and older radiocarbon dates may be attributed to the redeposition of shells derived from preexisting sediments. The top 6 m of core 92-03 must have formed in the past 650 years. Thus the average deposition rate for the past 1,000 years is greater than 9 mm/year near the present shore of Mazoma Lagoon. Such a large sedimentation rate may be attributed to the rapid tectonic subsidence of the eastern side of the isthmus.

A calibrated radiocarbon date of 3350-2930 B.P. comes from a charcoal sample found at a depth of 5.3-5.4 m (2.4 m below modern sea level) in core 92-02. The sample is from the top of the marine unit. This date indicates that the shore of the lagoon or shallow marine embayment was beyond the location of core 92-02, 550 m inland from the present shore,

i67


TABLE 5.1. RADIOCARBON DATES FROM THE AMBRACIAN EMBAYMENT

Depth below Conventional 28 Max. Calibrated Age MSL Sedimentary Radiocarbon (CaL Age Intercepts)

Lab No.a Core (m) Material Facies Age (B.P) Min. CalibratedAge (B.P.)b

UCR-3201 NC92-02 2.4 charcoal estuary (mud basin) 3430 ? 70 3350 (3160) 2930* UCR-3202 NC92-03 5.7 shell estuary (mud basin) 4940 ? 80 5320 (5070) 4840 *

UCR-3219 NC92-03 5.9 charcoal estuary (mud basin) 1060 ? 70 650 (530) 440* UCR-3218 NC92-04 +0.8 charcoal estuary (mud basin) 6430 ? 70 6950 (6750) 6600* UCR-3220 NC92-07 6.6 charred roots estuary (mud basin) 4290 ? 70 4440 (4240) 4010 *

UCR-3221 NC92-08 3.7 charred roots estuary (mud basin) 4010 ? 70 4080 (3850) 3640 *

UCR-3203 NC92-09 4.2 charcoal estuary (mud basin) 970 + 60 560 (490) 330 *

UCR-3204 NC92-09 6.3 charcoal estuary (mud basin) 840 ? 60 490 (370) 260 *

UCR-3205 NC92-09 7.1 wood estuary (mud basin) 880 ? 70 520 (420) 270 *

UCR-3206 NC92-09 7.5 wood estuary (mud basin) 1250 ? 80 880 (670) 540* UCR-3222 NC92-10 4.5 charred roots estuary (mud basin) 1600 ? 70 1210 (1030) 890 * UCR-2691 NC93-07 1.0 peat swamp (marsh) 4810?60 5650 (5590) 5330 UCR-2692 NC93-09 3.6 peat swamp (marsh) 1670 ?60 1710 (1550) 1410 UCR-2693 NC93-11 5.8 peat swamp (marsh) 5900?70 6890 (6720, 6700, 6690) 6500 UCR-2694 NC93-16 4.4 peat swamp (marsh) 2510 ? 70 2760 (2710,2630,2620,2560,2550) 2350 BK-94168 NC94-19 5.1 peat swamp (marsh) 4090 ? 80 4830 (4570,4560,4550, 4540, 4530) 4410 BK-94169 NC94-19 1.2 peat swamp (marsh) 1510? 80 1560 (1410, 1400, 1390) 1290

aDating laboratory: UCR = Radiocarbon Laboratory, University of California, Riverside; BK = Radiocarbon Dating Laboratory, Peking University.

bCalibrated ages obtained using CALIB 4.3, developed by Quaternary Isotope Laboratory, University of Washington. *Marine reservoir correction made for the estuary samples, using AR = 118 + 35, a regional average for the eastern Mediterranean

(Siani et al. 2000).

around 3000 B.P. Further inland, in core 92-04, a date of 6950-6600 B.P.

was obtained from a charcoal sample at a depth of 7.15-7.25 m (0.8 m above modern sea level)-near the top of the marine or paralic unit in which some microfauna (Loxoconcha, Elphidium) were found in core 93- 04. This circumstance suggests that the area of both cores 92-04 and 93- 04 was then covered by seawater.

From cross sections D-D' and E-E' we can see that the upper boundary of the marine or paralic unit is sloping upward away from the shore. The upper boundary in both cores 92-04 and 93-04 is ca. 1-2.5 m higher than modern mean sea level. The radiocarbon date from the top of the unit in inland core 92-04 is much older than that in core 92-02; moreover, a much younger date is situated at a much lower elevation in core 92-03 on the shore. Based on the radiocarbon dates on peat samples from buried coastal marsh and swamp in the coastal delta plain north of the Ambracian Gulf, we know that relative sea level was much lower prior to 6,000 or 7,000 years ago (see below). Therefore the relatively high elevation of the marine or paralic unit in cores 92-04 and 93-04 does not indicate that sea level was higher during the formation of the unit than exists today. A reasonable explanation would be that the marine or paralic deposits, formed in the Nikopolis isthmus during the early phase of marine transgression, were elevated by tectonic uplift. As mentioned earlier, the Preveza peninsula- in contrast to the subsiding Ambracian embayment-has been sub- 29. King, Sturdy, and Whitney jected to tectonic uplift since the Pleistocene.29 1993.

I68


Cross sections A-A', B-B', and C-C' (Figs. 5.7-5.9) are perpendicular to the axis of the Nikopolis isthmus and show the variation in stratigraphic units from the current shore to 1.2 km inland. The top alluvial and slope-wash sediments increase in thickness away from the shore. Cross section B-B', located ca. 500-550 m east of Mazoma Lagoon, is based on cores 92-02 and 93-03, only 150 m apart. Both cores show the same upper stratigraphy based on alluvial and slope-wash sediments interbedded with a 0.3-0.5 m thick mudflat mud, but they have completely different sediment assemblages in the lower stratigraphy. In core 93-03 the alluvial and slope-wash unit is underlain by more than 7 m of marine or paralic sediments, while in core 92-02 the underlying marine or paralic unit is only 1.55 m thick and lies on a paleosol developed on relatively early alluvial or slope-wash sediments. This stratigraphic pattern is not seen in cross sections A-A' and C-C'. At the northern end of cross section C-C' (in core 93-04) the marine or paralic unit rests on a Pleistocene deposit.

The radiocarbon date from the top of the marine or paralic unit in core 92-02 is 3350-2930 B.P. Based on stratigraphic evidence and radiocarbon dates, the low-lying slope-wash deposit may represent a fan that protruded from the hill slope in the north sometime in the early or middle Holocene but before 3,000 years ago. This fan would have dammed part of the lagoon or shallow marine embayment. The fan was subject to marine transgression only for a short period, probably between 3000 and 3500 B.P. The deepest part of this earlier lagoon or marine embayment probably lay on the south side of the isthmus, as indicated in cross sections B-B' and C-C'. Even the modern contour lines project further inland along the south side of the isthmus.

From the subsurface stratigraphy of the Nikopolis isthmus, a paleogeographic map can be made showing the changing shoreline over the past several thousand years (Fig. 5.10). The morphology of the isthmus might have been created by tectonic faulting during the Pleistocene or Pliocene period. In the middle Holocene, the eastern portion of the isthmus was submerged below a large embayment that extended much further inland than the present Mazoma Lagoon. The furthest inland core yielding Holocene marine or paralic sediments is core 93-04, 1.2 km west of Mazoma Lagoon, in the middle of the 2.3-km long isthmus. Here the isthmus has an elevation of 12 m, and the highest portion has an elevation of only 16 m. If the tectonic uplift of the Preveza peninsula has been rapid, the isthmus might have been low enough to be an open channel between the Ionian Sea and the Ambracian Gulf during the early Holocene (before 6500 or 7000 B.P.) . When marine regression occurred, the shoreline of the embayment on the Ambracian Gulf side prograded eastward. By 3000 B.P.

the shoreline was ca. 1 km inland of the current shore, and by 500 B.P. it was likely less than 400 m inland. The date of formation of the sandy barrier on the east side of Mazoma Lagoon remains unknown, as we were unable to core along its length. On analogy with the Salaora barrier, for which we do have some evidence (see below), the Mazoma barrier may not have formed completely until 1000 B.P. or later. The barrier at Mazoma is believed to have formed by littoral transport of the sand and gravel sediments eroded from the Pliocene-Pleistocene rocky shore along the north side of Ayios Thomas peninsula.

I69

0

m N 1 lKOpOllS

12-

-10 ///

I2^vyyy^ V 93-02 0

:1 i^^ ^ V ^ 311"^""1"'^ - -^

--92-01 ': ~w -4alluvium

*-10:':' -:*;"': -_ *mudflat -4-, -

0 200 400 600 800 1000 m

Figure 5.7. Stratigraphic cross section A-A', perpendicular to the axis of the Nikopolis isthmus. For core locations, see Figure 5.3; for legend, see Figure 5.5.

-XT'I-- -1r

100 200 300 400 500 600 700 800 m

Figure 5.8. Stratigraphic cross section B-B', perpendicular to the axis of the Nikopolis isthmus. For core locations, see Figure 5.3; for legend, see Figure 5.5.

m

12

10

8

6

4

2

0

-2

-4

-6

-8

-10

0

100 200 300 400 500 600 700 800 m

Figure 5.9. Stratigraphic cross section C-C', perpendicular to the axis of the Nikopolis isthmus. For core locations, see Figure 5.3; for legend, see Figure 5.5.

m

20

18

16

14

12

10

8

6

4

2

0

-2

0


Figure 5.10. Paleogeographic reconstruction of the eastern side of the Nikopolis isthmus showing the shorelines at different periods: 6500/ 7000 B.P., 3000 B.P., and 500 B.P.

30. Rapp 1986.

In the past 5,000 years or so, at least two distinct phases of alluvial or

slope-wash sediments and their associated soil development occurred on the isthmus. Natural factors, such as tectonic movement and climatic fluc- tuation, as well as human impact might have caused hill erosion on both sides of the isthmus. To determine the exact timing of hill erosion and associated deposition would require more data. These slope-wash and alluvial sediments covered the preexisting marine or paralic unit, elevating the surface of the isthmus by ca. 5-10 m.

During the Roman period, the city of Nikopolis might have been closer to the shore than today, facing a larger area of water. The barrier of Mazoma Lagoon might not have been fully formed and thus the lagoon could have been a very well sheltered harbor for the city. During the occupation of Nikopolis, the surface of the isthmus northwest of the city might have been 3-6 m lower than today. Continued uplift of the Preveza peninsula may have led to frequent earthquakes, which could have destroyed many structures in Nikopolis as well as in neighboring towns.30

From the above discussion, we can see that shoreline changes on the Nikopolis isthmus are the result of dynamic interactions among a variety of factors, including a rise in relative sea level, tectonic movement and subsidence, and hill erosion and associated deposition. The rapid tectonic uplift of the Preveza peninsula constitutes the dominant factor leading to the gradual shoreline progradation during prehistoric and historical periods.

I73


ORMOS VATHY

Ormos Vathy is a narrow bay protruding from the Ambracian Gulf into the junction between the Preveza and the Ayios Thomas peninsulas (Figs. 5.1:7, 5.4). The bay is ca. 1 km long and 120-450 m wide, and is narrowest at its middle. It is surrounded by hills of relatively low relief composed of Pliocene and Pleistocene deposits. A low ridge with an elevation of 12 m projects into the north end of the bay, creating extensions (arms) of the inlet on either side of the projecting ridge. The northern edges of both arms are fringed by coastal marsh. The east arm has a larger area of fringe marsh than the west arm as it has a much wider area enclosed within contour lines of low elevation (below 4 m). Given its location and size, the bay must have been a well-protected harbor before a modern bridge and causeway were built across its opening to the Ambracian Gulf.

Four drill cores were taken in the two arms of the bay. Three of them were drilled along the west arm, and one on the east arm (Fig. 5.4). Core 93-07 was located in the marsh, 360 m north of the current shore. The core penetrated only 5.17 m, and consists mainly of bluish gray (5B 5/1) and very dark gray (N3/) organic mud and muddy peat that formed in a dominantly coastal marsh environment. The decayed grass from a muddy peat layer at a depth of 1.88-2.00 m (ca. 1.0 m below modern sea level) yielded a calibrated radiocarbon date of 5650-5330 B.P. The relatively high position for a date of this age may be attributed to the tectonic uplift of the Preveza and Ayios Thomas peninsulas. Part of the coastal marsh area is more than 2 m higher than the current sea level while the Ambracian Gulf has a very low tidal range (5 cm on average). Tectonic uplift must have been a significant factor in the retrogradation and progradation of the marine bay during prehistoric and historical times. Assuming that the two arms of the bay had the same rate of uplift as the surrounding ridge, relative sea level around 5500 B.P. was ca. 1 m lower than today.

Figure 5.11 is a cross section based on three cores: 94-10, 94-06, and 93-08. The stratigraphic sequence begins with a Pleistocene deposit, seen only in the furthermost inland core, 94-10, ca. 600 m from the current shore of the bay. The deposit is composed of olive (5Y 5/4) sandy mud and silt containing many calcium carbonate nodules. Overlying the Pleistocene deposit is greenish gray (5G 5/1 and 5BG 5/1) and very greenish gray (5GY 4/1) soft muddy sand, sandy mud, and mud representing a shallow marine embayment deposition comparable to that seen in the modern bay to the south. Few microfauna were found in the samples collected from this marine unit, but some foraminifera species (Elphidium, Ammonia beccarii) were identified in core 94-10. The upper boundary of the marine unit is 1-1.5 m higher than current sea level as a result of continuous tectonic uplift during the Holocene period.

The marine unit is overlain by a 1-m thick coastal marsh deposit consisting of bluish gray (5B 6/1) and dark greenish gray (5G 4/1) organic mud and humified grass. The final unit of the sequence is a slope-wash deposit consisting of yellowish brown (10YR 5/6), dark yellowish brown (10YR 4/4), and olive brown (2.5Y 4/4) silt and mud with common

I74

12

41

- 1 m - .:- m - _---

o 200 300 400 500- slope wash600 700 800

Figure 5.11. Stratigraphic cross section along the swamp of Ormos Vathy

"-4 t-q"


Neolithic Period (ca. 5500 B.P.) Roman Period (ca. 2000 B.P.)

SK^^y^^ = t/ Figure 5.12. Paleogeographic I'V~^:^^%/ |>^ 0 400 M reconstructions of Ormos Vathy

: indicating shoreline changes from

... .^ ............ . .....the Neolithic through modern

periods. Tectonic activity was a dominant factor in shoreline

Modern Period progradation.

calcium carbonate nodules and some pebbles. These calcium carbonate nodules are not in primary pedogenic context and thus do not indicate a

well-developed soil associated with the slope-wash deposit. The nodules

were derived from a Pleistocene soil on the surrounding ridges and were

deposited in the bay as a result of hill erosion. A few small red ceramic

fragments were found in the middle of the unit. A Turkish limestone struc-

I76


ture and some red tile fragments were encountered at a depth of 0.5 m in core 93-08.

Although no radiocarbon date is available from the stratigraphic section, a preliminary estimate on the timing of the Holocene deposition can be made. As discussed above, sea level relative to the surrounding ridge was ca. 1 m lower around 5500 B.P. than at present. This level is a little higher than the bottom of the marine unit in core 94-10, suggesting that the embayment extended beyond the core along the west arm before 5500 B.P. After the sea-level rise slowed, the rate of tectonic uplift apparently exceeded the rise in relative sea level. As a result, the shore of the bay moved gradually seaward, and a coastal marsh started to form along the fringe of the bay. The majority of the slope-wash deposit seems to have formed after the extensive occupation in the Roman period.

Figure 5.12 is a paleogeographic reconstruction of Ormos Vathy, showing the changing shorelines and fringe marshes from the Neolithic through the present. By 5500 B.P. the marine transgression had extended more than 750 m inland of the current shore along the west arm of the bay, but the east arm was occupied by fringe marsh instead of bay water. In other words, the bay did not extend very far inland along its eastern arm. As the rise in sea-level slowed, the continuous tectonic uplift moved the shore gradually seaward. By the Roman period, the shore of the bay may have prograded halfway to the gulf along the west arm. The Roman harbor town was built along the west shore of the bay. Because it was well protected, the bay may have been one of the most important harbors serving the city of Nikopolis. In addition to the major portion of the bay, the extended west arm might have been wide enough to provide good anchorage. Extensive habitation since the Roman period on the hills surrounding Ormos Vathy has led to increasing erosion. As a consequence, slope-wash sediments started to fill in the west arm of the bay, accelerating the progradation of the shoreline. Much of the east arm has remained coastal marsh, having received much less sediment from slope-wash processes. This may be attributed to the gentle slope of the surrounding ridge and to the possibility that sparser occupation on this side of the bay resulted in less erosion.

GRAMMENO PLAIN

31. Archaeological survey undertaken by the Nikopolis Project in 1992 and 1993 suggests that sites of various periods, including a large Roman site near the modern village of Archangelos and two Byzantine and Turkish sites, are distributed mainly along the hilly edge of the plain.

The Grammeno plain is a tectonic lowland that cuts through the Preveza peninsula near its northern end. It is covered with Holocene alluvial and slope-wash sediments. To the east it merges with the floodplain of the Louros River. Two cores were drilled on the east side of the plain to determine how far westward the maximum marine transgression extended and to show the changing landscape of the plain (Fig. 5.2).31

Figure 5.13 is a stratigraphic cross section (II-II') based on cores 93- 12 and 93-13. The marine estuarine or lagoonal unit occurs only in the lower part of core 93-13, ca. 800 m east of the Louros River. The estuarine or lagoonal unit is composed of greenish gray (5GY 4/1) soft mud with some olive (5Y 5/3) oxidized mottles. It contains many microfaunal remains including ostracods (Cyprideis, Loxoconcha) and foraminifera (Elphi- dium, Ammonia). The upper boundary of the unit is clear but more or less transitional.

I77


m

12

10

8

6

4

2

0

-2

-4

-6

-8

-10

0 500

- -

1000 1500

The marine estuarine or lagoonal unit is capped by 3.75 m of alluvium

showing a moderately developed soil profile. The top 1.2 m of the profile is light brownish gray (2.5Y 6/2) to dark gray (5Y 4/1) silt and sandy silt with fine blocky structure. The silt contains decayed rootlets and many iron oxide mottles. The top layer is an A horizon. It is underlain by a 1.5- m thick weak B horizon that is an olive gray (5Y 4/2) blocky silty clay containing calcium carbonate nodules. Underlying the B horizon is olive

gray (5Y 5/2) massive clay and sandy mud with some iron oxide mottles. The sediments of this alluvial unit could be derived from both the Louros River and small streams flowing on the Grammeno plain.

Core 93-12 is located ca. 870 m southwest of core 93-13. No estuarine or lagoonal deposit occurs in this core. The sequence begins with pale olive (5Y 6/4) and yellow (5Y 7/4) firm to very firm clay and clayey silt with a 0.15-m thick very dark gray (N3/) clay on top. This unit occurs at a

depth of 5.95 m, and may represent a soil formed in a Pleistocene alluvial

deposit. The basal unit is covered by two phases of Holocene alluvial de-

posits, both of which show moderate soil development characterized by carbonate-enriched Bk horizons. These two alluvial deposits and associated soil profiles may correspond to those top slope-wash units observed on the Nikopolis isthmus in terms of the timing of the hill erosion that

provided the sediments.

Figure 5.13. Stratigraphic cross section near the Grammeno plain. For core locations, see Figure 5.2 (section is labeled II-II'); for legend, see Figure 5.5.

2000 2500 m

I78

- E


The stratigraphic sequence observed in these two cores indicates that the Holocene marine transgression did not extend very far into the Gram- meno plain because of the relatively high topography formed by the preexisting alluvium. The maximum transgression reached somewhat beyond the location of core 93-13, perhaps as far as 1 km from the current channel of the Louros River. As in the area to the north of the Ambracian Gulf, the marine transgression in the Grammeno plain might have reached its maximum around 4500 B.P., with gradual progradation after 1500 B.P. as alluvial and slope-wash sediments filled in the estuary at an increasing rate. The increased sediment supply might be due to accelerated hill erosion caused by human impact.

KASTRO ROGON AND STRONGYLI

32. Hammond 1967, p. 427. 33. Dakaris 1971, p. 42. 34. Dakaris 1971, p. 178;

Hammond 1967, p. 427. 35. The Kastro Rogon site and its

surrounding floodplain were surveyed by the Nikopolis Project in 1992 and 1994. Archaeological finds range from the Classical through post-medieval periods (early 5th century B.c.-19th

century A.C.).

Kastro Rogon is ancient Bouchetion, an urban site situated on the top of a

Jurassic limestone hill near the northern edge of the coastal delta plain of the Ambracian Gulf (Figs. 5.1:15, 5.14). It was first built as a colony around 700 B.C.,32 but the earliest wall might not have existed until the early 5th

century B.c.33 The site was occupied until the medieval period, and additions and repairs to the circuit walls were made throughout its history. Bouchetion was one of four important walled settlements of the colonists from Elis during the Classical and Hellenistic periods (late 5th century to 168/7 B.c.), and it remained an important site after the Roman conquest because of its strategic position.34

The hill of Kastro Rogon, ca. 65 masl, is located south of Mt. Rokia and stands isolated from other hills to the east and north. The Louros River flows along the southwest side of the hill and then turns to the north at the hill's northwest corner. Across the Louros River to the south and west are a floodplain and reclaimed swamp (previously brackish estuarine marsh) with a relatively low elevation (0-4 masl). To the east of Mt. Rokia, the delta floodplain rises gradually northward toward the deeply incised

valley of the Louros River.35

Strongyli, a Roman-period villa rustica, is situated on a small ridge north of Koryphi, the northernmost protuberance of Mt. Mavrovouni located in the low-lying estuarine swamp (Figs. 5.1:16, 5.14). The Stron-

gyli site is ca. 3.2 km southwest of Kastro Rogon. In addition to Roman remains, there are remains of earlier and later times, including the Helle- nistic and Late Byzantine or Turkish periods.

A total of seven cores were drilled in the area around Kastro Rogon and Strongyli. Based on these cores, three stratigraphic cross sections were constructed to interpret the evolving sedimentary environments in this

historically strategic location. Cross section C-C' (Fig. 5.15) consists of two cores, 94-15 and 94-19, near the hilltop site of Kastro Rogon. Both cores are on the right bank of the Louros River (see Fig. 5.14). The bottom unit of the stratigraphic sequence shown in the cross section is a very dark gray (N3/) muddy peat consisting mainly of humified grass. The peat layer is found at a depth of 7.5 m (about 5 m below current sea level) in core 94-19, located only 130 m from the 10-m contour on the hill to the northeast. The layer has been radiocarbon dated to 4830-4410 B.P. and is believed to represent coastal fringe swamp before the maximum transgres-

I79


|I I

I d Mesozoic limestone v v vV Neogene-Pleistocene 1 9211 geologic swamp A cross

hill at 20 m contour v v v M hill at 20 m contour ? core f (reclaimed) [ sectio

0 1 2 3 km _T - - _

sion around 4500 B.P. This peat layer is overlain by greenish gray (5GY 6/ 1, 5G 5/1, and 5BG 5/1) soft estuarine or lagoonal mud containing some brackish ostracoda species (mainly Cyprideis torosa) and a few freshwater

species (Candona) in its upper portion. Within the estuarine or lagoonal unit there are two intercalated layers of dark brown (7.5YR 3/2) muddy peat and peaty mud seen in core 94-15, located 150 m north of the Kastro

Rogon hill. These two intercalated peat layers indicate that part of the foothill area of Mt. Rokia could have been intermittently swampy during the period of maximum marine transgression, ca. 4500-1500 B.P. In core 94-19 the estuarine or lagoonal unit is overlain by 1.4 m of olive (5Y 5/4) gravelly sandy mud containing many pebbles and secondary calcium carbonate nodules along with some plaster fragments. The gravelly unit has very sharp boundaries with both underlying and overlying sediments, and it may represent a small colluvial deposit derived from slope washes. Al-

ternatively, this unit may form part of a causeway built to connect the Kastro Rogon site to the mainland. If this was the case, the stratigraphy indicates that the causeway was built before 1500 B.P., most likely during the Roman period. More evidence would be needed to substantiate this hypothesis.

On top of the estuarine and lagoonal unit lies a greenish gray (5G 6/1 and 5GY 6/1) muddy peat and peaty mud containing abundant

Figure 5.14. Map of Kastro Rogon and vicinity showing the location of geologic cores and cross sections

I80

4I'- SE

3 9415 <94-1519

*2.' - ., .. * . ? slope wash < ... - . . -- - - - ''. ."."

r,~ _..... .. ai__ alluvium - S . ?' . '- ':- - - Sea Level

'_| , - '-v '-'_ _.:__ _ _--. '.swamp',' ..-- " 1 560- 1,290 B.P. swamp

2 1 \^, \ \ \ X \\ [....i-.- ~ , ',estuary /lagoon\ \ \ "'---. ' ''. \ \ \s l op" \ wx - " -2 I c.\lt\ura\l<..\ .X \\ -

...?..\\.... \\\ _.

3 '~%~~~~7 & '...... ' H \estuary/lagoon . '-f o . , , *,

..-5 | ,

-- -.....- ._ _._^ .. . ' .-, ~'~ '\ ~ , . ..-. 4,830 -4,410 B.P.

' swamp -- a,

Figure 5.15. Stratigraphic cross section C-C' at Kastro Rogon. For core locations, see Figure 5.14; for legend, see Figure 5.5.

00

r


decayed grass debris (very pale brown, 10YR 7/4). The unit is ca. 20-60 cm thick and is found in both cores. Ostracoda from this unit include both brackish and freshwater species (Cyprideis, Candona). Based on a calibrated radiocarbon date of 1560-1290 B.P., we interpret this layer as a brackish

swamp deposit formed along the northern fringe of the ancestral lagoon associated with the Ambracian Gulf after the end of maximum marine

transgression, probably around 1500 B.P. Clearly, the Kastro Rogon hill was an island in a marine estuary during the period of maximum trans-

gression, from 4500 B.P. to 1500 B.P.

The swamp unit is capped by a 2.5-m thick floodplain alluvium, con-

sisting of light olive brown (2.5Y 5/4) and olive (2.5Y 5/4) silt and silty clay. The lower boundary of the unit is gradual, as indicated by the downward increase of light greenish gray (5GY 7/1) mottles. The lower part of the alluvial unit yielded some freshwater gastropoda and ostracoda species (Candona). The top part of the unit shows weak blocky structure and contains some fine calcium carbonate nodules, representing a weak cumulative soil profile. Overlying the floodplain alluvium is 0.75-0.9 m of strong brown (7.5YR 5/6) silt and silty clay with rootlets and some limestone

pebbles. This top strong brown deposit occurs only on the Rogon side of the

Louros River; the top deposit on the floodplain side of the river is olive

(2.5Y 5/4) silty clay and clay (see Fig. 5.17). The strong contrast in color

suggests different origins for these deposits. The olive deposit on the flood-

plain side must have developed from overbank deposition of the Louros River, while the strong brown silt and silty clay might derive from the weathered hill slopes by sheet erosion. The strong brown deposit must have formed after the Louros River flowed in its current channel because it occurs only on the Rogon side of the river. Thus, we are able to determine when the Louros River started flowing in the current channel. We know that the floodplain alluvium overlying the post-transgression swamp (dated to 1560-1290 B.P.) formed after 1500 B.P. A certain amount of time was needed to form the 2.5-m thick alluvium underlying the top strong brown deposit. It seems reasonable to suggest that the diversion of the Louros River into the current channel did not occur until the 10th century A.C. or later.

On the floodplain side of the river, many cultural remains of the post- medieval period (middle 15th-19th century A.C.) were found in a plowed field 100 m from the river. These post-medieval remains enclosed in the

top floodplain deposit suggest that the Louros River was diverted into the current channel before the 15th century A.C. On the Rogon side of the

river, many artifacts were found in the low-lying area surrounding the fortified site. Although most of these artifacts are post-medieval in date, some

belong to the Classical through medieval periods. We believe these older artifacts were eroded from Kastro Rogon after the Louros River was diverted. A channel diversion sometime between the 10th and 15th centuries might explain the distribution of archaeological remains on the two sides of the river.

Figure 5.16 is a stratigraphic cross section (B-B') located in the central part of the deltaic floodplain formed by the Louros River as it enters into the Ambracian embayment (see Fig. 5.14). Along the central part of the floodplain, the surface rises rapidly toward the northeast. Core 94-16

I82


m

12

- ,- SW

:al luvi*um J .y ^1-N J^^ (deltaic).i-|:<

_W_ , '

: . _ W__, r<^ --, A_-

. .:> <. . ? . A

. -4 j -^ - ; /. aluiu 1 5 i A '

4'

< 4' ' t

- < ? 4 . 4 4 , _ s A

<,,4,,-~~~~~~~~~~~~~~~. .;i.<. .4._-. .,-;.-..v..

*- v > w > 4 S q 4 ^ - e . 41,

,, ,, - 7 . j ie A . s A

;

~ swamp p j (reclaimed)

_sf < , , *^ k * >-

^ .^ ^ ~ ^- s -" ^1-, - s~~~~~~~~~~- - 11-

- ll

IN,s -

lagoon / estuary

,k ', S- as? I _,V_

A IP -IL_ -Al -!1k.

k swamp I v ^~ ~ ~~~.. W.. _W_

-Al- - st 11

-W sbw Al 1

_ -Lr _ v ,

* - nearshore .^-

X

. 4A . n , *A 'J . '

AA

"if Av A A A l:->t^

~.- - - - u llll - llll - ll

2400 3200 4000 4800 5600 m

Figure 5.16. Stratigraphic cross section B-B' near Kastro Rogon. For core locations, see Figure 5.14; for legend, see Figure 5.5.

was drilled in reclaimed swamp with an elevation of 1 masl. The top part of the core consists of a 1.9-m thick light yellowish brown (2.5Y 5/4) massive silt, sandy silt, and silty sand with some mud laminations. Within the top unit, gray (N6/) mottles increase downward. The unit is inter-

preted as a swamp deposit formed at the front of the deltaic floodplain. In core 94-18 the top unit is a deltaic floodplain deposit composed of

3.4-m thick olive yellow (2.5Y 6/6) and dark grayish brown (2.5Y 4/2) silt, clayey silt, sandy silt, and silty sand containing many decayed grass rootlets and freshwater ostracoda species (Candona). Underlying the swamp or deltaic floodplain deposits in both cores is a 4-5 m thick estuarine or

lagoonal unit that consists of greenish gray (5BG 5/1) interbedded sandy mud, mud, and muddy sand. The unit contains both brackish and freshwater ostracoda species (Cyprideis, Candona), but more freshwater species appear in the upper part of the unit. Based on its stratigraphic context, the estuarine or lagoonal unit must be the product of maximum marine trans-

gression (4500-1500 B.P.).

The estuarine or lagoonal unit rests on a swamp deposit composed of

greenish gray (5G 6/1) and bluish gray (5B 5/1) peaty mud and dark brown (7.5YR 3/2) peat and peaty mud. This swampy unit is believed to have formed before maximum marine transgression and it may extend quite far into the previous valley of the Louros River. A 0.7-m thick gray (N6/0)

I

I

I

I

0

10

8

6

4

2

0

-2

-4

-6

-8

-10

800 1600

-- 1 - A- - I

_ -n

..

I83

- A

-,4


gravelly sand is interlayered within the swampy unit in core 94-16. The gravelly sand layer contains many brackish and freshwater microfauna including gastropoda, ostracoda (Cyprideis, Candona), and foraminifera (Elphidium). It may represent a nearshore deposit formed during a relatively short period of sea transgression.

Stratigraphic cross section A-A' (Fig. 5.17) is based on four cores, 94- 19, 92-11, 92-07, and 92-08, between Mt. Rokia in the north and Mt. Mavrovouni in the south. The surface along this traverse dips gently toward the south. The northern half of the section is covered with floodplain alluvium, and the southern half is occupied by reclaimed swamp. Kastro Rogon is located at the northern end of the section, and Strongyli lies at the southern end.

The stratigraphic sequence begins with a basal swamp unit in core 94- 19 at the northern end of the section (see above). Corresponding to the swamp unit is a nearshore deposit at the southern end near Strongyli. The nearshore deposit lies at a depth of 3.6 m in core 92-08 and 4.15 m in core 92-07 and it consists of dark gray (N4/0), gray (N5/0), and greenish gray (5GY 5/1) interbedded sandy mud, muddy sand, silt, and mud with some thin shelly sand lenses. This nearshore unit is very rich in marine and brackish gastropoda (Monodonta, Cyclope), bivalves, foraminifera (Trocham- mina, Elphidium, Protelphidium, Ammonia), and ostracoda (Cyprideis torosa, Loxoconcha). Two radiocarbon dates were measured on charred grass samples from the nearshore unit. The sample from a depth of 6.68-6.75 m in core 92-07 gave a calibrated date of 4440-4010 B.P. The sample from 3.8 to 3.9 m in core 92-08 dated to 4080-3640 B.P. Both dates suggest that the nearshore deposit formed during maximum marine transgression beginning around 4500 B.P.

As stated earlier, an estuarine or lagoonal deposit rests on the basal swamp unit in core 94-19. This estuarine or lagoonal unit is seen in all the cores across the section. It consists of greenish gray (5GY 5/1 and 5BG 5/1), gray (N5/0), and dark gray (N4/0) soft mud that contains some brackish ostracoda species (Cyprideis), foraminifera (Elphidium), and very few freshwater ostracoda (Candona). The major portion of the estuarine or lagoonal unit formed during the period of maximum transgression, probably between 4500 B.P. and 1500 B.P. In core 92-08 the estuarine or lagoonal unit is only 0.6 m thick and is covered by a 1.8-m thick nearshore deposit composed of sand, shelly sand, and sandy mud with abundant brackish shells. The dominance of nearshore facies in core 92-08 may be attributed to its location on the edge of Mt. Koryphi. During the Roman period, the seashell-enriched nearshore environment could have provided important food resources for the inhabitants of Strongyli.

Overlying the estuarine or lagoonal unit is a swamp deposit that is buried by floodplain alluvium in cores 94-19 and 92-11 and crops out southward in both cores 92-07 and 92-08. In core 94-19 the swampy unit is muddy peat and peaty mud dated to 1560-1290 B.P. The swamp formation started at the northern end of the section after the end of maximum marine transgression (ca. 1500 B.P.) and moved gulfward as fluvial sediments from the Louros River filled in the estuary and gradually covered the swamp.

I84

Mavrovouni

- m

400 800 1200 1600 2000 2400 2800 3200 3600

Figure 5.17. Stratigraphic cross section A-A' near Kastro Rogon. For core locations, see Figure 5.14; for legend, see Figure 5.5.

4000 m

-0 - Rokia m

12

10

8

6

4

2

0

-2

-4

-6

-8

-10

0


Before discussing the paleogeographic evolution of the Kastro Rogon- Strongyli area and its archaeological implications, we first need to examine the stratigraphic cross section across the whole coastal plain-lagoon- barrier system to the north of the Ambracian Gulf so that we can place our interpretation in the context of the whole embayment (see Fig. 5.2). Fig- ure 5.18 is a cross section based on five cores: 93-11, 93-09,92-10, 92-09, and 93-16. The northern end of the section is at the foothill of Mt. Rokia and the southern end is at the Salaora barrier. This traverse shows a very gentle topography from the foothill in the north to the edge of Rodia Lagoon.

The stratigraphic sequence begins with a 2-m thick gravelly sand with many angular to subangular pebbles seen only at the base of core 93-11, located 350 m south of the 10-m contour of Mt. Rokia. The basal gravelly sand is of fluvial or colluvial origin and constituted the pre-transgression surface along the edge of the tectonic embayment. This sand layer is overlain by a 0.9-m thick swamp layer comprised of dark yellowish brown (10YR 3/4) peat and peaty mud. The peat deposit dates to 6890-6500 B.P., ca. 2,000 years earlier than the date obtained from the basal peat layer in core 94-19 (4830-4410 B.P.). The age difference may be due to a lower elevation associated with the former date (see Table 5.1). Moreover, the large range of dates for this basal peat layer implies that relative sea level rose very slowly from 7000/6500 B.P. to 4500 B.P., allowing peat to develop in the coastal fringe swamp.

On the basal peat layer lies an estuarine or lagoonal unit composed of a dark greenish gray (5BG 4/1 and 5G 4/1) soft mud interbedded with sandy mud and muddy sand. The estuarine or lagoonal unit contains variable amounts of decayed plant remains and marine and brackish fauna such as foraminifera (Ammonia, Elphidium) and ostracoda (Cyprideis, Xesto- leberis, Basslerites). From this unit more microfauna are found in landward cores, particularly in cores 93-11 and 93-10 (not shown in cross section), and more plant remains are seen in cores on the lagoon side, especially in core 92-09. A date of 1210-890 B.P. was measured on a wood sample at a depth of 5.15-5.25 in core 92-10. Core 92-09 yielded four radiocarbon dates. The core was drilled in the swamp on the edge of Rodia Lagoon. The top 1.75 m of this core is dark reddish brown (5YR 3/2) peat and muddy peat. Underlying the peat unit are dark gray (N4/), gray (N5/), and greenish gray (5GY 5/1) interbedded sandy mud, muddy sand, and sand containing plant remains. All four radiocarbon dates are younger than 900 B.P., indicating increasing sedimentation rates from the foothills to the lagoon with gradual infilling of the lagoon.

The estuarine or lagoonal unit is overlain by a swamp unit. The lower boundary of this swamp unit rises gradually southward, suggesting a gradual progradation of post-transgression swamp with increased estuarine infilling. At the northern end of the section, a layer of peaty mud (0.6-0.9 m thick) seen in both cores 93-09 and 93-11 constitutes the bottom part of the swamp unit. The peaty layer yielded many brackish ostracoda (Cyprideis) and foraminifera (Elphidium, Trochammina, Cribroelphidium). A radiocarbon sample from core 93-09 dates the peaty layer to 1710-1410 B.P. As discussed above, the top peat deposit in core 94-19 yielded a calibrated

i86

Salaora Barrier

Rodia 92-10 Lagoon _-2- JA 92-09

L

e

- " ,19 ~ _ -- - - - - - - - E -= swamp

,(partially reclaimed)

,710-1,410 B..P v . " '

4 6 8 10

93-16 Ambracian Tsoukalio Lagoon - Gulf [ -:. -""?. ii:.:..: :\

back barrer swamp <,

^^^:$. \\ 2,760-2,350 B.P.

12 14 km

Figure 5.18. Stratigraphic cross section north of the Ambracian Gulf showing sedimentary sequences and environments across the entire coastal plain-lagoon-barrier system. For core locations, see Figure 5.2 (section is labeled I-I'); for legend, see Figure 5.5.

m

18

16

14

12

10

8

6

4

2

0

-2

-4

-6

-8

-10

-12

-14 0 2 ? I I ? I ? i I ?

1' S

\PU

0 2

c0

"-4


radiocarbon date of 1560-1290 B.P. (see Fig. 5.15). Given their sedimen-

tary and stratigraphic context, these dates suggest that maximum marine

transgression ended around 1500 B.P.

In both cores 93-11 and 93-09, the upper part of the swamp unit is

composed of greenish gray mud (5BG 5/1) with some muddy sand and sandy mud laminations. It is overlain by a top floodplain alluvium. Toward the south the swamp unit crops out and extends to the edge of Rodia

Lagoon. Top alluvium, 4.0-5.5 m thick, is found only in cores 93-11 and 93-

09 on either side of the Louros River. It thins seaward and merges into the swamp in the south. The upper part of the alluvial unit is olive brown (2.5Y 4/4) and light olive brown (2.5Y 5/4) silt and silty clay with a very weakly developed soil profile on the top. The lower part consists of olive (5Y 5/6) and olive yellow (5Y 6/6 and 2.5Y 6/6) silt, sandy silt, and silty clay with gleying mottles increasing downward (5GY 4/1 and 5Y 6/1). The alluvial unit started forming after the end of maximum marine trans-

gression, probably around 1500 B.P. The top part of the unit likely formed from overbank sedimentation of the Louros River that started flowing along the northern edge of the embayment after the 10th century A.c.

Other streams emerging from mountain valleys to the north and northwest might also have contributed a significant amount of sediment to the formation of the lower part of the alluvial unit.

Core 93-16 was taken on the lagoon side of the Salaora barrier. The barrier projects landward. At the east end it is connected to Salaora Island; at the west end it is attached to the Preveza peninsula. The top deposit in the core is composed of 2.2 m of shelly sand, sand, and silty sand contain-

ing abundant shells. The next 2.1 m is dark greenish gray (5GY 4/1) silty mud and mud with common shells and some decayed plant remains. At a depth of 4.3-5.0 m is a back barrier swamp deposit consisting of dark brown (7.5YR 3/3) peat layers interbedded with dark greenish gray (5GY 4/1) muddy sand and sandy mud. The back barrier swamp unit is superimposed on an estuarine or lagoonal unit composed of dark greenish gray (5G 4/1) interbedded sandy mud and muddy sand with common thin sand laminations. Marine and brackish fauna are common in the estuarine or lagoonal unit, including ostracoda (Basslerites, Loxoconcha, Xestoleberis, Cyprideis) and foraminifera (Trochammina, Ammonia, Elphidium).

A radiocarbon date of 2760-2350 B.P. was determined on a peat sample in the back barrier swamp unit at a depth of 4.5-4.7 m, suggesting that the overlying barrier unit started forming after 2500 B.P. Alongshore deposition rather than offshore deposition is most likely responsible for the barrier's formation. Thus the barrier started developing from either or both ends by alongshore transport of the sand and gravel sediments eroded from the Preveza peninsula and Salaora Island, probably around 4500 B.P. when maximum marine transgression was reached. The barrier migrated laterally as the sea level gradually rose. The radiocarbon date from the back barrier swamp unit in core 93-16 suggests that the barrier might not have migrated to the location of core 93-16 until 2500 B.P.36

36. Core 93-16 was drilled in the middle of the central barrier island. A Turkish military map published in 1900 shows that at that date there was still a large opening in the western part of the barrier. We believe that the previous lagoon or estuary was open to the Ambracian Gulf.

I88

COASTAL EVOLUTION OF THE AMBRACIAN EMBAYMENT 189

37. Strab. 7.7.5 (C 324), trans. H. L. Jones, Cambridge, Mass. [1924] 1954.

38. Hammond 1967, pp. 61-63; Dakaris 1971, pp. 57, 178, 180.

39. Hammond 1967. 40. Dakaris 1971. 41. Dakaris 1971, p. 6: "Pseudo-

Scylax (Periplous 32) wrote in 380-360 BC that the shore between the mouths of Louros and Arachthos was 40 stadia wide (approximately 8 km)." See also Dakaris 1971, fig. 9.

With all available subsurface data from the coastal plain-lagoon area north of the Ambracian Gulf, we can reconstruct the paleogeographic set-

ting of both Kastro Rogon and Strongyli. Strabo describes Kastro Rogon (Bouchetion) as follows: "Near Cichy-

rus is Buchetium, a small town of the Cassopaeans, which is only a short distance above the sea; also Elatria, Pandosia, and Batiae, which are in the interior."37 It is easy to understand that Elatria (Palaiorophoros), Pan- dosia (Kastri), and Batiae (Kastro Rizovouni) are "in the interior": Palaio-

rophoros and Kastro Rizovouni are situated in mountainous highlands and Kastri is a hilltop site located well inside the Acheron valley (see Fig. 5.1). But it is harder to reconcile the description of Bouchetion as lying "only a short distance above the sea." The hilltop site of Kastro Rogon is

currently located well inland. The direct distance between Kastro Rogon and Salaora is ca. 13 km, and the distance along the Louros River is more than 20 km.

Historically, Kastro Rogon was believed to be a port serving two urban settlements-Batiae and Elatria-during the Classical and Hellenis- tic periods.38 Neither Hammond39 nor Dakaris40 suggests that the port was located on the sea coast. Instead, both scholars believe that the port was linked to the Ambracian Gulf by the Louros River, and that the lower

portion of the river was navigable. This belief is based on the assumption that current geomorphic elements existed in antiquity as well, an assumption we have shown to be incorrect.

Figure 5.19 shows the evolution of the paleogeographic setting near Kastro Rogon and Strongyli based on the subsurface stratigraphic data discussed above. During maximum marine transgression, ca. 4500 B.P. (Fig. 5.19:b), the shoreline was at the foot of Mts. Stavros and Rokia, at the northern edge of the Ambracian embayment, thus making islands of previously inland hills. Mt. Mavrovouni was the biggest of these islands. Kastro

Rogon also became an island during this period, but it was very close to the mountainous mainland. The town of Bouchetion was situated on the

top of the island, 65-75 masl, during the Classical, Hellenistic, and Ro- man periods. This geographic setting fits well with Strabo's statement that "Buchetium... is only a short distance above the sea."Thus, Kastro Rogon was a logical site for a seaport and it held a strategic position within the

embayment. Our analysis also revealed evidence for the changing course of the

Louros River. During maximum marine transgression, the marine embayment probably extended inland along the river channel after it emerged from the deeply incised valley in the north (Fig 5.19:b). During the historical periods, however, the position of the channel was in some dispute. In an attempt to reconcile ancient sources about the Louros River, includ-

ing an account by Pseudo-Scylax, Dakaris proposed that the river flowed to the east of Mt. Mavrovouni for the Classical through Roman periods.41 This placement may be appropriate for the period around 1500 B.P. and later but not for the Classical, Hellenistic, and Roman periods. According to our paleogeographic reconstruction, the shoreline was well north of


a

4500 B.P. 0 1 2 3 km

b

Figure 5.19. Paleogeographic reconstructions of Kastro Rogon and vicinity showing the changing coastlines and environments from 7000/6500 B.P. through 1000/500 B.P.: a) 7000/6500 B.P.; b) 4500 B.P.; c) 1500 B.P.; d) 1000/500 B.P.

I90o

COASTAL EVOLUTION OF THE AMBRACIAN EMBAYMENT 191

C

d


Mavrovouni during these periods and Pseudo-Scylax's measurement of the distance between the mouths of the Louros and Arachthos Rivers (40 stadia or 8 km) was likely correct. After the end of maximum marine trans-

gression around 1500 B.P., the deltaic floodplain began to develop toward the south and southwest as more and more sediments entered the estuary (Fig. 5.19:c).The Louros River flowed in a relatively stable channel at this time. Based on the trend of the contour lines in the deltaic floodplain, the river likely flowed south or southwest directly into the lagoon or the Ambracian Gulf during the early phase of estuarine infilling. The river was not diverted into the current channel until sometime between the 10th and 15th centuries A.C. (Fig. 5.19:d). This channel diversion was cultural rather than natural.

COASTAL LANDSCAPE CHANGE OF THE AMBRACIAN EMBAYMENT

Major environmental changes have occurred in the Ambracian embayment. On the basis of subsurface stratigraphy and its implied sedimentary environments in archaeologically and geologically important locations, we can reconstruct the changing coastal landscape of the Ambracian embayment during the Holocene epoch (10,000 B.P. to present).

RELATIVE SEA LEVEL AND LOCAL TECTONISM

Change in relative sea level during the Holocene and the preceding Wiirm glaciation was the single most important element in shaping the morphology of the coastal landscape. Many studies have shown that there was a rapid rise in eustatic sea level from the end of the Wiirm glaciation (15,000- 20,000 B.P.) to 6000-7000 B.p.42 However, the change in eustatic sea level over the past 6,000-7,000 years has remained in dispute. It has been shown that relative sea level is more useful and appropriate than eustatic sea level for paleogeographic reconstruction with archaeological implications.43 The change in relative sea level is controlled mostly by eustatic level, tectonic movement, sedimentation, and compaction of the preexisting sediment column.

Local tectonic subsidence or uplift has been widely considered more critical than eustatic effects to the development of the coastal landscape in Greece over the past 6,000-7,000 years.44 During the evolution of the Ambracian coast, both tectonic uplift and subsidence have played a significant role in shaping the configuration of the embayment. The Preveza peninsula has been subjected to continuous tectonic uplift, as clearly indicated by the subsurface stratigraphic sequence. Thus the small embayments projecting into the Preveza peninsula, such as Ormos Vathy, have witnessed shoreline progradation for 6,000-7,000 years. As a result, much of the previously deposited marine or estuarine strata have been elevated above sea level.

The Ambracian embayment itself has a different history of marine transgression and regression due to tectonic subsidence. Here the maxi-

42. E.g., Fairbanks 1989. 43. Kraft, Aschenbrenner, and Rapp

1977; Kraft, Rapp, and Aschenbrenner 1980; Kraft, Kayan, and Aschenbrenner 1985; Rapp and Kraft 1994.

44. Flemming 1968,1972; Flemming and Webb 1986; Kraft, Aschenbrenner, and Rapp 1977.

I92


0

-2

o A

- -4

maximum -6 [/-'. transgression

- --

dated peat samples from swamp

7 5 004------------400300 00 1oo

deposits north of the Ambracian 7000 6000 5000 4000 3000 2000 1000 0

Gulf Calibrated Radiocarbon Age (B.P.)

mum marine transgression lasted from 4500 to 1500 B.P., with a subse-

quent regression resulting from the dominance of sediment infill over the rise in relative sea level or tectonic subsidence. Radiocarbon dates on peat samples from the coastal swamp deposits in the northern part of the Ambra- cian embayment indicate a gentle rise in relative sea level over the past 7,000 years (Fig. 5.20). Recalling the fact that the Ambracian embayment is a tectonic graben that has been subsiding since the Pliocene and Early Pleistocene, this rise in relative sea level is most likely attributed to continuous tectonic subsi dence of the embaymen t itself Marin e t ransgression and regression are dictated by the change in relative sea level. Thus, rela-

tive sea level change should be used in the interpretation of subsurface strat igraphy in the Ambracian embayment. Obviously such a generalized relative sea-leveltonic graben thatrend cannot be applied to an area subject to tectonic uplift, such as the Nikopolis isthmus. Any use of a relative sea-level curve for paleogeographic reconstruction must be made in an appropriate tec-

tonic and sediment ary context. It is important to know that the rise in relative sea level is with refer-

ence to a geodetic datum. A change in relative sea level does not mean the same change in eustatic sea level, which is measured in reference to the center of the earth. Based on observations of submerged remains, Ham- mond states:

T here are indications in the coast of Epirus that the level of th e sea was at least three or four feet lower in th e fourth century than it is tod ay...s The lower sea-level in antiquity affected, for instance,

the entry to the Gulf of Arta, and it may have reduced the area of swamps which are found today near the mouths of the Louros....

I93


The fertile plain on the north shore of the Gulf of Arta may have been more extensive in antiquity.45

This is an example of how a misunderstanding of relative sea level can lead to an inappropriate interpretation of paleogeographic change. It is true that relative sea level was more than three or four feet below current sea level due to tectonic movement, but this does not imply that absolute sea level was necessarily lower in the 4th century than today. Furthermore, the extent of transgression was not determined solely by the absolute sea level.

Instead, as mentioned previously, it was a result of the combination of sea

level, tectonics, sediment supply, and compaction of the preexisting sediment column. Contra Hammond, the embayment saw its maximum sea

transgression during the 4th century, with the sea reaching the foothills of the mountains in the north.

PALEOGEOGRAPHIC DEVELOPMENT

The Ambracian embayment is a shallow backarc basin, initially shaped by Oligocene-Miocene compressional folding and faulting followed by Pliocene-Quaternary extensional faulting. At the end of the Wiirm glaciation, ca. 15,000 B.P., sea level was 100-120 m below its present level.46 The shoreline of the Ionian Sea lay about 5 km west of the Preveza peninsula. Isolated from the Ionian Sea, the Ambracian embayment was

mostly exposed subaerially; only a small portion might have been under water, forming small isolated lakes, particularly in the southern part of the basin.47

As a large volume of glacial ice melted, the sea level started rising very rapidly around 13,000 B.P. By 10,000 B.P., sea level had risen to approximately 45 m below current sea level and the Ionian Sea began to intrude into the Ambracian embayment through the narrow channel at the south end of the Preveza peninsula.48 Sea level continued to rise, and the water

body in the embayment gradually expanded. Previously inland hills, such as Mt. Mavrovouni and Salaora, were left in the embayment as islands.

Apparently, eustatic sea level played a dominant role in the development of coastline change and geomorphic configurations from 13,000 to 6500 B.P. After 6500 B.P. or the beginning of the Neolithic period, the rise in eustatic sea level diminished greatly or ceased. As a result, the shoreline

migrated at a much slower rate, creating a favorable condition for the formation of coastal fringe swamp (Fig. 5.21:a). From 6500 B.P. onward, local tectonic movement became the primary element in the further evolution of the embayment. Relative sea level continued to rise because of tectonic subsidence, and the embayment migrated landward as transgression proceeded. By 4500 B.P. or the beginning of the Bronze Age, the embayment had gained maximum marine transgression, and the sea had extended to the northern edge of the embayment leaving no or a very narrow passage along the foothills of Mts. Rokia and Stavros (Fig. 5.21:b). As tectonic subsidence was still proceeding at a rate greater than sediment infill from the rivers and streams in the north and northwest, relative sea level continued to rise until 1500 B.P., about the end of the Roman period.

45. Hammond 1967, pp. 42-43. 46. Chappell and Shackleton 1986;

Nakada and Lambeck 1988; Fairbanks 1989.

47. The analysis of 3.5-kHz seismic reflection profiles suggests that small water bodies existed in the south of the Ambracian Gulf, particularly within the eastern part, during the late Wurm glaciation; see Poulos, Lykousis, and Collins 1995; Tziavos 1997.

48. Tziavos 1997, p. 428.

I94


49. Dakaris 1971, p. 5; Hammond 1967, p. 19.

During the period of maximum marine transgression, the entire embayment was flooded by the sea. The Louros River flowed directly into the

embayment and formed a subaqueous delta near the mouth of the deeply incised valley of the Louros River. The Salaora barrier started developing as relative sea level rose, but the full barrier did not form until the post- medieval period. In other words, the entire embayment was basically open water. Kastro Rogon hill, previously inland, was an island in the embayment, but it was separated from the mainland by a very narrow stretch of water. Structures could have been built to connect the hill to the mainland. In addition, the hill was very close to the mouth of the Louros River. This

environmentally advantageous location gave Kastro Rogon strategic significance during the Classical, Hellenistic, and Roman periods. During Classical and Hellenistic times, Bouchetion, one of four important walled towns of the Elean colonists, was located on an island hill. A seaport could have been associated with the settlement that served other towns-in-

cluding Batiae (Kastro Rizovouni) and Elatria (Palaiorophoros)-in the mountainous hinterland. Because of its strategic position, Bouchetion remained an important urban site during the Roman period when other Elean settlements were destroyed and abandoned.

Owing to a different tectonic context, the area surrounding Nikopolis has witnessed marine regression instead of marine transgression since 6000 B.P. The Preveza peninsula has been subjected to tectonic uplift since the Pleistocene. From 13,000 to 6000 B.P. the rapid rise of eustatic sea level was greater than the tectonic uplift of the peninsula. As a result, the tectonic lowlands projecting into the uplifting peninsula were gradually submerged as the sea level rose. By 7000-6500 B.P. these lowland areas, including the Nikopolis isthmus and Ormos Vathy, had witnessed maximum transgression. The west arm of Ormos Vathy extended 750 m inland of the current shore and the Nikopolis isthmus may have been an open channel between the Ambracian embayment and the Ionian Sea. After 6000 B.P., eustatic sea level rise ceased or greatly slowed, and tectonic uplift became the primary factor controlling shoreline change. With continued uplift, the shorelines in the small embayments migrated seaward. Con- tinuous tectonic uplift also led to increased slope erosion. Deposition of slope-wash sediments affiliated with increased slope erosion accelerated marine regression. Sometime after 6000 B.P. the Nikopolis isthmus had been elevated to a level so that no possibility of a channel remained. By 3000 B.P. the shoreline of the Mazoma embayment had migrated seaward to within ca. 1 km of the current shore. During the Roman period, however, both the Mazoma embayment and Ormos Vathy were still well- sheltered harbors serving the city of Nikopolis and other towns on the Preveza peninsula.

From late antiquity onward, beginning ca. 1500 B.P., the rate of sediment supply from the rivers exceeded the rate of relative sea-level rise and the estuarine embayment began to fill in, moving the shoreline seaward. The increased rate of sediment supply is likely related to human-induced erosion since the Roman period.49 The Louros River continued to enter directly into the estuarine embayment with a delta developing at its front.

I95

I96

c d

Figure 5.21. Paleogeographic reconstructions of the Ambracian embayment showing the shoreline changes from 7000/6500 B.P. through 1000/500 B.P.:

a) 7000/6500 B.P.; b) 4500 B.P.; c) 1500 B.P.; d) 1000/500 B.P.


As the delta advanced, the alluvial plain aggraded and the river flowed across the plain (Fig. 5.21:c). In addition to the Louros River, the streams flowing out from the mountains in the north and northeast also contributed sediments for estuarine infilling. By 1000 B.P. the shoreline had moved to the vicinity of Mt. Mavrovouni. Because the sediment supply from the rivers was not enough to develop an extensive floodplain, much of the area north of the embayment was left as swamp (Fig. 5.21:d).

Sometime between 1000 and 500 B.P., during the medieval period, the Louros River was diverted near Kastro Rogon and started flowing west along the foothills and then south along the west flank of the embayment (Fig. 5.21:d). Estuarine infill and sea regression had left the seaport far inland, separated by a wide swampy zone from the lagoon or embayment to the south. The channel diversion could have served two purposes: (1) draining swamps for agricultural use, and (2) establishing a transportation connection between Kastro Rogon and the Ambracian Gulf. The area along the northern and western flanks of the embayment was a logical route to dredge a channel as it was covered primarily by alluvial sediments. No doubt at least in part a result of this diversion, Kastro Rogon remained an important town during the medieval and post-medieval periods.

CONCLUSION

Subsurface stratigraphy and paleogeographic reconstruction have provided a picture of the changing landscape context of archaeological sites of the coastal zone of the Ambracian embayment. Around 10,000 B.P., the sea level had risen to about 45 m below current sea level and the Ionian Sea had intruded into the graben-like Ambracian embayment. After 6000 B.P.,

the rate of eustatic sea-level rise greatly slowed or ceased, but relative sea level continued to rise. By 4500 B.P. maximum marine transgression had occurred and the shoreline stood more than 12 km north of its current position. The entire embayment was flooded by the sea. This geomorphic configuration did not change significantly until the end of the Roman period when human-induced erosion increased sediment supply for estuarine infilling. By 1500 B.P. much of the fringe area in the embayment was exposed but remained swampy. The changing geomorphic configuration of the Ambracian embayment was critical to human exploitation of this region.

I98

CHAPTER 6

THE LOWER ACHERON RIVER

VALLEY: ANCIENT ACCOUNTS

AND THE CHANGING LANDSCAPE

by Mark R. Besonen, George (Rip) Rapp, and ZhichunJing

INTRODUCTION

Recognizing that the earth's coastal systems have undergone profound change since the end of the Pleistocene (about 10,000 years ago), the Nikopolis Project set as one of its objectives the interpretation and understanding of the changing geomorphology, topography, and paleoenvironments in the lower Acheron River valley from the middle Holocene through the present (Fig. 6.1).1 Archaeological remains in the valley are abundant, and literary and historical references go back at least to the 8th century B.C., when Homer and his contemporaries considered the Acheron to be an infernal river and held that the valley was an entrance to the Under- world (Od. 10.508-515).

Various other ancient literary and historical sources also make reference to the valley, and provide details of a landscape configuration that is inconsistent with the current physiography. The inconsistencies pose a problem for archaeologists trying to equate ruins in the valley with particular settlements mentioned in ancient accounts. Are these ancient authors mistaken in their descriptions of the valley, or can a natural sequence of landscape evolution account for these discrepancies? There are three conspicuous inconsistencies whose explanation and resolution have provided a focus for this component of the Nikopolis Project: 1) the size of the Glykys Limen (modern Phanari Bay); 2) the nature, geometry, and evolution of the Acherousian lake; and 3) the course of the Acheron River with respect to Kastri during the classical period.

1. This chapter is summarized and updated from Besonen 1997, a Masters thesis completed by the senior author at the University of Minnesota, Duluth. An electronic version of Besonen 1997 in Adobe Acrobat PDF format is freely available over the Internet at http:// www.paleoenvironment.org, or by requesting a copy from the author via e-mail ([email protected]).

THE SIZE OF THE GLYKYS LIMEN (MODERN PHANARI

BAY)

The small marine harbor located at the mouth of the Acheron River is known today as Phanari Bay (Fig. 6.2). Well protected by a series of high limestone cliffs, and continuously flushed out by the high discharge of the Acheron River and its tributaries, the bay has characteristics that make for an ideal marine harbor. Unfortunately, it is very small, measuring only 700 x 350 m, with a depth of less than 10 m. In ancient times, the embayment

M. R. BESONEN, G. RAPP, AND Z. JING

0 30 60 90km I..... I

was known as the Glykys Limen ("Sweet Harbor"). According to the Greek

geographer and historian Strabo (7.7.5 [C 324]), who lived through A.D. 21, this was because the influx of fresh water from the Acheron and its tributaries caused a dilution of the marine water filling the bay.

Strabo's account is not singular; many other ancient authors also mention the Glykys Limen, indicating that it was a well-known feature along the Epirote coastline. Three of these authors provide evidence for a dis-

crepancy between the ancient and modern landscape: while the modern harbor is quite small, the ancient harbor was apparently quite large. The

late-5th-century B.C. Greek historian Thucydides (1.46.1-5) wrote in his

Figure 6.1. Area map of Epirus

200

LOWER ACHERON RIVER VALLEY

Area Map of the Lower Acheron Valley

(I _ _1 2 3 kim

I ...... . - , - ; X X~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

Kicisoura

i .' Mot.zakaiika ;

! / ^

1 Nii: ,' :. ,,% 9 , ,Xirolophos I &:

"/ IF,,îgure 6.3

J Observation Point . ... '''--- '

\ c - .. } .

l /....-----. lThen.elon ..l Stav, rochorion ,

N \^ x \'. c. \ EPHYRA '- "

NDOSIA l, ; w E P FI Y R A! >_ ' . ,v ... of i. ohl. \ -" .? ^.." CICHYRUSV X NEK()OMNTI[ON' 'I

i Kastri

!.,' ,.,v '~

..........%Xylokastro r'--/

GI/.}ALSIIiJJ A*.\ '>r.A Xi^Â1nioudia D'N - / iVel' I)rontos l//.t. nl

" \ \W'- \ Skalaiatos

w ; -\^?^^ Acheron /i ,, v v ,E, v.<,, , Pountas

lonian - Vilanidorraehii .. r

Sea ^ X ,, < } i/ l,-/

(I .... . . 1' . -- I iTsotknida V Cd.s'Ksc'i.s.t ; ,Bie . .. ..

. . .. ... A BC iancienlt nalmli

a bc illoce' Id almCe _ ^ ^ ^ ^_ _ .. . . . . . . . . . .... ..... ................. ........... ............... .. ..

^,;Acherousia

. ,Narkissos

Kanal lakion

Figure 6.2. Area map of the lower Acheron valley

2. Hammond 1967, p. 69. 3. Dakaris 1971, p. 5. 4. Dakaris 1971, p. 5.

history of the Peloponnesian War that the Corinthians and their allies anchored 150 of their ships in the Glykys Limen before the Battle of Sybota in 433 B.C. Dio Cassius (50.12.2), another Greek historian and Roman official of the 2nd and 3rd centuries A.C., reported that in the summer of 31 B.C., Octavian moored 250 of his ships in the harbor a few days before his confrontation with Mark Anthony and Cleopatra in the Battle of Actium. Finally, Anna Komnena (Alexiad 4.3) recorded in the 12th cen-

tury A.C. that in A.D. 1081/1082, nearly 1,100 years after the Battle of Actium, the Norman Robert Guiscard and his large fleet wintered over at the Acheron delta. Modern Phanari Bay could not possibly accommodate such large naval fleets.

In his account of his travels through the region, the British historian Nicholas Hammond briefly suggested that the bay had silted up since ancient times.2 Sotirios Dakaris, an archaeologist who did extensive work in the area, addressed the topic more thoroughly. Motivated by the accounts

ofThucydides, Dio Cassius, and Anna Komnena, he supplied two further lines of geologic evidence that definitively indicate the harbor was once much larger. Dakaris noted the existence of a strip of ancient beach sand, similar to the white sand beach that surrounds Phanari Bay today, ca. 1.5 km east (inland) of the village of Ammoudia (Fig. 6.3).3 This strip of sand, in conjunction with "a boring near the confluence of the Cocytus and the Acheron [that] brought to light a layer of sand with sea shells at a depth of 17.5 m from the present surface,"4 provides unequivocal geologic evidence

0 1 . . I

201


Mesopotainon

.. :; ..

Tsouknida

* 7

Valanidorrachi

I A2mrnoid ia I'lanari Bav

,. .. X

that the Glykys Limen formerly extended further inland at some unknown Figure 6.3. View of concentric point in time. accretionary beach ridges surround-

point in tgme. Phanari Bay, looking south. This Dakaris's observations are significant, but they lack chronological con- photograph was taken from the

trol and thus cannot be used to verify the accuracy of the ancient literary bedrock highlands on the north side and historical accounts. They provide only a snapshot of the landscape of the valley; see Figure 6.2 for the configuration at an unknown moment in time, and do not afford the ar- location of the observation point. chaeologist an understanding of the changing landscape. Therefore, our The Acheron River is delineated by first objective was to develop a detailed picture and absolute chronology the faint dark band of trees visible in for the evolution of the Glykys Limen. the background. Photo M. Besonen

THE NATURE, GEOMETRY, AND EVOLUTION OF THE

ACHEROUSIAN LAKE

A second significant discrepancy between ancient references to the valley and the observable modern landscape concerns the nature, geometry, and evolution of the extinct Acherousian lake (Fig. 6.4). The existence of the lake is not in question, for its final swampy remnants persisted until just after the First World War, at which time they were drained and backfilled for agriculture.5 During Greek and Roman times, the lake was apparently a conspicuous feature given that many authors make reference to it (Thuc. 1.46.3-4; Pseudo-Scylax 30; Strab. 7.7.5 [C 324]; Plin., HN4.1.4; Livy 8.24; Paus. 1.17.5). By medieval times, it was referred to as the Acherousian swamp, apparently reflecting a natural infiUing.6 Though the number of references to the lake-swamp is significant, few provide any detailed topographic information that is useful in determining its location and nature.

Several modern authors have considered the existence of the lake in the valley. William Leake, who traveled through the region in 1809, left a fairly detailed description of the marshy valley bottom with its few, shallow, isolated pools.7 He concluded that the marsh-lake present below the hill of Kastri was the Acherousian swamp known from antiquity, seem- ingly not considering the possibility that it might previously have had a different nature or proportions (Fig. 6.4, upper left). Alfred Philippson and Ernst Kirsten presented a different scenario in their survey of the Greek landscape, suggesting that the swampy, marshy ground which rep-

5. Hammond 1967, p. 68. 6. Hammond 1967; Dakaris 1971. 7. Leake 1835, I, p. 232; IV, pp. 51-

54.

202

203

o,

0 0

o

.i

CD

-4


resented the lake had expanded areally, but become shallower, since ancient times.8 One of their maps shows a dotted outline of what is presumably the Acherousian lake (the Acheron River enters one side and exits the other). This lake stretches north from Kastri up the valley almost to the point where the Acheron River exits from the bedrock uplands that bound the valley to the east (Fig. 6.4, upper right).

By the time Hammond passed through the valley in the middle of the 20th century, the final remnants of the lake had been filled in. He indicated more definitive boundaries for the Acherousian lake based on ancient literary and historical references, the descriptions of Leake, and some earlier work by Dakaris.9 The boundaries he indicated are the Mesopo- tamon/Tsouknida valley constriction to the west, the bedrock highlands to the south, and the Pountas ridge and Kastri to the east (Fig. 6.4, lower left).

Dakaris presented the most careful consideration of the subject, bas- ing his theory on ancient literary and historical references as well as his own observations. His reconstruction of the lake's size and location is similar to that given by Hammond, but he extended the eastern boundary of the lake past the Pountas ridge and Kastri toward Kanallakion (Fig. 6.4, lower right).10 The basis for this eastward extension was the chance find of ten wooden beams during the excavation of a drainage canal east of Pountas ridge and southwest of Kanallakion. Dakaris interpreted these beams as part of the keel of an ancient boat that had once plied the lake; he also noted that a spot on the eastern side of Pountas ridge is still referred to as "Dromos Skalamatos," which means "port" or "place of embarkation" (Fig. 6.2).11

Dakaris, Hammond, and others based their reconstructions primarily on indirect evidence, but were also greatly influenced by their observations of the modern landscape in the valley. Their reconstructions overestimate the size of the lake at least as an open body of water, and lack definitive chronological control. They provide no information about the lake during pre-classical times, an item of interest to the Nikopolis Project. A complete and detailed chronology of the lake's development and evolution based on geologic evidence has never been prepared. Particularly important issues to resolve include when the lake came into existence, the mechanism by which this occurred, the nature of the lake, and its geometry and dimensions through time. These questions framed our second objective.

THE COURSE OF THE ACHERON RIVER WITH RESPECT

TO KASTRI DURING THE CLASSICAL PERIOD

The course of the Acheron River, like that of most rivers in their lower stretches, is constantly shifting. Our third objective was to determine the location of the course of the Acheron River with respect to the hillock Kastri during the first millennium B.C. (Fig. 6.2). This is particularly important to help resolve the long-standing problem of identifying the ruins 8. Philippson and Kirsten 1956, II, on that hillock with those of Pandosia, a fortified urban settlement often 9 Hammond 1967, p. 69 referenced in ancient literary and historical sources (Dem. 7.32;Justin 12.2; 10. Dakaris 1971, pp. 4-5 and fig. 7. Livy 8.24; Plin. HN 4.1.4; Strab. 7.7.5 [C 324]). The major discrepancy 11. Dakaris 1971, p. 57.

204


frustrating this identification has been that Kastri is located to the north of the Acheron River, but ancient sources indicate Pandosia was located to its south (Dem. 7.32; Strab. 7.7.5 [C 324]); hence the difficulty in equat- ing the two.

In ancient times, the Acheron River served as a political boundary dividing the territory of Thesprotia to the north from the territory of

Cassopaia to the south (Fig. 6.2). Ancient sources indicate that, in addition to Pandosia, a fortified urban settlement named Ephyra also existed in the valley. Geographic references situate Ephyra north of the Acheron River in the territory of Thesprotia, close to the sea, and near the Ache- rousian lake (Paus. 1.17.4-5; Strab. 7.7.5 [C 324];Thuc. 1.46.4). Pandosia, in turn, was located further inland, and within the territory of Cassopaia (Dem. 7.32; Strab. 7.7.5 [C 324]).

Besides the ruins at Kastri, the remains of a second fortified urban settlement can be found in the valley today. This second site is located just north of Mesopotamon on the ridge known as Xylokastro (Fig. 6.2). Since the Xylokastro site is closer to the sea, and the Kastri site is further inland, one might immediately suggest the first site should be identified with Ephyra, and the second site with Pandosia. Combined with simple differences of opinion, however, the issue of the river course has prevented consensus about the identification of the ruins in the valley. For example, Hammond did not consider the Kastri/Pandosia identification as appropriate, suggesting instead that the ruins at Gourana, much further upvalley, were actually those of Pandosia.12 Dakaris did prefer the Kastri/Pandosia identification, and he reconciled the issue of the river course by suggesting that the river had shifted since ancient times.

Wherever the river banks are not supported, or when the river overflows, it could result in a change in course.... The slight inclination of the Acheron plain, the swamps, and the lake, formed by the river to the south of Kastri hill, contributed to the change in the river bed, which, in ancient times, had the hill with the ruins to its south, at [sic] Cassopaia.13

While Dakaris's suggestions concerning the dynamic nature of the river course are correct, he did not provide any geologic evidence to show that the river had indeed shifted its course from the northern to the southern side of Kastri since the first millennium B.C. Hence, our third objective was to examine the changing course of the Acheron River with respect to Kastri during the past 2,000 years, and either confirm or deny the shift proposed by Dakaris.

GEOLOGY AND THE NEOTECTONICS OF THE ACHERON VALLEY AREA

12. Hammond 1967, p. 478. Jean Aubouin presented two detailed studies interpreting the stratigraphy 13. Dakaris 1971, pp. 136-137. and tectonics of Epirus which have served as the foundation for subse- 14. Aubouin 1959 and 1965. quent work.14 These studies were followed by Etude g6ologique,15 another 15. Etude geologique. major monograph on the geology of Epirus, which resulted from petro-

205


leum exploration work. The major features of the landscape of Epirus are structural in origin. The region consists of a series of north-northwest/ south-southeast trending folds and thrust fault blocks that have been formed in a sequence of compressional orogenic events since the Late Jurassic period.16 These folds and fault blocks form a series of parallel limestone mountain ranges with intervening flysch basins that can be delineated in a satellite image of the region (Fig. 6.5). Some of the ranges reach over 2,000 m in elevation, but on average range from 1,200 to 1,700 m.17 The marked and varied relief noted between the ranges and basins is a direct function of the structure and contrasting lithologic properties of the limestone and flysch.'8 Relief is even more spectacular along the coasts, where bedrock cliffs rise directly from the sea, or very flat coastal river plains give way in abrupt topographic discontinuity to carbonate bedrock valley walls.

A map of the simplified geology of the lower Acheron valley is shown in Figure 6.6. Recent alluvium floors the flat valley bottom, which is flanked

by the steep, carbonate bedrock valley walls. These valley walls are com-

posed, for the most part, of Mesozoic and some Eocene limestones. The limestones are cherty, range from fine-grained to sublithographic, are usually fossiliferous (with the remains of calcareous algae, radiolarians, rudist clams, ammonite cephalopods, and globigerinid and other foraminifera), and in places are dolomitized and/or brecciated. Upper Eocene to Lower Miocene (Aquitanien) flysch crops out at the base of the eastern valley wall. The flysch is composed primarily of alternating soft micaceous sandstones and shales with intercalated, thinly bedded biogenic limestones and

Figure 6.5. Satellite image of Epirus. North of the Acheron valley, the structural configuration of the region is delineated by a series of parallel limestone mountain ranges (trending north-northwest to south-southeast) with intervening flysch basins. Note the elongate geometry of the Thyamis and Arachthos river deltas with fringing delta top/front "barrier" beaches.

16. Etude g6ologique. 17. King, Sturdy, and Whitney

1993. 18. Etude g6ologique.

l--IC I - - l

206


Quaternary all uviuLtl

scree slopes and talus cones %% )

U

thrust fault--teeth on ovcrthrust block

nornmal faull--U (Up) and D (down) show relative motion of blocks

Plioccnc Arkhangclos Formation-m-ixcd marine and continental conglomerates, muddy sands, and lignilic and marine shalcs

Lower Miocene (Aquitanien) flysch--soft, micaccous sandstones and shales with intercalatcd thinly-beddcd biogenic limestones and zmails near top

Mesozoic and 1 locene carbonates--cherty. fossilifierous, fine-grained to sublithographic limestones; occasionally dolomitic or brccciatcd

microcrystallinc gypsum--age uncertain

Qal

SC

(:i;iii ii!iii:iii ii;

Figure 6.6. Simplified geology of the lower Acheron valley

207

'X Nt-4


marls near the top. The top of this flysch unit effectively marks a large shallow thrust fault over which the more competent Mesozoic limestone has ridden to create one of the limestone ranges seen in Figure 6.5. Recent talus and scree slopes cover the contact and most of the flysch unit. A small strip of the Pliocene Arkhangelos Formation crops out in the southern valley wall to the east of Pountas ridge. This formation is a mixed marine and continental unit that consists of conglomerates, muddy sands, and lignitic and marine shales. Finally, an inferred active, east/west-trend-

ing normal fault exists along the south valley wall,19 though this fault was not recognized by earlier work.20

While Etude geologique is comprehensive through the Pliocene, the tectonic history of the Pleistocene and more recent periods was not included. Fortunately, a recent dissertation by David Waters fills the gap.21 Waters provides an inventory of geologic evidence (e.g., incised river gorges, wave-cut notches, and raised shell burrows) that suggests mainland Epirus and much of the coast has been undergoing uplift since the Pliocene. At the same time, the evidence suggests that certain areas, such as the Ambra- cian Gulf, lower Acheron valley, and lower Thyamis valley (Figs. 6.1, 6.5), are subsiding; very thick deposits of Quaternary sediments are found at these locations. Subsidence also seems to be occurring along the northwest coast of the mainland opposite Corfu, a process indicated by the steep, rocky shorelines (with numerous small coves and islets) and the lack of beach platforms.22

Waters attributes modern subsidence of the lower Acheron valley bottom to movement on the inferred active normal fault along the southern valley wall (hanging wall to the north) (Fig. 6.6).23 The existence of this fault would make the valley configuration that of a half-graben, though Waters never explicitly describes it as such. While the alluvial valley bottom appears to be subsiding, there is some evidence to indicate uplift of the carbonate valley walls; Waters identified a wave-cut platform, 1.7 masl, on the north side of Phanari Bay.

HOLOCENE RELATIVE SEA LEVEL IN THE REGION OF EPIRUS

Coastal evolution is intimately tied to relative sea level, itself determined

by eustatic sea-level changes, isostasy, and vertical tectonic movements. A record of relative sea-level change for a particular region must be compiled from local evidence. A relative sea-level curve for the southwestern Epirote coast is shown in Figure 3.24. Particularly important to note is that, during the last 5,000 years, relative sea-level rise along the southwestern Epirote coast has been less than 2 m. The rate of sedimentation at river mouths, however, has been much greater. A significant implication of this relationship is that the physical sedimentology and microfossil assemblages contained in the stratigraphy are more important indicators for reconstructing shoreline position than the local sea-level curve.

19. Waters 1994, figs. 5.7, 5.10. 20. Cf. Etude geologique. 21. Waters 1994. 22. Waters 1994, p. 197. 23. Waters 1994, figs. 5.7, 5.10.

208


FIELD AND LABORATORY METHODS

24. Folk 1980. 25. Colors were described using the

Munsell Soil Color Chart (rev. ed. 1994).

26. Dean 1974. 27. Folk 1980.

Twenty-eight sediment cores were retrieved from various points in the lower Acheron valley during summer field seasons from 1992 to 1994 (Fig. 6.7; Appendix). All cores were retrieved by a hand-operated, 3 cm diameter Eijkelkamp gouge auger with the exception of cores 94-02 and 94-03, which were taken with a 7 cm diameter Edelman auger bit. Cores were described and logged on site using terminology following Folk.24 Field- observable parameters that were recorded include lithology and approximate grain-size distribution, color when wet,25 sediment consistency, plant and animal macrofossils, pedogenic characteristics (structure, sesquisoxide/reduction mottling, and calcium carbonate filaments or nodules), and chance finds such as pottery fragments. Sediment samples were collected for laboratory analysis in the U.S. with approximately 300 taken during the 1994 field season, and a much smaller number during the 1992 and 1993 seasons.

Laboratory analyses were focused mostly on sediment samples from the 1994 season. All 1994 samples were analyzed for dual-frequency magnetic susceptibility and anhysteretic magnetization along their length us-

ing facilities at the Limnologic Research Center and Institute for Rock Magnetism at the University of Minnesota,Twin Cities. Microfossil assemblages were determined in fifty samples, from ten different cores, all but one of which was collected during the 1994 season. In most cases, the total microfossil population, including ostracods, foraminifera, gastropods, pelecypods, and charophyte oogonia, was picked and identified. Relative percentages of fresh and brackish water microfossils were calculated from species counts to provide an approximation of the salinity of the environment of deposition. Eight cores from the 1994 season were analyzed along their length for organic carbon and carbonate using the method of Dean.26 Grain-size distribution by pipette analysis was determined for twenty- three samples using the method of Folk27 to supplement the field-based approximation of grain size. Eight samples of organic material were radiocarbon dated by the accelerator mass spectrometer (AMS) method at either the Radiocarbon Laboratory at the University of California, River- side, or Beta Analytic Laboratories Inc. of Miami, Florida (Table 6.1).

Results from field observations and laboratory analyses were examined together to determine the probable environments of deposition for each lithostratigraphic unit. A summary of these data for each sediment core can be found in the appendix to this chapter, with the full comple- ment of primary data available in Besonen 1997. A study of early maps of the area, as well as literary and historical references by both ancient and more recent authors (in particular, Homer, Thucydides, Strabo, Anna Komnena, and Leake), was undertaken to supplement and provide a context for the geological data. Finally, three cross sections through the valley were drawn (see Figs. 6.9-6.11), and reconstructions showing the evolution of the landscape in the valley at eight points in time during the past 5,000 years were constructed (see Figs. 6.12-6.15).

209

TABLE 6.1. RADIOCARBON DATES FROM THE ACHERON RIVER VALLEY

Depth below 13C/2C Conventional 14C CalibratedAge LabNo.a Core Surface (m) Material Ratio Age (B.) (s.e.J)b

UCR-3217 NC92-20 5.30-5.40 charcoal, root fragments n/a 2470 ?60 2650 +70/-290 UCR-2695 NC93-18 0.70-0.75 peat with organic material -27.18%o 2890 + 40 2980 +90/-30 UCR-2696 NC93-19 7.00-7.20 peat with wood fragments -20.36%oo 4520 +?60 5140 +160/-100 UCR-2697 NC93-21 5.80-6.10 peat with organic material -28.24%o 3460 ? 60 3690 +140/-60 Beta-80531 NC94-04 2.95-3.00 wood -26.00oo 340 ?50 380 +90/-70 Beta-80532 NC94-13 5.35-5.40 plant material -23.3%o 950 ? 50 850 +80/-60 Beta-80533 NC94-20 6.05-6.20 wood -39.9%oo 1740+60 1670 +40/-120 Beta-80534 NC94-23 10.35-10.55 plant material -27.0%o 3700 +?60 4030 ?100

aDating laboratory: UCR = University of California, Riverside; Beta = Beta Analytic Laboratories Inc. of Miami, Florida.

bCalibration from conventional 14C age to calendar years was performed using the CALIB Rev. 3.0.3c computer program available from M. Stuiver and P. Reimer of the Quaternary Research Center at the University of Washington, Seattle. All

options were set at their default values. The data set used to make the calibrations was the INT93CAL bidecadal

dendrochronologic calibration curve. A decadal calibration is also available, but is meant for use with high-precision dates

(a<40 years).

210 M. R. BESONEN, G. RAPP, AND Z. JING

Figure 6.7. Core locations in the lower Acheron valley


Figure 6.8. Topographic map of the lower Acheron valley bottom. Contour lines may not be continuous at valley edges where they are compressed. Sediment core locations are marked by black dots; see Figure 6.7 for labels. Heavy black lines show the locations of cross sections illustrated in Figures 6.9, 6.10, and 6.11.

28. Neale 1964; Phleger 1960.

Topographic control for all elevations cited in this study was provided by detailed 1:5,000 topographic maps produced by IGME (the Greek In- stitute of Geology and Mineral Exploration) in 1981. These maps are contoured at 1-m intervals over the flat valley bottom, but shift to 4-m intervals for the steep bedrock valley walls. The maps also record hundreds of

individually surveyed point elevations throughout the valley bottom where

topography is slight. Figure 6.8 presents a very reduced set of this topographic data, contoured at 2-m intervals to simplify presentation.

MICROFOSSIL ASSEMBLAGES AND RELATED ECOLOGY

Ostracoda and foraminifera have been used with great success as indicators of paleoenvironments in marginal marine systems because they are

extremely sensitive to salinity and temperature, among other factors.28 We examined the microfossil assemblages in fifty sediment samples, paying particular attention to ostracods and foraminifera for paleosalinity information. Identification of the ostracods was achieved down to the species level for twenty-four forms, down to the genus level for one form, and left undetermined for one form. Identification of the foraminifera was less rigorous: down to the species level for three forms, to the genus level for four forms, and to the family level for one larger, well-known group.

211


Reference works based generally on Mediterranean localities or similar marginal marine settings were used to identify the microfossils and to

gather information about their ecologic and environmental preferences.29 This allowed us to define two microfossil assemblages indicative of

paleoenvironments with differing salinities. The first microfossil assem-

blage is characteristic of shallow, freshwater environments, while the second assemblage is indicative of shallow, nearshore, brackish to marine waterconditions. Microfossil identifications and paleoecological interpretations based on the assemblages were later confirmed and corrected by micropaleontologist Frederick Swain of the University of Minnesota, Twin Cities.30

Whereas microfossils were sparse or absent in some freshwater samples, all shallow, nearshore brackish to marine sediments showed high total abun- dances. Certain forms were present in nearly all samples from brackish to marine deposits, and in some cases occurred in extreme abundance.

The shallow, freshwater assemblage is comprised of fourteen ostracod species: Candona albicans, Candona cf. caudata, Candona compressa, Candona cf. lactea, Candona neglecta, Candona truncata, Cyclocypris cf. laevis, Darwinula stevensoni, Herpetocypris cf. reptans, Ilyocypris gibba, Limnocythere cf. inopinata, Limnocythere sp., Potamocypris cf. villosa, and Ostracod sp. A

(possibly Prionocypris zenkeri). Twelve ostracod species and a series of foraminifera characterize the

shallow, nearshore, brackish to marine water assemblage: Cushmanidea elon- gata, Cyprideis torosa, Cytheridea neapolitana, Cytheridea cf. sorbyana, Cyth- eromorphafuscata, Leptocythere bacescoi, Leptocythere cf. castanea, Loxoconcha

elliptica, Loxoconcha cf. granulata, Loxoconcha ovulata, Paracytherois cf. acuminata, and Tyrrhenocythere amnicola. The foraminifera in the assemblage include Ammonia beccarii, Bolivina sp., Bulimina sp., Cribrononion translucens, Elphidium crispum, Fursenkoina sp., and members of the family Miliolidae (including several Quinqueloculina spp. and Triloculina sp.).

While mixing of fresh and brackish to marine assemblages may be significant, especially in regions with large tidal fluxes,31 it was minimal in the sediment samples from the Acheron valley. This is not unexpected given the small tidal variation (20 cm) in the region. Cathleen Villas noted some mixing of marine microfossils in the freshwater environments of the Acheloos delta, just 150 km to the south of the Acheron valley,32 but the Acheloos delta plain is totally unprotected and experiences the unbuffered assault of storm waves. This is not an issue in the Acheron valley where Phanari Bay is enclosed and well sheltered by the large carbonate cliffs.

SEDIMENTOLOGY AND ENVIRONMENTS OF DEPOSITION

The modern sedimentary environments in the lower Acheron River valley are very similar to those found at other spots along the Greek coast.33 For simplicity, we divide them into two broad depositional systems. The first system, herein referred to as the fluvial depositional system, consists of all the sedimentary environments landward of the shoreline. Six distinct environments can be identified: river channel, natural levee, crevasse splay,

29. Ascoli 1964; Bhatia 1968; Devoto 1965; Ellis and Messina 1952- 2000; Puri, Bonaduce, and Gervasio 1969; Puri, Bonaduce, and Malloy 1964; Sars 1928; Tassos 1975; Tziavos 1977; Villas 1983; Wagner 1957; Yang 1982.

30. A complete summary of this work, including scanning electron microscope plates of the microfossils encountered in the Acheron valley, is freely available online; see note 1 regarding the availability of Besonen 1997.

31. Kilenyi 1969. 32. Villas 1983, p. 54. 33. Tziavos 1977; Villas 1983.

212


34. Middleton 1973. 35. See Chapter 5 for a short

discussion of Walther's Law.

floodplain, backswamp, and shallow freshwater lake. The second depositional system is a deltaic nearshore association, composed of the environments located seaward of the shoreline and within the marine embayment of the Glykys Limen. Eight distinct environments can be identified in this system: fresh to brackish water delta top marsh, delta distributary channel, distributary channel mouth bar, subaqueous levee, lower delta front, prodelta, interdistributary bay, and accretionary beach.

On the modern landscape, many of these environments grade into one another laterally, making it difficult to place exact boundaries between them. This difficulty is further magnified when attempting to reconstruct paleoenvironments based on a finite number of 3-cm diameter sediment cores. However, Walther's Law of the correlation of facies34 provides a powerful tool to interpret the subsurface stratigraphy.35 This is especially true in a marginal marine environment like the lower Acheron valley, when work is grounded in physical sedimentology and the analysis of microfossil assemblages. What follows is a brief sedimentological and geomorphic description for each of the fourteen environments of deposition mentioned above. We begin with the environments of the fluvial depositional system, and then discuss those of the deltaic nearshore association.

River channel deposits are composed of the coarsest sediments found in the fluvial depositional system, and include lag deposits and bars that form directly in the river channel. Deposits are usually tan or buff in color, but may exhibit a reduced color if trapped in an environment such as an oxbow lake. In the lowest reaches of the valley, where the river channel and bar system grades into the deltaic environment, sands and gravels may also have a gray color. Ostracods and other microorganisms do not generally inhabit such environments, and the occasional carapaces of detrital origin that do make it to the river channel are quickly destroyed in the high- energy environment, or are diluted in the abundant clastic material. Re- worked microfossils from the local bedrock occur in extreme abundance in deposits of this environment since it is the main transport agent for such material.

Subaerial natural levees are wedge-shaped ridges of sand and muddy sand that are deposited directly adjacent to a river along its length, and thin away from the river. These deposits may form significant topographic highs, and are created when coarse sediments carried overbank by a flooding river are dropped out of suspension. They are generally finer-grained than channel deposits, become increasingly fine away from the river channel, and eventually grade into floodplain or backswamp. Because these deposits are exposed subaerially, they tend to be tan to orange to brown in color, and may exhibit weak pedogenic features such as sesquisoxide mottles and nodules, and carbonate filaments and small nodules. Occasional ostracod carapaces of detrital origin and abundant reworked microfossils lib- erated from the local bedrock are found in these deposits. Natural levees have played a significant role as agents of geomorphic evolution in the lower Acheron valley through the middle and late Holocene, and their significance will be discussed below.

Crevasse splay deposits form during periods of exceptional flooding, when channels are cut through the natural levee system allowing water and bedload sediment to escape onto the adjacent floodplain or into

2I3


backswamp or interdistributary bay environments. They occur as lobe-

shaped wedges of sand- to mud-sized sediment that thin away from the river channel. A modern lobe-shaped crevasse splay deposit, delineated by the 2-m contour line, can be seen in the floodplain to the southwest of ancient Ephyra (Fig. 6.8). Although these deposits are similar in composition to natural levee deposits, they can be distinguished by their geometry and by the fact that they appear as abrupt pulses of coarser sediment within the mud and silt of floodplain, backswamps, or interdistributary bays. Such

deposits contain abundant reworked microfossils from the local bedrock, and relatively small amounts of organic matter.

The floodplain environment consists of the flat, low ground adjacent to a river channel and natural levee system that acts as a settling basin for fine-grained suspended sediment carried over the river's bank during flooding. Floodplain deposits consist mostly of silt and clay with occasional fine sand laminae. This environment is exposed subaerially, so its sediments tend to be tan to orange to brown in color, exhibit slight to moderate pedogenic development, and tend to be more compact and stiffer than sediments from other environments. Occasional modern ostracod carapaces of detrital origin, common fragments of terrestrial gastropod shells, and abundant reworked microfossils from the local bedrock are found in such deposits. These deposits contain moderate amounts of

organic carbon. The backswamp environment represents a transitional step between

floodplain and shallow lake environments. It commonly occurs in low, poorly drained areas adjacent to the river channel or valley walls, and consists of nearly perennially saturated swampy and marshy ground. In the Acheron valley, it also occurs in the low swales between the spectacular accretionary beach ridges east and northeast of Phanari Bay (Fig. 6.3). Backswamp deposits are composed of dark gray to brown, organic-rich mud and clays, though sandy intervals may be present depending on the proximity of the river channel. In some cases, vegetation is so abundant that the backswamp is essentially a freshwater marsh, and deposits consist of peat and peaty mud. Such deposits are composed of up to 25% (by weight) organic carbon. Members of the freshwater ostracod genus Candona occur in common to abundant quantities in backswamp deposits, while other freshwater forms occur in lesser quantities.

Shallow freshwater lakes and pools are no longer present in the lower Acheron valley because they have been filled in for agriculture, but they occupied a significant portion of the valley bottom in the past. These lakes are commonly transitional with backswamp and marsh environments. Deposits from such lakes are generally gray in color, and extremely rich in clay-sized particles; they have a moderate organic content, ranging from 3 to 8% by weight. Microfossils present in these deposits include relatively sparse numbers of freshwater ostracods and gastropods. Microfossil abundance increases when the deposit is transitional with backswamp and marsh deposits, and this is probably the result of the greater organic content (food supply) of shallower environments. The most significant mechanism for the creation of these shallow lakes in the lower Acheron valley involves the impingement of a river channel and levee system against the bedrock valley walls, as described below. Oxbow lakes, which are very common in

214


other coastal river plain localities,36 are infrequent in the lower Acheron

valley at the present day. The only example that currently exists is the horse-

shoe-shaped loop immediately north of the Acheron River, ca. 1.25 km to the east-southeast of Phanari Bay (Figs. 6.2, 6.8).

The fresh to brackish water delta top marsh is a thick accumulation of reeds and marsh grasses fringing the shoreline on the delta top, such as that which exists at present on top of the Acheron delta to the south and southeast of Phanari Bay. The marsh is situated at approximately sea level and receives input of water and sediment from the fluvial and marine systems. Deposits consist of peat and peaty mud with occasional sand layers, and are composed of up to 25% (by weight) organic matter. The microfossil assemblages in delta top marsh deposits grade upward from extremely abundant shallow brackish water forms (especially Cyprideis torosa, Lepto- cythere cf. castanea, Loxoconcha elliptica, Ammonia beccarii, and Cribrononion translucens) to abundant freshwater forms. This distribution of microfossils reflects its location at the transition from the marine to freshwater

system during an overall regressive sedimentary regime. The delta distributary channel, distributary mouth bar, and subaque-

ous levee are active delta front environments within the marine embayment where the majority of deposition and delta progradation occurs. All three environments are essentially subaqueous continuations of the subaerial fluvial channel and natural levee environments. The coarsest sediments in the system are generally sands and sandy gravels that floor the delta distributary channel. Subaqueous levees border the distributary channel and are composed mostly of sand and silt. As currents in the distributary channel lose competence, sediment is dropped out of suspension and forms a broad sandy apron around the distributary known as the distributary mouth bar. All recognized active delta front deposits from the lower Acheron valley are gray to dark gray in color; however, Villas reports that both gray and tan components exist in the subaqueous levee deposits of the Acheloos River.37 Microfossils present in such deposits consist primarily of abundant numbers of brackish to marine water organisms, and abundant reworked microfossils from the local bedrock that were carried to the delta by the fluvial system. Deposits from these environments grade basinward into the laminated clays, muds, and fine sands of the lower delta front and

prodelta, and laterally into the interdistributary bay environment. The lower delta front and prodelta environments are located basinward

of the active delta front, and act as a settling basin for suspended sediment. Deposits from both environments consist of gray to dark gray laminated clays, muds, and fine sands, but the sediments of the lower delta front are noticeably coarser since they are a distal extension of the active delta front. Deposits from these environments have a low to moderate organic carbon content (3-4% by weight), and their microfossil assemblages consist strictly of abundant brackish to marine water organisms without any freshwater forms.

The interdistributary bay is a shallow open body of water located to the side or partially behind the active delta front. At present, there are no interdistributary bays on top of the Acheron delta because Phanari Bay is

36. Russell 1954; Villas 1983. almost entirely filled in, but such bays did exist in the past. Deposits from 37. Villas 1983, p. 75. this environment are composed of gray to dark gray silts and clays that

2I5


settle out of suspension, and sandy material washed in over the natural levees surrounding the fluvial distributary channels of the delta top. Cre- vasse splay deposits are also commonly found interbedded in deposits of interdistributary bays. Delta top marshes, which surround these bays, contribute to their moderate to high organic carbon content (4-8% by weight). Deposits from this environment also contain extremely abundant brackish to marine water microfossil assemblages, as well as reworked microfossils from the local bedrock in common to abundant quantities.

A spectacular series of concentric accretionary beach ridges and intervening swales surrounds modern Phanari Bay (Fig. 6.3). The Acheron delta top and front provide a constant source of sandy sediment that is reworked by normal wave activity, and then gently piled up over the regular wave base by spring and winter storm waves. Longshore currents that could keep the system in equilibrium by removing excess sand do not exist or are very weak because Phanari Bay is so well sheltered. As a result, these ancient beach ridges have accreted one by one, continuously decreasing the size of Phanari Bay. The sands that comprise these ridges are generally coarse-grained with occasional small pebbles, and are tan to buff in color. The intervening swales are flooded seasonally because of their low elevation, and often accumulate backswamp and marshy deposits. This beach ridge and swale environment is laterally transitional with the delta top and delta front environments.

MIDDLE AND LATE HOLOCENE GEOMORPHIC EVOLUTION OF THE GLYKYS LIMEN

We have documented the relative sequence of geomorphic evolution indicated by subsurface stratigraphy, and supplemented this with eight radiocarbon dates which provide absolute chronological control. Overall, the middle and late Holocene sedimentary record in the valley is regressive in nature reflecting alluviation during a period of very slowly rising relative sea level. Dakaris suggested that the Glykys Limen was formerly much larger, extending back to near the Mesopotamon/Tsouknida valley constriction at "a certain geological period."38 The suggestion was based on his observation of a fossil beach ridge 1.5 km east of the village of Ammoudia (on Phanari Bay), and the presence of fossil marine macrofauna encountered in a boring near the confluence of the Acheron and Vouvos Rivers. The Mesopotamon/Tsouknida valley constriction is a natural obstruction in the valley both areally and in the subsurface because of shallow bedrock (Figs. 6.9, 6.10), and logically might have served as a natural boundary to transgressing Holocene seas. Our results indicate, however, that marine influence reached even further inland than Dakaris suggested; rising Ho- locene seas stretched at least to the location of core 94-17 (Fig. 6.7), several hundred meters east of the valley constriction, around 2100 B.C. (see Fig. 6.12). Several radiocarbon dates provide absolute chronological control for this and other shoreline positions during the past 4,000 years.

The reconstructed shorelines we present should be taken only as generalized locations of the shoreline position. Because wave and tidal energy 38. Dakaris 1971, p. 5.

216


are low along the Epirote coast, and even lower in well-protected Phanari Bay, the Acheron delta is dominated by fluvial processes. Such fluvially dominated deltas display an elongate geometry, space permitting. Unfor- tunately, since Phanari Bay is almost entirely filled in, this elongate geometry is not apparent at present. It can be seen, however, in subsurface cross section C-C' (Fig. 6.11), as well as in other river deltas in Epirus, such as those of the Thyamis and Arachthos Rivers (Fig. 6.5). Additionally, the Acheron delta may have had multiple distributary channels. These ambi- guities preclude the possibility of reconstructing the exact shoreline configuration at any moment in time.

Cores 94-17 and 94-23 (Figs. 6.7,6.10; Appendix) have a similar stratigraphy and clearly illustrate the overall regressive nature of the sediments laid down in the lower Acheron valley during the middle and late Ho- locene. Both cores consist of deposits from the following environments given in normal stratigraphic order: 1) delta top to front, 2) brackish water delta top marsh grading upward into freshwater marsh, 3) shallow freshwater lake, and 4) floodplain. A radiocarbon date on peat from the bottom of the brackish water delta top marsh of core 94-23 returns a calibrated lo range of ages from 4030 +?100 B.P., or 2080 +?100 B.C. This peat belongs to the extensive subsurface delta top marsh deposit seen in the A-A' and B- B' cross sections through the Mesopotamon/Tsouknida valley constriction (Figs. 6.9, 6.10). The radiocarbon date from core 94-23 indicates that the interface between the delta top to front and brackish water delta top marsh environments found today at Phanari Bay has migrated at least 5.3 km seaward at the expense of the Glykys Limen since approximately 2100 B.C. (Fig. 6.12).

Brackish water conditions also existed at the locality of core 94-17, which is located further inland, ca. 5.7 km from Phanari Bay. A radiocarbon date was not obtained from this core, but it is reasonable to assume that the base of the delta top marsh is approximately the same age or slightly older than that of core 94-23. The maximum post-glacial extent of the marine embayment is not known because only a few relatively shallow cores are available east of cores 94-17 and 94-23.

Core 93-21 (Figs. 6.7, 6.9; Appendix) shows a basal stratigraphy that is similar to cores 94-17 and 94-23, and provides another radiocarbon date that further helps to constrain the position of the ancient shoreline. Delta top and front sediments directly overlie bedrock, and are in turn succeeded by a delta top marsh environment. However, since 93-21 is located ca. 0.5 km to the west of the other two cores, within the Mesopo- tamon/Tsouknida valley constriction where several fluvial systems coa- lesce, subaerial fluvial deposits dominate the stratigraphy above the delta top marsh. A radiocarbon date from the delta top marsh peat returns a calibrated la range of ages from 3690 +140/-60 B.P. (1740 +140/-60 B.C.).

This age is approximately 350 years younger than the 14C date from core 94-23, which is appropriate given that core 93-21 is closer to the modern shoreline.

West of the Mesopotamon/Tsouknida valley constriction, both geologic evidence and historical documents provide information about the changing size of the Glykys Limen. Core 94-13 (Figs. 6.7, 6.10;

2I7

o00

North South A A'

1200 - P 12)0

1100 - - j

Mesopotamon ridge Tsukid ridge I I 00 _ / o Tsouknida ridge - - I lOO

I000 - - 1000

t 900 - Acheron - 9(0 c

8 800 -- / River /- 8o . 700 - - 700 >

600 -- , - 600 o _. N;C-94-20 NC-94-12 NC-(93-21

500 - - .........00

- 400 - 3 | / fluvial cihannel - 400) c

4 M and subaerial o

;' 300 - fluvial channel and floodplai natu levee -- 300 < > - a subacrial natural Icvcc -. o 200 200 5

al,? ? _n<*:$backswamp , 100 - - t ('e-.1. 4: . . /- 1670 - 1 E lu K Cal. C-14: 1670 ............... -- 100 . C) noo _~ -40/-120BP \ ?:-40 120 liP 1c4 .' . iuvial channel l 0

- .>' """""- - -140 -601BP 1

O^ ~ ~ ~ ~ ~ ~ ~ < -100 ------ ~'":"~-=- delta top - I -10

-200 - - 7arsh 200 ( fluvio-deltaic fill / \ -300 -- [ - - -300

-400 - | I delta ti-ont distributary A \ delta - I j channel, channel mouth \ distributary

I T T bar. and subaqueous levee delta top and friont and channel interdistributlar bay

0 100 200 300 111 . . 50(x vertical exaggeration(

Figure 6.9. North-south cross section through the Mesopotamon/Tsouknida valley constriction

2I9

Elevation above sea level in centimeters

IC i \ C

C5

i . "0

K, I

! c; ;> sW \' ^ - ." .

0

>, . R < 0a- '::: , d

i . Cit::: : Ei (

o 1 Et i o CD

-j C D \ :......... ""

mi : ----- B~ ,,; =

::' , ,

tri i r I_ :

s 2 SD I^R'> ? ?C

Ill ^ c d CCo

*C 0 0 -' 0 0 C) 00 C 00 0 0

C st l rCt 1 - vC oc

SJOa9LUHUU3 LUl [AOl B1S 3AOqb UOl}3A31j

Northeast

C 1200 -

1100 -

1000 -

900 -

800 -

700 -

600 -

500 -

400 - NC-

300 -

200 -

100 - ...

0 --

-100 -

-200- -

-300- .

-400- -:

Soutt

NC-93-1 7

floodplain

. -

0 300 600 900m ' '' _ .

nearshore (reworked)

accretionary (reworked) beach

NC- 92-16

I.. . . .............. -cNC.'-93-19v_ X, ..... '-- .--- - ----- . . .. ..i

el I floodplain erial' J

-. g__,~~,~~ ~ _

_~d,, ,`

. # ~~

I-

hwest -"i

- 1200

- 1100

- 1000

- 900

- 800

- 700

- 600

- 500

)- 400

- 300

- 200

- 100

0

-100

- -200

-300

- -400

shallow marine embayment a. C-14 5140 (Glykys Limen) + 60/- oo00 BP

500x vertical exaggeration

Figure 6.11. Northeast-southwest cross section through the valley bottom (area of former marine embayment)

0

cfl

I-. C) C)

P. C) C) P.

C)

-0

0

C)

D

crJ

EC

0

co CD CA

CD

S CD

ID CD

.. ..... ...


Figure 6.12. Paleogeographic ( ? . < ^ i V \

~

reconstructions of the lower Acheron \ \? valley for 2100 B.C. and the 8th

century B.C. Small black squares mark core locations; see Figure 6.7 for labels.

Appendix), ca. 3.5 km from modern Phanari Bay, is composed from the base upward of shallow marine deposits of the Glykys Limen which are overlain by delta front sediments. The delta front sediments grade upward into deposits of a distributary mouth bar, and then an interdistributary bay. The sequence is capped by subaerial fluvial sediments. A marsh reed retrieved from the distributary mouth bar deposit was radiocarbon dated and returns a calibrated la range of ages from 850 +80/ -60 B.P. (A.D. 1100 -80/+60). The vertical sequence in this core indicates that it is not directly in front of the prograding delta, but on its flank.

22I


Therefore, it is not appropriate to use this as an indicator of the actual delta front position, which would have been somewhat seaward of this location. A hypothetical delta front position for this time is illustrated in

Figure 6.14. Several historical documents, in particular early maps of the region,

provide information that helps to reconstruct the evolution of the Glykys Limen since A.D. 1100.39 The maps are clearly not geographically accurate, but they do indicate that the Glykys Limen was still of significant size through the 15th and 16th centuries A.C. Leake's description of the

valley as he passed through the region in 1809 provides important infor-

Figure 6.13. Paleogeographic reconstructions of the lower Acheron valley for 433 B.C. and 1 B.C. The dashed line in the 433 B.C. panel indicates a possible alternative course for the Vouvos River. Small black squares mark core locations; see Figure 6.7 for labels.

39. Besonen 1997 shows 16 maps; see note 1 regarding the availability of Besonen 1997.

222


Figure 6.14. Paleogeographic \ reconstructions of the lower Acheron K>j ( valley for A.D. 1100 and A.D. 1500. > v Small black squares mark core ' Acherousian Swamp locations; see Figure 6.7 for labels.

mation about the landscape configuration east of the Mesopotamon/ Tsouknida constriction, but there are few details about the coastline and actual delta front.40 The modern village of Ammoudia, which surrounds present day Phanari Bay, did not come into existence until after Leake's time, in the early part of the 20th century. Therefore, the position of the shoreline in 1809 must have been a bit further to the east (Fig. 6.15).

There is one radiocarbon date from the area of the Glykys Limen that seems anomalously old, given its location and the type of deposit from which it was obtained. Core 92-20 (Figs. 6.7, 6.11; Appendix) is situated

40. Leake 1835. in the middle of the area of the Glykys Limen, ca. 1.6 km from Phanari

223


..1 ^"/. -f, * M I \ y^^ Figure 6.15. Paleogeographic ^ X, ,~ " jf-Ai)^ * ' f,. \

~ reconstruction of the lower Acheron

xS^L^ 7 C f/ '

"^/^^ y valley forA.D. 1809 and a map of the modern landscape. Small black

c d~~ squares mark core locations; see

-> _ Figure 6.7 for labels.

Bay. It consists of inferred floodplain and natural levee deposits that directly overlie either bedrock or gravel. A radiocarbon date on organic material retrieved 50 cm above the base of the core returns a calibrated la range of ages from 2650 +70/-290 B.P., or 700 +70/-290 B.C. Such a date would suggest that the delta top was located here as early as 700 B.C.,

forcing the delta front position even further basinward. This is problematic since it is in gross contrast with the coherent sequence of coastal evolution documented by the rest of this study.

The anomalously old radiocarbon date from core 92-20 is likely due to reworking of older deposits. The stratigraphy in the core is rather pecu-

224


liar, and is only similar to that seen in core 92-16, less than 600 m away. Though the deposits in these cores are apparently subaerial (according to their color), they occur up to 5 m below sea level. Bedrock in the area is

very shallow, as indicated by limestone knobs that stick up through the alluvium just 500 m to the south, and 700 m to the west (Fig. 6.8). These bedrock knobs are covered with red sediment and vegetation at present, and would have been small islands before the infilling of the Glykys Limen.

Consequently, it seems probable that the deposits around the bedrock knobs, such as retrieved in core 92-20, may represent reworked older sediment and material shed off of the islands.

CONTROLS ON SHORELINE PROGRADATION IN THE ACHERON VALLEY

41. Waters 1994, p. 197. 42. Tziavos 1977. 43. Besonen 1997.

Our data indicate that the rate of shoreline progradation in the lower Acheron valley varied significantly through time; it was slow earlier on, but then much more rapid over the last millennium. In the 3,200 years from 2100 B.C. to A.D. 1100, the shoreline position prograded just 2 km

(Figs. 6.12-6.14). In the 850 years from A.D. 1100 to the present, however, almost 3.5 km of shoreline progradation has occurred (Figs. 6.14, 6.15). Rapid recent progradation is also supported by a detailed consideration of the 3-km wide system of beach ridges noted east of modern Phanari Bay (Fig. 6.3). As Waters has shown, the valley bottom is subsiding;41 if these ridges were accreting slowly over time during subsidence, one would expect to find them at progressively lower elevations moving inland. This is not the case, however. Careful examination of surveyed point elevations on the 1:5,000 topographic maps shows that all the ridges are no higher than 1 masl, and no particular progression of ridge elevations can be noted moving inland. This suggests that the beach ridges have accreted rapidly.

What were the controls on the rates of shoreline progradation? The simple dynamics of basin infilling were probably important factors. Fol- lowing stabilization of sea level in the middle to late Holocene, sediment deposition would have been directed toward filling the deeper parts of the Glykys Limen. As the basin grew continuously shallower, an increasingly larger proportion of the sediment load could be dedicated to the shoreline, leading to the increased rate of progradation we have documented. A similar phenomenon has been noted for the Spercheios delta on the eastern coast of Greece, where delta growth seems to be occurring at a continuously increasing rate.42

One significant local geomorphic control that moderated sediment delivery to the coastline was the formation of the Acherousian lake. As will be discussed below, the lake probably did not come into existence until sometime between the 8th century B.C. and 433 B.C. (Figs. 6.12, 6.13). It then served as an efficient sediment trap, capturing material transported by the Acheron River that would otherwise have been carried to the coast. The lake's ability to trap sediment was further enhanced by a spillway that was built increasingly higher, and subsidence of the lake floor.43 These

225


factors allowed the lake to accommodate nearly 9 m of sediment infill before being breached, probably after A.D. 1100 but before Turkish times

(Fig. 6.14). Once this occurred, the Acheron River was again able to de- liver its sediment load directly to the shoreline.

Several larger-scale controls, operating over more than just the Acheron valley, may also have had the ability to significantly alter the quantity of sediment delivered to the coast, thereby moderating rates of shoreline progradation. In particular, anthropogenic influence has been implicated as responsible for a profound change in landscape stability and associated erosion/alluviation events over Greece as a whole beginning around 4500 B.P.44 In Epirus, pollen studies from two sites located ca. 80 km to the north of the Acheron valley also recognized several erosive events during the middle to late Holocene. Using pollen data from Gramousti lake and Rezina marsh (Fig. 6.1), Katherine Willis recognized erosive events from ca. 6300-5000 B.P., 4300-3500 B.P., and finally at 2500 and 2000 B.P.45

Though both climatic shifts and anthropogenic influence were cited as possible causes for these periods of increased erosion, anthropogenic influence was the more favored explanation, especially for the event dating to 4300-3500 B.P. We do not recognize any of these erosional events in the geomorphic evolution of the Acheron valley, despite its proximity to Willis's study area, but even adjacent regions may have different erosional histories.46

A second large-scale control that may have moderated sediment delivery to the coast is a change in climate, in particular, a change in moisture balance. Though the system of responses and feedbacks may be complex, changes in moisture balance affect vegetation cover, and thus could easily alter the effectivity of erosion. Unfortunately, there is little paleoclimatic information available for Greece, and that which does exist is predominantly pollen work.47 Some effort, however, has been focused on interpreting changes in moisture balance based on lake-level fluctuations.48 A low- resolution record of interpreted lake-level fluctuations exists for Lake Ioannina, just 55 km to the northeast of the Acheron valley (Fig. 6.1),49 but the data from the last 5,000 years are too sparse to relate to geomorphic changes in our area. A new record of lake-level fluctuations exists for Lake Xinias,50 just 160 km to the east of the Acheron valley, but the lake is located on the other side of the Pindos Mountains, a major orographic boundary, and thus a comparison to our area is not appropriate. Further- more, the data from Lake Xinias-like that from Lake Ioannina-are very sparse for the middle and late Holocene.

While middle and late Holocene paleoclimatic information from Greece may not be the most impressive, there does appear to be increasingly robust evidence for a significant, abrupt aridification event around 4200 B.P. over the eastern half of the Mediterranean and West Asia.51 Pre- sumably this event would have affected Greece as well, and may have led to a reduction in vegetation cover, thus increasing the effectivity of erosion and resulting in a higher flux of sediment being delivered to the coast. However, this issue cannot be adequately addressed with the present body of Greek paleoclimatic information (e.g., mostly pollen analyses). Fur- thermore, this event may be impossible to recognize with a proxy like pollen because of the strong overprint of anthropogenic influence that begins

44. Davidson 1980; van Andel, Runnels, and Pope 1986; van Andel, Zangger, and Demitrack 1990.

45. Willis 1992. 46. In particular, see the comparison

of the Southern Argolid and Argive Plain regions examined in van Andel, Zangger, and Demitrack 1990.

47. See reviews in Roberts and Wright 1993, and Willis 1994.

48. Harrison and Digerfeldt 1993; Digerfeldt, Olsson, and Sandgren 2000.

49. Harrison and Digerfeldt 1993. 50. Digerfeldt, Olsson, and

Sandgren 2000. 51. Weiss et al. 1993; Dalfes, Kukla,

and Weiss 1997; Cullen et al. 2000; Weiss 2000.

226


at this time.52 To resolve the issue, development of proxy records for moisture balance that are unaffected by human activity (e.g., an oxygen stable isotope record) would be more suitable.

In summary, local factors such as the dynamics of basin infilling and the formation of the Acherousian lake certainly played a role in moderating the progradation of the shoreline in the Acheron valley. Larger-scale factors such as anthropogenic influence and climate change were capable of affecting the amount of sediment delivered to the coast, but the data currently available are not yet sufficient to link either factor to the changing rate of shoreline progradation.

MIDDLE AND LATE HOLOCENE EVOLUTION OF THE ACHEROUSIAN LAKE

52. See note 44 above. 53. Dakaris 1971; Hammond 1967.

We have documented the development and evolution of the Acherousian lake (Fig. 6.4), which until now was most thoroughly considered by Dakaris followed by Hammond.53 The absolute chronology for our study is based partly on radiocarbon dates, and partly on an analysis of literary and historical references. Unfortunately, the reconstructions of Dakaris and Hammond were based primarily on indirect evidence, the modern landscape configuration in the valley, and the assumption that the lake filled in gradually, becoming shallower and areally less expansive over time. How- ever, the mechanism responsible for the impoundment of the lake was dynamic, and thus it did not experience a typical lacustrine infill sequence and evolution. Instead, the lake maintained a shallow profile but grew continuously larger, spreading upvalley through time. As a result, Dakaris, Hammond, and others overestimated the size of the lake, at least as an open body of water.

The development and evolution of the lake is best recorded by the stratigraphy around and to the east of the Mesopotamon/Tsouknida valley constriction (Fig. 6.2). Sediment cores 94-23 and 94-17 (Fig. 6.7; Appendix) document the overall regressive nature of the middle and late Holocene stratigraphy in the valley, and the entire history of the Ache- rousian lake. As described above, the cores consist from the base upward of deposits from the following environments: 1) delta top to front, 2) brackish water delta top marsh grading upward into freshwater marsh, 3) shallow freshwater lake, and 4) floodplain. The shallow freshwater lake deposit is from the Acherousian lake. A radiocarbon date on peat from the bottom of the fresh to brackish water delta top marsh of core 94-23 returns a calibrated la range of ages from 4030 ? 100 B.P., or 2080 + 100 B.C.

We therefore conclude that the Acherousian lake came into existence at some point after ca. 2100 B.C.

The stratigraphy in cores 94-23 and 94-17 indicates that the marsh was essentially drowned as the lake came into existence directly on top of it. Some mechanism to the west of these cores was therefore responsible for the impoundment of the lake. Analysis of the stratigraphy in cores 94-20, 94-12, and 93-21 (Fig. 6.7; Appendix), located ca. 600 m to the west in the Mesopotamon/Tsouknida valley constriction, shows what this mechanism might have been.

227


The A-A' cross section based on these cores (Fig. 6.9) shows a massive plug of fluvial sediments filling the valley at this point. The stratigraphy in these cores consists of delta top and front sediments that are immediately overlain by fluvial channel, subaerial natural levee, and floodplain sediments. In contrast, core 94-23 consists of the same delta top and front sediments overlain by 7.5 m of sediment from the Acherousian lake. From this relationship, it is clear that the lake was impounded to the east of the Mesopotamon/Tsouknida valley constriction because of fluvial sediments that essentially plugged the constriction.

This fluvial plug records the migration of the channel and levee system of the Acheron River and/or one of its tributaries. As the channel/ levee system built south-southwestward from the eastern side of the Meso- potamon ridge, it eventually impinged onto the bedrock promontory near Tsouknida (Figs. 6.12, 6.13). As a result, a shallow, closed depression was pinched off to the east behind this channel/levee/proximal floodplain system and water ponded up, drowning the delta top marsh to form the Acherousian lake (Fig. 6.13).

The stability and longevity of this fluvial plug system are important points to emphasize. Following the initial impoundment of the lake, fluvial sedimentation has dominated in the area of the valley constriction until the present day, as illustrated by cross section A-A' (Fig. 6.9). Thus, as the channel/levee/proximal floodplain system slowly aggraded through time, it caused a progressive rise in the surface elevation of the lake. Be- cause the lake was also receiving sediment input, this process allowed it to accommodate 9 m of sediment infill while simultaneously maintaining a shallow profile.54 More information regarding the progressively rising surface elevation of the lake and its areal expansion will be discussed below.

This mechanism of river channel and levee migration is an extremely important agent of geomorphic evolution in the valley, and its effects can be seen in the topography at other points in the valley today. Three excellent examples include the topographic depression to the west of Koroni, the depression between Kastri and Kanallakion, and the small depression to the east-southeast of Ephyra (Fig. 6.8). In these cases, the migrating course of the Kokytos and Acheron Rivers impinged onto a bedrock highland and pinched off a shallow, closed basin upvalley of the constriction. Richard Russell noted a similar process in his study of the Meander River in western Anatolia.55 In this case, a rapidly prograding delta front/coastal plain built across the entrance to a marine embayment, essentially trap- ping a standing pool of water within the embayment. He also recognized shallow lakes ("levee-flank depressions") that had formed on the delta top in the area behind/between the intersection of two stream channel/levee systems.56

Though the radiocarbon date from core 94-23 indicates that the impoundment of the Acherousian lake must have occurred after ca. 2100 B.C., a more tightly constrained chronology could be determined by another 14C date at the top of the freshwater marsh deposit. Unfortunately, limited resources did not provide this option. Consequently, a closer dating of the lake's inception will be based on an analysis of literary and historical references by ancient authors. Differing opinions about the accu-

54. Besonen 1997. 55. Russell 1954. 56. Russell 1967, p. 17.

228


racy and validity of topographic references made by ancient authors are certain. However, if such references, taken in chronological order, present a coherent and logical sequence of events, they may be useful. On the contrary, if they present a sequence of events that is clearly impossible, or if various references contradict one another, one may be inclined to question their validity. This, however, is not the case in the Acheron valley. A detailed analysis of ancient literary and historical references in chronological order presents a logical and coherent picture of the probable evolution and development of the Acherousian lake. That these references fit nicely within the story developed by geological and sedimentological evidence lends them some credence.

The earliest reference to the valley comes from the Odyssey of Homer. Current thought suggests that the Odyssey may have been written in the 8th century, but describes some events and settings that go back to the 12th century B.C. in the Late Bronze Age. Homer writes:

And when in your ship you have traversed Oceanos, Where the scrubby strand and groves of Persephone are, Both tall poplars and willows that lose their fruit, Beach your ship there by deep-whirling Oceanos; But go on yourself to the moldy hall of Hades. There into Acheron flow Puriphlegethon And Cocytus, which is a branch of the Styx's water, And a rock and a concourse of the two resounding rivers.57

Homer makes no mention of the lake; in fact, he strictly describes a scene in which several tributaries feed into the Acheron River. The adage "lack of evidence doesn't constitute evidence for a lack" is applicable here, but it may be suggested that Homer does not mention the lake because it did not exist at the time a contemporary witnessed the topography in the valley. The lake then probably formed at some point between the writing of the Odyssey (in the 8th century B.C.) and the time of Thucydides' account of the valley (about 400 years later), when the lake is mentioned for the first time.

Thucydides, who wrote contemporary history, gives a description of a recently nascent Acherousian lake in his account of the Battle of Sybota in 433 B.C.

It is a harbour, and above it lies a city away from the sea in the Eleatic district of Thesprotia, Ephyra by name. Near it is the outlet into the sea of the Acherusian lake; and the river Acheron runs through Thesprotia and empties into the lake, to which it gives its name.58

Of interest here is the fact that Thucydides strictly states "near it is the outlet into the sea of the Acherousian lake," as if the lake empties directly into the sea. This seems to imply that the Acherousian lake and the sea

57. Od 10.508-515, trans. A. Cook, (actually the Glykys Limen) are very close-the two are split by only a New York, 1967. ' very narrow barrier of land on which is situated the lake spillway (Fig.

58. Thuc. 1.46.4, trans. C. F. Smith, 6.13). This narrow barrier of land is the channel and levee system of Cambridge, Mass., [1928] 1956. the Acheron (or one of its tributaries) that caused the impoundment of

229


the lake, as explained above. Thucydides clearly identifies the Acheron River as flowing into the lake, but says nothing about its exit from the lake.

His account is distinct from all later references in that it suggests the extreme proximity of the Acherousian lake and the sea. Later accounts suggest that more than just a lake spillway is present, and that the channel carrying water from the lake to the sea is long enough to be identified as that of the Acheron River. For example, Strabo writes:

Then comes Cape Cheimerium, and also Glycys Limen, into which the River Acheron empties. The Acheron flows from the Acherusian Lake and receives several rivers as tributaries, so that it sweetens the waters of the gulf.59

It appears that the strip of land separating the lake from the sea had grown sufficiently wide in the 400 years between the accounts of Thucydides and Strabo that the channel draining the Acherousian lake could be identified as that of the Acheron River. Furthermore, several tributaries fed into the Acheron after it exited from the lake.

Based on the radiocarbon date from core 94-23, the Acherousian lake must have formed after 2100 B.c. And, since Homer, Thucydides, and Strabo all present a chronologically coherent picture of the development of the lake in the valley, it is probable that the lake formed sometime after the 8th century B.C. but before 433 B.c. An additional bit of circumstantial evidence supports the notion that the lake did not come into existence until this time. Dakaris noted that the archaeological record from the valley indicated a decrease in population during the Archaic period (ca. 700- 500 B.c.),60 and the data from the diachronic survey of the Nikopolis Project tend to support this conclusion. Dakaris suggested that the population decline might have been related to malaria, which has always been a problem in the low-lying coastal areas of Epirus. Why malaria would have flared up at this particular time was unknown, since Dakaris probably assumed the lake had been present in the valley following the post-glacial rise of sea level. Our analysis seems to indicate, however, that the Ache- rousian lake came into existence at the same time as the Archaic-period population decline. While the timing of these events may be coincidental, the birth of the lake and associated swampy areas may have given rise to the malarial epidemic postulated by Dakaris.

Because Dakaris and others did not recognize the mechanism responsible for the lake's impoundment, they assumed that it followed a typical lacustrine infill sequence and became increasingly shallower and areally less expansive through time. In contrast, Philippson and Kirsten suggested that the lake had become larger since ancient times,61 but they did not explain why they considered this to be the case nor did they provide evidence to support their conclusion. They also placed the lake too far upvalley (Fig. 6.4, upper right). The results from our study suggest that their assertion regarding the size of the lake is correct, but we furthermore document the mechanism and details of the lake's evolution as well as its true location.

59. Strab. 7.7.5 [C 324], trans. H. L. Jones, Cambridge, Mass., 1960.

60. Dakaris 1971, p. 12. 61. Philippson and Kirsten 1956, II,

p. 105.

230


62. Besonen 1997. 63. Dakaris 1971; Hammond 1967;

Leake 1835; Philippson and Kirsten 1956.

Soon after its formation, the lake existed as a shallow body of open water surrounded by a fringe of marshy ground (Fig. 6.13). Sediment carried by the Acheron would have quickly filled it in were it not for the slowly aggrading spillway mechanism discussed above. This allowed the lake to accommodate an increasingly larger volume of sediment vertically and areally, as the lake expanded upvalley because of its slowly increasing surface elevation.62

Evidence for the initial small size of the lake, and for the subsequent expansion of marshy, swampy ground upvalley, can be seen by comparing the stratigraphy in cores 94-23 and 94-17 with that of core 93-22 (Fig. 6.10; Appendix). Cores 94-23 and 94-17 are located just to the east of the fluvial plug in the valley constriction and contain 7.5 and 5.9 m, respectively, of lacustrine mud and clay from the Acherousian lake. These lake deposits begin at 3.1 and 1.7 m below sea level, and run to 4.4 and 4.2 masl, respectively. Core 93-22 is located ca. 1 km east of cores 94-23 and 94-17, in the area considered by Dakaris and others to be the ancient lake. At this locality, however, a much thinner sequence (3.5 m) of mixed lacustrine and marsh deposits occurs between 1.2 and 4.7 masl. The lake deposits in the core are underlain by a very stiff floodplain alluvium with some pedogenic development. Thus, this package of lacustrine and marshy deposits shows stratigraphic onlap upvalley, and its transgressive nature confirms the gradual increase of the lake's surface level and its areal expansion through time. The lake probably never extended much further upvalley than the location of core 93-22 because the mixed lacustrine and marsh deposit in this core is indicative of the lake edge and shore.

This information also helps constrain the size and location of the lake, at least as an open body of water. Mixed lacustrine and marsh sedimentation at the location of core 93-22 could not have begun until the surface level of the lake had reached at least 1.2 masl (i.e., the base of the lacustrine material in that core). When did the lake surface level reach this elevation? By ignoring factors such as subsidence and changes in the rate of sedimentation or spillway aggradation, we can loosely base it on the chronology from core 94-23. Elevationally, 1.2 masl corresponds approximately with the middle of the lacustrine sedimentation sequence of core 94-23. From our preceding analysis of core 94-23, we concluded that the lake probably came into existence after the 8th century B.C., but before 433 B.C. Continuous, uninterrupted deposition occurred there until after the First World War, at which time the final remnants of the swamp were backfilled. Assuming that the surface level of the lake rose at a constant rate, it would have reached 1.2 masl in the middle of this time span, or roughly A.D. 850. Thus, we estimate that the expanding lake and marsh ground reached the locality of core 93-22 around the 9th century A.C.

This evidence suggests that Dakaris, Hammond, and others greatly overestimated the size of the lake (Fig. 6.4), especially considering that their reconstructions are supposed to show the extent of the lake during the classical period.63 In some of the reconstructions, the shape and location of the lake contradict the modern topography. For example, Dakaris and Hammond suggest that the lake had a northeast/southwest-trending

231


shore between Mesopotamon and Kastri. However, the topographic lines that would have defined the lake shore in this area have a northwest/southeast trend, exclusive of the elevated subaerial natural levees which flank the Acheron River (Fig. 6.8).

Dakaris's reconstruction also suggests that a branch of the lake extended to the east between Pountas ridge and the villages of Kastri, Kanal- lakion, and Acherousia, but this is not correct. This area is a closed depression (Fig. 6.8) that came into existence by the same mechanism which caused the impoundment of the Acherousian lake. In this case, the Acheron river channel and levee system pinched off the depression against the tip of the Pountas ridge, which projects up from the south. This depression, therefore, would not have come into existence until the course of the Acheron shifted to the south of Kastri. As we discuss below, this shift probably occurred very recently, perhaps around the end of the 16th century A.C.

There is additional geologic evidence to suggest that the main body of the Acherousian lake to the west of Pountas ridge was not confluent with the water body to the east of the ridge. Laminated lacustrine silts and clays do indeed occur in this small basin, but they form a relatively thin layer and are too high topographically to have been deposited by the Acherous- ian lake. Core 94-03 (Fig. 6.7; Appendix), taken from the center of this small depression, is composed of a backswamp deposit overlain by a freshwater marsh deposit, which is in turn succeeded by floodplain deposits. Core 94-21 (Fig. 6.7; Appendix), located just 450 m to the west, exhibits identical stratigraphy but bottoms out with a floodplain deposit as well.

Though core 94-03 did not penetrate these lower floodplain sediments, its

proximity to core 94-21 and the fact that it is shorter support the inference that further penetration of core 94-03 would have encountered the same

floodplain deposit. The Acherousian lake would have necessarily had a surface elevation

at or below the elevation of the fluvial plug sediments that impounded it. This fluvial plug was continuously aggrading, but never reached more than 5.0 masl, the present elevation at the Mesopotamon/Tsouknida valley constriction. Thus, sediments from the Acherousian lake could only have been

deposited up to this height. But the backswamp and freshwater marsh deposits in cores 94-03 and 94-21 occur between 5.1 and 7.7 masl. There- fore, the body of standing water in which these sediments were deposited could not possibly have been confluent with the Acherousian lake as the

standing water had a significantly higher surface elevation. This conclu- sively proves that the body of ponded water that once existed here was not a branch of the larger lake as Dakaris indicated.

By Turkish times, the Acherousian lake had become a swamp with a few isolated pools of water (Fig. 6.14).64 Continued growth of the Acheron river channel and levee system split the remains of this swamp. This interpretation is supported by the broad topographic high of the river channel and levee system to the east of Mesopotamon, and by the closed depression directly to the east of Ephyra, created when the channel and levee system impinged against the bedrock ridge (Fig. 6.8). Leake provided an 64. Hammond 1967, p. 39.


excellent description of the marshy valley bottom from his travels through the region in the spring of 1809, and he noted that several pools of open water still existed (Fig. 6.15).65 After the First World War, the final marshy remnants of the former Acherousian lake were filled in for agriculture.66

THE CHANGING COURSE OF THE ACHERON WITH RESPECT TO KASTRI

65. Leake 1835, I1, p. 232; IV, pp. 51-54.

66. Hammond 1967, p. 68. 67. Dakaris 1971, pp. 136-137.

In order to reconcile the archaeological remains in the valley with the accounts of ancient authors, Dakaris suggested that the Acheron River had shifted its course to the south of Kastri since classical times.67 Unfortu- nately, he could not provide geologic evidence with chronological control to support his theory. Cores 94-02 and 94-04 provide the evidence to document this shift.

Core 94-02 (Fig. 6.7; Appendix) was retrieved north of Kastri, between it and the larger of the two hillocks named Xirolophos (Fig. 6.2). The core consists from the base upward of deposits from the following environments: 1) floodplain, 2) backswamp, 3) floodplain, 4) fluvial channel, and 5) floodplain. At the interface between the lowest floodplain unit and the backswamp, a small reddish pottery fragment was encountered. The fragment is abraded and lacks diagnostic features, but ceramic specialists on the project have suggested that the texture of the sherd should place it some time in the classical period. Since this pottery fragment occurs below the deposits of a fluvial channel, it provides a terminus post quem for the existence of the river channel at that location. Therefore, at some point past the beginning of the classical period, a fluvial channel existed north of Kastri.

Core 94-04 (Fig. 6.7; Appendix) was also retrieved north of Kastri, between the hillock of Koronopoulos and the larger of the two Xirolophos hillocks (Fig. 6.2). From the base upward, deposits from the following environments occur in succession: 1) floodplain, 2) backswamp, 3) fluvial channel, 4) backswamp, and 5) floodplain. The fluvial channel sediment is over 1.5 m thick, and contains gravel clasts up to 1 cm in diameter. This deposit is from a significant river channel, like that of the Acheron, and not from a smaller stream. A radiocarbon date on a piece of wood from the base of the fluvial channel deposit returns a calibrated la range of ages from 380 +90/-70 B.P., or A.D. 1570 +70/-90. For radiocarbon dates this young, however, the calibration curve is relatively irregular and the speci- men could date to almost any time during the last 500 years. Nevertheless, the radiocarbon date shows that a river channel, probably that of the Acheron River, was operating to the north of Kastri within the last 500 years. When Leake passed through the region in 1809, he recorded that the Acheron River followed a course to the south of Kastri, as it does today. Therefore, if the fluvial channel sediments in core 94-04 are indeed from the Acheron River, it would suggest that the course of the Acheron shifted from the north of Kastri to its south sometime between ca. 1500 and 1809.

233


CONCLUSIONS

Numerous ancient authors, beginning with Homer in the 8th century B.C.,

make reference to the lower Acheron valley and indicate a landscape con-

figuration that is significantly different from at present. Three notable dis-

crepancies between the ancient and modern landscape exist. The first problem concerns the size of the Glykys Limen (modern Phanari Bay), which at present is very small, but was much larger in ancient times. The second

significant discrepancy concerns the evolution of the extinct Acherousian

lake, which ancient sources indicate was a conspicuous feature in the val-

ley. The final discrepancy concerns the course of the Acheron River, which

today flows to the south of Kastri but was once located to the north of that site.

Geologic evidence based on twenty-eight gouge auger sediment cores taken at various locations in the valley indicates that significant geomorphic change has occurred in the valley during the last 4,000 years. The shoreline of the Glykys Limen has prograded nearly 6 km in that time,

doing so at varying rates. The Acherousian lake developed relatively late in the Holocene probably between the 8th century B.C. and 433 B.C. Since that time it has been filled in by natural alluvial processes, modified by a

constantly aggrading spillway. Finally, the Acheron River appears to have

occupied a channel to the north of Kastri, and has only shifted to the south of that hillock in the last 500 years.

It appears that the discrepancies between the ancient accounts and the modern landscape are not due to errors in the ancient sources, but are instead the result of a natural sequence of landscape evolution in the valley. Furthermore, careful examination of the ancient accounts may in some cases provide details and information for paleogeographic and paleoenvironmental reconstructions that are not recoverable from the geologic record. The disciplines of geology and archaeology find a natural interface here, both contributing to, and benefiting from, one another. Indeed, the dynamic geomorphic evolution seen in the Acheron valley during the last

4,000 years reaffirms the need for multidisciplinary archaeological investi-

gations that strive for a broad understanding of the dynamics of environmental change.


APPENDIX: CORE STRATIGRAPHY AND LITHOLOGY

This appendix contains the sediment core stratigraphy from all twenty- eight cores taken in the Acheron valley. Width of the core, lithologic patterns, and a "Sediment Type" description reflect the grain size and type of sediment based on field and laboratory observations. Organic matter present in the stratigraphy is indicated by one of the symbols in the legend below. Locations of calibrated 14C AMS dates are indicated by arrows. "Color" (according to the Munsell Soil Color Chart), weight percent of organic matter determined by loss on ignition analysis ("% Organic Content"), and results of the microfossil analyses ("Microfossils") are also included (see legend below). The "Environment of Deposition" field represents our interpretation of the stratigraphy based on all available data. All primary data, including results from magnetic analyses, pipette grain-size analysis, microfossil plates and counts, and any data not included here, can be found in Besonen 1997, which is freely available in Adobe Acrobat PDF format (see note 1 for details).

Symbol Explanation w\il common coarse-grained organic matter

abundant coarse-grained organic matter

(1J_) few to trace coarse-grained organic matter _~ common fine-grained organic matter

abundant fine-grained organic matter (_i) few to trace fine-grained organic matter

C-100 44BP calibrated C-14 AMS date in years B.P.

qty. 615: 1.3% F, 91.4% B, 7.3% R qty. XXX = quantity/total number of freshwater, brackish to marine water, and reworked microfossils in the sample

1.3% F = percentage of freshwater forms in quantity XXX 91.4% B = percentage of brackish to marine water forms in

quantity XXX 7.3% R = percentage of reworked microfauna in quantity XXX

235

slightly sandy silt

slightly silty clay

Color % Organic Content

0 0j I I I I

not recorded

5Y5/2

5Y4/2

5Y5/2

fine sand

5Y5/1

clay 2.5Y; N5/

7.5YR;_N-4f-

NC-92-16

a- -L C- a- CD

Sediment Type

(KJL/)

('JJ)- - - - - -

Microfossils

mud

o 190

50 /140

100 / 90

150 /40

200 /-10

2501/-60

300 /-110

350 /-160

400 /-210

4501/-260

500 / 310

5501/-360

600 / -410

650 /-460

700 /-510

7501/-560

cr,

0-

Cr

0- a)

0

0z 01

Environment of Deposition

natural levee

floodplain

natural levee

floodplain

Zs a a-= a a 0 a-L

Color % Organic Content Microfossils

I I I I

mixed beach sand not recorded

Environment

of Deposition

accretionary beach

NC-92-1 7

Sediment Type

0 /100

50 /50

100 /0

150 / -O

200 /-100

250 /-150

300 / 200

350 /-250 cj

-

a) a)-

a)

~0 C

C.)

a1)

C) 0

C)

0/100 _ _ _

150/ -50

- 200 /-100 ''- i ^' , () mud -

C) 0 CD

Sediment Type

mud

mixed beach sand


not recorded

Microfossils

I I I I

2.5Y5/2


topsoil

accretionary beach

- 2.5Y4/4

~0 110

0 0

0

NC-92-18

00

CD Ct C

sea level - - - - - - - - - - - - - - -

. . .. . .. . .. . .. . ~ ~ (\ l L /

- F : 3~~~~~C

CD/

Sediment Type

- t-i w ) 0 ) 0 ) 0

fine sandy mud

muddy fine sand fine sand with

clay interlayers

not recorded-- 5Y3/1

4E-25Y6/6 with

7.5YR5/0 with 5Y3/2 mottles

2.5Y5/6 mottles 2. 5Y4/4 with 41-2.5Y4/0 mottles

5Y4/1

clay grading upward to slightly

clayey silt 2.5Y3/0

5Y3/1

NC-92-19


-100 /40 -

-50 /-10 -

0 / -60

50 /-110 -

100 /-160 -

150 /-210 -

200 / -260 -

250 /-310 -

300 /-360 -

350 /-410 -

400 /-460 -

4501/-510 -

500 /-560 -

5501/-610 -

600 / -660 -

1n

0-

tf

0)

0

0 0D

C.


floodplain

a, - ~3 n- 7

- J i 4 - - - -

(\JL/)

Sediment Type

clay

fine sand to muddy fine sand

\jIL

interbedded clay, \IIC-14: 2650 mud, silt, and muddy

= +70/ 290 BP fine sand

BEDROCK OR GRAVEL


t j wJ C) C) 0

not recorded

2.5Y6/5 with IOYR5/6 mottles

2,5Y4/4 mottles

E-5Y5/ Iwith -- 5Y6/6 mottles

2.5Y3/0

2.5YR3/0

NC-92-20

Microfossils

o 90 50 /40

100 /-10

150 /-60

200 /-110

250 /-160

300 /-210

350 /-260

400 /-310

450 /-360

500 /-410

550 /-460

600 /-510

t6)

0

CA

0-

a) 04- 0

0~ 0)

Q

0-- 01 0)

0

2 0 P := a l0 C,C a a


floodplain

natural levee

floodplain

BEDROCK OR GRAVEL

241

30 -

20 -

10 -

OS- S a o J

V 20-

p e- 0 pebble

c. sand -c-

m. sand - - m. sand f. sand- ,- f. sand

silt - 1 - silt clay - I a

o o o o o o o o o o o - Cl oC

C o) O C C>

? ) ? m m t t ^ e n _

(sl3mj3 luo33) JlA3l 3oS OAOqt UOlthA3jp / JlOO Ul qldaI

~~ -

e level

,

_ _

~~~~~~~~~.._ --C~~~~~~~~C

---sea level ~.-. _::_-.._...k,

_.. : '. _'.'.n '.m : 7 4 _'fl '.." v. B..".: .'. (_ _ _)

_ (_I/

:::...'-':':. /::'.o io. - ':?*' rf '.' -

- -se levelo

_: ::: :/: _ : - __ _- _ _

^lll?~~(1~

?_,.- . . _--L

. . _ .. _ .. _ .. _ ......--

:' '-/: ~.' ".: ' .-: .' _ '

... ...

~~~~~(iJ/

- J

i'a'|Xt; ?. X'Xa (\JI)

.:' ':g------- *;: * -:*.: -- - . ; * ,;* ; - -

Sediment Type Color % Organic Content Microfossils

I I I I

_ N? (- o0 o 0o 0

silt

mud

slightly silty clay

2.5Y6/6

2.5Y5/4

mud

clay

slightly sandy silt fining upward to mud

clay

interbedded gravelly sands to fine sands

fine sandy mud grading upward to fine sand

laminated clays, muds, and silts with

several cm-thick muddy fine sand layers

N3/0 and N5/0 4- 5Y5/2

5G5/1

5BG5/1

5G5/1 to 5G4/1 with 20-50%

2.5Y5/6 mottles

5G4/1

N5/0 to N4/0

N3/0

5GY4/1

5G4/1

Environment


crevasse splay

floodplain

interdistributary bay

subaqueous levee

delta distributary channel

delta distributary mouth bar

shallow marine (Glykys Limen)

t)

53-

0/300

50/250 -

100/200 -

150/ 150 -

200 /100 -

250/50

300/0 -

350 /-50 -

400 / -100

450 /-150 -

500/-200 -

550 /-250 -

600 /-300 -

650 /-350 -

700/-400 -

750 /-450 -

800 /-500 -

850 / -550 -

900 /-600 -

950 /-650 -

1000 / -700 -

1050/ -750 -

_/3

r( 0

-?

0

0

o

cn

Q2 .-

* -

1100 / -800 -

1150 / -850 -

1200 / -900 -

1250/ -950

NC-93-14

sea level

.-. : .. . .,.,

.- .. :.-'

- ' _.. - . .:._ ... : . i._. '

-.'. --.'' '.-_.-". -. '~..

',':',., ' ' : '. v

._.?1 . i

-_c. :

Sediment Type Color % Organic Content

gravelly sand fining upward to sandy mud

-1a-YR33:: 5Y5/6

5Y6/6

5G4/1

_I

s

I o0 o 0o 0

N4/0

interbedded fine sands and muddy sands with

some mud layers

N4/0 interbedded with 5Y4/1

5BG4/1

mud

fine sand


accretionary beach and nearshore reworked

nearshore reworked delta front

5G4/1

5GY4/1

NC-93-15

0/50

50/0

100/ -50

150 /-100

200 / -150

250 / -200

300/ -250

350 / -300

400/ -350

450/ -400

500/ -450

550/ -500

600 / -550

650/ -600

700/ -650

0

13

ra

0

-

o0 {D 0D

(D

0

0

0

0

CL

Microfossils

a a- a- Q.

.............. sea level xl/ .. _ . _ . .. - .. . . ...

. . .. ... . - ..

? . : . _ . -..

a , a

Sediment Type

slightly clayey silt grading upward to

fine sandy silt

mud muddy fine sand

gravelly sand

muddy fine sand

Color

10OYR4/4 and 5Y6/4

5Y6/6

% Organic Content Microfossils

I I I I o0 o 0o 0

5G4/1

5Y5/1

5G5/1


floodplain

backswamp

accretionary beach and nearshore reworked

NC-93-17

0/50

50/0-

100 / -50

150 / -100-

200 / -150-

250 /-200-

300 / -250

-a

a)

Da)

*-4

o

c)

0

c~

.. _ _ - 0- CL sr Ii 1 1 1 1

.*-* - .**-** \jL/ -.. ..- .... sea level

= _== -- - --- - -- i Cr g -

---I ---r~ z ^ +90/-30BP

";- .:.: D;.Jl -.'- ~?.'i-: , .. '.'- B'~.- ,,- '*;g ri f * - :.C' .@ ; " .'-, '.- '

e. * | ?/ -

.- -.i****^' t- r. _.**** .n

- ;-i-i-I li

o--__ ((i-i 0

** * 3 2

p.g....

- . ? 2 -~

_ _ _

_ _ _ _ _ 0

Sediment Type

slightly clayey silt mud and peaty mud

interbedded clay, mud, and gravelly sand

Color % Organic Content Microfossils Environment

10Y3/2 i- interi6edded-N2/70

-

-andlQOYRl5/6 --. 5Y4/2 and 5Y7/3

interbedded 5Y8/3

and 5G4/1

5Y8/2

I I I I

5GY4/1 mud coarsening

upward to fine sand 5BG5/1

5BG5/1 with some 10YR5/3

laminae

laminated clays, muds, and silts


floodplain

backswamp

accretionary beach

nearshore reworked


5G4/1

NC-93-18

0/50

50/0

100 /-50 -

150 /-100 -

200 / -150

250 /-200

300 /-250

350 /-300

400 /-350 -

450 /-400 -

500 /-450 -

550 /-500

600 /-550 -

650 /-600 -

700 /-650 -

750 / -700 -

800 /-750 -

850/ -800

+-,

0

_0

0

C

0


Sediment Type

_ _ _ _ _ fine to coarse sand with some muddy layers

Color

5Y6/4

5GY4/1


-- I I I I

..- N2/0

--- 5Y5/4

5B4/1

5BG4/1 to 5GY4/1

laminated muds, silts and fine sands


intercalated floodplain, backswamp,

accretionary beach

nearshore reworked


5GY4/1

\L muddy fine sand

\AZ/ C-14: 5140 mud and peaty mud +160/-100 BP

clay N4/0

NC-93-19

.- r 0 c Pz 0 0-

c C. 0 CD

0/90

50/40

100 / -10

150/ -60

200 / -110

250 /-160

300 /-210

350 / -260

400 / -310

450/ -360

500 / -410

550 / -460

600 / -510

650/ -560

700 / -610

750 / -660

sH

4--4

0

0

a)

u 0 co

0

g

0

W-

0

0u

0

i

0) *- <u

. V.

0 " p -

ac a CD

250 cm above sea level

slightly clayey silt

mud and peaty mud

BEDROCK

NC-93-20

Sediment Type Color

0 /500

50 /450

100 / 400

150 /350

200 /300

250 /250

% Organic Content

0 0 0

IOYR6/6

Microfossils

2.5Y5/6

5GY4/ land 2.5Y6/6 N3/0

A


floodplain

backswamp

BEDROCK a-)

a-) 0

a-

a-) a-) a-) 0

a-)4 a-) a-)

0

0

a-)

Sediment Type Color

ca 0-- - & a

. ' ... ? .. -. ... .

. ... . . . . ... .

. ... ......... .

(_L)

e-'.? ,' '-' '-.' ? ( .40.60 BP

-. . _ - C'.


I I I I

slightly clayey silt with some sand

muddy fine sand with interbedded silty layers

slightly clayey silt and mud with some

interbedded silty and sandy laminae

sand and gravelly sand with some interbedded

muddy layers

mud

peat and peaty mud

interbedded muds and silts with some fine to gravelly sand layers

gravel BEDROCK

10YR6/6

2.5Y5/6

2.5Y5/8

5GY4/1

interbedded N4/0 and N3/0

5GY4/1 with some N4/0 layers

5Y5/2 and 5Y5/4 5GY4/1 and 5Y5/4

7.5Y3/2 10YR4/1

---: N30 Oand_N4ZG0:::

5GY4/1 with some 5Y4/1

layers


floodplain

natural levee

floodplain

shallow freshwater lake

(Acherousian lake)

fluvial channel or proximal crevasse splay

floodplain

delta top marsh

delta top and front


BEDROCK

NC-93-21

0t

0/500

50 / 450

100 / 400

150/350

200 / 300

250/250

300 / 200

350 / 150

400 / 100

450 / 50

500/0

550 / -50

600 / -100

650 / -150

700 / -200

750 / -250

800/ -300

0

0D 0

0-

0

a)

0 I

C :0 - L- ?

sea level

_ _a || o< ac',

Sediment Type


slightly silty clay


slightly silty clay


o0 o 0o 0

2.5Y5/2 and 5Y4/3

5Y4/2 to 5Y4/4

5GY5/1

5GY4/1

5BG4/1

5GY4/1 with 10OYR5/6 mottles

10 OYR5/6 with 5BG5/1 mottles

qty. 116: 3.5% F, 1.7% B, 94.8% R


floodplain


(Acherousian lake) with some freshwater

marsh horizons

i floodplain

NC-93-22

0 / 520

50 / 470

100 / 420

150 / 370

200 / 320

250/270

300 / 220

350/ 170

400 / 120

450 / 70

500 / 20

550 / -30

rA

a)

c-i-

0

0

a1)

a)

a) 5-4 0

a)

0 ?

Sediment Type

800 cmw above sea level

i _i i i i . . ' ?_

_ CD


0 0 0 I I I I

_I

2.5Y4/3


peat and peaty mud

2.5Y5/3

: 5BG6/1-- 5BG6/1

5B5/1 qty. 41: 97.6% F, 2.4% B, 0.0% R

mud


floodplain

freshwater marsh

backswamp

5GY5/1 qty. 7: 100.0% F, 0.0% B, 0.0% R *


mud with some fine sand interlayers

poorly sorted silty fine sand

5Y4/2

10OYR4/6

5B5/1 qty. 132: 1.5% F, 0.0% B, 98.5% R

qty. 210: 0.0% F, 0.0% B, 100.0% R

qty. 220: 0.0% F, 0.0% B, 100.0% R 5Y5/3

5Y4/3 & 10Y5/4

floodplain

backswamp

natural levee

NC-94-01

0 / 1280

50 / 1230 -

100/ 1180

150/ 1130 -

200 / 1080 -

250/ 1030 -

300/980 -

350/930 -

400/ 880 -

450 /830

500/780 -

550/730

600 /680

650/630 -

700/580 -

750/530

800/480

o-o

0

0

0 ?

I

0

0

+j

0 0)

u

0 0

C0 (D CA -? ?

1.

V-4

CD

....... ... ... . .

.... .. .. ._. . .

. ... ... ... ... . .... ..._...._ . .

.... . . . _ . .. . _ . -:

. . .. ... ... ...

n- . .

_ _ _ _ _~ ~ f2

Sediment Type

slightly clayey silt with some fine sand interlayers

gravelly -sand fini-/ig ... upward to fine sand slightly clayey silt

with some muddy, fine sand interlayers

mud


NC-94-02 t-I

(-/a


I I I I 0/ 1580

50 /1530

100/ 1480

150/ 1430

200 /1380

250 /1330

300/ 1280

350/ 1230

400 /1180

450/1130

500/ 1080

550/ 1030

600/980

650/930

700 /880

750/830

c)

0-

0

13 0

a) c)

*=-

2.5Y5/3

2.5Y7/4

2.5Y6/6

5BG4/1

5GY4/1

5Y5/4


floodplain

fluvial channel

floodplain

backswamp

floodplain

Sediment Type c CD

AD Z

. . .. . . . .. . ..


0 -CD 0- c


o> w ON ~lc k

2.5Y4/3

slightly clayey silt i.

Microfossils

qty. 12 1:I00.0% F,00% B, 00% R

5Y5/2

peat and peaty mud

mud

qty. 76: 1 00.0% F, 0.0% B, 0.0% R I 5G4/1

5BG4/1

5BG5/1


floodplain

freshwater marsh

backswamp

qty. 121: 89.3% F, 0.0% B, 10.7% R 'V

NC-94-03

o / 970

50 /920

100 /870

150 /820

200 /770

250 /720

300 /670

350 /620

400 /570

450 /520

500 /470

tF)

0

0

0 0 .

0 0

0

Sediment Type

- - - - J/ C-14: 380

. . .. " . : . . '." ._. . .

. . .. -..-.. .. _. . _ . .

1150 cm above sea level I i i i * *-

_^ ___C

_ _ _ _2

slightly clayey silt with some fine sand interlayers

very poorly sorted muddy, fine to gravelly sand

mud


NC-94-04


2.5Y5/3

0/1780

50/1730-

100/1680

150/1630

200/1580-

250/1530-

300/1480-

350/1430 -

400/1380-

450/1330-

500/1280-

550/1230-

600/1180-

650/1130-

2.5Y4/3

2.5Y5/4 5Y5/2

5G5/1

rJ

a

c)

0

0

u

ct


floodplain

backswamp

fluvial channel

backswamp

floodplain

5G4/1

5GY5/1 5Y4/2

I I I I

.oo .e B p r

y- ,n vL 0-


i i i-^ i i T

...... . . . p. . *, :--*- :-' .-

'.'- ':, - ' .-::-." (_~z) ::. .:: .':._. '.:_--.:;.

_ . . _ . . _<. . _ .(_ _

_ ._._

1300 . above sea .eve._

i i i i i i

, ?


- o o

slightly sandy silt


mud

slightly silty clay

mud with a few silt and fine sand interlayers

poorly sorted muddy sand

slightly clayey silt with some sand near top

2.5Y4/3

5Y4/3

transitional

5GY6/1

5GY5/1

5BG4/1

5Y4/2


floodplain

backswamp

natural levee and floodplain

NC-94-05

0/ 2210

50/2160 -

100/2110 -

150 /2060 -

200/2010 -

250/ 1960

300/1910 -

350/1860 -

400/1810 -

450/1760 -

500/ 1710 -

550/1660 -

600/1610 -

650/1560 -

700/1510 -

750/1460 -

800/1410 -

850/1360 -

900/1310 -

950/1260 -

,Ia

rA

0

au

o

C)

?~

Q ..


0 ) -P < C 00

5Y5/3

2.5Y4/4

slightly clayey silt qty. 800: 0.0% F, 0.0% B, 100.0% R 2.5Y3/2

5Y4/2

5Y4/4

muddy fine sand

mud

( I ) poorly sorted muddy,

fine to coarse sand

interbedded muds and silts with some fine sand

and clay interlayers

2.5Y4/4


floodplain

crevasse splay

qty. 559: 1.6% F, 0.0% B, 98.4% R interdistributary bay

5G4/1

5BG4/1


delta front

NC-94-08

(WI) 0/425

50/375 -

100/325 -

150 /275 -

200/225 -

250/ 175 -

300/ 125 -

350/75 -

400/25 -

450 /-25 -

500 / -75 -

550 /-125

600 /-175 -

650 /-225 -

700 /-275 -

750 /-325 -

800 /-375 -

850 /-425 -

900 / -475 -

, _ .. .. _ .. .. _

sea level

, 0 CD.

o _ .. ..

? _ _ _ _

.. _ .. .

a)

E au

0

c) 0D

(D

-

a)

a) 0

0 0

* _

- - - - - - - - _


o= FJ 4~- O~ 00 I I I I I

2.5Y4/3

2.5Y4/4

IOYR3/2 slightly clayey silt

5Y4/3

qty. 276: 0.4% F, 0.0% B, 99.6% R

qty. 90- 21.1% F, 0.0% B, 78.9% R

qty. 192: 3. 1 % F, 0.5% B, 96.4% R


floodplain

qty. 58. O0% F,0.O% B,I 0O.O%R w

muddy fine sand

mud

coarse, gravelly sand grading upward to silt

5Y5/4

5BG5/1

5G4/1

qty. 46: 2.2% F,O0.0% B, 97.8% R T qty 50: 56.2% F, 13.8% B, 30.0% R

crevasse splay


delta distributary channel grading upward

to subaqueous levee

NC-94-09

kL/)0

tF) (ON

0 /500

50 /450

100 / 400

150 /350

200 /300

250 /250

300 /200

350 / 150

400 /100

450 /50

500/ 0

550 / -o

600 / -100

650 / -iO

700 / -200

750 /-250

800 /-300

850 /-350

CD

. . . . . . . . .

.... ..... s a le e

~ -CD

00

0

C.)

C) a)

a)

0 10

0)

- - - - - - - - -

. _.._. ... . . . ... . .

. .._ . .._. .._. .._. .

.. ... .........

...........

- sea level

..-.... ......

:. .?':..' ... . . . .. ..

. ..... ... _ ..

K _ . . . . _ .. _ . . ?

_ . . _ . . . .:_z.:. , .


I I I I I

2.5Y6/4


mud


fine sandy silt


interbedded muds and silts with some fine to medium sandy layers

very poorly sorted, slightly muddy, fine to

gravelly sand

2.5Y5/3

2.5Y4/3

5Y8/3

5Y4.5/2

2.5Y5/4

2.5Y4/4

5BG5/1

5B4/1


floodplain

natural levee

floodplain

delta front and interdistributary bay


NC-94-11

0/ 600

50/550

100/500

2 150/450

e. 200 /400

.. 250 /350

0g 300 / 300 o

- 350 /250

> 400 / 200

450/ 150

m 500/100 o > 550 / 50

'*3 600/0

O 650/ -50 '

700/-100

, 750 /-150

800/ -200

O 850/-250

900 /-300

, 950/ -350

; 1000 /-400

1050/-450


(= " ~- Cl~ 00 0>

2.5Y5/4

5Y4/3

2.5Y4.5/4


2.5Y5/4

peat and peaty mud

interbedded muds, silts, and muddy sands with some peaty horizons

laminated clays,

muds, and silts

BEDROCK

transitional N4/

5G5/1

5G4/1

qty. It: 90.9% F, 9. 1 B, 00% R

qty. 39: 7.7% F, 71.8% B, 20.5% R

qty. 146. 1.4% F, 39.7% B, 58.9% R

qty. 11 3: 1.8%F, 98.2% B, 0.0% R

qty. 11 1: 2.7% F, 90. 1 %B, 7.2% R

qty. 143: 0.0% F, 100.0% B, 0.000 RI


floodplain

delta top marsh

delta top and front and interdistributary bay

BEDROCK

NC-94-12

o 500

50/450-

100/400-

150/350-

200 300-

250/250-

300/200-

350 150-

400/100-

450 /50-

500 /0-

550 / -o

600/-100-

650 -150-

700/-200-

750 /-250-

800 /-300-

850 /-350-

0 . . . .

. . .. .. . . . .. ..............n

sea. ..level.

. . .. .. . . . 0 .

. . .. .. . . . .. .. . . <. . .

00-1

a) $a)

a) 0D

a)

15

a) 00

a)-

0U

0D

0

a)

a)

Sediment Type

0/390

50/340 -

100/290 -

150 / 240 -

200 / 190 -

250/ 140 -

300/90 - s'-' 350/40 -

> 400 /-10 -

450 /-60 -

m 500 /-110 -

> 550 /-160 -

' 600 /-210 -

= 650/-260 -

700 / -310 -

, 750 / -360 -

- 800/-410

850 /-460 -

900/-510 -

950/ -560 -

1000 /-610 -

1 1050 /-660 -

.. ... ......

..- .- _ sea level _ _ - (

- )

', c -14: 850 -+80/-60 BP

. , '.-. . _.._. , .. . .

_ ~ ? _-I

*_ - 3 U .:_::=:: CD x,C-4 5


o I - ON 00

2.5Y5/3


slightly silty clay

mud with silty laminae toward bottom

fine sandy mud grading upward to mud

2.5Y4/4

2.5Y4/3

transitional

5GY4.5/1

5BG5/1

5GY3/1

mud with interbedded clay laminae

interbedded muds, silts, and muddy sands

5GY4/1

laminated clays, muds, and silts

qty. 21: 42.9% F, 0.0% B, 57.1% R

qty. 48: 43.8% F, 0.0% B, 56.2% R

qty. 11: 90.9% F, 0.0% B, 9.1% R

qty. 0: 0.0% F, 0.0% B, 0.0% R I

qty. 10: 0.0% F, 0.0% B, 100.0% R

qty. 423: 4.0% F, 13.3% B, 82.7% R

qty. 119: 0.0% F, 100.0% B, 0.0% R

qty. 101: 0.0% F, 100.0% B, 0.0% R


floodplain


delta distributary mouth bar

delta front


NC-94-13 n

',D

i

- sea level

I I I I I I

C--~. _ _ ,.

_ _ _ _ _~ ~ ~~~C _ _ _ _ _~ ~~~~C

Sediment Type


(_/) mud

Color

2.5Y4/2

2.5Y5/2


'I I I I I

A

I I I I I

/

( )

clay 5BG4/1

'IL,

peat and peaty mud 5GY4/1


floodplain


(Acherousian lake)

brackish delta top marsh grading upward into freshwater marsh

qty. 292: 0.0% F, 60.6% B, 39.4% R v

poorly sorted sand delta top

NC-94-17

Oc 0

--4

a)

-

(D

o *

0

-0

0

. Pg

5~

0/ 480

50 /430 -

100/380 -

150/330 -

200/280 -

250/230 -

300 / 180 -

350 / 130 -

400 / 80 -

450/30 -

500 /-20 -

550 / -70 -

600 /-120 -

650 / -170

700 /-220 -

750 /-270 -

800 /-320 -

850 /-370 -

900 / -420

________

Sediment Type C o Cn C)D

. _. .. _. .. . .. .. .

.... ... ... - ... ( )

_. ... ... . _. a ._. .

sea level _ ._- - ._ .- - -

C-14. 1670

40 /-120 BP


o Ki . oC oo0 I I I I I

2.5Y5/3


silty fine sand grading upward to

sandy silt

-mud

coarse, gravelly sand grading upward to

silty fine sand

2.5Y4/3

2.5Y5/4

qty. 54: 0.0% F, 0.0% B, 100.0% R

5BG5/1

qty. 2: 0.0% F, 100.0% B, 0.0% R I

BEDROCK


floodplain

subaerial natural levee

delta distributary channel and

distributary mouth bar grading upward into

F subaqueous levee BEDROCK

NC-94-20

0/ 470

50 /420 -

100/370 -

150/320 -

200/270 -

250/220 -

300/ 170 -

350/ 120 -

400 / 70 -

450/20 -

500 /-30 -

550 /-80 -

600 /-130 -

650 /-180 -

au ca

r0

0

a)

(D

0 -a o

Sediment Type

. ._. ._._ .. ... .

. _. ._ .. ._. .. .


CL _ . . .

i0 cm ibv se leve

_ _ _ _ , , , , _ _ _ _ _


oI o I I I I I

2.5Y5/4


peat and peaty mud

2.5Y5/3

transitional

5B5/1

5G5/1

slightly silty clay

5BG5/1

slightly clayey silt 5Y6/4


floodplain

freshwater marsh

backswamp

I floodplain

NC-94-21

0 / 1000

50 / 950

100 / 900

150/850

200 / 800

250 / 750

300 / 700

350 / 650

400 / 600

450 / 550

500 / 500

550/450

600 / 400

CA

r4I

01 0

0


I.- wJ C) 0> 0 I I I i


mud

slightly silty clay

2.5Y4/2

5Y4/3

5Y5/2

5Y4/3

clay

qty. 16: 6.3% F, 0.0% B, 93.7% R

qty. 229: 10.50 F, 0.00 B, 89.50 R

qty. 24: 12.50 F, 0.00 B, 87.50 R

qty. 53. 3.8% F, 0.0% B, 96.2% R

qty. 11: 36.4% F, 0.0% B, 63.6% R

qty. 6: 100.000 F, 0.000 B, 0.000 R

qty. 7: 14.300 F, 0.0% B, 85.70 R

qty. 21: 71.4% F, 0.000B, 28.6%R

qty. 122: 100.000 F, 0.000 B, 0.0% R


floodplain


(Acherousian lake)

5BG65/1

peat and peaty mud

laminated clays, silts, and fine sands

5G5/1 for sediment; and 2.5YR2.5/1

for peat

5G5/1

qty. 96: 100.000 F, 0.000 B, 0.000 R

qty. 33: 100.000 F, 0.000 B, 0.000 R

qty. 68: 95.600F, 4.400B, 0.000R

qty.879: 010I%F, 889% B, 1100oR 1

qty. 615: 1.30oF, 91.40oB, 73% R

brackish delta top marsh grading upward into freshwater marsh

delta top and front

qty. 579: 1.4% F, 92.2% B, 6.40 R

NC-94-23

(\iI/)

a) 0

15 a)

a)

a1)

ra

0) 0

la

0/520

50/470-

100/420-

150/370-

200/320 -

250/270 -

300 220 -

350/170-

400/120-

450 /70-

500 /20-

550 /-30-

600 /-80-

650 /-130-

700 /-180

750 -230-

800/-280-

850 -330-

900/-380

950 -430-

l000/-480

1050 -530 -

1100 -580 -

1150/-630-

1200 -680

U) CD

sea level

O .- CD0

\LI C-1: 03

+10 10 B

(-ML.

qty. 5: 100.000F, 0.000B, 0.000R

k

Sediment Type Color % Organic Content

- - - - - - - -

0

0

0

CHAPTER 7

SUMMARY OBSERVATIONS

by James Wiseman and Konstantinos Zachos

1. Including all the questions listed in Chapter 1, pp. 8-9.

This first volume of the results of the Nikopolis Project provides the theoretical and methodological underpinnings of the research and describes the changes in the landscape of southern Epirus from the time of the earliest human inhabitants (more than 250,000 years ago) up to the present. These reports constitute the framework into which will be set the remaining results of the project (to be presented in volume 2), including discussions of the changing patterns of settlement and land use revealed by the diachronic survey. At the same time, the reports in this volume provide in themselves contributions both in substantive results and in the several cri- tiques of methodologies. The authors have endeavored in all cases to be explicit about the aims of the research, how the investigations were carried out (what worked and what did not), the constraints on the fieldwork and analyses (however imposed), and the significance of the results. A final assessment of the significance of the project's results, however, must await the publication of volume 2.

The broad aim of the Nikopolis Project-to explain the changing relationships between humans and the landscape they inhabited in southern Epirus-required an intensely interdisciplinary approach. In order to study humans in their landscape, it was essential to determine early in the investigation just what that landscape was, and to develop parameters of as many other environmental factors as possible. The collaboration of archaeologists and geologists was vital not only to the general aim of the project, but also to many of the research questions concerned with problems or issues belonging to specific time periods.1 It is important to stress this close collaboration because it affected almost all aspects of the project, from conception to publication. Geologic and geographic/political con- siderations, for example, played a greater role in determining the boundaries and size of the zone to be investigated than the likely area that could be walked by archaeological survey teams. We were well aware from the beginning that 1,200 km2 constituted too large an area for intensive survey over even most of it, much less all of it, and such a survey was never intended. It was our aim instead to test by survey all the different kinds of environmental zones, and eventually to focus the diachronic survey on a few regions of particular interest, as determined both by cultural and envi-


ronmental factors. As discussed in Chapter 2, two of those regions were the Ayios Thomas peninsula and the lower Acheron valley. Still, the diachronic survey was a component of the project, not the definition of the project itself. In geologic terms, the size of the project area was no impedi- ment; the boundaries enclosed a reasonably coherent area based on the lower courses of the two principal rivers, the Acheron and the Louros, while still providing diverse environmental zones (a desideratum of the research design), ranging from coastlines and marshes to inland valleys, upland plains, and rugged mountains.2 The survey zone, essentially the modern nomos of Preveza, includes most, perhaps all, of the territorium controlled directly by Nikopolis in Roman times,3 a useful unit of analysis for a time period of particular interest to the project. A minor consideration was that we were also able to test the applicability of remote-sensing imagery to a large regional study.

The papers in this volume provide substantial evidence of the utility of this combined geologic-archaeological approach. Tartaron's report on methodology (Chapter 2) details several of the ways that the operation of the survey benefited from the geologic components, ranging from selection of areas for survey to the interpretation of certain phenomena observed in the field. The theoretical and methodological discussions in that same chapter place the conduct of the survey in its historical context, and provide sufficient detail, we hope, for readers to assess the significance of the survey both in its relationships to other components of the project and (ultimately) in understanding the cultural and environmental factors affecting the distribution of artifacts across the landscape.

The discovery early in the 1991 field season by Runnels and van Andel of the first Lower Palaeolithic tool-an Acheulean handaxe-found in southeast Europe in a secure geologic context was the precursor to a series of discoveries and analyses that made possible their presentation of a coherent evolution of early human habitation in Epirus (Chapter 3). Their explanation of the creation and evolution of poljes and loutses in the dynamic karst landscape of Epirus is both an important contribution to geomorphology and a basis for understanding the attractions of the region through time. Small bands of Lower Palaeolithic hunter-gatherers were drawn to the lakes and ponds that accumulated in the karst depressions, not only as a source of water but also for the birds and other animals that gathered there. Erosion of the surrounding limestone and along the associated streams also made available concentrations of flint. The Middle Palaeolithic inhabitants, more numerous than their predecessors millennia before, were equally attracted to the lakes, marshes, and swamps scattered across the landscape from the Louros gorge to the ridges that (now) overlook the Ionian Sea. The specialized nature of many of the Middle Palaeolithic sites, which were seasonally revisited, suggests that the foraging groups-Neanderthals or archaic Homo sapiens-were following logistical patterns.

The Early Upper Palaeolithic (EUP), the period during which modern Homo sapiens replaced Neanderthals, is represented more sparsely. Runnels and van Andel point out that the small scatters of EUP artifacts at polje and loutsa sites indicate a pattern of land use different from that of

2. See the discussion in Chapter 1, pp. 2-3, and in Chapter 2.

3. The principal question concerns the eastern boundary. On the northeast it terminated at the territory of Photike; the ancient town has been identified as the archaeological site some 3 km south of the town of Paramythia in the plain of Chrysauge, but its territorial limits are uncertain (see Samsaris 1988). The territory of Nikopolis may have included part of the deltaic plain of the Arachthos River west of Ambracia, the ancient Corinthian colony, which was evidently in ruins at the time of the founding of Nikopolis; see Doukellis 1990.

266

SUMMARY OBSERVATIONS

earlier periods. Anomalous for this time period is the site of Spilaion near the (present) mouth of the Acheron, perhaps the largest lithic site in Greece. It is an open-air Aurignacian site, rare in Greece, and extraordinarily rich in lithic artifacts, which are estimated to total some 150,000. In Chapter 4, Runnels, Karimali, and Cullen report on their spatial analysis of the

material, which demonstrates the existence of specific areas of activity within the site. The success of the spatial analysis is testimony to its utility in the

study of artifact-rich sites. No new sites of the Late Upper Palaeolithic, which was of short duration in Epirus, were discovered by the project, but six Mesolithic sites were added to the small number previously known in Greece. All these new sites, discussed by Runnels and van Andel, were

along the new Holocene coast of the Ionian Sea.

Chapter 5 is a detailed account by Jing and Rapp of the geomorphologic changes over the past 10,000 years of the north coast of the Ambracian Gulf and the Nikopolis peninsula. Their extensive program of geologic cores, laboratory analyses, and repeated examinations of the landscape has resulted in an understanding of the dramatic changes over time in these

regions, which are displayed in a series of maps showing paleogeographic reconstructions of the Nikopolis peninsula and the northern coast of the Ambracian Gulf from ca. 6500 to 500 B.P. They have also determined the

principal geologic and environmental forces that brought about the changes, as well as some of the cultural influences that were also at work. They demonstrated that eustatic sea-level rise from the melting of glaciers was the dominant force in determining coastal change from ca. 13,000 to ca. 6500 B.P., after which tectonic subsidence caused the sea level to continue to rise, but more slowly, until ca. 4500 B.P. when maximum marine trans-

gression was reached; at that time, the Ambracian embayment extended north to Mts. Rokia and Stavros, leaving at most a narrow passage at their base. Relative sea level continued to rise for the next 3,000 years, until about A.D. 500, because tectonic subsidence proceeded at a rate greater than the accumulation of sediment from rivers entering from the north. At that time, the amount of sediment from rivers and streams, and from erosion, exceeded the relative sea-level rise, and the northern shoreline moved gradually into the gulf, incorporating islands and creating lagoons and swamps.

Jing and Rapp correlate these dramatic changes with some of the notable archaeological sites in the region, especially Nikopolis and its harbor on the peninsula and Kastro Rogon, the ancient Bouchetion, near the mouth of the Louros River gorge. They show, for example, that the latter was

originally a town on a small island near the coast (the gap was bridgeable), and so could have served as a regional port town itself. In late antiquity it became an inland town, and was only reconnected to the sea when the Louros River was diverted by human means in the medieval period; as a result of this diversion, the river flowed beneath the town walls in a new channel that led along the mountains and the Grammeno plain to enter the gulf near Nikopolis.

Besonen, Rapp, and Jing in Chapter 6 document the equally dramatic changes during the Holocene in the lower Acheron River valley and in Phanari Bay, the ancient Glykys Limen (or "Sweet Harbor") at the present

267


mouth of the river. Geologic cores, again, proved indispensable to the paleogeographic reconstructions, and to the understanding of factors involved in coastal change and the evolution both of the bay and the lower Acheron valley. The much larger bay that existed during the Bronze Age and classical antiquity, with two entrances from the Ionian Sea, now makes under- standable the accounts in ancient literature of the great fleets that could be accommodated there. The size of the earlier bay enhanced the strategic location of the Bronze Age site identified as Ephyra, on the hill of Xylo- kastro, which closes the eastern side of the earlier main portion of the bay. Other results include the resolution of the long-standing problem of the location of the Acherousian lake, mentioned by ancient writers, and its relationship over time with the Acheron River and the fortified urban settlement upstream, known locally as Kastri and which may be the ancient Pandosia.

We close this chapter with a modest disclaimer: as editors, we have here presented summary observations on the reports in volume 1, not a set of final conclusions, which will follow in volume 2. That volume will also contain reports on the study of the cultural remains, along with the integration of the results of the diachronic survey with the geomorphologic investigations that have been presented here.

268

REFERENCES

Adam, E. 1989. A Technological and TypologicalAnalysis of Upper Palaeolithic Stone Industries of Epirus, Northwestern Greece (BAR-IS 512), Oxford.

Alcock, S. E. 1991. "Urban Survey and the Polis of Phlius," Hesperia 60, pp. 421-463.

.1993. Graecia Capta: The Land- scapes of Roman Greece, Cambridge.

Alcock, S. E.,J. F. Cherry, andJ. L. Davis. 1994. "Intensive Survey, Agricultural Practice, and the Classical Landscape of Greece," in Classical Greece: Ancient Histories and Modern Archaeologies, ed. I. Morris, Cambridge, pp. 137-170.

Allsworth-Jones, P. 1986. The Szeletian and the Transition from Middle to Upper Paleolithic in Central Europe, Oxford.

Ammerman, A.J. 1981. "Surveys and Archaeological Research," Annual Review of Anthropology 10, pp. 63- 88.

. 1985. "Plow-Zone Experi- ments in Calabria, Italy,"JFA 12, pp. 33-40.

. 1993. Rev. of Cherry, Davis, and Mantzourani 1991, inJFA 20, pp.367-372.

Arsebiik, G. 1993. "Yarimburgaz, a Lower Paleolithic Cave Site near Istanbul," in Between the Rivers and Over the Mountains: Archaeologica anatolica et mesopotamica Alba Palmieri dedicata, ed. M. Frangi- pane, H. Hauptmann, M. Liverani, P. Matthiae, and M. Mellink, Rome, pp. 23-36.

. 1996. "The Cave of Yar- imburgaz (the Oldest Stratified Site Yet Known in Turkey)," in Vom Halys zum Euphrat (Alter- tumskunde des vorderen Orients 7), ed. T. Beran, Miinster, pp. 1- 13.

Ascoli, P. 1964. "Preliminary Eco- logical Study on Ostracoda from Bottom Cores of the Adriatic Sea," Pubblicazioni della Stazione

Zoologica di Napoli 33 (suppl.), pp.213-246.

Ashmore, W., and A. B. Knapp, eds. 1999. Archaeologies of Landscape: Contemporary Perspectives, Malden, Mass.

Aubouin, J. 1959. "Contribution a l'etude geologique de la Grece septentrionale: Les confins de l'Epire et de la Thessalie," Annales

geologiques des pays helleniques 10, pp. 1-525.

. 1965. Geosynclines, Amsterdam.

Bailey, G. 1992. "The Paleolithic of Klithi in Its Wider Context," BSA 87, pp. 1-28.

, ed. 1997. Klithi: Palaeolithic Settlement and Quaternary Land- scapes in Northwest Greece, 2 vols., Cambridge.

Bailey, G. N., E. Adam, E. Pana- gopoulou, C. Perles, and K. Zachos, eds. 1999. The Palaeo- lithicArchaeology of Greece and AdjacentAreas: Proceedings of the ICOPAG Conference, Ioannina, September 1994 (BSA Studies 3), London.

REFERENCES

Bailey, G. N., P. L. Carter, C. S. Gam- ble, and H. P. Higgs. 1983a. "Epirus Revisited: Seasonality and Inter-Site Variation in the Upper Palaeolithic of North-West Greece," in Hunter- Gatherer Economy in Prehistory: A European Perspective, ed. G. Bailey, Cambridge, pp. 64-78.

. 1983b. "Asprochaliko and Kastritsa: Further Investigations of Palaeolithic Settlement and Economy in Epirus (North-West Greece)," PPS 49, pp. 14-42.

Bailey, G., G. King, and D. Sturdy. 1993. "Active Tectonics and Land- Use Strategies: A Palaeolithic Example from Northwest Greece," Antiquity 67, pp. 292-312.

Bailey, G. N.,J. Lewin, M. G. Mack- lin, and J. C. Woodward. 1990. "The 'Older Fill' of the Voidomatis Valley, Northwest Greece, and Its Relationship to the Palaeolithic Archaeology and Glacial History of the Region,"JAS 17, pp. 145- 150.

Bailey, G., V. Papaconstantinou, and D. Sturdy. 1992. "Asprochaliko and Kokkinopilos: TL Dating and Reinterpretation of Middle Palaeolithic Sites in Epirus, Northwest Greece," CAJ2, pp. 136- 144.

Barbaroux, L., and B. Bousquet. 1976. "Contribution a l'etude des sols mediterraneens: Les sols rouges du domaine hellenique de Grece occidentale," Geologie mdditerraneenne 3, pp.237-250.

Bard, E., M. Arnold, R. G. Fairbanks, and B. Hamelin. 1993. "23?Th-234U and 14C Ages Obtained by Mass Spectrometry on Corals," Radio- carbon 35, pp. 191-199.

Bard, E., B. Hamelin, and R. G. Fair- banks. 1990. "U-Th Ages Obtained by Mass Spectrometry in Corals from Barbados: Sea Level during the Past 130,000 Years," Nature 346, pp. 456-459.

Bard, E., and B. Kromer. 1995. "The Younger Dryas: Absolute and Radiocarbon Chronology," in The Younger Dryas: Proceedings of a Workshop at the Royal Netherlands Academy of Arts and Sciences on 11-13 April 1994 (Verhandelingen der Koninklijke Nederlandse

Akademie van Wetenschappen, Afdeling Natuurkunde, Eerste Reeks 44), ed. S. R. Troelstra, J. E. van Hinte, and G. M. Ganssen, Amsterdam, pp. 161-166.

Barker, G. 1991. "Approaches to Archaeological Survey," in Roman Landscapes: Archaeological Survey in the Mediterranean Region, ed. G. Barker and J. Lloyd, London, pp. 1-9.

Barker, G., and D. Mattingly. 1999. "General Editors' Introduction: The POPULUS Project," in Reconstructing Past Population Trends in Mediterranean Europe (3000 B.C-A.D. 1800) (The Archaeol- ogy of Mediterranean Landscapes 1), ed. J. Bintliff and K. Sbonias, Oxford, pp. iii-ix.

Bar-Yosef, 0. 1998. "Early Coloniza- tions and Cultural Continuities in the Lower Palaeolithic of Western Asia," in Early Human Behaviour in Global Context: The Rise and Diversity of the Lower Paleolithic Record, ed. M. D. Petraglia and R. Korisettar, London, pp. 221-279.

Bennett, K., P. C. Tzedakis, and K. Willis. 1991. "Quaternary Refugia of North European Trees," Journal of Biogeography 18, pp. 103- 115.

Besonen, M. 1997. "The Middle and Late Holocene Geology and Landscape Evolution of the Lower Acheron River Valley, Epirus, Greece" (M.S. thesis, University of Minnesota).

Bhatia, S. B. 1968. "Pleistocene Ostra- codes from the Upper Karewas of Kashmir, India," Micropaleontology 14, pp. 465-483.

Binford, L. R. 1964. "A Consideration of Archaeological Research Design," AmerAnt 29, pp. 425-441.

Bintliff, J. 1985. "The Boeotia Survey," in Archaeological Field Survey in Britain andAbroad, ed. S. Macready and F. H. Thompson, London, pp.196-216.

. 1997. "Regional Survey, Demography, and the Rise of Complex Societies in the Ancient Aegean: Core-Periphery, Neo- Malthusian, and Other Interpretive Models,"JFA 24, pp. 1-38.

270

REFERENCES

Bintliff, J., and A. Snodgrass. 1985. "The Cambridge/Bradford Boeotian Expedition: The First Four Years," JFA 12, pp.123-161.

. 1988a. "Off-Site Pottery Distributions: A Regional and

Interregional Perspective," CurrAnthr 29, pp. 506-513.

. 1988b. "Mediterranean Sur- vey and the City," Antiquity 62, pp. 57-71.

Birkeland, P. W. 1984. Soils and

Geomorphology, New York. . 1999. Soils and Geomorphology,

3rd ed., New York. Bockheim, J. G., J. G. Marshall, and

H. M. Kelsey. 1996. "Soil-Forming Processes and Rates on Uplifted Marine Terraces in Southwestern Oregon, USA," Geoderma 72, pp.39-62.

Boero, V., and U. Schwertmann. 1987. "Occurrence and Transformation of Iron and Manganese in a Col- luvial Terra Rossa Toposequence of Northern Italy," Catena 14, pp.519-531.

. 1989. "Iron-Oxide Mineral- ogy of Terra Rossa and Its Genetic Implications," Geoderma 44, pp. 319-327.

Bommelj6, S., and P. K. Doom. 1987. Aetolia and the Aetolians: Towards the Interdisciplinary Study of a Greek Region, Utrecht.

Bordes, F. 1992. Lefons sur le Paleolith-

ique 2: Le Pal6olithique en Europe, Paris.

Bottema, S. 1974. "Late Quaternary Vegetation History of Northwest- ern Greece" (diss. University of Groningen).

. 1978. "The Late Glacial in the Eastern Mediterranean and the Near East," in The Environ- mental History of the Near and Middle East since the Last Ice Age, ed. W. C. Brice, New York, pp. 15-28.

. 1994. "The Prehistoric Environment of Greece: A Review of the Palynological Record," in Kardulias 1994b, pp. 45-68.

.1995. "The Younger Dryas in the Eastern Mediterranean," Quaternary Science Reviews 14, pp. 883-892.

Bradley, R., T. Durden, and N. Spen- cer. 1994. "The Creative Use of Bias in Field Survey," Antiquity 68, pp. 343-346.

Buidel, J. 1973. "Reliefgenerationen der Poljebildung im Dinarischen Raum," Geographische Zeitschrift 32, pp.134-142.

Bull, W. B. 1972. "Recognition of Alluvial Fan Deposits in the Strati- graphic Record," in Recognition of Ancient Sedimentary Environments (Society of Economic Paleontolo- gists and Mineralogists Special Publication 16), ed. J. K. Rigby and W. K. Hamblin, Tulsa, pp. 68-83.

. 1977. "The Alluvial Fan Environment," Progress in Physical Geography 1, pp. 223-270.

Cavanagh, W., R. Jones, and A. Sarris. 1996. "The Phosphate and Geo- physical Surveys," in The Laconia Survey: Continuity and Change in a Greek Rural Landscape 2: Archaeo- logical Data (BSA Suppl. 27), ed. W. Cavanagh et al., London, pp.235-261.

Chappell, J., A. Omura, T. Esat, M. McCulloch, J. Pandolfi, Y. Ota, and B. Pillans. 1996. "Reconciliation of Late Quaternary Sea Levels Derived from Coral Terraces at Huon Penin- sula with Deep-Sea Oxygen Isotope Records," Earth and Planetary Science Letters 141, pp. 227-236.

Chappell,J., and N.J. Shackleton. 1986. "Oxygen Isotopes and Sea Level," Nature 324, pp. 137-140.

Cheddadi, R., and M. Rossignol- Strick. 1995. "Eastern Mediterra- nean Quaternary Paleoclimates from Pollen and Isotope Records of Marine Cores in the Nile Cone Area," Paleoceanography 10, pp. 291- 306.

Chen, J. H., H. A. Curran, B. White, and G. J. Wasserburg. 1991. "Precise Chronology of the Last Interglacial Period: 234U/230Th Data from Fossil Coral Reefs in the Bahamas," Geological Society of America Bulletin 103, pp. 82-97.

Cherry, J. F. 1982. "A Preliminary Definition of Site Distribution on Melos," in An Island Polity: The Archaeology of Exploitation in Melos, ed. C. Renfrew and M. Wagstaff, Cambridge, pp. 10-23.

.1983. "Frogs Round the Pond:

Perspectives on Current Archaeo- logical Survey Projects in the Mediterranean Region," in Keller and Rupp 1983, pp. 375-415.

. 1984. "Common Sense in Mediterranean Survey?"JFA 11, pp.117-120.

. 1994. "Regional Survey in the Aegean: The 'New Wave' (and After)," in Kardulias 1994b, pp. 91- 112.

Cherry, J. F., J. L. Davis, A. Demitrack, E. Mantzourani, T. F. Strasser, and L. E. Talalay. 1988. "Archaeological Survey in an Artifact-Rich Land- scape: A Middle Neolithic Example from Nemea, Greece," AJA 92, pp.159-176.

Cherry, J. F., J. L. Davis, and E. Mant- zourani. 1991. Landscape Archaeology as Long-Term History: Northern Keos in the Cycladic Islandsfrom Earliest Settlement until Modern Times (Monumenta archaeologica 16), Los Angeles.

Cherry, J. F., J. L. Davis, E. Mantzour- ani, and T. M. Whitelaw. 1991. "The Survey Methods," in Cherry, Davis, and Mantzourani 1991, pp. 13-35.

Chrysos, E., ed. 1987. Nicopolis I: Proceedings of the First International Symposium on Nicopolis (23-29 September 1984), Preveza.

Clark, G. A. 1994. "Migrations as an Explanatory Concept in Paleolithic Archaeology,"Journal of Archaeologi- cal Method and Theory 1, pp. 305- 343.

Clews, J. 1989. "Structural Controls on Basin Evolution: Neogene to Qua- ternary of the Ionian Zone, Western Greece,"Journal of the Geological Society of London 146, pp. 447-457.

Coussot, P., and M. Meunier. 1996. "Recognition, Classification, and Mechanical Description of Debris Flows," Earth Science Reviews 37, pp.225-252.

Culiberg, M., and A. Sercelj. 1996. "Slovenia," in Palaeoecological Events during the Last 15,000 Years: Regional Syntheses of Palaeoecological Studies of Lakes and Mires in Europe, ed. B. E. Berglund, H.J. B. Birks, M. Ralska-Jasiewicowa, and H. E. Wright, London, pp. 687-700.

271

REFERENCES

Cullen, H. M., P. B. deMenocal, S. Hemming, G. Hemming, F. H. Brown, T. Guilderson, and F. Sir- ocko. 2000. "Climate Change and the Collapse of the Akkadian Empire: Evidence from the Deep Sea," Geology 28, pp. 379-382.

Dakaris, S. I. 1958. "AvocoxapLtxal ?p?ovoa ?tL; TYv OuYqp(LXqV E(p6pav xao TO N?xooyaovT?sov rq{ Apxocaa; OerpwOrxToc;," Prakt, pp. 107-113.

. 1960. "Avacxawpy Too N?XOo- acvT?-ro Too Ax?povTO; xac OoXto-

TO6 TacDpoo 7tXfOLOv Tj; IIapyao," Prakt, pp. 114-127.

. 1961. "AvaoxacpYi ect; TO Nexoo[avrTsLOv TOO Ax?povTO;," Prakt, pp. 108-119.

. 1962. "The Dark Palace of Hades," Archaeology 15:2, pp. 85-93.

. 1963. "Avaaoxaqocp etg TO N?XuoOiaVT?lOV Too AxpoVTO;," Prakt, pp. 89-92.

. 1964. "Avaoxacpy ?L<; TO N?xooV[acVT6oV Too Ax?pOVTO;," Prakt, pp. 44-53.

. 1971. Cassopaia and the Elean Colonies (Ancient Greek Cities 4), Athens.

. 1972. O?TrpcoTLr'a (Ancient Greek Cities 15), Athens.

. 1975a. "Avaoxaocpyl E TO

N?xooiaVT?Lrov Too AXE?pOVTOg," Prakt, 146-152.

. 1975b. TheAntiquity of Epirus: TheAcheron, Necromanteion, Ephyra, Pandosia, Cassope, Athens.

. 1977. "N?xooatxvt-r?Lov. E6popa. Kaooo&TrC," Prakt, pp. 140-148.

. 1978. "Avacxacpq oaTrv Koco- aoTrct H7?ipoo," Prakt, pp. 99-106.

.1979. "Avaoxawp oTYv Kao- XctW7n H7?rtpoo," Prakt, pp. 114-118.

.1980. "AvaoxaxpY oaqv Kac- cOytrn Htr?(poo," Prakt, pp. 21-32.

.1981. "Avacxawp] OTYMv Kao- ornqT," Prakt, pp. 72-77.

.1982. "Avao7xaoqp7 Kaoc7trqTM," Prakt, pp. 79-84.

.1983. "Avaooxaocp) oTMV Ka7aon7t," Prakt, pp. 69-77.

. 1993. The Nekyomanteion of the Acheron, Athens.

Dakaris, S., E. S. Higgs, and R. W. Hey. 1964. "The Climate, Environment, and Industries of Stone Age Greece: Part I," PPS 30, pp. 199-244.

Dalfes, H. N., G. Kukla, and H. Weiss, eds., 1997. Third Millen- nium B.C. Climate Change and Old World Collapse (NATO ASI Series 1, Global Environmental Change 49), Berlin.

Darlas, A. 1989. "H Optv6xota a6o- T?rXvca too EXLtoxoptoo Axxociat," ArchEph 128, pp. 137-159.

. 1995. "The Earliest Occupa- tion of Europe: The Balkans," in The Earliest Occupation of Europe: Proceedings of the European Science Foundation Workshop at Tautavel (France), 1993, ed. W. Roebroeks and T. van Kolfschoten, Leiden, pp. 51-59.

Davidson, D. A. 1980. "Erosion in Greece during the First and Second Millennia B.C.," in Timescales in

Geomorphology, ed. R. A. Culling- ford, D. A. Davidson, and J. Lewin, Chichester, N.Y., pp. 143-158.

Davis,J. L., S. E. Alcock,J. Bennet, Y. G. Lolos, and C. W. Shelmer- dine. 1997. "The Pylos Regional Archaeological Project, Part I: Overview and the Archaeological Survey," Hesperia 66, pp. 391-494.

Dean, W. E., Jr. 1974. "Determina- tions of Carbonate and Organic Matter in Calcareous Sediments and Sedimentary Rocks by Loss on Ignition: Comparison with Other Methods,"Journal of Sedi- mentary Petrology 44, pp. 242-248.

Debenath, A., and H. L. Dibble. 1994. Handbook of Paleolithic Typology 1: Lower and Middle Paleolithic of Europe, Philadelphia.

Demitrack, A. A. 1986. "The Late Quaternary History of the Larissa Plain, Thessaly, Greece: Tectonic, Climatic, and Human Impact on the Landscape" (diss. Stanford University).

D'Errico, F., J. Zilhao, M. Julien, D. Baffier, and J. Pelegrin. 1998. "Neanderthal Acculturation in Western Europe? A Critical Review of the Evidence and Its Interpretation," CurrAnthr 39, pp. 1-44.

Devoto, G. 1965. "Lacustrine Pleis- tocene in the Lower Liri Valley," Geologica romana 4, pp. 291-368.

Digerfeldt, G., S. Olsson, and P. Sand-

gren. 2000. "Reconstruction of

272

REFERENCES

Lake-Level Changes in Lake Xin- ias, Central Greece, during the Last 40,000 Years," Palaeogeography, Palaeoclimatology, and Palaeoecology 158, pp. 65-82.

Doelle, W. H. 1977. "A Multiple Sur- vey Strategy for Cultural Resource Management Studies," in Conserva- tion Archaeology:A Guidefor Cultural Resource Management Studies, ed. M. B. Schiffer and G.J. Gumerman, New York, pp. 201-209.

Doukellis, P. 1990. "Evoa Ox-ruo aypo- TLXW)v opLtov oJmv Tst&8ioaa -M 'Ap-rac," MeA&E,yaTa x roo KE.R.A.

10, pp. 269-286. Doutsos, T., N. Kontopoulos, and

D. Frydas. 1987. "Neotectonic Evolution of Northwestern- Continental Greece," Geologische Rundschau 76, pp. 433-450.

Douzougli, A. 1998a. "Kotv6orrra EzpoyyuoX;," ArchDelt 48, B' (1993), pp. 282-285.

.1998b. "Mta p(oPalixYl aoypot-

xoc c-ri; acxxr; T-oo Ali3paxxtxo6 x6X7ou," ApXatoaoyia xcat TXvye, 68, pp. 74-78.

Downum, C. E., and G. B. Brown. 1998. "The Reliability of Surface Artifact Assemblages as Predictors of Subsurface Remains," in Surface Archaeology, ed. A. P. Sullivan III, Albuquerque, pp. 111-123.

Dunnell, R. C., and W. S. Dancey. 1983. "The Siteless Survey: A Re- gional Scale Data Collection Strat- egy," in Advances in Archaeological Method and Theory 6, ed. M. B. Schiffer, New York, pp. 267-287.

Dunnell, R. C., and J. F. Simek. 1995. "Artifact Size and Plowzone Processes,"JFA 22, pp. 305-319.

Edwards, L. R.,J. H. Chen,T. L. Ku, and G. J. Wasserburg. 1987. "Pre- cise Timing of the Last Intergla- cial Period from Mass-Spectromet- ric Determination of Thorium-230 in Corals," Science 236, pp. 1547- 1553.

Efstratiou, N. 1985. Agios Petros, A Neolithic Site in the Northern Sporades: Aegean Relationships during the Neolithic of the 5th Millennium (BAR-IS 241), Oxford.

Ellis, B. F., and A. R. Messina. 1952- 2000. Catalogue of Ostracoda, New York.

Etude geologique = Etude g6ologique de

l'Epire (Grece nord-occidentale) (Edi- tions Technip 2), 2 vols., Paris 1966.

Fairbanks, R. G. 1989. "A 17,000 Year Glacio-Eustatic Sea Level Record: Influence of Glacial Melting Rates on the Younger Dryas Event and Deep Ocean Circulation," Nature 342, pp. 637-642.

Fish, S. K., and S. A. Kowalewski, eds. 1990. The Archaeology of Regions: A Casefor Full-Coverage Survey, Washington, D.C.

Fitzgerald, J. 1996. "Soil Profile Development Indices for Soils and Paleosols of the Mount Olympus Piedmont" (M.S. thesis, Ohio University).

Flemming, N. C. 1968. "Holocene Eustatic Movements and Eustatic Sea Level Change in the Pelopon- nese," Nature 217, pp. 1031-1032.

.1972. "Eustatic and Tec- tonic Factors in Relative Vertical Displacement of the Aegean Coast," in The Mediterranean Sea: A Natural Sedimentation Laboratory, ed. D. J. Stanley, Stroudsburg, Pa., pp. 189-201.

Flemming, N. C., and C. D. Webb. 1986. "Tectonic and Eustatic Coastal Changes during the Last 10,000 Years Derived from Archaeological Data," in Dating Mediterranean Shorelines (Zeitschrift fur Geomorphologie Suppl. 62), ed. A. Ozer and C. Vita-Finzi, Berlin, pp. 1-29.

Foley, R. 1981. "Off-Site Archaeology: An Alternative Approach for the Short-Sited," in Pattern of the Past: Studies in Honour of David Clarke, ed. I. Hodder, G. Isaac, and N. Hammond, Cambridge, pp. 157- 183.

Folk, R. L. 1980. Petrology of Sedimen- tary Rocks, Austin.

Ford, D., and P. Williams. 1989. Karst

Geomorphology and Hydrology, London.

Fotiadis, M. 1992. "Units of Data as Deployment of Disciplinary Codes," in Representations in Archaeology, ed. J.-C. Gardin and C. S. Peebles, Bloomington, pp. 132-148.

Foucault, A., and J.-F. Raoult. 1980. Dictionnaire de geologie, Paris.

Gaffney, V., and M. Tingle. 1985. "The Maddle Farm (Berks.) Project and Micro-Regional Analysis," in Arch- aeological Field Survey in Britain and Abroad, ed. S. Macready and F. H. Thompson, London, pp. 67-73.

Galanidou, N., P. C. Tzedakis, I. T. Lawson, and M. R. Frogley. 2000. "A Revised Chronological and Palaeoenvironmental Framework for the Kastritsa Rockshelter, North- west Greece," Antiquity 74, pp. 349- 355.

Gams, I. 1973. "Die zweiphasische quartarzeitliche Flachenbildung in

Polje und Blindtalern des nordwest- lichen Dinarischen Karstes," Geo-

graphische Zeitschrift 32, pp. 143- 149.

.1978. "The Polje: The Problem of Its Definition," Zeitschriftfur Geomorphologie 22, pp. 170-181.

Hammond, N. G. L. 1967. Epirus: The Geography, theAncient Remains, the History, and the Topography of Epirus andAdjacent Areas, Oxford.

Harden, J. W. 1982. "A Quantitative Index of Soil Development from Field Descriptions: Examples from a Chronosequence in Central California," Geoderma 28, pp. 1-28.

Harden, J. W., and E. M. Taylor. 1983. "A Quantitative Comparison of Soil Development in Four Climatic Regimes," Quaternary Research 20, pp. 342-359.

Harden, J. W., E. M. Taylor, C. Hill, R. K. Marl, L. D. McFadden, M. C. Reheis, J. M. Sowers, and S. G. Wells. 1991. "Rates of Soil Development from Four Soil Chronosequences in Southern Great Britain," Quaternary Research 35, pp. 383-399.

Harris, D. R., and C. Vita-Finzi. 1968. "Kokkinopilos-A Greek Badland," Geographical Journal 134, pp. 537- 546.

Harrison, S. P., and G. Digerfeldt. 1993. "European Lakes as Palaeo- hydrological and Palaeoclimatic Indicators," Quaternary Science Reviews 12, pp. 233-248.

Harrold, F. B. 1993. "Mousterian, Cha- telperronian, and Early Aurignacian in Western Europe: Continuity or Discontinuity?" in The Human Evo- lution Source Book, ed. R. L. Ciochon

273

REFERENCES

and J. G. Fleagle, Englewood Cliffs, NJ., pp. 585-603.

Hawkins, R. 1998. "Coming Full Circle: Plowzone Assemblages and the Interpretation of Fort Ancient Settlement Structure," in Surface Archaeology, ed. A. P. Sullivan III, Albuquerque, pp. 91-109.

Hemans, F. P., J. W. Myers, and J. Wiseman. 1987. Remote Sensing from a Tethered Blimp in Greece (Center for Remote Sensing Tech- nical Paper 2, Boston University), Boston.

Higgs, E. S., and C. Vita-Finzi. 1966. "The Climate, Environment, and Industries of Stone Age Greece: Part II," PPS 32, pp. 1-29.

Higgs, E. S., C. Vita-Finzi, D. A. Harris, and A. E. Fagg. 1967. "The Climate, Environment, and Industries of Stone Age Greece: Part III," PPS 33, pp. 1-29.

Higgs, E. S., and D. Webley. 1971. "Further Information Concerning the Environment of Palaeolithic Man in Epirus," PPS 37, pp. 367- 380.

Holliday, V. T. 1989. "Paleopedology in Archeology," in Paleopedology: Nature and Application of Paleosols (Catena Suppl. 16), ed. A. Bronger, K. Catt, andJ. A. Catt, pp. 187- 206.

Hope Simpson, R. 1984. "The Analy- sis of Data from Surface Surveys," JFA 11, pp. 115-117.

Huntley, D.J., D. I. Godfrey Smith, and M. L. W. Thewalt. 1985."Opti- cal Dating of Sediments," Nature 313, pp.105-107.

Huxtable, J. A., J. A. J. Gowlett, G. N. Bailey, P. L. Carter, and V. Papaconstantinou. 1992. "Ther- moluminescence Dates and a New Analysis of the Early Mousterian from Asprochaliko," CurrAnthr 33, pp.109-114.

Imbrie,J.,J. D. Hays, D. G. Martin- son, A. C. McIntyre, A. C. Mix, J. J. Morley, N. G. Pisias, W. L. Prell, N.J. Shackleton. 1984. "The Orbital Theory of Pleistocene Climate: Support from a Revised Chronology of the Marine 180O Record," in Milankovitch and Climate. Understanding the Response to Astronomical Forcing (NATO ASI

Series C, Mathematical and Phys- ical Sciences 126), ed. A. L. Berger, J. Imbrie, J. D. Hays, G. J. Kukla, and B. Saltzman, Dordrecht, pp.269-306.

Innes,J. L. 1983. "Debris Flows," Progress in Physical Geography 7, pp.469-501.

Isager,J., ed. 2001. Foundation and Destruction, Nikopolis and North- western Greece: The Archaeological Evidencefor the City Destructions, the Foundation of Nikopolis, and the Synoecism (Monographs of the Danish Institute at Athens 3), Athens.

Jacobshagen, V., ed. 1986. Geologie von Griechenland, Berlin.

Jameson, M. H., C. N. Runnels, and T. H. van Andel. 1994. A Greek Countryside: The Southern Argolid

from Prehistory to the Present Day, Stanford.

Jones, K. P N., I. N. McCave, and P. D. Patel 1988. "A Computer- Interfaced Sedigraph for Modal Size Analysis of Fine-Grained Sediment," Sedimentology 35, pp.163-172.

Kahle, H. V., M. V. Muller, S. Miller, and G. Veis. 1993. "The Kepha- lonia Transform Fault and the Rotation of the Apulian Platform: Evidence from Satellite Geodesy," Geophysical Research Letters 20, pp.651-654.

Kardulias, P. N. 1994a. "Paradigms of the Past in Greek Archaeology," in Kardulias 1994b, pp. 1-23.

, ed. 1994b. Beyond the Site: Regional Studies in the Aegean Area, Lanham.

Keay, S.J., and M. Millett. 1991. "Surface Survey and Site Recog- nition in Spain: The Ager Tarraco- nensis Survey and Its Background," in Schofield 1991b, pp. 129-139.

Keller, D. R., and D. W. Rupp, eds. 1983. Archaeological Survey in the Mediterranean Area (BAR-IS 155), Oxford.

Kelletat, D. 1974. Beitrage zur regionalen Ku'stenmorphologie des Mittelmeerraumes (Zeitschriftfur Geomorphologie Suppl. 19), Berlin.

Kelletat, D., G. Kowalczyk, B. Schr6der, and K. P. Winter. 1976. "A Synoptic View on the

274

REFERENCES

Neotectonic Development of the Peloponnesian Coastal Regions," Zeitschrift der Deutschen Geologischen Gesellschaft 127, pp. 447-465.

Keraudren, B., and D. Sorel. 1987. "The Terraces of Corinth (Greece): A Detailed Record of Eustatic Sea- Level Variations during the Last 500,000 Years," Marine Geology 77, pp. 99-107.

Kilenyi, T. I. 1969. "The Problems of Ostracod Ecology in the Thames Estuary," in The Taxonomy, Morphology, and Ecology of Recent Ostracoda, ed. J. W. Neale, Edin- burgh, pp. 251-267.

King, G., and G. Bailey. 1985. "The Palaeoenvironment of Some Arch- aeological Sites in Greece: The Influence of Accumulated Uplift in a Seismically Active Region," PPS 51, pp. 273-282.

King, G., D. Sturdy, and J. Whitney. 1993. "The Landscape Geometry and Active Tectonics of Northwest Greece," Geological Society of America Bulletin 150, pp. 137-161.

Kintigh, K. W. 1990. "Comments on the Case for Full-Coverage Survey," in Fish and Kowalewski 1990, pp. 237-242.

Kirkby, A., and M. J. Kirkby. 1976. "Geomorphic Processes and the Surface Survey of Archaeological Sites in Semi-Arid Areas," in Geo- archaeology: Earth Science and the Past, ed. D. A. Davidson and M. L. Shackley, London, pp. 229-253.

Kirsten, E. 1987. "The Origins of the First Inhabitants of Nikopolis," in Chrysos 1987, pp. 91-98.

Kotzambopoulou, E., E. Panago- poulou, and E. Adam. 1996. "H raXaitoXLO0txr Epeova car-v Mrt6'Xa," ApxatoAoyax xatc TiXveq 60, pp. 30-35.

Koumouzelis, M., J. K. Kozlowski, M. Nowak, K. Sobczyk, M. Kacza- nowska, M. Pawlikowski, and A. Pazdur. 1996. "Prehistoric Settle- ment in the Klisoura Gorge, Argo- lid, Greece (Excavations 1993, 1994)," Prehistoire europeenne 8, pp. 143-173.

Kourtesi-Philippakis, G. 1996. "Nea rTotLXsca raX atoXtotAXY; xactotxqcOTg

ouT Zocxovo," ApXaioAoy6a xaL

Tixveq 60, pp. 77-80.

Kozlowski, J. K. 1992. "The Balkans in the Middle and Upper Paleolithic: The Gate to Europe or a Cul-de- Sac?" PPS 58, pp. 1-20.

. 1999. "The Evolution of the Balkan Aurignacian," in Dorothy Garrod and the Progress of the Palaeo- lithic: Studies in the PrehistoricArch- aeology of the Near East and Europe, ed. W. Davies and R. Charles, Oxford, pp. 97-117.

Kraft,J. C. 1985. "Marine Environ- ments: Paleogeographic Reconstruc- tions in the Littoral Region," in Archaeological Sediments in Context, ed. J. K. Stein and W. R. Farrand, Orono, Maine, pp. 111-125.

Kraft, J. C., and S. E. Aschenbrenner. 1977. "Paleogeographic Reconstruc- tions in the Methoni Embayment in Greece,"JFA 4, pp. 19-44.

Kraft, J. C., S. E. Aschenbrenner, and G. Rapp Jr. 1977. "Paleogeographic Reconstructions of Coastal Aegean Archaeological Sites," Science 195, pp. 941-947.

Kraft, J. C., and M. J. Chrzastowski. 1985. "Coastal Stratigraphic Sequences," in Coastal Sedimentary Environments, ed. R. A. Davis Jr., 2nd ed., New York, pp. 625-663.

Kraft,J. C., I. Kayan, and S. E. Asch- enbrenner. 1985. "Geological Studies of Coastal Plains Applied to Archaeological Settings," in Archaeological Geology, ed. G. Rapp Jr., and J. A. Gifford, New Haven, pp. 57-84.

Kraft, J. C., G. Rapp Jr., and S. E. Aschenbrenner. 1980. "Late Holo- cene Paleogeographic Reconstruc- tions in the Area of the Bay of Navarino: Sandy Pylos,"JAS 7, pp. 187-210.

Kraft,J. C., G. Rapp Jr., G.J. Szemler, C. Tziavos, and E. W. Kase. 1987. "The Pass at Thermopylae, Greece," JFA 14, pp. 181-198.

Kromer, B., B. Becker, P. Trimborn, and M. Spurk. 1995. "Tree-ring Based 14C Calibration and Stable Isotope Series at the YD/BP Boundary," in The Younger Dryas: Proceedings of a Workshop at the Royal Netherlands Academy of Arts and Sciences on 11- 13 April 1994 (Verhandelingen der Koninklijke Nederlandse Akademie van Wetenschappen, Afdeling Nat-

uurkunde, Eerste Reeks 44), ed. S. R. Troelstra, J. E. van Hinte, and G. M. Ganssen, Amsterdam, pp. 167-172.

Ku, T. L., W. B. Bull, S. Freeman, and K. G. Knauss. 1979. "Th23?-U234 Dating of Pedogenic Carbonates in Gravelly Desert Soils of Vidal Valley, Southeastern California," Geological Society of America Bulletin 90, pp. 1063-1073.

Ku, T. L., M. Ivanovich, and S. Luo. 1990. "U-Series Dating of Last Interglacial High Sea Stands: Barbados Revisited," Quaternary Research 33, pp. 129-147.

Ku, T. L., and Z. C. Liang. 1983. "The Dating of Impure Carbonates with Decay-Series Isotopes," Proceedings, Seminar on Alpha Particle Spectrometry and Low Level Measurement, Harwell, pp. 563-571.

Kubilay, N. N., A. C. Saydam, S. Yem- enicioglu, G. Kelling, S. Kapur, C. Karaman, and E. Ak;a. 1997. "Seasonal Chemical and Mineral- ogical Variability of Atmospheric Particles in the Coastal Region of the Northeast Mediterranean," Catena 28, pp. 313-328.

Kuhn, S. L. 1995. Mousterian Lithic Technology:An Ecological Perspective, Princeton.

Kuhn, S. L., G. Arsebiik, and F. C. Howell. 1996. "The Middle Pleistocene Lithic Assemblage from Yarimburgaz Cave, Turkey," Paleorient 22, pp. 31-49.

Kuhn, S. L., M. C. Stiner, and E. Gii- leg. 1999. "Initial Upper Palaeolithic in South-Central Turkey and Its Regional Context: A Preliminary Report," Antiquity 73, pp. 505-517.

Kukal, Z. 1990. "Rates of Geological Processes," Earth Sciences Reviews 28,pp. 7-284.

Kvamme, K. L. 1996. "Investigating Chipping Debris Scatters: GIS as an Analytical Engine," in New Methods, Old Problems: Geographic Information Systems in Modern ArchaeologicalResearch, ed. H. D. G. Maschner, Carbondale, Ill., pp. 38-71.

Kyparissi-Apostolika, N. 1996. "EtcYjXLo OT6c-zpa;: OL 7CaXXaLO-

XtL0LX?; srcLtX)cL;g," ApXatoAoy(a xat Td?ve; 60, pp. 37-41.

275

REFERENCES

Labeyrie, L. D.,J.-C. Duplessy, and P. L. Blanc. 1987. "Variations in Mode of Formation and Tempera- ture of Oceanic Deep Waters over the Past 125,000 Years," Nature 327, pp. 477-482.

Laj, C., A. Mazaud, and J. C. Duplessy. 1996. "Geomagnetic Intensity and 14C Abundance in the Atmosphere and Ocean during the Past 50 kyr," Geophysical Research Letters 23, pp. 2045-2048.

Lambeck, K. 1995. "Late Pleistocene and Holocene Sea-Level Change in Greece and Southern Turkey: A Separation of Eustatic, Isostatic, and Tectonic Contributions," Geophysical

JournalInternational 122, pp. 1022- 1044.

. 1996. "Sea-Level Change and Shore-Line Evolution: A General Framework of Modelling and Its Application to Aegean Greece since Upper Palaeolithic Time," Antiquity 70, pp.588-611.

Lambeck, K., and M. Nakada. 1992. "Constraints on the Age and Duration of the Last Interglacial Period and on Sea-Level Varia- tions," Nature 357, pp. 125-128.

Lang, A., and G. A. Wagner. 1996. "Infrared Stimulated Luminescence Dating of Archaeosediments," Archaeometry 38, pp. 129-141.

Leake, W. M. 1835. Travels in Northern Greece, 4 vols., London.

Le Pichon, X., N. Chamot-Rooke, S. Lallemant, R. Noomen, and G. Veis. 1995. "Geodetic Determi- nation of the Kinematics of Central Greece with Respect to Europe: Implications for Eastern Mediterra- nean Tectonics,"Journal of Geophysi- cal Research 100, pp. 12675-12690.

Lloyd, J. A., and G. Barker. 1981. "Rural Settlement in Roman Molise: Problems of Archaeological Survey," in Archaeology and Italian Society: Prehistoric, Roman, and Med- ieval Studies (BAR-IS 102), ed. G. Barker and R. Hodges, Oxford, pp.375-416.

Machette, M. N. 1985. "Calcic Soils of the Southwestern United States," in Soils and Quaternary Geology of the Southwestern United States (Geologi- cal Society of America Special Paper 203), ed. D. L. Weide, Boulder, pp. 1-21.

MacLeod, D. A. 1980. "The Origin of Red Mediterranean Soils in Epirus, Greece,"Journal of Soil Science 269, pp.573-577.

MacLeod, D. A., and C. Vita-Finzi. 1982. "Environment and Prov- enance in the Development of Recent Alluvial Deposits in Epirus, Greece," Earth Surface Processes and Landforms 7, pp. 29-43.

Magaritz, M., A. Kaufman, and D. H. Yaalon. 1981. "Calcium Carbonate Nodules in Soils: 180/160 and 13C/12C Ratios and 14C Contents," Geoderma 25, pp.157-172.

Martinson, D., N. G. Pisias,J. D. Hays, J. Imbrie, T. C. Moore Jr., and N.J. Shackleton. 1987. "Age Dating and the Orbital Theory of the Ice Ages: Development of a High-Resolution 0-300,000 Year Chronostratigraphy," Quaternary Research 27, pp. 1-29.

McFadden, L. D.,J. B. Ritter, and S. G. Wells. 1989. "Use of Multi- parameter Relative-Age Methods for Age Estimation and Correlation of Alluvial Surfaces on a Desert Piedmont, Eastern Mojave Desert, California," Quaternary Research 32, pp.291-293.

McFadden, L. D., and R. J. Weldon. 1987. "Rates and Processes of Soil Development in Quaternary Terraces in Cajon Pass," Geological Society of America Bulletin 98, pp.280-293.

Mee, C., and H. Forbes. 1997. "Survey Methodology," in A Rough and Rocky Place: The Land- scape and Settlement History of the Methana Peninsula, Greece, ed. C. Mee and H. Forbes, Liverpool, pp. 33-41.

Mellars, P. 1992. "Archaeology and the Population-Dispersal Hypothesis of Modern Human Origins in Europe," Philosophical Transactions of the Royal Society of London 337, pp.225-234.

.1996. The NeanderthalLegacy: An Archaeological Perspectivefrom Western Europe, Princeton.

Middleton, G. V. 1973. "Johannes Walther's Law of the Correlation of Facies," Geological Society of America Bulletin 84, pp. 979-988.

276

REFERENCES

Millett, M. 1991. "Pottery: Population or Supply Patterns?" in Roman Landscapes: Archaeological Survey in the Mediterranean Region, ed. G. Barker and J. Lloyd, London, pp. 18-26.

Mirabella, A., and M. S. Carnicelli. 1992. "Iron Oxide Mineralogy in Red and Brown Soils Developed on Calcareous Rocks in Central Italy," Geoderma 55, pp. 93-110.

Miracle, P. T. 1995. "Broad-Spectrum Adaptations Re-Examined: Hunter- Gatherer Responses to Late Glacial Environment Changes in the Eastern Adriatic" (diss. University of Michigan).

Monopolis, D., and A. Bruneton. 1982. "Ionian Sea (Western Greece): Its Structural Outline Deduced from Drilling and Geophysical Data," Tectonophysics 83, pp. 227-242.

Moore, M. 2000. "Surveying Epirote Pottery: Ceramics, Cuisine, and Social History in Southern Epirus, Greece, 300 B.C.-A.D. 500" (diss. Boston University).

Morrison, R. B. 1976. "Quaternary Soil Stratigraphy: Concepts, Methods, and Problems," in

Quaternary Soils: Papers of the 3rd Symposium on Quaternary Research, ed. W. C. Mahaney, Norwich, pp. 77-108.

Murray, P. M., and P. N. Kardulias. 1986. "A Modern-Site Survey in the Southern Argolid,"JFA 13, pp. 21-41.

Murray, W. M., and Ph. M. Petsas. 1989. Octavians Campsite Memorial

for the Actian War (TAPS 79), Philadelphia.

Nakada, M., and K. Lambeck. 1988. "The Melting History of the Late Pleistocene Antarctic Ice Sheet," Nature 333, pp. 36-40.

Neale, J. W. 1964. "Some Factors Influencing the Distribution of Recent British Ostracoda," Pubbli- cazioni della Stazione Zoologica di Napoli 33 (suppl.), pp. 247-307.

Nicod, J. 1992. "Recherches sur l'6volution du karst du Massif de Montrieux (Basse Provence)," in Geomorphology and Geoecology: Karst (Zeitschriftfur Geomorphologie Suppl. 85), ed.J. Nicod, K. H. Pfeffer, and M. Sweeting, Berlin, pp. 39-57.

Niemi, T. M. 1990. "Paleoenviron- mental History of Submerged Ruins on the Northern Euboean Gulf Coastal Plain, Central Greece," Geoarchaeology 5, pp. 323-347.

Odell, G. H., and F. Cowan. 1987. "Estimating Tillage Effects on Artifact Distributions," AmerAnt 52, pp.456-484.

Olszewski, D. I., and H. L. Dibble. 1994. "The Zagros Aurignacian," CurrAnthr 35, pp. 68-75.

Papagianni, D. 1999. "Middle Palaeo- lithic Occupation and Technology in Northwest Greece: Evidence from Open-Air Sites" (diss. Univer- sity of Cambridge).

.2000. Middle Palaeolithic Occupation and Technology in Northwestern Greece (BAR-IS 882), Oxford.

Papazachos, B. C., and P. E. Commi- nakis. 1971. "Geophysical and Tectonic Features of the Aegean Arc," Journal of Geophysical Research 76, pp.8517-8533.

Parsons, J. R. 1990. "Critical Reflec- tions on a Decade of Full-Coverage Regional Survey in the Valley of Mexico," in Fish and Kowalewski 1990, pp. 7-31.

Perles, C. 1987. Les industries lithiques taillees de Franchthi (Argolide, Grece) 1: Presentation generale et industries

pal6olithiques (Franchthi 3), Bloomington.

.1990. Les industries lithiques taillees de Franchthi (Argolide, Grece) 2: Les industries du mdsolithique et du

neolithique initial (Franchthi 5), Bloomington.

Petruso, K. M., B. B. Ellwood, F. B. Harrold, and M. Korkuti. 1994. "Radiocarbon and Archaeomag- netic Dates from Konispol Cave, Albania," Antiquity 68, pp. 335-339.

Pfeiffer, D. 1963. "Die geschichtliche Entwicklung der Anschaungen tiber das Karstgrundwasser," Geologisches Jahrbuch 57, pp. 1-111.

Philippson, A., and E. Kirsten. 1956. Die griechischen Landschaften, 2 vols., Frankfurt.

Phleger, F. B. 1960. Ecology and Distribution of Recent Foraminifera, Baltimore.

Piper, D. J. W., N. Kontopoulos, and A. G. Panagos. 1988. "Deltaic Sedimentation and Stratigraphic

Sequences in Post-Orogenic Basins, Western Greece," Sedimentary Geology 55, pp. 283-294.

Piper, D. J. W., A. G. Panagos, and N. Kontopoulos. 1982. "Some Observations on Surficial Sedi- ments and Physical Oceanography of the Gulf of Amvrakia," Thalasso-

graphica 5, pp. 63-80. Pirazzoli, P. A., S. C. Stiros,J. Laborel,

F. Laborel-Deguen, M. Arnold, S. Papageorgiou, and C. Morhange. 1994. "Late-Holocene Shoreline Changes Related to Palaeoseismic Events in the Ionian Islands, Greece," The Holocene 4, pp. 397- 405.

Plog, F. 1990. "Some Thoughts on Full-Coverage Surveys," in Fish and Kowalewski 1990, pp. 243-248.

Plog, S., F. Plog, and W. Wait. 1978. "Decision Making in Modern Surveys," in Advances in Archaeologi- cal Method and Theory 1, ed. M. B. Schiffer, New York, pp. 383-421.

Pope, K. O., C. N. Runnels, and T. L. Ku. 1984. "Dating Middle Palaeolithic Red Beds in Southern Greece," Nature 312, pp. 264-266.

Pope, K. O., and T. H. van Andel. 1984. "Late Quaternary Alluviation and Soil Formation in the Southern Argolid: Its History, Causes, and Archaeological Implications,"JAS 11, pp. 281-306.

Poulos, S. E., V. Lykousis, and M. B. Collins. 1995. "Late Qua- ternary Evolution of Amvrakikos Gulf, Western Greece," Geo- Marine Letters 15, pp. 9-16.

Purcell, N. 1987. "The Nicopolitan Synoecism and Roman Urban Policy," in Chrysos 1987, pp. 71-90.

Puri, H. S., G. Bonaduce, and A. M. Gervasio. 1969. "Distribution of Ostracoda in the Mediterranean," in The Taxonomy, Morphology, and Ecology of Recent Ostracoda, ed. J. W. Neale, Edinburgh, pp. 356-411.

Puri, H. S., G. Bonaduce, and J. Malloy. 1964. "Ecology of the Gulf of Naples," Pubblicazioni della Stazione

Zoologica di Napoli 33 (suppl.), pp. 87-199.

Pye, K. 1992. "Aeolian Dust Transport and Deposition over Crete and Adjacent Parts of the Mediterra- nean Sea," Earth Surface Processes and Landforms 17, pp. 271-288.

277

REFERENCES

Pyle, D. M., T. H. van Andel, P. Pas- chos, and P. van den Boogaard. 1998. "An Exceptionally Thick Middle Pleistocene Tephra Layer from Epirus, Greece," Quaternary Research 49, pp. 280-286.

Rapp, G., Jr. 1986. "Assessing Archaeo- logical Evidence for Seismic Catastrophes," Geoarchaeology 4, pp.365-379.

Rapp, G.,Jr., and Z. Jing. 1994. "Reconstruction of the Changing Holocene Coastal Landscape in the Nikopolis Region, Northwestern Greece" (paper, Washington, D.C. 1993), abstract in AJA 98, p. 316.

Rapp, G.,Jr., andJ. C. Kraft. 1994. "Holocene Coastal Change in Greece and Aegean Turkey," in Kardulias 1994b, pp. 23-35.

Rathjens, C. 1960. "Beobachtungen an hochgelegenen Poljen im stidlichen dinarischen Karst (ein Beitrag zur Frage der Entstehung and Datier- ung der Poljen)," Zeitschriftfur

Geomorphologie 4, pp. 141-151. Redman, C. 1973. "Multistage Field-

work and Analytical Techniques," AmerAnt 39, pp. 61-79.

Reisch, L. 1980. "Pleistozan und Urgeschichte der Peloponnes" (habilitationsschrift, Friedrich- Alexander Universitat).

Retallack, G. J. 1988. "Field Recogni- tion of Paleosols," in Paleosols and Weathering through Geologic Time: Principles andApplications (Geo- logical Society of America Special Paper 216), ed. J. Reinhardt and W. R. Sigleo, Boulder, pp. 1-20.

. 1990. Soils of the Past:.An Introduction to Paleopedology, London.

Rick, J. W. 1976. "Downslope Move- ment and Archaeological Intrasite Spatial Analysis," AmerAnt 41, pp. 133-144.

Roberts, B. K. 1987. "Landscape Archaeology," in Landscape and Culture: Geographical andArchaeo- logical Perspectives, ed. J. M. Wag- staff, Oxford, pp. 77-95.

Roberts, N., and H. E. Wright Jr. 1993. "Vegetational, Lake-Level, and Climatic History of the Near East and Southwest Asia," in Global Climates since the Last Glacial Maximum, ed. H. E. Wright Jr.,

J. E. Kutzbach, T Webb III, W. F. Ruddiman, F. A. Street- Perrott, and P. J. Bartlein, Minne- apolis, pp. 194-220.

Roebroeks, W., and T. van Kolfschoten. 1995. "The Earliest Occupation of Europe: A Reappraisal of Artefac- tual and Chronological Evidence," in The Earliest Occupation of Europe: Proceedings of the European Science Foundation Workshop at Tautavel (France), 1993, ed. W. Roebroeks and T. van Kolfschoten, Leiden, pp.297-315.

Rolland, N. 1986. "Recent Findings from La Micoque and Other Sites in South-Western and Mediterra- nean France: Their Bearing on the 'Tayacian' Problem and Middle Palaeolithic Emergence," in Stone Age Prehistory: Studies in Memory of Charles McBurney, ed. G. N. Bailey and P. Callow, Cambridge, pp. 121- 151.

Rossignol, J. 1992. "Concepts, Methods, and Theory Building," in Rossignol and Wandsnider 1992, pp. 3-16.

Rossignol, J., and L. Wandsnider, eds. 1992. Space, Time, andArchaeological Landscapes, New York.

Rossignol-Strick, M. 1983. "African Monsoons: An Immediate Climate Response to Orbital Insolation," Nature 304, pp. 46-49.

Runnels, C. 1981. "On the Destruction of Archaeological Sites,"JFA 8, pp. 91-93.

. 1988. "A Prehistoric Survey of Thessaly: New Light on the Greek Paleolithic,"JFA 15, pp.277-290.

. 1994. "Pleistocene and Early Holocene Archaeological Sites in Southern Epirus, Greece" (paper, Washington, D.C. 1993), abstract in AJA 98, pp. 315-316.

. 1995. "The Stone Age of Greece from the Palaeolithic to the Advent of the Neolithic," AJA 99, pp.699-728.

. 1996. "The Palaeolithic and Mesolithic Remains," in Wells and Runnels 1996, pp. 23-35.

Runnels, C., and T. H. van Andel. 1993a. "The Lower and Middle Palaeolithic of Thessaly (Greece)," JFA 20, pp. 299-318.

278

REFERENCES

. 1993b. "A Handaxe from Kokkinopilos, Epirus, and Its Implications for the Paleolithic of Greece,"JFA 20, pp. 191-203.

Russell, R. J. 1954. "Alluvial Morphol- ogy of Anatolian Rivers," Annals of the Association of American Geogra- phers 44, pp. 363-391.

. 1967. River Plains and Sea Coasts, Berkeley and Los Angeles.

Rutter, J. B. 1993. "The Prepalatial Bronze Age of the Southern and Central Greek Mainland," AJA 97, pp. 745-797.

Samsaris, D. 1988. "H pLt OCLXYJ aTotLxia

rt tI)TcLXYlq," IzToptxoye?oypacxpoxd 2,pp.23-31.

Sarris, A., and R. E. Jones. 2000. "Geophysical and Related Tech- niques Applied to Archaeological Survey in the Mediterranaean: A Review,"JMA 13, pp. 3-75.

Sars, G. 0. 1928. An Account of the Crustacea of Norway 9: Ostracoda, Bergen.

Schiffer, M. B. 1987. Formation Processes of the Archaeological Record, Albuquerque.

Schiffer, M. B., A. P. Sullivan, and T. C. Klinger. 1978. "The Design of Archaeological Surveys," WorldArch 10, pp. 1-28.

Schneider, H. E. 1977. "Problems of Terrestrial Red Beds in the Neo- gene and Quaternary of Greece," in Proceedings of the VIth Colloquium on the Geology of the Aegean Region, ed. G. Kallergis, Athens, pp. 477- 493.

Schofield, A. J. 1991a. "Interpreting Artefact Scatters: An Introduction," in Schofield 1991b, pp. 3-8.

, ed. 1991b. InterpretingArtefact Scatters: Contributions to Ploughzone Archaeology (Oxbow Monograph 4), Oxford.

Schroder, B. 1986. "Das postorogene Kanozoikum in Griechenland/ Agais," in Geologie von Griechenland, ed. V. Jacobshagen, Berlin, pp. 208- 240.

Schroder, B., and D. Kelletat. 1976. "Geodynamical Conclusions from Vertical Displacement of Quater- nary Shorelines in the Peloponne- sos, Greece," Neues Jahrbuchfur Geo- logie und Paltontologie, Monatshefte, pp. 174-186.

Schwertmann, U. 1971. "Transforma- tion of Hematite to Goethite in Soils," Nature 232, pp. 624-625.

Sever, T., and J. Wiseman. 1985. Conference on Remote Sensing: Potentialfor the Future, NASA, Earth Resources Laboratory.

Shackleton, N.J. 1987. "Oxygen Iso- topes, Ice Volume, and Sea Level," Quaternary Science Reviews 6, pp.183-190.

Shott, M. J. 1995. "Reliability of Archaeological Records on Culti- vated Surfaces: A Michigan Case Study,"JFA 22, pp. 475-490.

Siani, G., M. Paterne, M. Arnold, E. Bard, B. Metivier, N. Tisnerat, and F. Bassinot. 2000. "Radiocarbon Reservoir Ages in the Mediterranean Sea and Black Sea," Radiocarbon 42, pp.271-280.

Singer, R., B. G. Gladfelter, and J. J. Wymer. 1993. The Lower Paleolithic Site at Hoxne, England, Chicago.

Smit, A., and T. A. Wijmstra. 1970. "Application of Transmission Elec- tron Microscope Analysis to the Re- construction of Former Vegetation," Pollen et spores 13, pp. 615-621.

Smith, G. W., R. D. Nance, and A. N. Genes. 1997. "Quaternary Geologi- cal History of Mount Olympus, Greece," Geological Society of America Bulletin 109, pp. 809-824.

Soil Survey Staff. 1975. Soil Taxonomy (U.S. Department of Agriculture Handbook 436), Washington, D.C.

Sordinas, A. 1968. "The Prehistory of the Ionian Islands: The Flints and Pottery" (diss. Harvard University).

. 1969. "Investigations of the Prehistory of Corfu during 1964- 1966," BalkSt 10, pp. 393-424.

. 1970. Stone Implements from Northwestern Corfu, Greece, Mem- phis.

. 1983. "Quaternary Shorelines in the Region of Corfu and Adjacent Islets," in Quaternary Coastlines and Marine Archaeology: Towards the Pre- history of Land Bridges and Continen- tal Shelves, ed. P. M. Masters and N. C. Flemming, London, pp. 335-344.

Sorel, D. 1989. "L'evolution structurale de la Gr&ece nord-occidentale depuis le Miocene dans le cadre geodyna- mique de l'arc Egeen" (diss. Univer- site de Paris-Sud XI).

Spaargaren, 0. C. 1979. Weathering and Soil Formation in a Limestone Area near Pastena, Italy (Publicaties van het Fysisch Geografisch en Bodemkundig Laboratorium, University of Amsterdam 30), Amsterdam.

Spate, A. P., J. N. Jennings, D. I. Smith, and M. A. Greenaway. 1985. "The Micro-Erosion Meter: Use and Limitations," Earth Surface Processes and Landforms 10, pp. 427-440.

Stein, C., and B. Cullen. 1994. "Satellite Imagery and Archaeology: A Case Study from Nikopolis" (paper, Washington, D.C. 1993), abstract in AJA 98, p. 316.

Stiner, M. C., G. Arsebiik, and F. C. Howell. 1996. "Cave Bears and Paleolithic Artifacts in Yarimburgaz Cave, Turkey: Dissecting a Palimpsest," Geoarchaeology 11, pp. 279-327.

Stiner, M. C., N. D. Munro, T. A. Surovell, E. Tchernov, and 0. Bar-Yosef. 1999. "Paleolithic Population Growth Pulses Evi- denced by Small Animal Exploita- tion," Science 283, pp. 190-194.

Stirling, C. H., T. M. Esat, M. T. McCulloch, and K. Lambeck. 1995. "High-Precision U-Series Dating of Corals from Western Australia and Implications for the Timing and Duration of the Last Interglacial," Earth and Planetary Science Letters 135, pp. 115-130.

Stiros, S. C., M. Arnold, P. A. Pirazzoli, J. Laborel, and F. Laborel- Deguen. 1994. "The 1953 Earth- quake in Cephalonia (Western Hellenic Arc): Coastal Uplift and Halotectonic Faulting," Geophysical Journal International 117, pp. 837- 849.

Stoddart, S. K. F., and N. Whitehead. 1991. "Cleaning the Iguvine Stables: Site and Off-Site Analysis from a Central Mediterranean Perspective," in Schofield 1991b, pp. 141-148.

Sturdy, D. A., and D. P. Webley. 1988. "Palaeolithic Geography: or Where are the Deer?" WorldArch 19, pp. 262-280.

Sturdy, D. A., D. P. Webley, and G. Bailey. 1997. "The Palaeolithic Geography of Epirus," in Bailey 1997, pp. 2:587-614.

279

REFERENCES

Sullivan, A. P., III, ed. 1998. Surface Archaeology, Albuquerque.

Sumner, W. M. 1990. "Full-Coverage Regional Archaeological Survey in the Near East: An Example from Iran," in Fish and Kowalewski 1990, pp. 87-115.

Sweeting, M. M., ed. 1985. Karst Geomorphology, Stroudsburg, Pa.

Sweeting, M. M. 1993. "Reflections on the Development of Karstgeomor- phology in Europe and a Compari- son with Its Development in China," in Klimagenetische Geomor-

phologie (Zeitschriftfuir Geomorph- ologie Suppl. 93), ed. R. W. Fair- bridge and K.-H. Pfeffer, Berlin, pp. 127-136.

Tartaron, T. 1994. "Prehistoric Settle- ment in Southern Epirus: Prelimi- nary Results from Survey" (paper, Washington, D.C. 1993), abstract in AJA 98, p. 316.

. 1996. "Bronze Age Settlement and Subsistence in Southwestern Epirus, Greece" (diss. Boston University).

. 2001. "Glykys Limin: A Mycenaean Port of Trade in Southern Epirus?" in Prehistory and History: Ethnicity, Class, and PoliticalEconomy, ed. D. W. Tandy, Montreal, pp. 1-40.

Tartaron, T., and C. Runnels. 1992. "The Discovery of a Palaeolithic Handaxe at Kokkinopilos in Epirus," Context 9:3-4, pp. 1, 4-7.

Tartaron, T., C. Runnels, and E. Kari- mali. 1999. "Prolegomena to the Study of Bronze Age Flaked Stone in Southern Epirus," in Melete- mata: Studies in Aegean Archaeology Presented to Malcolm H. Wiener as He Enters His 65th Year III (Aegaeum 20), ed. P. P. Betancourt, V. Kara- georghis, R. Laffineur, and W.-D. Niemeier, Liege, pp. 819-825.

Tartaron, T., and K. Zachos. 1999. "The Mycenaeans and Epirus," in H wrcptipeta ro .uVox7vdixo6 xoayovo: A' AelOvg Ai?tcoaTcLo- vtxo Svoyrr6oto Aai(a, 25-29

e7rerftSppioV 1994, Lamia, pp. 57-76.

Tassos, S. T. 1975. "Holocene Sedi- ments and Some Oceanographic Parameters of Kalloni Bay, Lesvos Island, Greece" (M.S. thesis, University of Minnesota).

Thomas, D. H. 1975. "Non-Site Sampling in Archaeology: Up the Creek without a Site?" in Sampling in Archaeology, ed. J. W. Mueller, Tucson, pp. 61-81.

Trigger, B. G. 1989. A History of Arch- aeological Thought, Cambridge.

Turner, C., and M. Sanchez-Gofii. 1997. "Late Glacial Landscape and Vegetation in Epirus," in Bailey 1997, pp. 2:559-585.

Tzedakis, P. C. 1993. "Long-Term Tree Populations in Northwest Greece through Multiple Quaternary Climate Cycles," Nature 364, pp. 437-440.

. 1994. "Vegetation Change through Glacial-Interglacial Cycles: A Long Pollen Sequence Perspec- tive," Philosophical Transactions of the Royal Society of London B 345, pp. 403-432.

Tziavos, C. C. 1977. "Sedimentology, Ecology, and Paleogeography of the Sperchios Valley and Maliakos Gulf, Greece" (M.S. thesis, University of Delaware).

. 1997. "Paleogeographic Evo- lution of the Amvrakikos Gulf, Western Greece," in The Engineer- ing Geology of Ancient Works, Monu- ments, and Historical Sites: Proceed- ings of an International Symposium I, ed. P. G. Marinos and G. C. Koukis, Athens, pp. 425-430.

Underhill, J. 1989. "Late Cenozoic Deformation of the Hellenide Foreland, Western Greece," Geological Society of America Bulletin 101, pp. 613-634.

van Andel, T. H. 1989. "Late Quater- nary Sea-Level Changes and Arch- aeology," Antiquity 63, pp. 733-745.

. 1998a. "Paleosols, Red Beds, and the Old Stone Age in Greece," Geoarchaeology 13, pp. 361-390.

. 1998b. "Middle and Upper Palaeolithic Environments and the Calibration of 14C Dates beyond 10,000 B.P.,"Antiquity 72, pp. 26- 33.

van Andel, T. H., and C. Runnels. 1987. Beyond the Acropolis:A Rural Greek Past, Stanford.

van Andel, T. H., C. N. Runnels, and K. 0. Pope. 1986. "Five Thousand Years of Land Use and Abuse in the Southern Argolid, Greece," Hesperia 55, pp. 103-128.

280

REFERENCES

van Andel, T. H., and J. C. Shackleton. 1982. "Late Paleolithic and Meso- lithic Coastlines of Greece and the Aegean,"JFA 9, pp. 445-454.

van Andel, T. H., and P. C. Tzedakis. 1996. "Palaeolithic Landscapes of Europe and Environs, 150,000- 25,000 Years Ago," Quaternary Science Reviews 15, pp. 481-500.

. 1998. "Priority and Opportu- nity: Reconstructing the European Middle Palaeolithic Climate and Landscape," in Science in Archaeol- ogy:An Agendafor the Future, ed. J. Bayley, London, pp. 37-46.

van Andel, T. H., E. Zangger, and A. Demitrack. 1990. "Land Use and Soil Erosion in Prehistoric and Historical Greece,"JFA 17, pp. 379- 396.

van Andel, T. H., E. Zangger, and C. Perissoratis. 1990. "Quaternary Transgressive/Regressive Cycles in the Gulf of Argos, Greece," Quater- nary Research 34, pp. 317-329.

van Zeist, W., and S. Bottema. 1982. "Vegetational History of the Eastern Mediterranean and the Near East during the Last 20,000 Years," in Palaeoclimates, Palaeo- environments, and Human Commu- nities in the Eastern Mediterranean Region in Later Prehistory (BAR-IS 133), ed. J. L. Bintliff and W. van Zeist, Oxford, pp. 277-321.

Villas, C. A. 1983. "The Holocene Evolution of the Acheloos River Delta, Northwestern Greece: Associated Environments, Geo- morphology, and Microfossils" (M.S. thesis, University of Dela- ware).

Vita-Finzi, C. 1978. Archaeological Sites in Their Setting, London.

Vreeken, W. J. 1975. "Principal Kinds of Chronosequences and Their Significance in Soil History," Journal of Soil Science 26, pp. 378- 394.

Wagner, C. W. 1957. Sur les Ostra- codes du quaternaire recent des Pays- Bas et leur utilisation dans l'etude

geologique des depots holocenes, The Hague.

Wandsnider, L., and E. L. Camilli. 1992. "The Character of Surface Archaeological Deposits and Its Influence on Survey Accuracy," JFA 19, pp. 169-188.

Waters, D. W. 1994. "The Tectonic Evolution of Epirus, Northwest Greece" (diss. University of Cambridge).

Weiss, H. 2000. "Beyond the Younger Dryas: Collapse as Adaptation to Abrupt Climate Change in Ancient West Asia and the Eastern Medi- terranean," in Environmental Disas- ter and the Archaeology of Human Response, ed. G. Bawden and R. Reycraft, Albuquerque, pp. 75-98.

Weiss, H., M. A. Courty, W. Wetter- strom, F. Guichard, L. Senior, R. Meadow, and A. Curnow. 1993. "The Genesis and Collapse of Third Millennium North Mesopotamian Civilization," Science 261, pp. 995- 1004.

Wells, B. 1994. "A Prehistoric Environ- mental Catastrophe: The Case of Berbati and Limnes," in Archaeology in the Peloponnese: New Excavations and Research (Oxbow Monograph 48), ed. K. A. Sheedy, Oxford, pp. 65-76.

Wells, B., and C. Runnels, eds. 1996. The Berbati-Limnes Archaeological Survey, 1988-1990, Jonsered.

Wells, B., C. Runnels, and E. Zangger. 1990. "The Berbati-Limnes Archaeological Survey: The 1988 Season," OpAth 18, pp. 207-238.

Weymouth, J. W. 1993. "Preliminary Report of Geophysical Surveys, Nikopolis Project 1993," report on file at the Department of Archaeol- ogy, Boston University.

Weymouth, J. W., and R. Huggins. 1985. "Geophysical Surveying and Archaeological Sites," in Archaeologi- cal Geology, ed. G. Rapp Jr., and J. A. Gifford, New Haven, pp. 191- 235.

Whitelaw, T. M. 1991. "Investigations at the Neolithic Sites of Kephala and Paoura," in Cherry, Davis, and Mantzourani 1991, pp. 199-215.

Wijmstra, T. A. 1969. "Palynology of the First 30 Metres of a 120m-Deep Section in Northern Greece (Mace- donia)," Acta Botanica Neerlandica 18, pp.511-557.

Wijmstra, T. A., and A. Smit. 1976. "Palynology of the Middle Part of the 120m-Deep Section in Northern Greece (Macedonia)," Acta Botanica Neerlandica 25, pp.297-312.

Wijmstra, T. A., R. Young, and H. J. L. Witte. 1990. "An Evalua- tion of the Climatic Conditions during the Late Quaternary in Northern Greece by Means of Multivariate Analysis of Palyno- logical Data and Comparison with Recent Phytosociological and Climatic Data," Geologie en Mi/nbouw 69, pp. 243-251.

Wilkinson, T. J. 1982. "The Definition of Ancient Manured Zones by Means of Extensive Sherd- Sampling Techniques,"JFA 9, pp.223-233.

Williams, G. E., and H. A. Polach. 1971. "Radiocarbon Dating of Arid Zone Calcareous Palaeosols," Geological Society of America Bulletin 83, pp.3069-3086.

Willis, K. J. 1992. "The Late Qua- ternary Vegetational History of Northwest Greece: I. Lake Gramousti; II. Rezina Marsh; III. A Comparative Study of Two Contrasting Sites," New Phytology 121, pp.101-117, 119-138, 139- 155.

. 1994. "The Vegetational History of the Balkans," Quaternary Science Reviews 13, p. 769-788.

Willis, K. J., P. Siimeg, M. Braun, and A. Toth. 1995. "The Late Quater- nary Environmental History of Batorliget, N.E. Hungary," Palaeo- geography, Palaeoclimatology, Palaeo- ecology 118, pp. 25-47.

Wintle, A. G. 1996. "Archaeologically- Relevant Dating Techniques for the Next Century: Small, Hot, and Identified by Acronyms,"JAS 23, pp. 123-138.

. 1997. "Luminescence Dating: Laboratory Procedures and Proto- cols," Radiation Measurements 27, pp. 769-817.

Wiseman, J. 1987. "Suggestions for the Study of Nikopolis in Its Changing Context," in Chrysos 1987, pp. 411- 416.

.1991. "First Season's Results: Satellite Imagery, Survey, and the Earliest Humans in Greece," Nikopolis Newsletter 1, Boston, pp. 1-12.

. 1992a. "Archaeology and Remote Sensing in the Region of Nikopolis, Greece," Context 9:3-4, pp. 1-4.

28I

REFERENCES

. 1992b. "The Boston University Nikopolis Project, 1992: Surveys and Investigations by Land, Sea, and Air," Nikopolis Newsletter 2, Boston, pp. 1-12.

. 1993a. "Land, Sea, and Aerial Surveys in Northwestern Greece," Context 10:3-4, pp. 11-15.

. 1993b. "The Nikopolis Proj- ect: Interdisciplinary Archaeology in Southern Epirus, Greece," Nikopolis Newsletter 3, Boston, pp. 1-12.

. 1994. "Harbors, Towns, and Prehistory: Survey and Field School in Greece," Context 11:1-2, pp. 1-4.

. 1995a. "Landscape Archaeol- ogy in Northwestern Greece: The Nikopolis Project, 1994," Nikopolis Newsletter 4, Boston, pp. 1-12.

. 1995b. "People and the Changing Landscape in Southern Epirus, Greece," Context 12:1-2, pp. 18-21.

. 1996a. "Space Missions and Ground Truth," Archaeology 49:4, pp. 11-13.

. 1996b. "Wonders of Radar Imagery," Archaeology 49:5, pp. 14- 18.

. 1997a. "Study Seasons in Ioannina and the Survey Zone in Epirus, Greece," Context 13:1-2, pp. 10-11.

. 1997b. "Remote Sensing and Archaeology in Northwestern Greece" (paper, Gainesville, Flor- ida 1996), abstract in Workshop on Radar Imaging and Cultural Resource Management at the Angkor Eco-Site and Symposium on New Technologies and Global Cultural Resource Management, New York, pp. 32-33.

. 1998. "Rethinking the 'Halls of Hades,'" Archaeology 51:3, pp. 12- 16, 18.

Wiseman, J., and A. Douzougli- Zachos. 1994. "The Nikopolis Proj- ect, 1991-1993: Overview of the Multidisciplinary Study of Southern Epirus" (paper, Washington, D.C. 1993), abstract in AJA 98, p. 315.

Wiseman, J., C. Robinson, and C. Stein. 1999. "Archaeological Applications of Radar and SPOT Satellite Imagery in Northwestern Greece" (paper, Chicago 1999), abstract in Abstracts of the SAA 64th Annual Meeting, Washington, D.C., p. 309.

Wiseman, J., K. Zachos, and F. Kephal- lonitou. 1996. "EXX-qvoocxpepLxccvLxo rrpoypocqa ?e7tL(pq0av?LaxCov eocovt)V,"

ArchDelt 46, B' (1991), pp. 247-251. . 1997. "ZovepyocTLnx] zXXqvo-

aqeptLxavLXq Er?ctaVSLax-x epeovaX

aoTo vopt6 Ilp?p3?a;,"ArchDelt 47, B' (1992), pp. 293-298.

. 1998. "EX?XYvoacpi?pxocvLx6 8ts- TLto7TYrLOVLx6 tcp6ypaxuxla sTL(pqaVet-

axxcv epeovC0v oTq] N6oa 'HTcsLpo," ArchDelt 48, B' (1993), pp. 309-314.

Woodward,J. C., M. G. Macklin, and J. Lewin. 1994. "Pedogenic Weath- ering and Relative Age Dating of Quaternary Alluvial Sediments in the Pindus Mountains of Northwest Greece," in Rock Weathering and Landform Evolution, ed. D. A. Rob- inson and R. G. B. Williams, New York, pp. 259-283.

Wright,J. C.,J. F. Cherry,J. L. Davis, E. Mantzourani, S. B. Sutton, and R. F. Sutton Jr. 1990. "The Nemea Valley Archaeological Project: A Preliminary Report," Hesperia 59, pp.579-659.

Yaalon, D. H. 1987. "Saharan Dust and Desert Loess: Effect on Surround- ing Soils,"Journal of African Earth Sciences 6, pp. 569-571.

. 1997. "Soils in the Mediterra- nean Region: What Makes Them Different?" Catena 28, pp. 157-169.

Yamin, R., and K. B. Metheny, eds. 1996. Landscape Archaeology: Reading and Interpreting the American Historical Landscape, Knoxville.

Yang, L. C. 1982. "The Distribution and Taxonomy of Ostracodes and Benthic Foraminifers in Late Pleist- ocene and Holocene Sediments of the Troad (Biga Peninsula), Turkey" (M.S. thesis, University of Dela- ware).

Yassoglou, N., C. Kosmas, and N. Moustakas. 1997. "The Red Soils, Their Origin, Properties, Use, and Management in Greece," Catena 28, pp.261-278.

Zachos, K. 1998. "0(IY OpacyxoxxXy]- aia xoLv6'orqYt Pitb,v (aypoTSiaixto acpxov A0avawoio)," ArchDelt 48, B'

(1993), pp. 301-302. Zangger, E. 1991. "Prehistoric Coastal

Environments in Greece: The Vanished Landscape of Dimini Bay and Lake Lerna,"JFA 18, pp. 1-15.

. 1992. "Prehistoric and His- toric Soils in Greece: Assessing the Natural Resources for Agricul- ture," in Agriculture in Ancient Greece, ed. B. Wells, Stockholm, pp. 13-18.

. 1993. The Geoarchaeology of the

Argolid, Berlin. . 1994. "Landscape Changes

around Tiryns during the Bronze Age,"AJA 98, pp. 189-212.

Zhou, L. P., T. H. van Andel, and A. Lang. 2000. "A Luminescence Dating Study of Open-Air Palaeo- lithic Sites in Western Epirus, Greece,"JAS 27, pp. 609-620.

Zilhao,J., and F. D'Errico. 1999. "The Chronology and Taphonomy of the Earliest Aurignacian and Its Implications for the Understand- ing of Neandertal Extinction," Journal of World Prehistory 13, pp. 1-68.

282

INDEX

ACARNANIA, 3 Acheloos delta, 212 Acheloos River, 215 Acheron River, 4, 5, 8, 160,199,204-

205,233,234,266 Acheron River valley: and Acherousian

lake, 202-204, (Fig. 6.4) 203,205, 225,227-233,234,268; and Bronze Age, 140; and coastal geomorphological studies, 27; and dolines, 57; and field and laboratory methods, 209-211; and geological coring, 209-211, (Fig. 6.7) 210,235-263, 268; geology and neotectonics of, 205-208; and Glykys Limen, 199- 202,212,213,214,215,216-225, 229,234,267-268; and Kastri, 204- 205,233,234; and loutses, (Fig. 3.10) 63; map of, (Fig. 6.2) 201; and Mesolithic, 118, 119; and microfossil analyses, 209,211-212,213,214, 215,235-263; and Mousterian, 112; and Palaeolithic survey, 97, (Fig. 3.32) 100, 131; paleogeographic reconstructions of, (Fig. 6.12) 221, (Fig. 6.13) 222, (Fig. 6.14) 223, (Fig. 6.15) 224, and radiocarbon dating, 209, (Table 6.1) 210,217, 233; sea-level changes in, 208,216- 217; and sediments, 209-216, 225- 226,231; and shoreline progradation, 225-227; simplified geology of, 206, (Fig. 6.6) 207, 208; stratigraphic analysis of, 213,216-217, 227-233,235-263; and surface survey, 28, 30, (Table 2.1) 31,266; and tectonic activity, 50, 55; topographic map of, (Fig. 6.8) 211

Acherousian lake, 22, 199,202-204, (Fig. 6.4) 203,205,225,227-233,

234,268; paleogeographic reconstructions of, (Fig. 6.13) 222, (Fig. 6.14) 223

Acheulean, 98, 99, 126, 266 Actium, 160; Battle of, 3,201; Straits

of, 3,4 Adriatic Sea, 50, (Fig. 3.21) 76 aerial photography by tethered blimp,

7, 15, 16, 17, (Fig. 1.6) 18, (Fig. 1.7) 19,28

Aetolia, 3, 32 agriculture: and Acherousian lake, 202,

233; and artifact distribution, 136; and human-land relationships, 48; and Louros River, 198; and poljes, 58; and resurvey, 43, 44; and site/ scatters, 39; and southern Epirus, 4-5,8

Alonaki: and land use, 129; and Mesolithic, 118, 121, (Fig. 3.58) 124; and Mousterian, 100, 106, 107, 112,113, 126; and Neanderthals, 127; and Palaeolithic survey, 96, 98, 100-101, (Fig. 3.32) 100, (Fig. 3.33) 101, (Figs. 3.34, 3.35) 102, (Figs. 3.36, 3.37) 103,140; and paleosol, 82, 89, 103

Ambracian Gulf: and agriculture, 4; and coastal landscape change, 192- 198; and geological coring, 12, 13, 158, (Fig. 5.2) 159, 1594, 267; geomorphologic studies of, 27, (Fig. 5.1) 158, 160-161,267; and Holocene, 157, 169, 192; and land use, 131; and Mousterian, 113; and Nikopolis isthmus, 162, (Fig. 5.3) 163,164-173; and Ormos Vathy, 158, 162, 174-177; and paleogeographic development, 194-198, (Fig. 5.21) 196-197; and Preveza

INDEX

peninsula, 160,162-177; and radiocarbon dating, 167, (Table 5.1) 168; and rivers, 50; and sea-level changes, 78,157, 162, 192-195, (Fig. 5.20) 193, 198,267; subsidence of, 208; and subsurface stratigraphy, 157-158, 162-192, (Fig. 5.18) 187; and survey zone, 4; and tectonic activity, 54, 55, (Fig. 3.2) 55, 56, 157, 161, 162, 167, 169, 192-194; wetlands exploitation in, 9

American School of Classical Studies, 6 Ammerman, A. J., 24 Ammoudia, 112,118, 119, (Figs. 3.52,

3.53) 120, (Fig. 3.54) 121,131, 201, 223

Ammoudia Bay, 4, 9. See also Glykys Limen; Phanari Bay

Amphilochia, 3 Anatolia, 228 Anavatis, 80, 82, 106, 107,108, (Fig.

3.46) 111, 127 aqueduct, 17,18, (Fig. 1.7) 19, 70. See

also Ayios Georgios Arachthos River, 4, 50, 78, 160, 161,

217 Archaic, 4, 230 architectural features, 42 Argive Plain, 92 Argolid, 48, 92,106, 114,116, 125,

126,136 Argolid Exploration Project, 27 Arizona, 25 Arta (= Ambracia), 4, 2663; nomos of, 3 artifacts: density of, 38, 39, 40, 42, 43,

44, 115, 135,136, 137, 139, 141, 148-154,156; distribution of, 35, 37, 93, 136, 137, 139, 147,149, 153-154, 156; and Kastro Rogon, 182

artifacts, stone: and deposition, 95; Early Stone Age chronology, (Table 3.12) 98; and Palaeolithic chronology, 91-95, (Fig. 3.30) 94; and Palaeolithic survey, 95-124; and patination, 118, 140-141, 142,143, 148, 153; and tracts, 35, 95-96. See also chipped stone; stone tools

Arvenitsa, 114 Asia, 126 Asprochaliko: and backed-blade

industries, 116; excavation records of, 52; and hunting, 54, 127, 130; and later Palaeolithic, 114, 115, 117, 130; and Mousterian, 51, 53, 106, 113, 114, 127; and Mousterian/

Early Upper Palaeolithic transition, 129; and Palaeolithic research, 47, 50; and radiocarbon assays, 89; and Upper Palaeolithic, 51, 52

Aubouin, Jean, 205 Augustus, 2-3, 9. See also Octavian Aurignacian, 113, 114, 115, 142, 143-

144,154-155,267 Ayia: and flints, 96; stratigraphy of,

(Fig. 3.12) 65, (Fig. 3.19) 74, 75, (Fig. 3.20) 75, 76, 95; and Mous- terian, (Fig. 3.20) 75, 106, 107,108, (Figs. 3.43, 3.44) 110, 129; and Neanderthals, 127; site of, (Fig. 3.42) 109

Ayia Kyriaki, (Fig. 3.32) 100,112 Ayios Georgios, 17. See also aqueduct Ayios Petros, 137 Ayios Thomas peninsula: and later

Palaeolithic, 116; and Mesolithic, 121; and Mousterian, 107, 113; and Ormos Vathy, 174; and Palaeolithic survey, 98, 103,105, (Fig. 3.41) 105; paleosol of, 82; sampling strategy for, 8; and surface survey, 28, 30, 266

BACHOKIRIAN, 143,155 Bailey, G. N., 47, 52-56, 69, 85, 90, 93,

96, 114,116,130 Balkans, 83, 85, 98, 113,114,129,131-

132, 142,143-144, 155 Batiae, 189, 195. See also Kastro

Rizovouni Baugh, Timothy G., 16 Berbati, 116, 125, 137 Berbati-Limnes Archaeological Survey,

27,116 Besonen, Mark, 21, 22,209, 267 bivalves, 164, 167, 184. See also

microfossil analyses blind valleys, 57 Boeotia, 58, 116 Boila, 47, 114, 116 Bordes, Francois, 97 Boston University, 6 Bottema, S., 83, 85 Bouchetion, 179, 189,267. See also

Kastro Rogon bricks, 35, 42 Bronze Age, 5, 8, 52, 95, 123,132, 139,

140,142,144,154,194, 268 Buiidel, J., 59 Bulgaria, 155 Byzantine, 6, 17731, 179

284

INDEX

CALIFORNIA, 55 Cambridge/Bradford Boeotian

Expedition, 25, 27 Camilli, E. L., 25, 27 Cassopaia, 205 Cassope, 4, 5 chaine operatoire, 142, 143, 148 Chatelperronian, 129 Cheimadio, 59, (Fig. 3.11) 63, 112 Cherry, J. F., 38 chipped stone: chert, 101, 105; flint, 51,

52, 70, 93, 96, 97, 101,107, 108, 115,119, 121,126,127, 128, 129, 131,138, 140, 142-146, 147; obsidian, (Fig. 3.58) 124,131,131225

Chrysos, Evangelos, 6 Clactonian technique, 101,144 Classical, 4, 8, 17, 179, 17935, 189, 195 climate: and glacial-interglacial cycles,

74, 78, (Fig. 3.28) 84; and Ho- locene, 226; and landscape dynamics, 57; and Late Quaternary, 83-85; and Middle Palaeolithic, 129; and Mousterian, 106, 107,113,128; and paleoenvironments, 48, 131; and sea-level changes, 78; and surface survey, 28

coastal plains: and Adriatic Sea, (Fig. 3.21) 76; and Ambracian Gulf, 27, 157, 162, 192-198; and Arachthos River, 161; Epirus coast in last interglacial, 80-83; and findspots, 91; and human occupation, 50; and Late Quaternary, 76-83; and Middle and Late Quaternary paleoshorelines, 77-78, (Table 3.6) 78, (Fig. 3.23) 79; and Middle Palaeolithic, 127, 130; and Mous- terian, 80, 109; and Palaeolithic survey, 97; and spatial distribution, 137; surface survey of, 9; in western Epirus, (Fig. 3.8) 60

colonial activities, 8, 42, 179, 2663 Comarus, 162 computer-aided analysis, 6, 115 controlled collections, 24-25, 115 Corfu, 52, 78, 82, 92, 106, 114, 127,

132,208 Corinth, Gulf of, 50, 131 Corinthia, 137 Cullen, Brenda, 22,267 cultural resource management (CRM),

25

DAKARIS, SOTIRIOS, 5,42,66, 162, 189, 201,202,204,205,216,227,230,

231-232,233 data comparison, 26, 27, 32 databases, 19-20, 34 dating. See luminescence dating;

radiocarbon (14C) dating Dean, W. E., Jr., 209 Debenham, N., 90 deflation, 136, 137 deposition: and Acheron River valley,

209,212-216, 225; and Nikopolis isthmus, 173; and paleosol, 92-95, (Fig. 3.30) 94; and resurvey, 43; and terra rossa redeposition, 62, (Fig. 3.12) 64-65, 66, (Fig. 3.13) 66, 69- 70,72,74-76,96

D'Errico, F., 155 Dio Cassius, 201 documentary research, 28 Dodona valley, 4 dolines, 57, (Fig. 3.6) 58, 107 Douzougli, Angelika, 6 downslope movement, 149,153

EFSTRATIOU, N., 137

Elatria, 189, 195. See also Palaiorophoros

electrical resistivity, 44, 45. See also geophysical prospection

electromagnetic conductivity, 44, 45. See also geophysical prospection

Eli, 113 Elis, 106, 116, 155, 179 England, 100 environmental zones, 30, 32,265,266 Ephoreia of Caves and

Paleoanthropology, 114 Ephyra, 205,214, 228,232,268 Epigravettian, 51, 114, 116, 130 Epirus: and archaeological survey, 50;

cities of, 3; geological history of, 54- 56; Late Quaternary chronology, 85-95; Late Quaternary landscape, 54-85; maps of, (Fig. 1.1) 2, (Fig. 3.1) 49, (Fig. 6.1) 200; and Mousterian, 91-92, 97, 101,127, 128; and Palaeolithic chronology, 91-95; and Palaeolithic sites, 47-50, 125-126; Pleistocene, 47-48; previous Palaeolithic research, 50- 54; Roman intrusion into, 9; satellite imagery of, (Fig. 6.5) 206; surface survey of, 7, 9, 28; and tectonic activity, 54-56, (Fig. 3.2) 55, (Figs. 3.3, 3.4) 56, 205,208

Epirus, southern: choice of, 1, 2-3; and cultural remains, 8; economic basis

of, 9; erosional landscapes of, 27; fortified citadels of, 38; maps of, (Fig. 1.2) 3, (Fig. 2.1) 29, (Fig. 4.1) 136; and Pleistocene, 48; and surface survey, 27, 28, 32; and survey area, 3-5

erosion: and Ambracian Gulf, 157, 162; and artifact distribution, 136-137, 147; and Holocene, 226; and limestone landforms, 50; and Nikopolis isthmus, 173; and poljes, 59, 70, (Fig. 3.16) 72, 93; and Preveza area, 113; and Quaternary, 62; and redbeds, 53, 97; and resurvey, 43; and southern Epirus, 27; and surface survey, 27, 31; and tectonic activity, 48; and terra rossa, 119

Etude g6ologique, 205-206,208 Europe, 51, 98, 106, 114, 126, 129, 132 excavation: of sites in Epirus, 5, 47, 50,

52, 53, 89; and permit regulations, 18, 25; surface survey compared to, 26, 41-42; targeted, 24

FAUNA: and Ambracian Gulf, 186, 188; and artifact distribution, 136; and Asprochaliko, 127; and climate history, 85; and coastal plains, 76; and Grammeno plain, 177; and human population density, 129; and hunting specialization, 54; and karst landscapes, 129; and Kastro Rogon, 182, 183-184; and Klithi, 53; and land use, 107, 130; and last interglacial, 80, 82; and later Palaeolithic, 116, 117; and Mesolithic, 117,123; and microfossil analyses, 209,211-212; and Mousterian, 51, 107, 108, 128; and Neolithic, 123; and Nikopolis isthmus, 164, 167, 168; and Ormos Vathy, 174; and paleoenvironments, 48; and poljes, 125, 266; and sea- level changes, 78; and Upper Palaeolithic, 51, 52

field methods: and site revisits, 39, 43- 45; and site/scatters, 37-42; and surface surveys, 27, 30, 34-45; and tracts, 34-37

fieldwalking teams, 9, 31, 32 First International Symposium on

Nikopolis, 5-6 Fish, S. K., 23 flora: and Asprochaliko, 127; and Late

Quaternary, 83-85; and Mesolithic, 123; and paleoenvironments, 48,

285

INDEX

131; and transhumance, 54. See also vegetation

Folk, R. L., 209 foragers, 108, 113,127,128, 129 foraminifera, (Fig. 3.22) 77, 80, 1594,

164,167,174,177,184,186,188, 206, 209, 211-212. See also microfossil analyses

Fort Ancient, Ohio, 24 France, 113 Franchthi Cave, 92,106, 114, 116, 118,

119,155 Frangoklisia, 18

GALATAS, 59, 96, 106, 112, 113, 115, 116

gastropoda, 80, 82, 164, 167, 182, 184, 209, 214. See also microfossil analyses

geographic information systems (GIS), 13, 37, 138, 147-154

geological coring: and Acheron River valley, 209-211, (Fig. 6.7) 210,235- 263,268; and Ambracian Gulf, 12, 13, 158, (Fig. 5.2) 159, 1594, 267; and archaeological survey, 13; and Grammeno plain, 177-179; and Kastro Rogon, 179, (Fig. 5.14) 180; and Nikopolis isthmus, (Fig. 5.3) 163,164,16425, 167; and Ormos Vathy, (Fig. 5.4) 163,164, 174-177; and shoreline changes, 12-13; and Strongyli, 179; and surface survey, 31

geomorphologic studies: of Ambracian Gulf, 27, (Fig. 5.1) 158, 160-161, 267; of Epirus, 7, 12, 28; and extensive nonsystematic survey, 32; of Glykys Limen, 216-225,226; and paleoenvironments, 48; and selection of fields and tracts, 9; and site revisits, 43, 45; and surface scatters, 25; and surface survey, 27, 31

geophysical prospection: and Nikopolis Project, 6, 7; and site revisits, 44, 45; and subsurface features, 17-18

global positioning systems (GPS), 16, 37

Glykys Limen, 199-202,212,213,214, 215,216-225, (Fig. 6.12) 221, (Fig. 6.13) 222, (Fig. 6.14) 223, (Fig. 6.15) 224, 229, 234, 267-268. See also Ammoudia Bay; Phanari Bay

Gourana, 205 grain-size analysis: and Acheron River

valley, 209; and coastal sediments,

80, (Fig. 3.26) 81; and Kokkinopilos polje, 72, (Fig. 3.17) 73; and terra rossa, 66, (Fig. 3.14) 67, (Table 3.2) 67, (Table 3.3) 68-69, 69, 75

Grammeno, 24,44-45, (Fig. 2.4) 44; plain of, 177-179,1773 , (Fig. 5.13) 178

Grava Cave, 52, 114, 116, 117 Gravettian, 51, 114, 116, 130 Greece: and climate history, 83, 85,

226; and late entry model, 126; and Mesolithic, 123; and Mousterian, 113, 129; and nonsite surveys, 25; and Palaeolithic, 125; paleosol stratigraphy in, 86-89; regional studies in, 26-28; and surface survey, 26, 27, 28; and urban surveys, 42

Greek, 22, 41,200, 201,202 ground-penetrating radar, 44. See also

geophysical prospection Ground-Truthing Form (GTF), 19 ground-truthing of satellite imagery, 7,

13, 16, 17, 19, 28, 30 Guiscard, Robert, 201 Gulf of Arta. See Ambracian Gulf

HAMILTON, MICHAEL, 17

Hammond, N. G. L., 5, 162,189,193- 194,201,204,205,227,231-232

Hellenistic, 4, 8,17, 179, 189,195 Hey, R. W., 66,90 Higgs, Eric: and Asprochaliko, 47, 50,

51, 89, 117; and Kastritsa, 47, 50, 52, 89; and Kokkinopilos, 5, 47, 50, 51, 53, 69, 89, 99, 105; and later Palaeolithic, 114; and Morphi, 107; and survey of northern Greece, 125; and transhumance, 52, 53

Homer, 199,209, 229, 230, 234 Homo erectus, 98 Homo sapiens, 98, 114, 127, 129,266 human behavior: and coastal plains, 76;

and geological features, 49-50; and landscape properties, 132; and Mousterian, 128; and paleoenvironments, 48, 49; and political leagues, 8; and regional dynamics, 26, 27; and residential mobility, 108,112,114,127, 128, 129; spatial aspects of, 37; and transhumance, 52, 53, 54. See also agriculture; hunting; land use; settlement patterns

human-land relationships: and Nikopolis Project, 1-2, 30, 265; and paleoenvironments, 48-49, 126, 132

286

INDEX

human occupation: decline in, 123, 132; and Kokkinopilos, 51; and Late Pleistocene, 49; and Mousterian, 113; and Palaeolithic, 47, 50, 52, 117, 126,266; and paleoenvironments, 48; and sea-level changes, 78, 128; of Spilaion, 147; and terra rossa, 50

hunter-gatherers: and karst landscapes, 50; and poljes, 126, 266

hunting: and Asprochaliko, 54,127, 130; and Corfu, 78; and Kleisoura, 116; and Klithi, 53, 54,108; and land use, 117, 128, 129; and off- site human activity, 113; and redbeds, 105; and seasonal camps, 52, 54, 107, 108, 112,116-117, 127, 130

hydrology, 48

IBERIA, 129

Iliovouni, 112 Imbrie, J., 77 infrared stimulated luminescence

(IRSL) dating: 97; and Anavatis, 82, 8283; and Mesolithic, 118, 121; and Palaeolithic open-air sites, 85; and sediments, 90, 91, (Table 3.10) 91, 92. See also luminescence dating

Ioannina, 4, 6, 7,21 Ioannina, Lake, 83,226 Ionian Sea: and Ambracian Gulf, 160,

162, 169,194, 195,198; and contacts between peoples, 8; and geological coring, 12, 13; and Nikopolis Project, 4

Iron Age, 5,132 Italy, 83,109, 113, 128, 130, 131, 132

JAPAN,55

Jing, Zhichun, 21,22,267

KANALLAKION, 4,204,228 Karimali, Evangelia, 22,267 karst landscapes: and land use, 49, 59,

107, 125, 126, 129, 131; and Late Quaternary, 57-59, (Fig. 3.5) 57; and Mousterian, 107; and Preveza, 49, 50; and Spilaion, 138, (Fig. 4.4) 140; and tectonic activity, 50, 131

Karvounari, 105 Kastri: and Acheron River course, 199,

204-205,233,234; and Acherousian lake, 202,204,268; and urban survey, 12, 42. See also Pandosia

Kastritsa: and backed-blade industries, 116; excavation records of, 52; and hunting, 54, 130; and later Palaeolithic, 114, 116, 117; and Palaeolithic research, 47, 50, 52; and post-glacial period, 17; and radiocarbon assays, 89

Kastro Rizovouni, 4, 5, 17, 113. See also Batiae

Kastro Rogon: aerial photography of, 17, (Fig. 1.6) 18; and Ambracian Gulf, 159,267; and geological coring, 179, (Fig. 5.14) 180; and Louros River, 179, 182, 184,189, 192, 198; paleogeographic setting of, 189, (Fig. 5.19) 190-191,195; stratigraphic cross sections of, 179- 180, (Fig. 5.15) 181,182-184, (Fig. 5.16) 183, (Fig. 5.17) 185. See also Bouchetion

Kastrosykia, (Fig. 3.45) 111 Kephalari, 106, 114, 116, 155 Kephallinia, 106 Kephallonitou, Frankiska, 6 Kintigh, K. W., 23 Kirsten, Ernst, 202,230 Kleisoura (Epirus), 4 Kleisoura (Argolid), 115, 116, 119, 155 Klithi: and backed-blade industries,

116; excavation of, 47,52; and hunting camp, 108; and later Palaeolithic, 114, 116, 130; and post-glacial period, 117; and radiocarbon dating, 89; and transhumance, 53

Kokkinopilos: excavation records of, 47, 52, 53, 89; and flints, 96, 129; and handaxe, 99, 103; and Higgs, 5, 47, 50, 51, 53, 69, 89, 99, 105; and later Palaeolithic, 116; and Lower Palaeolithic, 98; and luminescence dating, 89,90; and Middle Palaeolithic, 155; and Mousterian, 50,51,93,105,106,107,108,112, 113; and Neanderthals, 127; as open-air site, 51; and Palaeolithic chronology, (Fig. 3.30) 94,95, (Fig. 3.31) 99; polje of, (Fig. 1.5) 16, (Fig. 3.12) 65, 70, (Fig. 3.15) 71, 72, (Fig. 3.16) 72, (Fig. 3.17) 73; and terra rossa, 66, 69, (Table 3.4) 70, 75-76

Kokytos River, 5,228. See also Vouvos River

Komnena, Anna, 201,209 Konispol Cave, 117-118 Kopais, Lake, 58 Koronopoulos, 233

Koryphi, Mt., 184 Koumouzelis, M., 114 Kowalewski, S. A., 23 Kranea, 59, 96, 106, 107, 108, 113, 129 Kuhn, S. L., 113,128 Kvamme, K. L., 138

LACONIA SURVEY, 24-25 Lambeck, K., 78, (Fig. 3.24) 79 land use: and artifact distribution, 137;

and erosion, 62; and foragers, 108, 113, 127, 128, 129; and geoarchaeology, 47; and karst landscapes, 49, 59, 107, 125, 126, 129, 131; and later Palaeolithic, 115, 116, 130; and Mesolithic, 117; and Middle Palaeolithic, 48, 126; and Moust- erian, 107, 108, 113, 127; and Neanderthals, 127, 128; and open- air sites, 48; and paleoenvironments, 48-49; and poljes, 115,125,126, 127,266-267; and surface survey, 30; and tectonic activity, 54; and tracts, 34; and transhumance, 52, 54

landscape archaeology: and Nikopolis Project, 1-2; and paleoenvironments, 48-49; and surface survey, 26, 31, 3150

Lang, Andreas, 90 Late Antique, 6, 9, 164, 195,267 Leake, W. M., 162, 164,202,204,209,

222,223,232-233 Leucas, 3 Levallois technique: and Asprochaliko,

113; and Mousterian, 52-53, 98, 106, 108, 109, 114, 129; and Neanderthals, 128; and Spilaion, 142

limestone. See karst landscapes Limnes valley, 137 Logarou Lagoon, 161 long-term replication studies, 24 Louros, 50 Louros River: and Ambracian Gulf,

160, 161,16110,198; delta, 97; glacial sediment load of, 78; gorge, 5,17; and Kastro Rogon, 179,182, 184, 189, 192, 198; and sea-level changes, 195; and survey zone, 4, 266; and tectonic activity, 50; water channel and aqueduct bridges, (Fig. 1.7) 19

Loutsa, (Fig. 3.10) 63, 106, 107, 108, 113,118, 119, (Fig. 3.55) 122, 129, 131

loutses: and Acheron River valley, (Fig. 3.10) 63; and findspots, 91, 93;

287

INDEX

and karstic peneplain, (Fig. 3.7) 59; and land use, 115, 125, 126, 127, 129, 266-267; and Mouste- rian, 107; and Palaeolithic chronology, 93-95, (Fig. 3.30) 94, 99; poljes distinguished from, 59; and terra rossa, 62, (Fig. 3.12) 65; uniform sediments of, 69; in western Epirus, (Fig. 3.8) 60, (Table 3.1) 61

luminescence dating: and coastal plains, 82; and Palaeolithic, 126; and paleosol stratigraphy, 86; of sediments, 89-91, (Table 3.10) 91; and terra rossa, 69, 85. See also infrared stimulated luminescence (IRSL) dating; thermoluminescence (TL) dating

MACEDONIA, 50, 83, 86, 126

MacLeod, D. A., 69

magnetometry, 44, 45. See also geophysical prospection

malaria, 230 Mantineia, 58 Martinson, D., 77 Mavri, Lake, 5, 112 Mavrovouni, Mt., 189, 190, 194, 198 Mazoma Lagoon, 161,162,164, 167,

169,173 medieval, 1, 9,22, 179,182,198, 202,

267 Mellars, P., 51 Mercouri, Melina, 6 Mesaria, 112 Mesopotamon/Tsouknida valley, 216,

217, (Fig. 6.9) 218, (Fig. 6.10) 219, (Fig. 6.11) 220,223,227, 228,232

Messenia, 106, 116 microfossil analyses: and Acheron River

valley, 209, 211-212,213,214,215, 235-263; and sea-level changes, 77, 208

Micromousterian, 51, 53, 106 Middle East, 55 mineral composition: of last interglacial

coastal sands, (Table 3.7) 83; of terra rossa, 66, (Table 3.4) 70, (Table 3.5) 71

Moore, Melissa, 8,21 Morphi, 72, (Fig. 3.18) 74,105, 107,

108,112,113, 129 Mousterian, 50, 51, 52, 53, (Fig. 3.20)

75, 80,89,91, 93,97, 98,100,101, 103, 105-114, (Figs. 3.43, 3.44) 110,126,127, 128, 129, 155

Murray, Priscilla, 7

Myers, Eleanor Emlen, 17 Myers, J. Wilson, 17 Mytikas, 82

NEANDERTHALS: and climate, 113; and Homo sapiens, 129,266; and land use, 127,128; and Mousterian, 98, 106, 107, 108, 114; and Mousterian/ Early Upper Palaeolithic transition, 129-130; population density of, 128

Near East, 98,144 Nekyomanteion, 5 Nemea, 116, 125 Nemea Valley Archaeological Project,

27 Neolithic, 5, 52, 62, 95, 123, 132, 137,

144,154, (Fig. 5.12) 176, 177,194 New Zealand, 55 Nikopolis: and Ambracian Gulf, 158;

territory of, 2663; city plan of, 7; and

core-periphery interactions, 30; founding of, 9; mapping of, 6; and Ormos Vathy, 39, 41,177; and Preveza peninsula, 162; regional context of, 5, 7; Roman period, 173, 266

Nikopolis isthmus: 62,164-173; and geological coring, (Fig. 5.3) 163, 164,16425, 167; map of, (Fig. 5.3) 163; and sea-level changes, 168, 173,193, 195; shoreline changes of, 169, 173, (Fig. 5.10) 173; stratigraphic cross sections of, 164, (Fig. 5.5) 165, (Fig. 5.6) 166, 167-169,

(Fig. 5.7) 170, (Fig. 5.8) 171, (Fig. 5.9) 172

Nikopolis Project: background and organization of, 5-8; daily work assignments, (Table 2.2) 33; documentation, 19-20; field school students of, (Table 1.2) 12; and human-land relationships, 1-2, 30, 265; methodologies of, 9,12-13, 15-18,265; and post-fieldwork analyses, 21; prehistoric survey contribution, 131-132; and presentation of results, 21-22; project staff of, (Table 1.1) 10-11; purpose of intensive survey, 28; and regional studies in Greece, 26-28; research aims of, 8-9; and sampling strategies, 29-31; and southern Epirus as study choice, 1, 2-3; survey zone of, (Fig. 1.1) 2, 3-5, (Fig. 1.2) 3, 7, (Figs. 1.3, 1.4) 14, 29, 47,266

288

INDEX

OBSIDIAN, (Fig. 3.58) 124, 131, 131225

Octavian, 9,201. See also Augustus Orchomenos, 58 Ormos Odysseos, 83, 103, (Figs. 3.38,

3.39, 3.40) 104, 112 Ormos Vathy: 158, 162, 174-177; and

Early Palaeolithic materials, 103; and geological coring, (Fig. 5.4) 163,164, 174; and Holocene, 177; map of, (Fig. 5.4) 163; and Mous- terian, 107, 127; and Nikopolis, 39, 41, 177; paleogeographic reconstructions of, (Fig. 5.12) 176, 177; and Roman period, 162, 164, 177; and sea-level changes, 174, 177, 195; stratigraphic cross sections of, 174, (Fig. 5.11) 175, 176-177; and surface scatters, 24; survey units of, (Fig. 2.3) 41; and tectonic activity, 177, 192

ostracoda, 1594, 164,167,177,180, 182, 183, 184, 186, 188,209,211- 212,213, 214, 215. See also microfossil analyses

oxygen isotope stages (OIS), 77, (Fig. 3.22) 77, 78, (Table 3.6) 78, (Fig. 3.24) 79, 80

PALAIOROPHOROS, 17. See also Elatria paleoshorelines, 77-83, (Table 3.6) 78,

(Fig. 3.23) 79, (Fig. 3.25) 81 paleosols: and Ayia, (Fig. 3.19) 74;

chronosequences of, 86-89; and coastal plains, 80, 82; and Kokkino- pilos, 70; maturity levels of, 80, 82, 85, 86, (Fig. 3.29) 87, (Table 3.8) 87, (Table 3.9) 88, 89, 90, 91, 92- 93; and Mesolithic, 119, 121; and Morphi, 72, (Fig. 3.18) 74; and Nikopolis isthmus, 17,164; and Palaeolithic dating problem, 85; and Palaeolithic sites, 48; and Palaeo- lithic survey, 95, 96; and Rodaki, 108; and terra rossa, 61, 62, (Fig. 3.12) 64, (Fig. 3.13) 66, (Fig. 3.30) 94

Paliourias River, 82, 108, 121 Pandosia, 204-205,268. See also

Kastri Papagianni, Dimitra, 21, 97, 127 Paramythia, 5 Parga, 4, 56, 97, 115, 118 Paschos, Panayiotis, 69 pastoral transhumance. See transhu-

mance patination, 118, 140-141,142, 143,

148, 153

Peloponnese, 50, 58, 86, 94, 114, 125, 131

Peloponnesian War, 201 peneplains: and Palaeolithic survey, 95;

and tectonic activity, 50, 55, 56, 58, (Fig. 3.7) 59

Perles, C., 114 Phanari Bay, 4, 160, 199-202, (Fig. 6.3)

202,212,213,214,215,216-225, 234. See also Ammoudia Bay; Glykys Limen

Philippias, 4 Philippson, Alfred, 202,230 phosphate studies, 24-25 Pindos: and Ambracian Gulf, 160; and

Arachthos River, 161; and glacial periods, 48; and karst landscapes, (Fig. 3.5) 57; and land use, 131; and paleosol stratigraphy, 86; and poljes, 59; and tectonic activity, 54, 55

Plog, S., 23 Pogonitsa, Lake, 116, (Fig. 3.51) 118

poljes: coastal plains compared to, 77; and erosion, 59, 70, (Fig. 3.16) 72, 93; and findspots, 91, 93; and land use, 115, 125, 126, 127,266-267; and Mousterian, 107-108; and Palaeolithic chronology, 93-95, (Fig. 3.30) 94, 99; and tectonic activity, 58, 59, (Fig. 3.7) 59; and terra rossa, 62, (Fig. 3.12) 64; in western Epirus, 58, (Fig. 3.8) 60, (Table 3.1) 61

post-medieval, 17935, 182, 195, 198 pottery fragments: and Acheron River

valley, 209, 233; and tracts, 35; and urban surveys, 42

Preveza, nomos of: chronostratigraphic diagram for, 90, (Table 3.11) 92; limestone landforms, 49; and paleoshore deposits, 82; and Pleistocene, 48; and survey zone, 3, 4,29,47,266

Preveza peninsula: and Ambracian Gulf, 160, 162-177; and Mesolithic, (Fig. 3.51) 118, 119, 121; and Nikopolis isthmus, 162, 164-173; and Ormos Vathy, 174-177; and Salaora barrier, 161; and tectonic activity, 168, 169, 173, 174, 192, 195

Pseudo-Scylax, 189, 192 Pylos, 125

RADIOCARBON (I4C) DATING: and

Acheron River valley, 209, (Table 6.1) 210,217,233; and Acherousian lake, 227,228; and Ambracian Gulf,

167, (Table 5.1) 168, 169,186,188, 193, (Fig. 5.20) 193; effective range of, 51, 89; and Glykys Limen, 216, 223,224; and Kastro Rogon, 179, 182, 184; of sediments, 89, 98; and Upper Palaeolithic, 114

Rapp, George (Rip), 7,21,22,267 redbeds: and archaeological survey, 50;

diversity of, 61-62; and erosion, 53, 97; and hunting camps, 105; and karst landscapes, 58; and Kokkinopilos, 53, 96; and Middle Palaeolithic, 52; and Mousterian sites, 50, 106, 107; and open-air sites, 50, 53; and Pliocene, 53, 76

regional dynamics, 26,27 remote sensing, 6, 7, 13-18, 24, 44,266 resurvey, 43-45 Rick, J.W., 137 rockshelters: and Argive Kleisoura, 119;

and Bailey, 52; and Higgs, 50,51; and hunting, 54, 117; and land use, 130; and later Palaeolithic, 115,116, 117; and Mousterian, 51,106; stratified deposits in, 89; and Upper Palaeolithic, 50, 51, 52

Rodaki, 66, 82, (Table 3.7) 83, 107, 108, 109, (Figs. 3.47, 3.48) 112

Rodia Lagoon, 161,186, 188 Rokia, Mt., 4, 160, 180, 186, 189, 194,

267 Rolland, N., 98 Roman, 1, 9, 17, 18, 22, (Fig. 2.3) 41,

41,44, 70, 7068, 158, 162, 164, 173, (Fig. 5.12) 176, 177, 17731, 179, 180, 184,189,194,195, 198, 202,266

Romia, 112 Runnels, Curtis, 7,22,266,267 Russell, Richard, 228

SALAORA BARRIER, 161,169,188,195 Salaora Island, 188 sampling strategies, and surface survey,

23-25, 26,29-31, 43 Sarris, Apostolos, 18 satellite imagery: effectiveness of, 13,

15; of Epirus, (Fig. 6.5) 206; and fieldwalking teams, 9; ground- truthing of, 7, 13, 16, 17, 28, 30; and landscape locations, 16; of survey zone, (Figs. 1.3, 1.4) 14

scouting: and extensive nonsystematic survey, 32, 43; and site/scatter, 41; and surface survey, 28, 30

sea-level changes: and Acheron River valley, 208, 216-217; and Ambracian Gulf, 78, 157, 162, 192-

289

INDEX

195, (Fig. 5.20) 193, 198, 267; and coastal sites, 137; and Holocene, 208; and Late Quaternary, 76-83; and later Palaeolithic, 116; and Nikopolis isthmus, 168, 173, 193, 195; and Ormos Vathy, 174,177, 195; and oxygen isotope stages (OIS), 77, (Fig. 3.22) 77, 78, (Table 3.6) 78, (Fig. 3.24) 79, 80; and paleoenvironments, 48; and Spilaion, 138

sediments: and Acheron River valley, 209-216,225-226,231; and Ambracian Gulf, 157,160,162, 192, 195,198; and Grammeno plain, 177, 179; and karst landscapes, 125; Late Quaternary and Palaeolithic chronology, 91-95; of loutses, 69; luminescence dating of, 89-91, (Table 3.10) 91, 92; and Nikopolis isthmus, 164,167,173; radiocarbon dating of, 98; and Spilaion, 138, 141, 147; and stone artifacts, 93; Walther's Law, 162, 16216, 213

Seidi, 116 settlement patterns: and Ambracian

Gulf, 157-158; and geoarchaeology, 47; and Higgs, 52; and later Palaeolithic, 115, 116, 117; and Mousterian, 107,108, 113; post- Pleistocene settlement history, 117- 124, 130-131; in southern Epirus, 8

Sidari, 52, 117-118, 123, 131 simulation studies, 24 site revisits: and field methods, 39, 43-

45; and Palaeolithic survey, 96; purpose of, 43, 4378

site/scatters (SS): and artifact distribution, 136-137; and concept of site, 1351, 136, 141; documentation and collection procedures, 39-40; explanation of, 20; and field methods, 37-42; and Palaeolithic survey, 97, (Fig. 3.31) 99, (Fig. 3.32) 100, 105, (Fig. 3.42) 109, (Fig. 3.45) 111, (Fig. 3.51) 118, 134; and taphonomy of surface sites, 137- 138; and tracts, 40-41, (Fig. 2.3) 41

Skepasto, 106, 107, 108 Sordinas, Augustus, 52, 108, 117 Spain, 113 spatial analysis, and Spilaion, 141, 147,

148-154, (Table 4.3) 150, 156,267 spatial coverage: and site/scatters, 40-

41; and surface survey, 33-34; and tracts, 34

spatial distribution of artifacts: causes of, 136, 137-138; and Palaeolithic survey, 97; and Spilaion, 135-136, 141,142, 147-148, 150, (Fig. 4.12) 151, (Fig. 4.13) 152, 153

spatial patterns: and Ambracian Gulf, 162; and artifact density, 137; and Spilaion, 148, 153, 156

Spercheios delta, 225 Spilaion: and Aurignacian, 115, 130,

142, 143, 154-155,267; categories of flintknapping debitage, 142, (Table 4.1) 143; and end scrapers, 115, (Fig. 3.50) 115, 142, 144, (Figs. 4.6, 4.7) 145, (Figs. 4.8, 4.9, 4.10, 4.11) 146, (Fig. 4.13) 152,153,154, 155; and geographic information systems (GIS) analyses, 147-154; and land use, 131; lithic assemblage of, 142-146, 155-156,267; maps of, (Fig. 4.1) 136, (Fig. 4.2) 139; sample grid, (Fig. 4.5) 141; site description, 138-141; and spatial analysis, 141,147, 148-154, (Table 4.3) 150, 156, 267; and spatial distribution, 135-136,141,142, 147-148, 150, (Fig. 4.12) 151, (Fig. 4.13) 152, 153; topographic map of, (Fig. 4.2) 139; views of, (Fig. 4.3) 139, (Fig. 4.4) 140

SPOT imagery, of survey zone, 13, (Figs. 1.3, 1.4) 14. See also satellite imagery

Stavros, Mt., 4, 160, 189, 194,267 Stein, Carol, 8, 15 Stephani, 50 Stiner, M. C., 128 stone tools: Acheulean, 98, 99, 126,

266; backed-blades, 51, 52, 53, 114, 115-116, (Fig. 3.50) 116, 117, 119, (Fig. 3.52) 120, 130; becs, (Fig. 3.52) 120, 144, (Fig. 4.10) 146; bifaces, (Fig. 3.41) 105; bifacial foliates (leafpoints), 51, 52, 113, 114, 144; bladelets, 118, 119, (Fig. 3.58) 124; blades, 53, 96, (Fig. 3.43) 110, 115, 118, 142, 143, (Fig. 4.12) 151, 153, 154; burins, 114, 115, (Fig. 3.52) 120, 129, 142, 144; choppers, (Fig. 3.35) 102; Clactonian technique, 101, 144; core-choppers, 19, 101, (Fig. 3.36) 103, (Fig. 3.48) 112, 126, 143; cores, 96, 101, (Fig. 3.37) 103, 106, 108, (Fig. 3.43) 110, 113, 115, 118, 119, (Fig. 3.55) 122, 123, (Fig. 3.58) 124, 128, 131,142-143, (Fig. 4.9) 146, (Fig. 4.12) 151, 153;

290

INDEX

cortical flakes, 142,144,150, (Fig. 4.12) 151, 153, 154; denticulates, 99, 101, (Fig. 3.34) 102, 117,119, 128, 142,144, (Fig. 4.6) 145, (Fig. 4.10) 146, (Fig. 4.13) 152, 153; Dufour bladelets, 154; end scrapers, 53, 109, (Fig. 3.43) 110, 114, 115, (Fig. 3.49) 115,117, 118, 119, (Fig. 3.52) 120, 121, 123, 129, 142, 144, (Figs. 4.6,4.7) 145, (Figs. 4.8,4.9, 4.10, 4.11) 146, (Fig. 4.13) 152, 153,154, 155; Epigravettian, 51, 114, 116,130; Gravettian, 51,114, 116, 130; handaxes, 98, 99, 103, (Figs. 3.38, 3.40) 104, 126,266; Levallois technique, 52-53, 98, 106, 108,109,113,114,128,129,142; microburins, 119; microliths, 117, 119, (Fig. 3.52) 120, (Fig. 3.58) 124; notched pieces, 99, (Fig. 3.34) 102, 109, 117, 119, 142, 144, (Figs. 4.9, 4.10) 146, (Fig. 4.13) 152, 153; perfoirs, 117,119, (Fig. 3.52) 120, 144, (Fig. 4.10) 146; pieces esquillees, 144; plain flakes, 142, 144, 150, (Fig. 4.12) 151,153, 154; points, 53, 101,106, 108, 113,114, (Fig. 3.53) 120, 128, 129; raclette, (Fig. 4.6) 145; retouched tools, 96,101,108, 114, 115, 117, 118, 119, (Fig. 3.52) 120, 121, (Fig. 3.55) 122, 123, (Fig. 3.58) 124, 127, 128, 129, 142, 144, (Table 4.2) 144, (Figs. 4.6-4.11) 145-146, (Fig. 4.12) 151, (Fig. 4.13) 152,153, 154; side scrapers, 51,101, (Fig. 3.34) 102, 106, 109, (Fig. 3.43) 110, (Fig. 3.48) 112, 113, 114, 128, 129, 144, (Fig. 4.7) 145; and silica gloss, 118, 119, 121, (Fig. 3.55) 122, (Fig. 3.58) 124; tanged arrowhead, (Fig. 3.58) 124; tranchet arrowhead, (Fig. 3.58) 124; trapezes, 117, 118, 119, (Figs. 3.52, 3.53) 120, 121, (Fig. 3.55) 122, (Fig. 3.58) 124,131

Strabo, 162, 189, 200,209,230 Strongyli, 18, 179, 184, 189 Stymphalos, 58 surface scatters, 24, 25, 39, 4378, 1351,

136-137 survey, archaeological: coordination of,

13; daily work assignments, (Table 2.2) 33; diachronic, 26, 95,265,266; of Epirus, 7, 9, 28; excavation compared to, 26, 41-42; extensive, 28, 32, 43; field methods, 27, 30, 34-45; and human activity, 17;

intensity and coverage of, 31-34, 265; intensive, 32; as less destructive technique, 2632; methodology of, 12, 25-28; and Nikopolis Project, 8, 30; and Palaeolithic survey, 95-124; purpose of, 28; and remote sensing, 13; and sampling strategies, 23-25, 26, 29-31, 43; and surface scatters/ subsurface relationship, 24-25

survey, geological, 9, 13 survey, geophysical, 28 survey, Palaeolithic: conclusions of,

125-131; Early Palaeolithic, 98- 105; and extensive nonsystematic survey, 32; geological setting of sites, 97-124; goals and procedures of, 95-97; later Palaeolithic, 114-117; Mesolithic, 117-124, 130-131; and methodology, 31; Mousterian (Middle Palaeolithic), 97, 103,105- 114; Neolithic, 132; and surface survey, 28

survey, topographical, 42 survey, urban, 12, 40, 41-42 survey zone: multispectral (MSS)

imagery of, 13, (Figs. 1.3, 1.4) 14; of Nikopolis Project, (Fig. 1.1) 2, 3-5, (Fig. 1.2) 3, 7, (Figs. 1.3, 1.4) 14, 29, 47,266; size of, 9

Swain, Frederick, 212 Sybota, Battle of, 201,229

TARTARON, THOMAS, 8, 12, 21,266 tectonic activity: and Acheron River

valley, 50,55; and Ambracian Gulf, 54, 55, (Fig. 3.2) 55, 56, 157, 161, 162, 167,169,192-195; and Epirus, 54-56, (Fig. 3.2) 55, (Figs. 3.3, 3.4) 56,205,208; and Ormos Vathy, 177, 192; and paleoenvironments, 48, 131; and poljes, 58, 59, (Fig. 3.7) 59; and Preveza nomos, 49; and Preveza peninsula, 168, 169, 173,174, 192, 195; and transhumance, 54

Tegea, 85 Tenaghi Philippon, 83 terra rossa: color of, 62; and grain-size

analysis, 66, (Fig. 3.14) 67, (Table 3.2) 67, (Table 3.3) 68-69, 69, 75; and karst landscapes, 58, (Fig. 3.7) 59; and Late Quaternary, 61-62; and luminescence dating, 69, 85; and Mesolithic, 119; mineral composition of, 66, (Table 3.4) 70, (Table 3.5) 71; and Palaeolithic survey, 95; and redeposition, 62, (Fig. 3.12) 64-65, 66, (Fig. 3.13) 66,

69-70, 72, 74-76, 96; and Spilaion, 138, 141; and tectonic activity, 50

Thematic Mapper (TM) satellite imagery, 13. See also satellite imagery

Theopetra, 116 thermoluminescence (TL) dating: 97;

and Asprochaliko, 89; and Mesolithic, 121; and Palaeolithic open-air sites, 85; and sediments, 90, 91, (Table 3.10) 91, 92. See also luminescence dating

Thesprotia, 205 Thesprotiko valley, (Fig. 3.31) 99,112-

113 Thessaly: and Aurignacian, 113,155;

and backed-blade industries, 116; and deposition, 94; landscape of, 48, 50; and Mousterian, 106; and Mousterian/Early Upper Palaeolithic transition, 129; and Palaeolithic, 114, 125, 126; and Palaeolithic chronology, 92; and paleosol stratigraphy, 86; and prehistoric periods, 2

Thucydides, 200-201,209, 229-230 Thyamis River, 217 Thyamis valley, 208 tiles, 35, 42, 177 Tippett, H., 66, 90 topographic maps: of Acheron River

valley, (Fig. 6.8) 211; as documentation, 16, 20, 36, 37, (Table 3.1) 61, 16426, 211,225; of Spilaion, (Fig. 4.2) 139

topography: and Acherousian lake, 202, 229; and spatial distribution of artifacts, 138, 148, 149; and surface survey, 28, 30; and tectonic activity, 54; and tracts, 34

Tourkovouni, 121 tracts (T): archaeological survey tract

form, 36-37, (Fig. 2.2) 36; database, 19; and field methods, 34-37; and Palaeolithic survey, 95-96; and site/ scatters, 40-41, (Fig. 2.3) 41; and surface surveys, 28, 30; and urban surveys, 42; and walkovers, 43

tractwalking, 30, 41 transhumance, 52, 53, 54 Tsarlambas, 82, (Fig. 3.27) 82, 121,

(Fig. 3.58) 124, 131 Tsoukalio Lagoon, 161 Tsouknida, 112, 118, (Fig. 3.52) 120,

131,228 Typical Balkan Aurignacian, 142,143,

154 Tzedakis, P. C., 83

29I

292

ULBRICH, 114, 116 United States: and cultural resource

management (CRM) surveys, 25; and surface scatters, 24; and surface sites, 137-138; and tectonic activity, 55

VALANIDORRACHI, 106, 107, 108,127 Valtos Kalodiki, 59, (Fig. 3.9) 63,107 van Andel, Tjeerd, 7,22,266,267 Vassiliko, 108 vegetation: and Acheron River valley,

214; and glacial-interglacial cycles, (Fig. 3.28) 84; and landscape dynamics, 57; and Late Quaternary, 83-85; and loutses and poljes, 125; and Mousterian, 107, 108; and resurvey, 43, 45; and spatial

INDEX

distribution, 138, 147; and Spilaion, 138. See also flora

villas, 18, 44, 179 Villas, Cathleen, 212 Vita-Finzi, C., 52 Voulista Panayia, 4, 17 Vouvopotamos, 112,115 Vouvos River, 5, 216. See also Kokytos

River

WALKOVERS (W), 20, 28, 43, 96, 141 Walther's Law, 162, 16216, 213 Wandsnider, L., 25,27 Waters, David, 55,56,208,225 Weymouth, John, 17-18 Willis, Katherine, 85,226 Wiseman, James, 6 Wiseman, Lucy, 7-8

Wuiirm glaciation, 192,194

XINIAS, LAKE, 226 Xirolophos, 233 Xylokastro, 205

YAALON, D. H., 69

Younger Dryas, 85, 89 Yugoslavia, 57, 58

ZACHOS, KONSTANTINOS, 6

Zaimis, 116 Zakynthos, 108-109 Zalongo, Mt., 4, 160 Zhou, Li-Ping, 90 Zilhao,J., 155 Ziros, Lake, 5

0-87661-532-9

Date post:	21-Jan-2016
Category:	Documents
Upload:	txus-devuelveme-mi-leica
View:	141 times
Download:	4 times

Wiseman&Zachos (Eds) - Landscape Archaeology in Southern Epirus, Greece

Documents